Qual Life Res (2015) 24:1–191 DOI 10.1007/s11136-015-1078-4

ABSTRACTS

22nd Annual Conference of the International Society for Quality of Life Research

 Springer International Publishing Switzerland 2015

Cutting Edge Research Plenary (1) Inflammatory gene SNPs as predictors of patient reported symptom related functioning in patients with multiple myeloma Xin Shelley Wang, MD, MPH, MD Anderson Cancer Center, Houston, TX, United States; Qiuling Shi, MD, PhD, MD Anderson Cancer center, Houston, TX, United States; Huei Kai Lin, PhD, MD Anderson Cancer Center, Houston, TX, United States; Guojun Li, MD, PhD, MD Anderson Cancer Center, Houston, TX, United States; Loretta Williams, PhD, UT MD Anderson Cancer Center, Houston, TX, United States; Robert Orlowski, MD, PhD, MD Anderson Cancer Center, Houston, TX, United States; Charles S. Cleeland, PhD, UT MD Anderson Cancer Center, Houston, TX, United States AIMS: Multiple myeloma (MM) is an incurable disease and most patients remain on therapy for extended periods of time. The disease and its treatment often produce significant functional impairment. We have previously shown a relationship between patient-reported symptom severity and markers of inflammation. In the current study, we expected that certain regulatory single-nucleotide polymorphisms (SNPs) in inflammation genes would be associated with functioning in patients with MM 1–5 years postdiagnosis. METHODS: In a crosssectional study, patients rated symptom-related functional impairment on the 6 interference items of the MD Anderson Symptom Inventory (MDASI) (symptom interference with general activity, work, walking, mood, enjoyment of life, relations with others). Symptom interference was rated on a 0–10 scale, with 0 = ‘‘does not interfere’’ and 10 = ‘‘interferes completely.’’ Patients also provided buccal-swab DNA samples. SNPs for 4 cytokine genes (IL6 -174G[C, IL1b -511C[T, TNFa -308G[A, IL10 -1082G[A) were tested. Twostep cluster analysis of a patient’s composite MDASI interference score indicated either good or poor functioning status. Logistic regression models, adjusted for age, sex, MM stage, performance status (PS), comorbidities, and body mass index, were used to identify SNPs that might predict poor functioning. RESULTS: Of 344 patients enrolled, 80 % received stem cell transplant, 21 % had progression/ relapse disease at time of study. Cluster analysis indicated that 31 % reported poor symptom-related functional status [mean 5.67 (SD 1.56)]. For non-Hispanic whites (n = 222, 64.4 %), the IL1b -511 CC genotype was associated with poor functioning (OR 2.14; 95 % CI 1.18–3.89; p = .012) and poor PS (OR 3.40; 95 % CI 1.55–7.46; p = .002). For other patients, no SNP was related to poor functioning, although poor PS (OR 5.36; 95 % CI 1.47–19.48; p = .011) and

advanced-stage disease (OR 2.69; 95 % CI 1.13–6.39; p = .025) were related to poor functioning. CONCLUSIONS: We found that symptoms were interfering with physical and affective functioning in one-third of patients with MM, from patient rated MDASI interferences. Further, genetic associations with inflammation were associated with the poor functioning in non-Hispanic whites. Our study suggests that racial/ethnic factors contribute to this association.

(2) A cluster-randomized study of clinician-patient shared vs standard reporting of symptomatic adverse events using PROCTCAE nested in a multicenter trial of multimodal therapy for rectal cancer (Alliance N1048 PROSPECT) Amylou C. Dueck, PhD, Mayo Clinic, Scottsdale, AZ, United States; Sandra A. Mitchell, PhD, CRNP, AOCN, National Cancer Institute, Rockville, MD, United States; Lauren J. Rogak, MA, Memorial Sloan Kettering Cancer Center, New York, NY, United States; Brenda Ginos, MS, Mayo Clinic, Scottsdale, AZ, United States; Daniel J. Sargent, PhD, Mayo Clinic, Rochester, MN, United States; Qian Shi, PhD, Mayo Clinic, Rochester, MN, United States; Jeffrey M. Farma, MD, Fox Chase Cancer Center, Philadelphia, PA, United States; Cathy Eng, MD, FACP, University of Texas M. D. Anderson Cancer Center, Houston, TX, United States; Christopher Crane, MD, University of Texas M. D. Anderson Cancer Center, Houston, TX, United States; Hagen Kennecke, MD, British Columbia Cancer Agency, Vancouver, BC, Canada; Ann M. O’Mara, PhD, RN, FAAN, National Cancer Institute, Rockville, MD, United States; Lori M. Minasian, MD, FACP, National Cancer Institute, Rockville, MD, United States; Deborah Schrag, MD, MPH, Dana-Farber Cancer Institute, Boston, MA, United States; Ethan Basch, MD, MSc, University of North Carolina, Chapel Hill, NC, United States AIMS: Treatment toxicity in cancer trials is assessed through clinician grading of adverse events (AEs) using the U.S. National Cancer Institute (NCI) Common Terminology Criteria for AEs (CTCAE); however numerous studies document underreporting of symptomatic AEs. Patient-reported outcomes (PROs) have been proposed to improve detection of symptomatic AEs. Recently, the NCI developed a PRO companion measure to the CTCAE (PRO-CTCAE). To evaluate whether sharing patient-reported AEs with clinicians alters toxicity detection, we nested a cluster-randomized study within a multicenter trial (NCT01515787; neoadjuvant chemotherapy followed by selective chemoradiation or upfront chemoradiation in

123

2 rectal cancer). METHODS: English- or Spanish-speaking participants completed PRO-CTCAE items measuring 13 symptomatic AEs weekly from home throughout neoadjuvant treatment using their choice of web or interactive voice response. Research staff followed up by telephone for missed surveys. Sites were randomized to standard AE reporting (no systematic provision of PROs to clinicians) or having PRO-CTCAE responses shared with clinicians at time of toxicity assessment. For each AE, the maximum clinician-reported grade for each patient was tabulated and compared between standard and shared groups using generalized linear models with a random intercept term for site clustering. RESULTS: 269/271 (99 %) eligible enrollees in PROSPECT agreed to participate. 2536 (91 %) of 2798 scheduled surveys were completed of which 431 required back-up data collection. CTCAE data were available for 181 patients (median age 56 [19–81 years], 67 % male, 40 % high school education or less; demographics/treatment assignment similar between groups). Clinicians in the shared versus the standard group reported a statistically significantly higher rate of CTCAE grades [0 for anorexia, anxiety, depression, dysphagia, and dyspnea; and higher ordinal grade for anorexia, anxiety, dysphagia, fatigue, and neuropathy (Fig. 1). CONCLUSIONS: Weekly PRO-CTCAE collection via home reporting and sharing of these data with clinicians at the point-of-care is feasible in a large multicenter trial. Sharing PRO-CTCAE alters clinicians’ AE reports, suggesting that investigators will use PRO data to inform their own AE reporting, thereby bringing clinician and patient assessment of symptomatic AEs into better alignment. Sharing PRO-CTCAE with clinicians appears to mitigate current underreporting of symptomatic AEs in cancer trials.

Qual Life Res (2015) 24:1–191 Support: HHSN261201000063C, U10CA180821, U10CA180882, 1UG1CA189823, CA077202, CA180888, CA180858, CA189820.

(3) Factor structure of the CDC healthy days core and symptoms modules and their association with stress, access to care, and social determinants of health John P. Barile, PhD, University of Hawaii, Honolulu, HI, United States; William W. Thompson, PhD, U.S. Centers for Disease Control and Prevention, Atlanta, GA, United States AIMS: This study assessed the factor structure of the Centers for Disease Control (CDC) Healthy Days Core Module (4 items) and Healthy Days Symptoms Module (5 items). While the four core Healthy Days items have been used extensively, less is known regarding the utility of the CDC’s 5-item Healthy Days symptom module. Analyses were conducted to determine (1) the number of factors associated with these nine items and (2) their association with measures of stress, access to care, and social determinates of health. METHODS: We used a sample of 1551 adults living an in urban county that had completed either an online or paper-based survey regarding social conditions and health. The survey included the 4-item Healthy Days Core Module and the 5-item Healthy Days Symptoms Module (CDC, 2000), the Perceived Stress Scale-4 (Cohen, Kamarck, & Mermelstein, 1983; Cohen & Williamson, 1988), an inventory of healthcare services and experiences, and a neighborhood environment measure (Mujahid et al., 2007). The sample was overrepresented by females (77 %), African Americans (53 %) and young adults (26 %). RESULTS: Our findings suggested that a three factor model (physical, mental, and vitality) using the nine CDC Healthy Days resulted in high loadings on all three factors (physical health [k = 0.51–0.80], mental health [k = 0.83–0.86], and vitality [k = 0.65–0.66]) When fitting the three factor HRQOL with the independent stress factor, the four factors were all moderately correlated with each other (r = 0.55–0.71) and the fit statistics suggested the overall model fit the data well (CFI = 0.93, RMSEA = 0.04). After controlling for demographic characteristics, the physical health factor was found to be significantly associated (P \ 0.05) with perceptions of support, fear of crime, and access to care; the mental health factor was associated with walkability, fear of crime, and access to care; the vitality factor was associated with walkability and access to care; and the stress factor was associated with fear of crime, walkability, and access to care. CONCLUSIONS: Researchers should consider using the expanded 5-item Healthy Days Symptoms Modules to assess a breadth of quality of life indicators and their associations with social determinants of health.

(4) National implementation of clinical PRO in Denmark Niels Henrik Hjollund, MD, PhD, AmbuFlex, Herning, Denmark; Troels Andersen, MSc Political Science, CEDI, Copenhagen, Denmark; Aleksander Bjerrum, Danish Regions, Copenhagen, Denmark; Nanna Skovgaard, Ministry of Health, Copenhagen, Denmark; Morten Freil, MSc, Danish Patients, Copenhagen, Denmark

Fig. 1 Clinician-graded adverse event frequencies during neoadjuvant treatment for patients at sites randomized to clinician-patient shared (96 patients at 45 sites) versus standard (85 patients at 34 sites) symptomatic AE reporting. All p values are two-sided and represent between-group comparisons of grade 0 versus [0 and ordinal grade

123

AIMS: In 2013, the Danish Government launched a national PRO initiative within its ‘Making eHealth Work: National Strategy for Digitalisation of the Danish Healthcare Sector 2013–2017’ in collaboration with the Danish Regions, whose members manage nearly all the hospitals in Denmark. The initiative involves a national analysis of the potentials of the nationwide systematic use of PRO in clinical practice. METHODS: The analysis was carried out by the PA Consulting group and involved separate evaluations of the potential of PRO in a range of diagnostic groups, technology, methods and

Qual Life Res (2015) 24:1–191 strategies of implementation and the calculation of the economic consequences of nationwide implementation. The initial phase included an iterative process of selecting diagnostic groups for further analyses. The process was closely monitored by the Government and the Danish Regions. The work was concluded in March 2015. During this process, the analysis was confined to telePRO where the patient report PRO from home before or instead of visiting the outpatient clinic. RESULTS: TelePRO experiences included asthma, epilepsy, chemotherapy, renal failure, prostatic cancer, and rheumatoid arthritis (29 clinics, 10,767 PRO assessments). National advantages of telePRO were related to transformation of traditional follow-up visits with fixed appointments to a periodic telePRO-screening for the actual need of a visit based on PRO-based reports of symptoms and the patients’ wish of a visit. The main conclusions were (1) The economic advantages are moderately, but robustly positive (2) Considerable gain in quality of care (3) Well documented clinical experiences in selected areas (4) Technical support of telePRO should be concurrently developed (5) Local involvement from clinicians is crucial. The economic gain originates from three sources of nearly equal sizes: reduction in patient visits, reduced transport reimbursement, reduced waste of medicine. The latter refers specifically to chemotherapeutic medicine made ready for the patient, but which must be destroyed because the patient turns out not to be fit when he shows up at the clinic. A stepwise approach for implementing telePRO is proposed and is at the moment the subject of negotiations between the Danish Government and Danish Regions. CONCLUSIONS: This is to our knowledge the first time that a national strategy for nationwide implementation of telePRO has been formulated by national health authorities.

Symposium Sessions Symposium #1: Training clinicians in how to use patient reported outcome measures in routine clinical care Chair: David Feeny, PhD, Professor, McMaster University, Hamilton, Ontario, Canada Panelists: Martha A Grootenhuis, PhD, Professor, Psychosocial Department of the Emma Children’s Hospital, Amsterdam, The Netherlands; Kate Absolom, PhD, Institute of Oncology, Leeds, United Kingdom Presenters: Lotte Haverman, PhD, Psychosocial Department of the Emma Children’s Hospital, Amsterdam, The Netherlands; Maria Jose Santana, PhD, Assistant professor, O’Brien Institute for Public Health, Cummings School of Medicine, University of Calgary, Calgary, Canada; Galina Velikova, PhD, MD, Professor of Psychosocial and Medical Oncology/Consultant Medical Oncology, University of Leeds, United Kingdom Using patient-reported outcome measures (PROMs) in routine clinical care identifies problems, enhances patient-clinician communication as well communication with and among family members, improves clinical management, and promotes share decision making. A key issue limiting successful implementation is clinicians’ lack of knowledge on how to effectively utilize PROs data in their clinical encounters. Therefore, clinicians need to be trained to interpret PRO data and make it actionable. The symposium describes three programs training clinicians to use PROMs in routine clinical practice of chronically ill patients, illustrating both successes and obstacles within the implementation process. The symposium includes presentations on: (1) Development and evaluation of an oncologist training program; (2) Training lung transplant specialists to use patient-reported outcome measures in routine clinical practice; (3) Using PRO measures in routine chronic care in paediatrics.

3 Individual Presentations Development and evaluation an oncologist training programme in the use of patient-reported outcome data Kate Absolom, PhD, Institute of Oncology; Galina Velikova, PhD, MD, Professor of Psychosocial and Medical Oncology/Consultant Medical Oncology, University of Leeds, Leeds, United Kingdom AIMS: Previous research has demonstrated the potential value of PROs in clinical practice. Health professionals have highlighted the importance of using appropriate measures and receiving training in incorporating and responding to the information. This study assesses the impact of a doctor training programme designed to support professionals using a PRO measure for cancer patients receiving chemotherapy. METHODS: The training structure and content was based on a national communication skills program for oncologists using role plays, providing a supportive environment to practise skills and share experiences. In order to have a more standardised content and to reduce the duration of the session, we used DVD/‘trigger tapes’ of simulated consultations with oncologists using PROs. The DVD was developed based on real-life patient scenarios from oncology outpatient settings. The DVD was used to design an interactive training session where oncologists could see examples of PRO data in use and discuss their value. Each scenario addressed different disease profiles including impaired physical functioning, emotional distress and social difficulties. A before-after design pilot study with 3 oncologists has been conducted to assess the impact of the training on how PRO data is utilised. A minimum of 10 patients per doctor were recruited before and after the training. Patients completed the PRO measure on touch-screens prior to routine outpatient consultations and graphic results were provided to the doctor. Consultations were audiorecorded and patients and doctors completed measures of patientcentred communication. RESULTS: 61 patient-consultations (31 pretraining/30 post-training) were content analysed. The PROs were explicitly referred to in significantly more consultations in the posttraining phase (48.4 vs 76.7 %, p \ 0.05). Physical functioning was raised more frequently in the post-training consultations (61.3 vs 86.7 %, p \ 0.05) as was pain (51.6 vs 86.7 %, p \ 0.05). Feedback from the oncologists suggested the training helped them consider different ways of integrating and sharing the patient reported data within consultations. CONCLUSIONS: Feedback received in the current study suggests a small group interactive session using simulated consultations is an acceptable way of training staff and engaging them in using the data effectively.

Training lung transplant specialists to use patient-reported outcome measures in routine clinical practice Maria Jose Santana, PhD, Assistant professor, O’Brien Institute for Public Health, Cummings School of Medicine, University of Calgary, Calgary, Canada; David Feeny, PhD, Professor, McMaster University, Hamilton, Ontario, Canada AIMS: Using PROs in routine clinical care identifies problems, enhances patient-clinician communication, improves clinical management, and promotes shared decision making. But clinicians need to be trained to interpret PRO data and make it actionable. In a randomized controlled clinical trial in chronic care management patients completed the Health Utilities Index (HUI) questionnaire at every out-patient clinic visit. Results were presented graphically as the Health Utilities Index score card (HUIsc). We describe how lung transplant specialists, transplant coordinators and pharmacists were trained in the use of the HUIsc, including the content of the training session with supporting material, subsequent monitoring of PROs use and evidence of adoption. METHODS: The selection of the PROs and

123

4 design of the HUIsc were done in consultation with the lung transplant team. The training session had three parts: (1) introduction to PROs and evidence supporting the use in clinical care; (2) familiarization with PROs; (3) interpretation of PRO data and what to do with it. Supporting material included the HUI proxy, case studies and audio-taped recordings from the encounter with patients. The training was based on real clinical scenarios. Clinical cases were selected to include patients with different problems, including hearing difficulties, cognitive deficits, and unexpected pain. Clinical parameters and HUI scores were compared. Training was conducted both formally at the weekly rounds and informally during regular clinics. RESULTS: Clinical scenarios were presented. For instance, a 69 years old, male lung recipient on a complex medication regimen. The HUIsc revealed serious memory problems. An action care plan was developed including change in immunosuppressant, addition of pillboxes, and transplant coordinators enlisted the help of a family caregiver. In another case, the HUIsc detected unexpected pain and revealed emotional distress/depression. In both cases medication doses were adjusted. CONCLUSIONS: Clinicians should be part of the PRO selection process and training design. We used real clinical cases as training tools; learning by doing. The HUIsc structured the consultation by prioritising important issues. The impact of the training sessions should be further tested in a randomized controlled clinical trial.

The effectiveness of an innovative web-based application to monitor health-related quality of life in pediatric clinical practice Lotte Haverman, PhD; Martha A. Grootenhuis, PhD, Psychosocial Department of the Emma Children’s Hospital, Amsterdam, The Netherlands AIMS: Children with a chronic disease are at risk for HRQOL problems. The use of electronic Patient Reported Outcomes (ePROs) to assess HRQOL in clinical practice is therefore indicated. We describe the implementation of ePROs and how paediatricians, nurses, social workers and psychologist are trained to use the ePROs in their daily paediatric practice. METHODS: In different children’s hospitals in the Netherlands, children (8–18 years old) or parents (0–7 years old child) visiting the outpatient clinic complete HRQOL questionnaires at home before the consultation via a website (KLIK program; www.hetklikt.nu). The resulting ePROfile is provided to the pediatrician and/or another medical professional. To improve the use and interpretation of ePROs in clinical practice we developed a training program for pediatricians, nurses, social workers and psychologists. Four psychologists and one pediatric oncologist were involved in the development of the training. RESULTS: The training consisted of a short theoretical component and an extensive practical component. For the practical component, we used movies containing short patient cases representing real consultations and actual ePROfiles. Up to now, 2200 patients have participated in the KLIK program. Different disease groups are included for example; rheumatology, haematology, nephrology, HIV, oncology, follow-up PICU and cystic fibrosis. So far, 90 paediatricians, 35 nurses, 12 social workers and 21 psychologists were trained. CONCLUSIONS: We recommend implementation of ePROfiles in pediatric clinical practice to detect HRQOL problems and enable patient-tailored psychosocial interventions at an early stage. The multidisciplinary team should be involved in ePRO development and training to optimally benefit from the use of ePROs in clinical practice. The training is an essential part of the implementation; however, the effect of the training should be investigated further.

123

Qual Life Res (2015) 24:1–191 Symposium #2: PRO assessment in clinical practice: PCORI’s perspectives and case examples to advance research and clinical care Moderators: Steven Clauser, PhD, PCORI, Washington, DC, United States; Ethan Basch, MD MSc, Cancer Outcomes Research Program, University of North Carolina - Chapel Hill, Chapel Hill, NC, United States AIMS: Research studies that involve use of PRO assessment in clinical settings have expanded over the last 5 years and several sustainable and scalable models exist. Despite this, several common challenges to PRO uptake and use remain. Research sponsored by the Patient Centered Outcomes Research Institute (PCORI) and others suggests bridging from research use to clinical use requires awareness of factors that impinge on feasibility of PRO administration, data management, analysis, and interpretation. Methodologic challenges of research use of PRO data, aggregated across samples, differ from the challenges of clinical use which requires both aggregate-level benchmark and/or interpretation guidelines and individual-level reports useful for patients, clinicians, and other end users of the information. METHODS: Through case examples and a review of recent PCORI funded research awards, presenters will provide background on the research and real-world clinical challenges PROs are intended to address. RESULTS: This symposium presents opportunities and challenges common to diverse clinical settings for both research and clinical PRO use, and lessons learned in overcoming those challenges. Through specific examples from different therapeutic areas and patient populations, an action agenda for improving accessibility of PRO data for both research and clinical purposes is forming. This agenda points toward establishing best practices in measure selection, usability testing particularly for administration across different platforms, data management principles useful across settings, and ensuring data interpretation guidelines differentiate between research needs and clinical use. CONCLUSIONS: The examples presented here provide a range of possibilities for increasing the potential for and relevance of PRO assessment for clinical care and research.

Individual Presentations Integrating PROs into routine clinical care: perceived value and impact Clifton O. Bingham III, MD, Johns Hopkins University, Baltimore, MD, United States; Ana-Maria Orbai, MD, Johns Hopkins University, Baltimore, MD, United States; Katherine C. Smith, PhD, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States; Elaine Deleon, MPH, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States; Susan Bartlett, PhD, Associate Professor of Medicine, Divisons of Clinical Epidemiology, Rheumatology and Respiratory Epidemiology and Clinical Trials Unit, Montreal, QC, Quebec, Canada In the second of two presentations concerning PRO integration in arthritis care, we will discuss the reactions (both positive and negative, and over time) of patients, providers, and clinic staff to the value of having PRO data available during the clinic visit. We will offer examples of how PROs results impacted conversations, medical decision-making, and immediate outcomes. A consistent call from both patients and providers has been the need for additional training on PRO interpretation and use. New findings from ongoing studies to define meaningful change and actionable (alert) levels for symptoms such as depression will also be presented.

Qual Life Res (2015) 24:1–191 Issues in using patient-reported outcomes for performance measurement Sarah Hudson Scholle, MPH DrPH, National Committee for Quality Assurance, Washington, DC, United States; Jessica B. French, National Committee for Quality Assurance, Washington, DC, United States; Mary Barton, National Committee for Quality Assurance, Washington, DC, United States AIMS: Performance measures based on patient-reported outcome measures (PROMs) may address the need for assessing whether care is both effective and patient-centered. Efforts are critically needed to identify and understand best approaches for introducing PROMs into the point of care for decision-making and how they could be used to benchmark performance at the delivery system or provider level. METHODS: We present a framework identifying key stakeholders, roles for and uses of the patient-reported outcome measure data for each stakeholder, and relevant incentives. We also report on field work in several projects involving the development and testing of performance measures based on patient-reported outcome measures, including care for depression among adolescents, adults receiving long-term services and supports (LTSS), and adults with diabetes. RESULTS: We identified key stakeholders to be patients and family members, clinicians, health care teams, and QI staff and the delivery systems in which they operate, as well as payers, employers and national policymakers. For example, patients need to understand the potential for PROMs to support better care and how PROMs can help them communicate more effectively with their clinicians to achieve their goals. Interviews with patients receiving LTSS showed that patients felt PROMs were duplicative. Clinicians need to see that the data add value to care and that the time required to review the PROMs enhances patient outcomes. Interviews with care managers and primary care staff suggest that PROMs may provide a structured process for gathering information on topics such as sleep, pain, social functioning. However, our field test on depression care demonstrated that the use of structured tools for monitoring symptoms was limited even when tools were available and could be captured in the EHR. Utilization of tools was higher in settings that required structured PROM to be completed at the time of antidepressant prescriptions. CONCLUSIONS: Understanding and addressing the concerns of diverse stakeholders involved in collecting, using and acting on PROMs results will be critical to successful implementation in clinical care and making data necessary for evaluating the performance of health care providers and organizations.

Integrating PROs into routine arthritis clinical care: process and feasibility Susan Bartlett, PhD, McGill University, Montreal, Quebec, Canada; Ana-Maria Orbai, MD, Johns Hopkins University, Baltimore, MD, United States; Michelle Jones, BA, Johns Hopkins University, Baltimore, MD, United States; Clifton O. Bingham III, MD, Johns Hopkins University, Baltimore, MD, United States Patient reported outcomes (PROs) querying health status, symptoms and impacts of chronic diseases and their treatments are powerful tools to help clinicians, administrators, and policy makers recognize the lived experiences of chronic disease patients. However, the collection of PROs as part of routine care visits has been largely relegated to collecting data for research purposes. In an era of growing emphasis on patient-centered care, there is considerable interest in using this information as part of chronic disease management to facilitate patient-physician communication, to better match patients to treatment, and to promote opportunities for shared decision-making. A number of process, measurement, and interpretation considerations are important to recognize as potential challenges for PRO integration in clinical care. Over two presentations, we will illustrate lessons learned from integrating a battery of PROs into routine visits for inflammatory arthritis management. In our PCORI Pilot

5 project, we evaluated the feasibility, acceptability and potential impact of integrating ‘‘universal’’ PROs (PROMIS measures) into routine arthritis care. We hypothesized that the systematic collection of patient-valued symptoms and impacts could help improve the quality of care by helping clinicians to identify unmet needs of patients and comorbidities, enhance communication between patients and their doctors and increase satisfaction. Over 2 years, 200 people with RA seen at a large, urban arthritis clinic completed computerized adapted tests [CAT] and short forms [SF] assessing arthritis symptoms and impacts (pain, fatigue, physical function, mood, and participation) along with legacy PROs in the waiting room prior to seeing their doctor. Results were made available to physicians and clinicians for use during the clinical visit. This process was repeated for up to six regularly scheduled visits. In the first of two presentations, we will discuss the practice aspects and outcomes associated with routine PRO collection in arthritis clinical care at Johns Hopkins. We will review the necessary infrastructure (personnel, computers, adaptations or disability), practical considerations (training, time requirements, work-flow considerations), performance of universal versus disease-specific measures, and impact on identification of comorbidities, referrals, etc.

PCORI’s improving PCOR methods program: building a patient-centered measurement portfolio Jason Gerson, PhD, Associate Director, CER Methods, PCORI, Washington, DC, United States AIMS: PCORI’s Improving PCOR Methods program is one of PCORI’s National Research Priorities. Its goal is to fund studies that address gaps in methodological research relevant to conducting PCOR, including in the area of patient-centered measurement. In November 2013, PCORI held a Patient Reported Outcomes (PRO) Infrastructure Workshop on Integrating PROs into Electronic Health Records (EHR). The workshop brought together a diverse group of individuals and organizations poised to make a positive change in uptake and use of PROs in EHRs for clinical care, research, and performance management. The workshop produced several recommendations in the area of improving methods, and was a catalyst for the Methods program establishing research on patient-centered measurement as a programmatic priority. METHODS: My presentation will highlight currently funded PCORI research projects to address key challenges and approaches taken to address the multiple dimensions of using PRO data in clinical care. RESULTS: Through a review of these projects, I will review the following elements of PRO use and integration in clinical practice: • Research on usefulness of patient-reported data in clinical care; issues affecting measure usefulness that may be included in this research are usability and variation by platform, concordance between patient report and other data sources, and respondent burden • Research on PRO score interpretability, with a focus on clinical meaningfulness • Research on the collection of PRO data for multiple purposes (clinical care, care quality assessment, effectiveness research) • Research on integrating PROs into clinical practice, including clinician use and acceptance of PRO data; research on the relationship between PROs and clinical outcomes (utilization, hospitalizations, medications); comparisons between clinic-based and home-based PRO reporting to understand differences; research on health systems-level incentives and disincentives to incorporating patient-reported data into clinical care may also be included • Research on the use of PROs in registries (feasibility assessment, approaches for enhancing completeness and reducing missing data, measuring improvement versus progression, estimating ideal follow-up time frames) CONCLUSIONS: PCORI, and the Improving PCOR Methods program in particular, is dedicated to funding rigorous research concerning PRO use and integration in clinical practice. This presentation will provide the audience with a clear articulation of PCORI’s programmatic interest in this area.

123

6 Symposium #3: Health preference research: enhancing the relevance of QOL evidence Moderator: Benjamin M. Craig, PhD, Moffitt Cancer Center, Tampa, FL, United States Panelists: Scott M. Gilbert, MD, Moffitt Cancer Center, Tampa, FL, United States, Axel Mu¨hlbacher, PhD, Hochschule Neubrandenburg, Neubrandenburg, Germany, Jeff A. Sloan, PhD, Mayo Clinic, Rochester, MN, United States The relevance of QOL evidence can be enhanced through the systematic assessment of health preferences on measured outcomes using choice experiments. The results of these choice experiments can aid in the dissemination and implementation of comparative QOL evidence by better informing policy makers, providers and patients faced with difficult decisions, such as market access, pricing, guidelines, and treatment choice, respectively. This symposium covers 3 exemplary applications demonstrating how patient preference research enhances QOL evidence: (1) Patient preferences included in decision aids can help navigate their care (BEACON Project; Jeff Sloan); (2) Patient preferences within a cancer center can enhance the summary of clinical QOL evidence (Bladder Cancer Index; Scott Gilbert); and (3) Patient preferences on diagnostic test outcomes (e.g., false positives) can inform health policy (Lung Cancer; Axel Mu¨hlbacher). This patient-centered symposium is most beneficial for researchers interested in understanding and demonstrating the relevance of QOL evidence, particularly in decision making and physician/patient/policymaker communication.

Individual Presentations Incorporating patient preferences on QOL into routine clinical practice: the BEACON project Jeff A. Sloan, PhD, Mayo Clinic, Rochester, MN, United States Clinical relevance of PROs and patient preferences in a surgical oncology clinic Scott M Gilbert, MD, Moffitt Cancer Center, Tampa, FL, United States Treatment preferences of patients with metastatic non-small cell lung cancer: a discrete choice experiment Axel C. Mu¨hlbacher, PhD, Hochschule Neubrandenburg, Neubrandenburg, Germany

Symposium #4: Developing and validating clinician reported outcomes: discussion of applied methods Moderator: Louise Humphrey, MSc, Abacus International, Manchester, United Kingdom Panelists: Emuella Flood, ICON PRO, Bethesda, MD, USA; Linda Abetz-Webb, Patient-Centred Outcomes Assessment (P-COA), Cheshire, United Kingdom; Cicely Kerr, MSc PhD, Oxfordshire, Oxford, United Kingdom; Stacie Hudgens, Clinical Outcomes Solutions, Tucson, AZ, USA Clinician-Reported Outcomes (ClinROs) fall under the umbrella term of Clinical Outcomes Assessments (COAs) and provide clinician insight into ‘‘a patient’s symptoms, overall mental state, or the effects of a disease or condition on how the patient functions’’ (FDA Guidance for Industry 2009). ClinROs are particularly important for capturing treatment benefit for patients unable to self-report or for concepts which require training and expertise to make the evaluation. Within the regulatory framework, ClinROs must be fit-for-purpose within the context of use to support product labeling. Currently, there is no specific guidance on ClinROs published by regulatory agencies, and a critical need within the field is to understand applied, practical methods that are specifically appropriate for the development and psychometric testing

123

Qual Life Res (2015) 24:1–191 of these instruments. Utilizing learnings from Patient-Reported Outcomes and Health-Related Quality of Life measure development and experiences in developing and applying ClinROs, this symposium intends to address this critical need and engage participants by providing real world examples of applied methodological considerations for ClinROs, including practical challenges, evaluation of existing ClinROs and development of new ClinROs.

Individual Presentations ClinROs: History and importance, regulatory guidance, selection and applied development methods Emuella Flood, ICON PRO, Bethesda, MD, United States Clinician Reported Outcomes (ClinROs) are assessments of a patients’ health status made by a trained health care professional (HCP). ClinROs are the most commonly used type of clinical outcomes assessment (COA) in interventional studies: In a review of FDA labels from 1997 to 2002, there were 215 products approved and 62 % included ClinRO endpoints (52 % included lab tests/device measurements; 30 % included PROs) whereas 28 % included ClinRO only (25 % lab/device only; 11 % PRO only). ClinROs are also widely used in observational studies and day to day clinical practice. Within the regulatory framework, ClinROs must be fit-forpurpose within the context of use to support product labeling. Currently, there is no specific guidance on ClinROs published by regulatory agencies. This presentation explores the history and importance of ClinROs, considers the defining characteristics of ClinROs and different types of ClinROs and the application to ClinROs of existing regulatory guidance to support product labeling to ClinRO selection and development. In the absence of ClinRO-specific regulatory or other best practice guidance, the presentation also considers applied methods for developing ClinROs, drawing on published examples and existing relevant methods including Delphi panel and nominal group techniques.

Pediatric ClinROs: special considerations and recommendations for pediatric ClinRO development methods Linda Abetz-Webb, Patient-Centred Outcomes Assessment, Cheshire, United Kingdom There is widespread interest in pediatric ClinROs, particularly in an age group or particular disease where it is difficult for a child to communicate. However, most pediatric ClinROs were adapted from adult ClinROs with limited scientific rigor applied in terms of content validity and psychometric reliability and validity for children across age groups. This presentation focuses on key considerations and methods to employ when designing pediatric ClinROs. When replying to a pediatric ClinRO, clinicians consider multiple elements: their own observations, lab or outside reports, child reports, and caregiver reports. The element they take as primary in their decision making can vary dependent on the child’s age and developmental stage, their disease, and the clinician’s own biases. Discrepancies between clinician and parent reports are well documented and these can often be due to the different elements and biases that each bring to the assessment. Pediatric ClinROs should be developed following similar methods to pediatric ObsRO and PRO development, that is with careful consideration given to the disease in the context of different age groups and developmental milestones across the development process (concept elicitation, cognitive debriefing, psychometric validation). Additional issues related to the analysis of studies will also be discussed.

Qual Life Res (2015) 24:1–191 ClinROs: applied methods for evaluating content validity of ClinROs Cicely Kerr, MSc. PhD. Oxfordshire, Oxford, United Kingdom This presentation considers the challenge of evaluating content validity of ClinROs, which in a regulatory context includes both how well the instrument measures the concept of interest and the link to meaningful benefit in terms of how patients feel and function in their everyday lives, which may more appropriately be defined as ecological validity. Evidence of good content validity is of core importance for patient reported outcomes (PROs) for which regulatory guidance is available, however approaches advocated for PROs are not directly applicable to ClinRO evaluation. The presentation focuses on key considerations for study design and planning relevant to the characteristics of the ClinRO instrument, the measurement concept and the patient population. This includes discussion of what to evaluate and how, including logistical and ethical challenges, practical considerations around observation and administration of the ClinRO, development of interview questions for clinicians and patients. The presentation also considers specific qualitative analysis challenges of handling a mixture of structured/ probed cognitive debrief and more open-ended concept elicitation data, variation in experience and life context, the meaning of data saturation in this research context and how this may be appropriately assessed.

ClinRO psychometric validation: analytic and research design considerations Stacie Hudgens, MA (ABD), Clinical Outcomes Solutions, Tucson, AZ, United States Development of a reliable and valid ClinRO endpoint for assessing patient outcomes is a stepwise process which begins with strong qualitative evidence substantiated and confirmed with equally strong psychometric data. This presentation focuses on the research design and analytic considerations when psychometrically evaluating a de novo or legacy ClinRO. Prior to designing a psychometric study to assess the measurement properties of a ClinRO, several considerations must be addressed including the content, structure, and format of the tool (e.g. number of items, domains, response options, scoring) as well as an external variables that may influence study design and possible outcomes (e.g. clinician experience, patient diversity). While we can retain most of the methods used for PRO/ObsRO validation, we need to introduce new methods and revisit study design to accommodate the fact the information is not coming from the patient directly. For example, test–retest reliability may now be replaced by intra- or inter-rater reliability and data collection through in-person patient assessments vs video consultation, photographs or a standardized patient scenario may have implications logistically and analytically. Utilizing a combination of classical (e.g. rank order or ANOVA) and novel (e.g. item level analysis model that allows for the evaluation of item performance as well as patient and clinician level responses) approaches to analytic methods will help to ensure appropriate validation of the ClinRO.

Symposium #5: Quality of life following stroke, traumatic brain injury and spinal cord injury: influences of person and environmental factors Moderator: Allen Heinemann, Feinberg School of Medicine Northwestern University, Chicago, IL, United States

7 The quality of life of persons living in the community with physical disabilities can be influenced by a variety of person and environment factors. This symposium reports findings from a large-scale study that recruited a community sample of people with stroke, traumatic brain injury and spinal cord injury. The aims of this symposium are to: (1) Describe the influence of health literacy on quality of life following stroke, traumatic brain injury and spinal cord injury; (2)Discuss how economic quality of life reflects resources and disability burden in a community sample; (3) Discuss the acceptability of outcome measurement for persons with disability; (4) Illustrate the relationships between emotional functioning and quality of life in a community sample of persons with disabilities; and (5) Model social aspects of quality of life as a function of cognition and motor performance in a community sample of persons with disabilities. The symposium draws on data collected as part of Rehabilitation Research and Training Center funded by the National Institute on Disability and Rehabilitation Research. The project recruited 604 persons with stroke, TBI and SCI living in Chicago, Ann Arbor, MI, and St. Louis, MO. They completed the NIH Toolbox for the Assessment of Neurological and Behavior Function, neuropsychological instruments, Patient Reported Outcomes Measurement Information System (PROMIS) item banks, and newly developed measures of environmental factors. Five presentations will highlight findings related to (1) computer-based measurement of health literacy, functional literacy, cognitive function, and patient-reported outcomes, (2) measuring economic quality of life, (3) modeling social outcomes with cognitive and environmental measures, (4) acceptability of outcome measurement for persons with disability and the implications for people with limited experience or ability interacting with datacollection technology, and (5) emotional functioning and quality of life. April 2015 Archives of Phys Med Rehabil Heinemann AW, Magasi S, Hammel J, et al. Environmental Factors Item Development for Persons with Stroke, Traumatic Brain Injury and Spinal Cord Injury Tulsky DS, Kisala PA, Lai JS et al. Developing an Item Bank to Measure Economic Quality of Life for Individuals with Disabilities Garcia S, Hahn E, Heinemann AW, Lai JS et al. Development of SelfReport Measures of Social Attitudes that Act as Environmental Barriers and Facilitators for People with Disabilities Magasi, Wong A, Gray DB, et al. Theoretical Foundations for the Measurement of Environmental Factors and their Impact upon Participation among People with Disabilities Hammel J, Magasi S, Heinemann AW, et al. Environmental Barriers & Supports to Everyday Participation: A Qualitative Insider Perspective from People with Disabilities.

Individual Presentations Computer-based measurement of health literacy, functional literacy, cognitive abilities and patient-reported outcomes among English-speaking adults with functional limitations due to spinal cord injury, stroke or traumatic brain injury Elizabeth A. Hahn, Northwestern University Feinberg School of Medicine, Chicago, IL, United States Literacy skills are critical for adults in the U.S. to function effectively in their daily lives. Health literacy represents a constellation of skills to perform health-related tasks, and may be significantly worse than functional literacy because of the unfamiliar context and vocabulary of the health care system. Limited health literacy is widespread and is associated with poorer health care access and outcomes; it may also be associated with poorer cognitive abilities. Few studies have measured health literacy and cognition in medical rehabilitation populations. The objectives of this study were: (1) to describe preferences

123

8 and experiences with computer-based testing; (2) to describe the functional and health literacy levels of individuals with functional limitations due to spinal cord injury, stroke or traumatic brain injury; and (3) to evaluate associations between health literacy, functional literacy, cognitive abilities and patient-reported outcomes (PROs). About 200 people in each of the three diagnostic groups completed cognitive function, literacy and PRO assessments using a variety of novel health information technologies. Most participants did not require assistance using the computer equipment. In terms of preferences for completing PROs, 38 % preferred the touch screen plus audio device, 37 % preferred text only and 25 % had no preference. The three injury groups differed in most measures of health literacy, functional literacy and self-reported health. The stroke group had the lowest levels of health literacy, functional literacy and cognitive functioning, and the poorest overall health (37 % poor/fair) compared to those with a spinal cord injury (17 %) or traumatic brain injury (19 %; p \ 0.001). Correlations between health literacy and functional literacy measures ranged from 0.569 to 0.653. Correlations among functional literacy measures were slightly higher (0.603–0.859). Correlations between health/functional literacy measures and cognitive function measures were much lower (0.247–0.463). Mean health literacy scores increased across levels of self-reported health in each group. Novel computer-based technologies to measure health literacy and PROs are acceptable to medical rehabilitation populations, and provide useful methods for self-administration of instruments. Consideration of health literacy in rehabilitation practice could enhance the effectiveness of the clientprovider relationship and identify strategies to improve health-related quality of life.

New PRO scales to measure economic quality of life (ECQ) David S. Tulsky, University of Delaware, Newark, DE, United States; Allen Heinemann Feinberg School of Medicine Northwestern University, Chicago, IL, United States; Pamela Kisala, University of Delaware, Newark, DE, United States; Sara Jerousek, Rehabilitation Institute of Chicago, Chicago, IL, United States; JinShei Lai, Northwestern University, Chicago, IL, United States; Noelle Carlozzi, University of Michigan, Ann Arbor, MI, United States; Patrick Semik, Rehabilitation Institute of Chicago, Chicago, IL, United States AIMS: Research has demonstrated that individuals with lower socioeconomic status have poorer health outcomes, however, the impact on quality of life (QOL) from disparities remains understudied. Limited economic resources affect individuals with disabilities and their QOL. Individuals with disabilities often have fewer resources because of unemployment, poorer health status, and greater healthcare needs. Previous talks have identified the need to develop a measure that describes the consequences of economic resources on QOL. This presentation presents the finalization of this work and validation of the item bank and short form (SF) on ECQ. METHODS: We used a mixed-methods design to develop an item bank based on patient feedback about economic issues and how they affect QOL. We had previously reduced the item pool to 32 using Rasch analyses. Currently, we report the results of an extensive field test using samples of community-dwelling individuals with the same conditions (N = 1195). We planned to conduct a 2-parameter IRT model to calibrate the items, remove poorly performing items, and develop a calibrated item bank that can be administered via SF and CAT. Finally, we sought validity evidence by correlating the new measure with self-reported personal and household income. RESULTS: In the field test, confirmatory factor analyses (CFA) indicated that the items did not fit a unidimensional model, a violation of the assumptions for

123

Qual Life Res (2015) 24:1–191 IRT. Then, we explored item fit to the Rasch model to identify misfitting items. We removed 6 items because of local dependence and misfit to the Rasch model. Because CFA did not support a unidimensional model for the 18 positively worded items, we developed a fixed scale that can be administered using the entire bank or as an 8-item SF. Spearman’s correlation with self-reported household income was .48 for the full bank and .44 for the short form, and correlations with personal income were .42 for both forms. CONCLUSIONS: The economic quality of life (ECQ) scale defines a unique construct for which there is no other measure. Its development addresses an issue highly salient to individuals with disabilities. The ECQ scale allows investigators to address economic and disabilityrelated health disparities.

Finding a focus for social aspects of quality of life with neurological disorders: cognitive, emotional, and environmental factors Alex Wong, Washington University School of Medicine, St Louis, MO, United States; Elizabeth A. Hahn, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; David S. Tulsky, University of Delaware, Newark, DE, United States; Susan Magasi, University of Illinois at Chicago, Chicago, IL, United States; Noelle Carlozzi, University of Michigan, Ann Arbor, MI, United States; Allen Heinemann, Feinberg School of Medicine Northwestern University, Chicago, IL, United States AIMS: Full and equal participation in community life is the ultimate social health outcome in people with neurological disabilities. Unfortunately, there is no gold standard for the measurement of social health outcomes, which are multifaceted, complex and individualized. Furthermore, participation in social or community activities can be influenced by multiple factors including opportunities, environmental barriers, social roles, and various aspects of body function. Defining the construct of social outcome and identifying the factors associated with this outcome could ultimately improve the quality of life of people with neurological disabilities. The purpose of this study was to test several latent models to investigate the complex relationships between cognitive, emotional, environmental factors and participation in social/community activities among persons with spinal cord injury (SCI), traumatic brain injury (TBI), and stroke. METHODS: We recruited a sample of 604 community-dwelling adults (SCI = 209, TBI = 184; stroke = 211; average age = 47.3 years; 64.1 % men; 58.6 % white; 67.2 % completed more than high school education) who were at least 1-year post-injury or disease to participate in the current study. We administered the NIH Toolbox—Cognitive and Emotional batteries, Executive Function Performance Test (EFPT) Test, along with the PROMIS—Social Health, Participation Enfranchisement, Economic Quality of Life, and the Environmental Factors Item Banks (EFIB). We used structural equation modeling (SEM) to evaluate their complex relationships. RESULTS: Final models for cognitive, emotional and environmental factors fit the data relatively well (CFA [ 0.9, TLI [ 0.9; RMSEA B 0.08). Specifically, higher environmental barriers were associated with lower participation (r = -0.68, p \ 0.001). Higher emotional functioning was associated with higher participation (r = 0.86, p \ 0.001). Whereas cognitive function measured by the EFPT and NIH Toolbox measures were not associated with participation (r = 0.06, p [ 0.05). CONCLUSIONS: Both emotional and environmental factors were associated with social aspects of quality of life. Future efforts for improving social health outcomes of these neurologic populations could be enhanced by utilizing patient-reported outcome measures and tailoring interventions dedicated to improve emotional functioning, and reduce/address the environmental barriers.

Qual Life Res (2015) 24:1–191 Reasonable accommodations affect the validity of scores on the NIH Toolbox Cognition Battery among people with neurological disabilities Susan Magasi, University of Illinois at Chicago, Chicago, IL, United States; Mark Harniss, University of Washington, Seattle, WA, United States; Apeksha Gohil, UIC, Chicago, United States AIMS: Principles of fairness in testing seek to ensure that all test takers, including those with disabilities, have the opportunity to demonstrate their true capacity or experience in constructs being measured. The Standards describe fairness in testing as a fundamental validity issue. Yet issues at the human–computer or assessment interaction can pose barriers to accurate measurement and interpretation of derived scores among individuals with physical, cognitive and sensory impairments. The purpose of this study was to evaluate the effects of barriers in the human–computer interaction had on test takers with spinal cord injury, traumatic brain injury and stroke’s (1) ability to complete the NIH Toolbox Cognition Battery and (2) the effects of accommodations on the validity of scores. METHODS: 604 people with neurological disabilities (209 Spinal cord injury, 184 traumatic brain injury, and 211 stroke) completed the computer-administered NIH Toolbox Cognition Battery as part of a comprehensive validation protocol. Upon completion of each measure, test administrators scored the measure as complete, incomplete or non-standard administration and provided a written description. The lead author, a disability and accessibility expert, used a content analysis approach to categorize the rationale for each non-standard administration. RESULTS: While the majority of the participants with neurological conditions were able to complete the NIH Toolbox Cognition Battery, 83 (13.7 % of the total sample) received accommodations for physical, cognitive or sensory impairments. Based on expert review of the accommodated assessments, 63 (or 10 % of the total sample) were deemed to pose threats to validity and test score interpretation. These cases were dropped from subsequent analyses. CONCLUSIONS: It is important to include people with disabilities in validation studies, yet administrator accommodations to address barriers at the human–computer interaction can limit the validity and interpretability of derived scores. It is critical that test administrators understand the impact of accommodations on scores and that interpretations be informed by actual practices. At a more fundamental level, test developers should consider accessibility at the design, administration and interpretation stages.

Health disparities in health-related quality of life among individuals with disabilities Noelle Carlozzi, University of Michigan, Ann Arbor, MI, United States; Ana Miskovic, Rehabilitation Institute of Chicago, Chicago, IL, United States While much research has been done identifying health disparities in healthcare, much less is known about health disparities as they relate to health-related quality of life for individuals with disabilities. To address this shortcoming, we examined self-reported HRQOL in individuals with traumatic brain injury (n = 184), stroke (n = 211), and spinal cord injury (n = 209) using the NIH Toolbox for the Assessment of Neurological Behavior and Function (NIHTB) Emotion Battery. Specifically, we examined 17 different patient reported outcomes measures across four subdomains of emotion: Psychological Well-Being, Social Relationships, Stress and Self-Efficacy, and Negative Affect. Demographic group comparisons using general linear models were conducted for gender, education, race, ethnicity and socioeconomic status (effect sizes are reported as partial Eta squared after adjusting for age and other additional demographic variables). Self-reported HRQOL did not differ according to gender, education or ethnicity. There were significant differences for race on

9 friendship (partial g2 = .011) and Negative Affect (Anger-Hostility partial (g2 = .031) and Fear Somatic arousal partial (g2 = .018); blacks indicated more distress and less friendship than whites; effect sizes were small. There were also significant differences for those with above versus below average income (i.e., $52 K annual household income) with regard to: (1) self-reported satisfaction with social relationships and social support (i.e., Friendship (partial g2 = .044, Loneliness partial g2 = .022, Emotional Support partial g2 = .035, Instrumental Support partial g2 = .053); (2), negative affect (Perceived Rejection partial g2 = .03, Anger-Affect partial g2 = .01, Anger-Hostility partial g2 = .013), general well-being (General Life Satisfaction partial g2 = .022), and self-efficacy (partial g2 = .013). Individuals with more economic means reported more satisfaction with friendships, emotional support, and instrumental support, as well as less loneliness than those of lesser means. In addition, those with greater income also reported less personal rejection, less anger, and more general life satisfaction than those with lower income. While results suggest some health disparities in HRQOL for race, they especially highlight health disparities for socioeconomic status. Future work is needed to better understand such disparities so that interventions might be developed to mitigate such effects.

Symposium #6: Response shift effects at the item level: cross-method insights for interpreting change over time Chair: Ve´ronique Se´bille, ScD, EA 4275, Faculty of Pharmacy, University of Nantes, France Discussant: Carolyn E. Schwartz, ScD, On behalf of the ISOQOL Response Shift Special Interest Group, DeltaQuest Foundation, Inc., Concord, MA; Tufts University School of Medicine, Boston, MA, United States Presenters: Mirjam Sprangers, PhD, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands; I-Chan Huang, PhD, Department of Health Outcomes and Policy, and Institute for Child Health Policy, College of Medicine, University of Florida; Mathilde Verdam, MSc, Child Development and Education, and Academic Medical Centre, University of Amsterdam, The Netherlands M.Sc., Myriam Blanchin, PhD; Jean-Benoit Hardouin, ScD, EA 4275, Faculty of Pharmacy, University of Nantes, France Interpreting change in patient-reported outcomes (PRO) can be obfuscated by response shift (RS). RS research focuses on subscale scores combining several items. It can be hypothesized that the prevalence and magnitude of RS effects at item-level may influence RS detection and true change estimation and that item-level analysis could help getting more insight into RS interpretation. Different RS detection methods may perform differently in identifying item-level RS effects. This symposium presents five studies of item-level RS detection methods to investigate these unanswered questions in different disease populations. Qualitative and quantitative analyses examine RS effects at itemlevel and their impact on interpreting change in PRO. Dr. Sprangers presents a cognitive interviewing study of item response to qualitatively examine then test assumptions for detecting RS in cancer patients. Dr. Huang presents item-level structural equation modelling (SEM) analyses of pediatric patients. Dr. Verdam presents SEM analyses for item-level RS detection in cancer patients undergoing radiotherapy. Dr. Blanchin presents Guttman errors and Item Response Theory (IRT) models to identify individual- and item-level RS in chronically ill patients. Dr. Hardouin presents non-parametric IRT for item-level RS detection. Thentest assumptions were not supported in qualitative analyses. Preliminary analyses suggest that item-level RS effects were detected in quantitative analyses. Item-level RS effects had a small to moderate impact on true change estimation. The discussant will present

123

10 results of a SIG survey examining the expected implications of itemlevel RS effects. This symposium provides a range of methods and implications for detecting item-level RS effects.

Individual Presentations Using a retrospective pretest instead of a conventional pretest is replacing biases: a qualitative study of cognitive processes underlying responses to thentest items Mirjam A.G. Sprangers, PhD; Elsbeth F. Taminiau-Bloem, PhD; Florence J. van Zuuren, PhD, Department of Medical Psychology, Academic Medical Centre, University of Amsterdam, The Netherlands; Margot A. Koeneman, PhD, Body@Work, Research Center for Physical Activity, Work and Health, TNO-VU University Medical Center, Amsterdam, The Netherlands; Mechteld R.M. Visser, PhD, Department of Medical Psychology, Academic Medical Centre, University of Amsterdam, The Netherlands; Carol Tishelman, PhD, Karolinska Institutet, Stockholm, Sweden; Caro C.E. Koning, MD PhD, Department of Radiotherapy, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands; Carolyn E. Schwartz, ScD, DeltaQuest Foundation, Inc., Concord, MA; Tufts University School of Medicine, Boston, MA, United States BACKGROUND: Cognitive interviewing techniques hold special promise for evaluating the cognitive processes relevant to response shift at the item level. One application would be to investigate whether similar cognitive processes over time are used across items reflecting different quality of life (QoL) domains, particularly within the thentest design, an approach to detect and control for recalibration response shift. This design assumes (1) more consistency in the content of the cognitive processes underlying patients’ QoL between posttest and thentest assessments than between posttest and pretest assessments; and (2) consistency in the time frame and description of functioning referenced at pretest and thentest. Our objective is to utilize cognitive interviewing to qualitatively examine both assumptions. METHODS: We conducted think-aloud interviews with 24 cancer patients prior to and on the last day of radiotherapy to elicit cognitive processes underlying their assessment at pretest, posttest and thentest of 7 EORTC items covering: walking, pain, fatigue, worry, social functioning, overall health, and overall QoL. Two independent raters used an analytic scheme based on the cognitive process models of Tourangeau et al. and Rapkin & Schwartz that yielded five cognitive processes. We subsequently used this input for quantitative analysis of count data. RESULTS: Contrary to expectation, the number of dissimilar cognitive processes between posttest and thentest was generally larger than between pretest and posttest across patients. Further, patients considered a range of time frames when answering the thentest questions. Moreover, patients’ description at the thentest of their pretest functioning was often not similar to that which was noted at pretest. Items referring to trouble taking a short walk, overall health, and QoL were most often violating the assumptions. CONCLUSIONS: Both assumptions underlying the thentest design appear not to be supported by the patients’ cognitive processes. Patients provide assessments that are personally meaningful but which do not necessarily reference the time frame or content intended by the thentest approach. Replacing the conventional pretest–posttest design with the thentest design may simply be replacing one set of biases with another.

Assessing item-level response shift related to the change of health state on the NIH PROMIS Paediatric Scales: a report from the NIH PROMIS Paediatric Asthma Study

123

Qual Life Res (2015) 24:1–191 Pranav K. Gandhi, PhD, Department of Pharmacy Practice, School of Pharmacy, South College, Knoxville, TN, United States; Carolyn E. Schwartz, ScD, DeltaQuest Foundation, Inc., Concord, and Tufts University School of Medicine, Boston, MA, United States; Bryce Reeve, PhD, Department of Health Policy and Management, Gillings School of Global Public Health, and Cecil G. Sheps Center for Health Services Research, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Heather Gross, MEd, Cecil G. Sheps Center for Health Services Research, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Darren A. DeWalt, MD MPH, Cecil G. Sheps Center for Health Services Research, and Department of Medicine, School of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; I-Chan Huang, PhD, Department of Epidemiology and Cancer Control, St. Jude Children’s Research Hospital, Memphis, TN, United States BACKGROUND: To examine response shift (RS) associated with the change of asthma-related health states for the items measuring asthma, fatigue, pain, and mobility domains of the NIH Patient-Reported Outcomes Measurement Information System (PROMIS) for asthmatic children, after accounting for measurement bias and confounding variables. METHODS: Study sample comprised 238 asthmatic children age 8–17 years who have completed the NIH PROMIS Pediatric asthma, fatigue, pain, and mobility scales at baseline and anytime in a 3-month window whenever asthma status was changed. Oort’s structural equation modelling approach was used to assess the item-level recalibration and reprioritization RS associated with change in asthma control and global rating of change in breathing problems. RS analyses specifically adjusted for the influence of measurement bias and confounding variables including child’s age, race, gender, comorbidity, and body mass index. Measurement bias reflects the unequal ratings on the items within domains for different levels of asthma-related health states given the same underlying domain scores. Effect sizes (ES) (Cohen’s d-index) were estimated to evaluate the impact of RS by testing the change of item- and domainlevel scores with and without accounting for RS and measurement bias. RESULTS: The PROMIS asthma domain evidenced recalibration RS in one item with a marginal impact on the item score change (ES = 0.18). The PROMIS fatigue and mobility domains evidenced reprioritization RS in one item each, with marginal and moderate impact on the item score change (ES = 0.17 and ES = 0.43, respectively). The RS led to a negligible increase in mean latent scores of the asthma domain (ES = 0.03). Similarly, the RS led to a negligible decline in mean latent scores of the fatigue domain (ES = 0.004) and a negligible increase in mean latent scores of the mobility domain (ES = 0.02). No items in the pain domain demonstrated RS and no items in any PROMIS domains shown the measurement bias. CONCLUSION: Recalibration and reprioritization RSs were identified in few items of the NIH PROMIS Pediatric asthma, fatigue, and mobility domains. However, the impact of these RSs on the changes of item and domain scores was negligible.

Item-level response shift detection using structural equation modeling Mathilde G.E. Verdam, MSc, Academic Medical Centre and Child Development and Education; Frans J. Oort, PhD, Child Development and Education; Mirjam A.G. Sprangers, PhD, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands PURPOSE: The structural equation modeling (SEM) approach for detection of response shift (Oort, 2005) is especially suited for continuous data, e.g., questionnaire scales. The present objective is to explain how the SEM approach can be applied to discrete data, and to illustrate response shift detection in items measuring health-related

Qual Life Res (2015) 24:1–191 quality of life (HRQL) of cancer patients. METHODS: The SEM approach for discrete data includes two stages: (1) establishing a model of underlying continuous variables that represent the observed discrete variables, (2) using these underlying continuous variables to establish a measurement model for the detection of response shift, and to assess true change. The proposed SEM approach was illustrated with data of 485 cancer patients whose HRQL was measured with the SF-36, before and after start of antineoplastic treatment. RESULTS: Response shift effects were detected in items of the subscales mental health, physical functioning, role limitations due to physical health, and bodily pain. Recalibration response shifts indicated that patients experienced relatively fewer limitations with ‘‘bathing or dressing yourself’’ (effect size d = 0.51) and less ‘‘nervousness’’ (d = 0.30), but more ‘‘pain’’ (d = -0.23) and less ‘‘happiness’’ (d = -0.16) after antineoplastic treatment as compared to the other symptoms of the same subscale. Overall, patients’ mental health improved, while their physical health, vitality, and social functioning deteriorated. No change was found for the other subscales of the SF-36. CONCLUSIONS: The proposed SEM approach to discrete data enables response shift detection at the item level. This will lead to a better understanding of the response shift phenomena at the item-level and therefore enhances interpretation of change in the area of HRQL.

A Guttman errors based method for response shift detection at the individual and item levels Myriam Blanchin, PhD; Ve´ronique Se´bille, ScD; Alice Guilleux, MSc; Jean-Benoit Hardouin, ScD; EA 4275 SPHERE ‘‘bioStatistics, Pharmacoepidemiology and Human Sciences Research Team’’, Faculty of Pharmacy, University of Nantes, France AIMS: Several approaches have been proposed for response shift (RS) detection and adjustment in the appraisal of change of PRO over time at the group level. Nevertheless, we can suspect that among a sample, only some individuals might be affected by RS, and that RS of different types might affect different items to different extent. To further assess whether item parameters vary at an individual level and identify RS at item level, Guttman errors (GE) may be valuable as they enable the detection of discrepancies in respondent’s answers to items compared to an expected response pattern. This study explores the benefits of using a GE based method for RS detection at the individual and item levels regarding the estimation and interpretation of observed differences in health-related quality of life over time (HRQOL) in a clinical study. METHODS: The analysis was performed on the SatisQoL study, aiming at studying links between satisfaction of chronically ill patients and HRQOL. The expected response pattern was computed by ordering all the possible answer’s categories (from answered items) from the easiest (most prevalent) to the most difficult (least prevalent) based on patients’ responses observed at the first time of measurement (T1). The number of GE (unforeseen answers given this expected pattern) has been determined for each individual at each time of measurement (T1 and T2). Individuals with many GE were suspected to interpret the items differently from the whole sample, and assumed to present RS. Individuals were set in four different groups depending on the number of GE that they had at T1 or/and T2. The Response Shift Algorithm in Item response theory (ROSALI) was then applied on the whole sample and on each of the four groups separately. RESULTS: Different types of RS (non-uniform recalibration, reprioritization) were more prevalent in groups composed of patients assumed to present RS based on GE. On the opposite, no RS was observed on patients having few GE. CONCLUSIONS: Guttman errors and Item Response Theory models seem to be relevant tools to discriminate individuals affected by RS from the others at the item level.

11 Use of non parametric item response theory to detect item-level response shift Jean-Benoit Hardouin, ScD; Myriam Blanchin, PhD; Ve´ronique Se´bille, ScD, EA 4275 SPHERE ‘‘bioStatistics, Pharmacoepidemiology and Human Sciences Research Team’’, Faculty of Pharmacy, University of Nantes, France AIMS: Non parametric Item Response Theory (NPIRT) is a set of methods allowing analysing the psychometric properties of a scale, in particular the discrepancies in the responses to the different items. These methods are mainly applied in validation of scales. The main advantage of the NPIRT compared to Parametric Item Response Theory Models is to avoid the estimation of a large number of parameters and the specification of many assumptions (for example regarding the form of the measurement model). In particular, trace lines are very simple graphical tools in NPIRT. They represent the mean responses to each item as a function of a measure of the latent variable (generally the score obtained from the subscales). They can be considered as a non-parametric version of the Item Characteristic Curves in IRT. From a psychometric perspective, Response Shift (RS) can be viewed as a modification of the response process of patients to items over a period of time. In presence of RS, it is likely that the Traces Lines will be modified, for example a shift of a trace line on the right or on the left between the time occasions could be interpreted as a uniform recalibration, and a variation of the slope of the trace line could be interpreted as a non-uniform recalibration or as a reprioritisation. METHODS: We propose methods for analysing the traces lines of items measured at different time occasions. Modifications of the trace lines measured using quantitative indexes such as area under the curves or mean slope will be used to determine which items might be affected by RS. Simulated responses to item presenting or not RS will be analysed in order to propose a way to detect RS by using NPIRT. RESULTS: First results suggest that items presenting uniform or non-uniform recalibration could be successfully detected using the trace lines. The reprioritisation process at the item level is more difficult to detect using the trace lines, and others indexes of the NPIRT like Loevinger’s H coefficients could be used. CONCLUSIONS: Trace lines seem to provide useful tools to determine items presenting uniform or non uniform recalibration.

Oral Sessions 101: Quality of Life Constructs (101.1) Psychological profiles of attitudes about health: does it make a difference to the self-rating of health and life satisfaction? Yash J. Jalundhwala, MS, University of Illinois at Chicago, Chicago, IL, United States; A. Simon Pickard, PhD, University of Illinois at Chicago, Chicago, IL, United States AIMS: The objective of this study was to examine the association between profiles of health-related attitudes and motivations (HRAMs) and the self-assessment of health -related quality of life (HRQL) and life satisfaction. METHODS: A secondary analysis was conducted using cross-sectional data from a wellness survey in community residents of Kirklees, UK (n = 10,401). Responses from the Healthy Foundations Life-Segmentation Questionnaire were used to create HRAM groups (profiles)—health conscious realists (HCR), balanced compensators (BC), live-for todays (LFT), hedonistic immortals (HI) and unconfident fatalists (UF). The self-rating of HRQL (measured by the EQ-VAS) and well-being (measured by life-satisfaction) was compared across the HRAM groups using regression models to adjust for current health status as defined by the EQ-5D-5L. RESULTS: When categorized based on HRAM, the sample consisted of 30.8 %

123

12 HCR, 14.4 % BC, 18.2 % LFT, 10.4 % HI and 26.1 % UF respondents. Adjusting for age and gender, compared to respondents classified as HCR: (i) UF were more likely to have medical conditions (OR obesity 1.93, 95 % CI 1.67–2.24; OR mental illness 3.33, 95 % CI 2.90–3.84) and greater odds of taking addictive substances (OR drug abuse 1.88, 95 % CI 1.38–2.54; OR tobacco use 1.92, 95 % CI: 1.64–2.25); (ii) BC exercised more (OR physical exercise 1.31, 95 % CI 1.15–1.50); and (iii) LFT exercised less (OR physical exercise 0.79, 95 % CI 0.69–0.89). Compared to VAS scores of HCR, UF respondents were 8.82 points lower (p \ 0.001), LFT respondents were 1.43 points lower (p = 0.002), and BC respondents were 1.25 points higher (p \ 0.01). Compared to HCR, BC respondents were less likely report low life-satisfaction (OR 0.76, 95 % CI 0.63–0.91), while UF were more than three times as likely to report lower lifesatisfaction (OR 3.41, 95 % CI 2.83–4.10). CONCLUSIONS: Profiles based on health-related attitudes and motivations are associated with systematic differences in health, as well as the valuation of health and life satisfaction. For patient populations characterized by a predominant profile, these systematic differences in health valuation may have important implications for the management of patients when routine outcome measures that value health are part of the decision making process.

(101.2) A single question about a respondent’s perceived financial ability to pay monthly bills explains more variance in health utility scores than absolute income and assets questions Janel Hanmer, MD PhD, University of Pittsburgh, Pittsburgh, PA, United States; Dasha Cherepanov, PhD, Partnership for Health Analytic Research, Beverly Hills, CA, United States AIMS: Health utility scores are associated with socio-economic status (SES), such as assets and income. Since direct SES questions are sometimes considered intrusive, we explored if a general question about ability to meet monthly payments is associated with health utility scores. METHODS: We used data from the National Health Measurement Study-a US nationally representative telephone survey collected in 2005–2006. These data include four distinct health utility measures (EuroQol-5D [EQ-5D], Health Utilities Index Mark 3 [HUI3], Short Form-6D [SF-6D], and Quality of Well-being Index [QWB]), household income and assets, and the question: ‘‘How difficult is it for you to meet the monthly payments on your bills?’’ (Extremely difficult, very difficult, somewhat difficult, slightly difficult, or not difficult at all?). Each utility score was regressed on: income and assets (model 1); difficulty paying bills (model 2); income, assets, and difficulty paying bills (model 3). All models used survey weights and adjusted for age, sex, race, and education. RESULTS: We use 3666 respondents who report black or white race. As income and assets increase, difficulty paying bills decreases. This question had fewer missing values (n = 30) than income (n = 311) or assets (n = 373). Models with difficulty paying bills (model 2) explained more variance in health utility than models with absolute measures (model 1). Including all measures (model 3) had very modest increases in R-square. For the EQ-5D, R-squared values were 0.112 (model 1), 0.166 (model 2), and 0.175 (model 3). For the HUI3, R-squared values were 0.109 (model 1), 0.149 (model 2), and 0.161 (model 3). For the SF-6D, R-squared valued were 0.116 (model 1), 0.157 (model 2), and 0.171 (model 3). For the QWB, R-squared values were 0.107 (model 1), 0.159 (model 2), and 0.162 (model 3). CONCLUSIONS: Changes in R-squared suggest the perceived financial ability question explains more variance in health utility than income and assets. This single question may yield more information than the traditionally used income and assets questions, has less missing values than the traditional questions, and should be considered as an alternative SES measure in future studies.

123

Qual Life Res (2015) 24:1–191 (101.3) Comparison of QOL between personalized and ‘‘one size fits all’’ measures Nancy Mayo, PhD, McGill University, Montreal, Canada; Ayse Kuspinar, McGill University Health Centre - RI, Montreal, QC, Canada; Ala’ S. Arubub, McGill University Health Centre - RI, Montreal, QC, Canada; Carolina Moriello, McGill University Health Centre - RI, Montreal, QC, Canada; Susan C. Scott, McGill University Heath center, Montreal, QC, Canada AIMS: Personalized measures of quality of life (QOL) have been in existence for decades and included in 140 studies and yet their use has not penetrated the field as much as ‘‘one size fits all’’ measures which have people answer the same set of questions regardless of health, demographic, or social situation. The purpose of this study is to compare and contrast drivers of QOL using these two approaches to measurement across four health conditions: stroke (n = 253), multiple sclerosis (MS; n = 185), advanced cancer (n = 192), and HIV+ (n = 275). METHODS: The Patient Generated Index (PGI), EQ-5D, and SF-6D, have been in consistent use by our research team for several decades providing an opportunity to compare results across measures and conditions. Data on the 905 subjects came from two randomized trials (stroke), a cross-sectional study (MS) and two longitudinal studies (cancer and HIV+). Measures are scaled to range from 0 to 1, with 1 indicating the highest health-related QOL. RESULTS: Across conditions, the number of domains nominated using PGI ranged from 60 (MS) to [100 (stroke, cancer), yet EQ-5D and SF-6D together cover only 9 unique domains. Common domains were work, mobility, memory, and fatigue but there were more differences than similarities with speech, dexterity, and driving named for stroke, sleep and appetite for named for cancer, and health, worry, and stigma named for HIV. The total score on PGI ranged from 0.37 (cancer) to 0.5 (MS, HIV+) whereas for EQ-5D the range was from 0.66 (cancer) to 0.81 (HIV+) with less variability across the SF6D (0.65–0.68). The lower value on PGI makes it an ideal measure to use in randomized trials of interventions targeting QOL as there is clearly more room for improvement. The structure of the PGI also makes it an ideal method of examining the potential for response shift in QOL. CONCLUSIONS: These three measures all yield a single index to summarize QOL, an advantage over profile measures. But, the two generic measures overestimated the impact of these health conditions with respect to the PGI. Given the heterogeneity of domains contributing to QOL, it is clear that ‘‘one size does not fit all’’.

(101.4) How is subjective well-being related to quality of life? Do we need two concepts and measures? Suzanne M. Skevington, PhD, CPsychol, FBPsS, University of Manchester, Manchester, United Kingdom AIMS: We investigated how the concepts of subjective wellbeing (SWB) and quality of life (QoL) are related, and considered the need for separate measures. Thirty-three WHOQOL-SRPB QoL facets operationalized an SWB model extended by eudaimonia (SWB+). METHODS: WHOQOL-SRPB data was collected from 2543 adults in 11 countries. Multiple hierarchical linear regression analysis examined how much variance in overall general quality of life (including life satisfaction) was explained by positive and negative feelings (hedonia), and meaning and purpose in life (eudaimonia), with health as a covariate. Significant QoL facets outside the SWB+ model were identified. Repeat testing with equal split-half samples provided cross-validation. RESULTS: All three tests accounted for *70 % of the total variance. Two SWB+ components accounted for a total of *25 % in separate tests, where hedonic elements of positive (25 %) and negative feelings (0.2 %) combined. Positive feelings (7.7 %) and purpose in life (0.1 %) but not meaning

Qual Life Res (2015) 24:1–191 in life, weakly joined hedonia with eudaimonia. Health influence was removed, accounting for *25 % of the total. Up to 13 additional facets were significant increasing the total explained to *68 %. Cross-validation prioritised seven of these: personal relations, financial resources, energy, health/social care, recreation, self-esteem and wholeness. CONCLUSIONS: SWB+ is nested within highly differentiated QoL concept. Happiness is the mainstay of SWB+ but explains only one-third overall, with negative feelings or purpose in life. Thirteen dimensions containing significant QoL facets, health and SWB+ components represent a novel subjective model of life quality and wellbeing (LQW), which could be evaluated by the WHOQOL-SRPB-BREF.

(101.5) Is there a relationship between spiritual wellbeing (SWB) and QOL? Findings from international validation of the EORTC QLQ-SWB32 Bella Vivat, University College London, London, United Kingdom; Teresa Young, Lynda Jackson Macmillan Centre, Mount Vernon Cancer Centre, Northwood, Middlesex, United Kingdom; Julie Winstanley, Patricia Ritchie Centre for Cancer Care and Research, Sydney, Australia; Juan I. Arraras, Oncology Departments, Complejo Hospitalario de Navarra, Pamplona, Spain; Kath Black, St. Gemma’s Hospice, Leeds, United Kingdom; Anne Bredart, Psychiatry and Psychooncology Unit, Institut Curie, Paris, France; Anna Costantini, Psychoncology Unit, Department of Oncological Sciences, Rome, Italy; Sheila Fisher, Leeds Institute of Cancer and Pathology, University of Leeds, Leeds, United Kingdom; Jingbo Guo, Palliative Ward, Shengjing Hospital of China Medical University, Liaoning, China; M. Elisa Irarrazaval, Centro Los Juncos, Santiago, Chile; Kunihiko Kobayashi, Department of Respiratory Medicine, Saitama International Medical Centre, Hidaka City, Japan; Renske Kruizinga, Medical Oncology, Academic Medical Centre, Amsterdam, Netherlands; Mariana Navarro, Instituto Nacional de Cancerologia, Mexico City, Mexico; Sepideh Omidvari, Health Metrics Research Centre, Iranian Institute for Health Science Research, ACECR & Cancer Research Centre, Cancer Institute of Iran, TUMS, Tehran, Iran; Gudrun Rohde, Faculty of Health and Sport Sciences/ Department of Clinical Research, University of Agder and Sorlandet Hospital, Kristiansand, Norway; Samantha Serpentini, Psychoncology Unit, IOV-IRCCS, Padua & Psychoncology Service, Azienda Ulss 3, Bassano del Grappa, Italy; Nigel Spry, Cancer Centre Radiation Oncology, Sir Charles Gairdner Hospital, Nedlands, Australia; Hanneke W. van Laarhoven, Medical Oncology, Academic Medical Centre, Amsterdam, Netherlands; Grace Yang, National Cancer Centre, Singapore, Singapore AIMS: The EORTC Quality of Life Group has developed a spiritual wellbeing (SWB) measure for palliative cancer patients; the EORTC QLQ-SWB32. Data collection for measure validation was completed in 2013, and analysis is ongoing. One element of the analysis explored relationships between participants’ responses to the provisional 36-item SWB measure and to the EORTC QLQ-PAL-C15 (‘‘PAL’’)—a validated version of the EORTC QLQ-C30 for palliative care. METHODS: Study documents were translated as needed. Participants completed the provisional SWB measure and PAL, then participated in debriefing interviews. We analysed response data in two groups: all participants (‘‘A’’) and a subset of those with a religious faith: ‘‘believers’’ (‘‘B’’). We conducted PCA analysis with Oblimin rotation, plus Rasch analysis, on 35 of the 36 SWB items (the 36th was a global SWB score), and identified four scales. We examined relationships between responses to these scales; the global SWB score; and PAL scales/items. RESULTS: Participants—451 in all; 368 in subset B—came from 14 countries: Australia, Austria, Chile, China, France, Iran, Italy, Japan, Mexico, Netherlands,

13 Table 1 Correlations between PAL global QOL and EF scale and SWB global item and scales for all participants A: All participants

SWB RSG global

RO

RS

EX

PAL correlation coefficient .200** .027

.154** .274** .499**

QOL Sig. (2-tailed)

.001

.000

.584

.000

.000

Global N 447 PAL correlation coefficient .100*

427 432 444 439 -.075 .128** .435** .406**

EF Sig. (2-tailed)

.034

.120

.008

.000

.000

N

438

426

431

443

438

Table 2 Correlations between PAL global QOL and EF scale and SWB global item and scales for ‘‘believers’’ B: ‘‘Believers’’

SWB global

RSG + RG

RO

RS

EX

PAL correlation coefficient

.210**

.040

.188**

.280**

.528**

QOL Sig. (2-tailed)

.000

.471

.001

.000

.000

Global N

345

321

336

342

340

PAL correlation coefficient

.114*

-.053

.145**

.435**

.443**

EF Sig. (2-tailed)

.035

.342

.008

.000

.000

N

344

320

335

341

339

Norway, Singapore, Spain and UK. The four scales identified were Relationship with Self (RS) (5 items); Relationships with Others (RO) (6 items); Existential (EX) (5 items); and Relationship with Something Greater (RSG) [5 items for all participants, plus one single-item scale for ‘‘believers’’: Relationship with God (RG)]. A further ten non-scoring items were also retained for the final 32-item tool. The global SWB item only very weakly correlated with the PAL Emotional Functioning (EF) scale and the PAL global QOL item. The RSG scoring scale did not correlate with either, and the RO scale only weakly. The RS scale moderately correlated with the PAL EF scale (Spearman’s Rho coefficient for both A and B = .435**), and only weakly with the PAL global QOL. The EX scale moderately correlated with both the PAL global QOL (A .499**; B .528**), and PAL EF scale (A .406; B .443**). **p B 0.001 CONCLUSIONS: This new EORTC measure of SWB extends QOL beyond that identified by the PAL, tapping a distinct construct for palliative care patients both with and without religious faith (Tables 1, 2).

(101.6) Appraisal assessment in patient-reported outcome research: methods for uncovering the personal context and meaning of quality of life Carolyn Schwartz, ScD, Tufts University Medical School; DeltaQuest Foundation, Inc., Oslo and Akershus University College, Concord, MA, United States; Bruce D. Rapkin, PhD, Albert Einstein College of Medicine, Bronx, NY, United States AIMS: Patient-reported outcomes have become increasingly prominent in clinical research. These increasingly sophisticated tools are developed in a zeitgeist of measurement minimalism, seeking streamlined approaches to data collection that minimize respondent burden and maximize the measurement precision. Consequently,

123

14 whether generic or disease-specific, these sophisticated items and scales convey little information about what a particular individual’s rating actually means. Individual differences in appraisal parameters are important and relevant to quality of life (QOL) research. This presentation highlights the advantages of integrating appraisal assessment into clinical research. METHODS: The most comprehensive method for assessing appraisal is the QOL Appraisal Profile (QOLAP). This measure includes open-ended (i.e., qualitative) and multiple-choice questions to assess four appraisal parameters: frame of reference, sampling of comparison, standards of comparison, and combinatory algorithm. We illustrate with empirical findings three classes of investigation that would benefit from appraisal assessment: Methodological, Interpretation of Change, and the Back-Story of Resilience. RESULTS: A methodological investigation of HIV/AIDS patients revealed a range of cognitive schemas induced by a commonly-used response shift detection method, only 15 % of which reflected the presumed process invoked. Two studies on interpreting change in multiple sclerosis and spine surgery patients revealed that the Minimally Important Difference metric of clinical success changes as a function of symptom experience. Two studies of reserve in multiple sclerosis patients revealed distinct patterns of appraisal among more resilient patients as compared to less resilient patients, reflecting differences in personality and other antecedent characteristics. One study of underserved cancer patients utilizes the QOLAP as a clinical tool to understand patients’ values and goals, and to guide treatment decision-making in the face of uncertainty. CONCLUSIONS: Integrating appraisal assessment can provide a more textured, person-centered understanding of how demographic, clinical, and psychosocial person factors come together in individual lives. The QOLAP can elucidate methodological strengths and weaknesses, interpretation of change over time, and an understanding and promotion of resilience despite chronic health conditions. While its scoring can be time-intensive, the QOLAP’s payoff is a level of narrative that is relevant and missing in more minimalist approaches to QOL assessment.

102: Quality of Life in Neurological Conditions (102.1) Exploring the relationship between patient-reported improvement in pain, function and satisfaction 12 months after spine surgery Sara Khor, MSc, University of Washington, Seattle, WA, United States; Annie Pugel, MD, University of Washington, Seattle, WA, United States, Danielle C. Lavallee, PharmD, PhD, University of Washington, Seattle, WA, United States; David R. Flum, MD, MPH, University of Washington, Seattle, WA, United States AIMS: To examine the interaction of three patient-centered outcomes domains: pain improvement, functional improvement, and patient satisfaction 12 months following spine surgery. METHODS: Data derived from medical records for spine surgery at 20 hospitals participating in the Surgical Care and Outcomes Assessment Program Spine Program (2013–2014) were merged with patient-reported outcomes (PROs) and satisfaction with spine surgery obtained through self-administered surveys collected through the Comparative Effectiveness Research Translation Network at baseline and 12 months. PROs captured include the Numeric Rating Scale for Pain (NRS) and the Oswestry Disability Index (ODI) or Neck Disability Index (NDI) for lumbar and cervical procedures respectively. We preformed a multivariate logistic regression analysis to determine if improvement in disability [defined as a 15-point improvement on a 100 point scale (ODI/NDI)] and pain (defined as a 2 point improvement on the 10-point scale) were independently associated with patient

123

Qual Life Res (2015) 24:1–191 satisfaction at 12 months. RESULTS: 800 patients (47 % male, mean age 62, 73 % lumbar procedure) who underwent spine surgery had pain and function data at baseline and completed survey questions in all three domains at 12 months. At 12 months, 66 % had improved pain scores, 54 % improved in function scores and 45 % improved in both pain and function scores. 96 % of those with improved pain and function reported being satisfied, while 58 % of the patients who did not improve in either pain or function reported being satisfied. Regression analysis showed that improvement in pain and function were associated with satisfaction, with odds ratio 1.22 [CI 1.16–1.29] for pain and 1.04 [CI 1.03–1.06] for functional improvement at 12 months. CONCLUSIONS: Improvements in pain and function are associated with a greater likelihood of being satisfied with surgery at 1 year, but nearly 6 out of 10 patients report being satisfied even when these metrics do not improve. Satisfaction with spine surgery appears to be a unique domain not well explained by other PRO changes. Identifying the relationship between PROs and the factors that improve long-term satisfaction is critical to delivering more value added care.

(102.2) Patient-reported and performance-based outcomes in persons after mild traumatic brain Nicole von Steinbuechel, University Medical Center Goettingen, Goettingen, Germany; Annette Groeneveld, University Medical Center Goettingen, Goettingen, Germany; Holger Schmidt, Elbekliniken Stade, Stade, Germany; Klaus von Wild, KvwNeuroscience Consulting GmbH, Muenster, Germany; Amra Covic, University Medical Center Goettingen, Goettingen, Germany; Joy Backhaus, University Medical Center Goettingen, Goettingen, Germany AIMS: Little is known with respect to generic health-related Quality of Life (HRQOL) of individuals after mild traumatic brain injury (mTBI) and its association with cognitive functioning. Generally, it is still assumed that persons after mTBI recover completely 3 months post traumatic injury. No investigations have yet been conducted concerning disease-specific HRQOL and the relation to possible additional medical symptoms, as well as cognitive and emotional deficits after mTBI. Thus, the investigation of the association of PROs (Patient-Reported Outcomes) and PERBOs (PERformance-Based Outcomes) is here of special interest. METHODS: This study compared 60 individuals after mTBI with 30 matched brain healthy controls in a cross-sectional retrospective design. The application of strict inclusion, exclusion and matching criteria resulted in equal groups concerning demographic characteristics and variables such as lack of CT abnormalities, depression, anxiety or alcohol abuse. RESULTS: We found that even though at least 3 months have passed since the mTBI diagnosis, all patients still suffered from post concussive symptoms. Interestingly, for all 12 administered PERBOS (neuropsychological tests) only two significant group differences were found. This seems to imply that the patients either did not develop any noticeable cognitive deficits or that the tests were not sensitive enough to detect them. However, as for the PROs HRQOL and psychosocial instruments proved to be very sensitive in identifying impairments related to mTBI as well as the RPQ, the BSI and the PCL. Analyses revealed significant differences for the SF-36 and for the disease-specific QOLIBRI (Quality Of LIfe after Traumatic BRain Injury). Assessment of the cognitive dimension of HRQOL in detail with the COGQOL resulted in significant differences for all subscales. CONCLUSIONS: Multidimensional correlation-sensitive analyses underlined the fact that mTBI patients are a heterogeneous population, which can suffer from a variety of long-term adverse consequences on HRQOL and psychosocial functioning.

Qual Life Res (2015) 24:1–191 (102.3) ‘‘When I saw walking, I just kind of took it as wheeling.’’ A qualitative exploration of how individuals with spinal cord injury perceive mobility-related items in generic health-related quality of life (HRQoL) instruments Yvonne Anne Michel, Dipl Psych, University of Oslo, Oslo, Norway; Lidia Engel, MSc, Simon Fraser University, Vancouver, BC, Canada; Liv Ariane Augestad, MD, PhD, Univeristy of Oslo; Akershus University Hospital, Oslo, Norway; Kim Rand-Hendriksen, Cand.psychol, PhD, Akershus University Hospital; University of Oslo, Oslo, Norway; David G. Whitehurst, PhD, Simon Fraser University, Vancouver, BC, Canada AIMS: Mobility-related items are included in all generic HRQoL instruments. However, the way in which respondents are asked to describe their physical functioning along these items differs substantially. Recent research has shown that individuals with spinal cord injury (SCI) have different overall perceptions towards the descriptive systems of HRQoL instruments. This study is an in-depth analysis of the perceptions and response strategies of individuals with SCI to the mobility-related items of generic HRQoL instruments. The objective is twofold: firstly, to compare the inherent features of mobility-related items of six HRQoL measures and, secondly, to explore how these items are perceived by individuals living with SCI. METHODS: First, we systematically compared mobility-related items of EQ-5D-5L, SF36v2, HUI, QWB-SA, 15D, and AQoL-8D with regard to response options, position in the questionnaire, time horizon, wording, mention of mobility aids, and the use of further instructions. Second, we performed content analysis using focus group transcripts (n = 15 individuals with SCI) from a 2013 study that explored the face validity and feasibility of the six HRQoL instruments. RESULTS: The descriptive comparison of mobility-related items showed fundamental differences in terms of wording and indication of the use of mobility aids and appliances. The same features were highlighted by participants in the focus groups. Some participants stated a preference for questions that used ‘getting around’ over questions that equated mobility to ‘walking’. Interestingly, some individuals reframed mobility items by replacing the word ‘walking’ by ‘wheeling’ to make them more accessible. Explicit reference to walking equipment, such as the use of wheelchair, was mentioned as being important. It enabled some of the participants to identify a response option that appropriately reflected their health status. CONCLUSIONS: This study showed that individuals with SCI interpreted the wording of the mobility items differently, across and even within questionnaires. The use of reframing, like substituting ‘wheeling’ for ‘walking’, can introduce construct-irrelevant variance, which effects cannot easily be disentangled. This poses a challenge to interpret responses to mobility-related items in HRQoL instruments. The appropriateness of different approaches to instruct respondents is an area for further consideration.

15 Pitie-Salpetriere University Hospital, Paris, France; Luc Taillandier, MD, Department of Neurology, University Hospital, Nancy, France; Didier Frappaz, MD, Department of Oncology, Leon Berard Centre for Fight against Cancer, Lyon, France; Herve´ Taillia, Department of Neurology, HIA Val de Grace, Paris, France; Roland Schott, Department of Oncology, Paul Strauss Centre for Fight against Cancer, Strasbourg, France; Michel Fabbro, Department of Oncology, Val d’Aurelle Center for Fight against Cancer, Montpellier, France; Isabelle Tennevet, MD, Department of Oncology, Henri Becquerel Center for Fight against Cancer, Rouen, France; Francois Ghiringhelli, Prof., Department of Oncology, GF Leclerc Center for Fight against Cancer, Dijon, France; Ce´cile Dalban, Biostatistics and Quality of life unit, Centre Georges Franc¸ois Leclerc, Dijon, France; Jerome Skrzypski, Biostatistics and Quality of life unit, Centre Georges Franc¸ois Leclerc, Dijon, France; Olivier Chinot, Prof., Department of Neuro-Oncology, University Hospital La Timone, Marseille, France; Bruno Chauffert, Prof., Department of Medical Oncology, University Hospital, EA 4666, Amiens, France; Franck Bonnetain, Prof., Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸onQuality of Life in Oncology National Platform, Besanc¸on, France AIMS: Two clinical trials investigating the addition of bevacizumab (BEV) to TMZ treatment (Chinot et al., NEJM 2014; Gilbert et al., NEJM 2014) for unresectable glioblastoma (GB) highlighted an improvement of progression-free survival (PFS) but no effect on overall survival. Health-related quality of life (HRQoL) results were divergent and compromised the conclusion about clinical benefit of bevacizumab. A phase II clinical trial evaluated BEV and irinotecan (IRI) in addition to TMZ-based chemoradiation for unresectable GB (Chauffert et al., Annals of Oncology 2014) was conducted. The primary objective 6-month PFS rate from 50 to 66 % was not reached. This present study focuses on the HRQoL analysis (secondary endpoint). METHODS: EORTC QLQ-C30 and its BN20 brain cancer module were used at baseline, during treatment and every 4 weeks until progression. Time until definitive HRQoL score deterioration (TUDD) was used as a modality of longitudinal HRQoL analysis, with a 10-point Minimal Clinically Important Difference, including or not death as an event. TUDD was estimated with the Kaplan–Meier method. Cox model was used to estimate Hazard Ratios (HR) and its 95 % confidence interval (CI). Multivariate Cox model investigated factors associated with TUDD. RESULTS: Among the 120 patients included from 2009 to 2011, 90 patients (75 %) filled the baseline HRQoL questionnaires (45 in each treatment arm). For TUDD, patients in BEV/IRI arm presented a longer TUDD than those of TMZ arm for 9/15 dimensions of the QLQ-C30 and 10/11 dimensions of the

(102.4) Impact of bevacizumab added to temozolomidechemoradiation on time to health-related quality of life deterioration in unresectable glioblastoma: results of a phase II randomized clinical trial Ame´lie Anota, PhD, Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸on, Besanc¸on, France; Tienhan Sandrine Dabakuyo, PhD, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France; Astrid Pozet, Methodology and Quality of Life in Oncology Unit (EA3181), University Hospital of Besancon, Besanc¸on, France; Brice Paquette, MD, Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸on; Department of Digestive Surgery and Liver Transplantation, University Hospital of Besanc¸on, Besanc¸on, France; Loic Feuvret, MD, Department of Radiotherapy,

Fig. 1 .

123

16 BN20. As an example, patients in BEV/IRI arm presented a longer TUDD of cognitive functioning than those of TMZ arm [median 3.8 months (95 % CI 2.5–NA) for BEV/IRI vs. 5.0 (95 % CI 3.0–NA) for TMZ, HR = 0.80 (0.41–1.57)]. CONCLUSIONS: Patients in BEV/IRI arm presented a longer TUDD than those of the TMZ arm. These HRQoL results are consistent to those obtained with AVAGLIO trial suggesting HRQoL benefit for patients with BEV (Fig. 1).

(102.5) Factors influencing health status and quality of life in Canadians living with a traumatic spinal cord Joel Finkelstein, MD, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada; Vanessa K. Noonan, PT, PhD, Rick Hansen Institute, University of British Columbia, Vancouver, British Columbia, Canada; Nader Fallah, PhD, Rick Hansen Institute, University of British Columbia, Vancouver, British Columbia, Canada; Chester Ho, MD, University of Calgary, Calgary, AB, Canada; Eve C. Tsai, MD, PhD, The Ottawa Hospital, Ottawa Hospital Research Unit, University of Ottawa, Ottawa, Ontario, Canada; Carly S. Rivers, PhD, Rick Hansen Institute, Vancouver, British Columbia, Canada; Catherine Truchon, PhD, MScAdm, Institut National d’Excellence en Sante´ et Services Sociaux (INESSS), Quebec City, Quebec, Canada; A. G. Linassi, MB, FRCPC, University of Saskatchewan, Saskatoon, Saskatchewan, Canada; Colleen O’Connell, MD, FRCPC, Stan Cassidy Centre for Rehabilitation, Dalhousie Faculty of Medicine, Fredericton, New Brunswick, Canada; Andrea Townson, MD, University of British Columbia, Vancouver, British Columbia, Canada; Henry Ahn, MD, PhD, FRCSC, St. Michael’s Hospital, University of Toronto, Toronto, Ontario, Canada; Brian Drew, MD, Hamilton General, McMaster University, Hamilton, Ontario, Canada; Marcel Dvorak, MD, FRCSC, University of British Columbia, Vancouver, British Columbia, Canada; Michael G. Fehlings, MD, PhD, FRCSC, Toronto Western Hospital, University Health Network, Toronto, Ontario, Canada; Carolyn Schwartz, ScD, Tufts University Medical School; DeltaQuest Foundation, Inc., Oslo and Akershus University College, Concord, MA, United States; Luc Noreau, PhD, Universite´ Laval, Centre for Interdisciplinary Research in Rehabilitation and Social Integration, Quebec City, Quebec, Canada AIMS: To analyze relationships among injury and demographic variables, health status, and life satisfaction in persons who have sustained a traumatic spinal cord injury (tSCI). METHODS: Participants from the Rick Hansen Spinal Cord Injury Registry (RHSCIR), prospectively recruited from 2004 to 2014 from 31 Canadian acute and rehabilitation specialized spine centres were studied. A path analysis was used to determine the relationships among tSCI severity [ASIA Impairment Score (AIS), AIS A (complete motor/sensory impairment to AIS D (incomplete motor/sensory impairment] and anatomical level (high cervical, low cervical, thoracic, lumbrosacral), participant age, education, number of complications, 1 year postdischarge physical (PCS) and mental (MCS) component scores of the SF-36v2, and Life Satisfaction-11 score (mean of LISAT-11 items). The path analysis was conducted using Mplus version 7.1 and model fit was assessed using a Chi square test, Comparative Fit Index (CFI), Root Mean Square Error (RMSE) and Standardized Root Mean Square Residual (SRMSR). RESULTS: The cohort included 599 participants: 79.5 % were male; mean age was 44.8 (±18.2); 33.2 % were motor/sensory complete (AIS A) and 66.8 % were incomplete (AIS B-D). Good fit was supported by: non-significant Chi square test (p = 0.11), CFI (=0.99) [ 0.95, RMSEA (=0.04) \ 0.05 and SRMSR (=0.02) \ 0.08. Higher age, less education, more severe injuries (AIS A–C), and more complications negatively correlated with PCS. Less education and more complications negatively correlated with MCS, however the most severe injury (AIS A) was positively associated with MCS. Having a motor complete or cervical

123

Qual Life Res (2015) 24:1–191 injury negatively affected LISAT, but a higher PCS and MCS was associated with a higher LISAT. CONCLUSIONS: Individuals who are older, less educated, experience more complications, and have a more severe tSCI report lower physical health at 1 year post-discharge. Having less education and more complications negatively affected self-reported mental health, however individuals sustaining a severe injury (AIS A) have better mental health. Being completely paralyzed may facilitate a response shift, while those with incomplete injuries may experience frustration at achieving less recovery than expected. Future research should explore response shift within-person assessments. Results from this study also suggest that initiatives targeting complications could positively improve health status and ultimately impact quality of life.

103: Quality of Life Across Health Conditions (103.1) Body image after bladder removal: are surgical reconstructions delivering intended patient-reported outcomes? Scott M. Gilbert, MD, Moffitt Cancer Center, Tampa, FL, United States; Joanna E. Bulkley, Kaiser Permanente Northwest, Portland, OR, United States; Michael C. Leo, Kaiser Permantente Northwest, Portland, OR, United States; Kim N. Danforth, Kaiser Permanente Southern California, Pasadena, CA, United States; Marilyn L. Kwan, Kaiser Permanente Northern California, Oakland, CA, United States; Robert S. Krouse, University of Arizona, Tucson, AZ, United States; Marcia Grant, City of Hope National Medical Center, Duarte, CA, United States; Chris Wendel, University of Arizona, Tucson, AZ, United States; James V. Davis, Kaiser Permanente Northwest, Portland, OR, United States; Maureen O’Keeffe-Rosetti, Kaiser Permanente Northwest, Portland, OR, United States; Mark C. Hornbrook, Kaiser Permanente Northwest, Portland, OR, United States; Carmit K. McMullen, Kaiser Permanente Northwest, Portland, OR, United States AIMS: Men and women with bladder cancer who are treated with cystectomy (bladder removal) face a host of challenges after surgery. Roughly 80 % must adjust to an altered body due to urinary diversion with urostomy (stoma). Alternative surgical reconstructions (neobladders) that allow for voiding through the urethra are thought to improve body image, but studies have not clearly supported this assumption. Neobladders can result in urinary incontinence and other complications that lower health-related quality of life. Therefore, patient-reported outcomes about body image are needed to ensure that neobladder surgery, which is more complex than urostomy, is likely to result in its expected comparative benefits. We describe the role of urinary diversion type on body image after cystectomy for bladder cancer with the goal of building evidence to support surgical decisionmaking. METHODS: In an ongoing mixed-methods study, members of Kaiser Permanente who had been treated with cystectomy receive mailed surveys that include validated QOL measures 6 months postsurgery (response rate = 66 %). The survey includes the Body Image Scale (BIS), a reliable, validated 10-item scale developed for cancer patients that measures patient-reported impact of cancer and treatment on body image. Scores range from 0 to 30. Higher scores correspond to lower satisfaction with body image. We have received surveys from 118 urostomy patients and 35 neobladder patients to date. We report mean BIS scores for patients by diversion type, and because gender could affect BIS scores, we explore sub-group differences by gender. RESULTS: The mean age of respondents was 71 years. The mean BIS score was 7.0. Respondents with urostomy had worse body image than those with neobladder (7.3 [SD = 6.9] vs. 5.8 [SD = 5.3]). Women with urostomy had worse body image than men did (8.5 [SD = 7.3] vs. 6.9 [SD = 6.8]), whereas women with neobladder had better body image than men did (1.7 [SD = 2.1] vs. 6.2 [SD = 5.4]). CONCLUSIONS: Our preliminary analysis showed

Qual Life Res (2015) 24:1–191 that bladder cancer patients with neobladders had slightly better body image than did patients with urostomy. Our results are consistent with previous reports. We are continuing to explore differences in body image by gender and urinary diversion type.

(103.2) Patient-relevant treatment goals in psoriasis and their association with quality of life Christine Blome, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Ramona Gosau, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Marc Radtke, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Kristian Reich, Dermatologikum Hamburg, Hamburg, Germany; Stephan J. Rustenbach, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Christina Spehr, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Diamant Thaci, University of Lu¨beck, Lu¨beck, Germany; Matthias Augustin, University Medical Center Hamburg-Eppendorf, Hamburg, Germany AIMS: Patient-oriented care requires that therapeutic decisions agree with the patient’s treatment needs and goals. This study addressed three questions: What is important to patients starting systemic treatment for psoriasis? How stable are these preferences within the first year of treatment? Can treatment goals be derived from quality of life impairment, or is an additional goal assessment reasonable? METHODS: Treatment goal importance was assessed in patients with moderate to severe psoriasis in the German psoriasis registry PsoBest at baseline (onset of a systemic treatment, n = 3066) and at 1-year follow-up (n = 1444), using the Patient Benefit Index (PBI). The association of goal importance and quality of life was assessed in a second study. 55 patients in outpatient psoriasis care completed the PBI and a quality of life questionnaire corresponding in content to the PBI treatment goal items, both at baseline and after 5 months. RESULTS: Patients with moderate to severe psoriasis pursued a wide range of different goals beyond skin clearance, particularly concerning the reduction of itching (83.9 % ‘‘quite’’ or ‘‘very’’ important), burning (70.6 %), and pain (60.6 %), a normal everyday life (78.4 %), and low treatment burden (64.2–77.9 %). In a substantial proportion of patients (30.3–54.7 %), goals changed within 1 year after onset of systemic treatment, but change in quality of life impairment was mostly (17 in 23 goals) not associated with a corresponding change in treatment goal importance. CONCLUSIONS: Treatment goal importance is informative beyond health-related quality of life and should be assessed in clinical practice on a regular basis.

(103.3) Health-related quality of life (HRQoL) and patient experience in advanced melanoma: 6-month results from the image study Susan D. Mathias, MPH, Health Outcomes Solutions, Winter Park, FL, United States; Srividya Kotapati, Bristol-Myers Squibb, Wallingford, CT, United States; T. Kim Le, Bristol-Myers Squibb, Hopewell, NJ, United States; Javier Sabater, Bristol-Myers Squibb, Madrid, Spain; Amy P. Abernethy, Duke University, Durham, NC, United States AIMS: To present patient-reported HRQoL data for patients with advanced melanoma who received ipilimumab (IPI) versus another approved treatment (non-IPI) after first-line therapy. METHODS: This multinational, prospective, observational cohort study (IMAGE) enrolled patients with unresectable or advanced melanoma receiving either IPI or non-IPI post first-line treatment. Patients completed the EORTC-QLQ-C30, EQ-5D, and Work Productivity and Activity Impairment Questionnaire (WPAI) at enrollment (baseline) and

17 quarterly for 1 year until study withdrawal or death. Adjusted mean change scores from baseline to Month 6 were compared by treatment group. ANCOVA regression was used to adjust mean scores for baseline score, demographics (e.g., age, gender, and race) and clinical characteristics (e.g., time since diagnosis, disease stage, and ECOG). Data collection is ongoing; results represent data from 261 versus 275 days of follow up for the IPI versus non-IPI treatment group, respectively. RESULTS: Compared with the non-IPI group (n = 59), the IPI group (n = 230) had fewer males (58 vs 63 %), slightly greater age (61.4 vs 60.3 years), similar time since diagnosis (58.2 vs 58.8 months), and more patients with ECOG performance status of 0 or 1 (81 vs 63 %). Disease stage at initial diagnosis was the same (17 % stage IV). At Month 6, ANCOVA-adjusted change scores from Table 1 Adjusted mean changes from baseline to Month 6 for EORTC-QLQ-C30, EQ-5D, and WPAI Mean ± SE (n)

IPI (n = 196)

Non-IPI (n = 100)

EORTC functioninga Global health -11.9 ± 5.2 (n = 101) status

-13.7 ± 6.1 (n = 28)

Physical functioning

-15.9 ± 5.7 (n = 101)

-22.8 ± 6.8 (n = 28)

Role functioning

-19.7 ± 7.5 (n = 99)

-24.9 ± 8.9 (n = 28)

Emotional functioning

-6.5 ± 5.3 (n = 101)

-10.2 ± 6.2 (n = 29)

Cognitive functioning

-14.7 ± 5.5 (n = 101)

-16.6 ± 6.5 (n = 29)

Social functioning

-18.1 ± 7.0 (n = 101)

-27.4 ± 8.2 (n = 29)

EORTC symptomsb Fatigue 15.2 ± 6.2d (n = 101)

29.5 ± 7.6 (n = 27)

-0.7 ± 4.1e (n = 101)

11.5 ± 4.8 (n = 28)

7.4 ± 6.4 (n = 103)

19.3 ± 7.5 (n = 29)

Nausea and vomiting Pain Dyspnea Insomnia

8.4 ± 6.2 (n = 101)

14.6 ± 7.4 (n = 28)

12.1 ± 7.3 (n = 98)

13.4 ± 8.3 (n = 28)

Appetite loss

15.0 ± 7.7 (n = 101)

23.5 ± 9.1 (n = 28)

Constipation

-1.7 ± 5.9 (n = 100)

3.0 ± 7.0 (n = 28)

Diarrhea

3.1 ± 5.5 (n = 101)

1.8 ± 6.6 (n = 29)

Financial difficulties

21.6 ± 5.8 (n = 101)

21.5 ± 6.8 (n = 27)

EQ-5 Da EQ-5D index -0.24 ± 0.08 (n = 92)

-0.32 ± 0.10 (n = 25)

EQ-5D VAS

-12.6 ± 5.9 (n = 21)

-6.0 ± 4.6 (n = 98)

WPAIb,c Activity impairment

11.5 ± 6.2d (n = 88)

26.9 ± 7.8 (n = 22)

a

+ mean change scores represent improvement; - mean change scores represent worsening

b

+ mean change scores represent worsening; - mean change scores represent improvement

c

Other WPAI scores were not calculated because only a small number of patients were working for pay

d e

P \ 0.05 versus non-IPI treatment P \ 0.005 versus non-IPI treatment

123

18 baseline for EORTC, EQ-5D, and WPAI worsened for IPI and nonIPI treated patients (Table 1). CONCLUSIONS: Patients with advanced melanoma experience HRQoL deteriorations during the first 6 months of second-line or subsequent-line treatment. However, patients receiving IPI showed a less pronounced deterioration in all sub-scale/item scores compared with non-IPI patients, with a pattern and statistically significant findings that suggest worsening functioning and symptoms in patients who do not receive IPI. This profile is consistent with worse personal experience due to progressive disease in the non-IPI group, and better experience when disease is controlled by effective treatments such as IPI. Of note is that patients receiving IPI are more likely to be of better performance status at the time of study entry; analyses were adjusted for these differences. Longer follow-up with larger sample sizes will demonstrate if results persist.

(103.4) Psychometric evaluation of emotional well-being in advanced basal cell carcinoma Karen Bartley, PhD, MPH, Genentech, San Francisco, CA, United States; Jonathan Gable, MPA, Clinical Outcomes Solutions, Tucson, AZ, United States; Stacie Hudgens, MA (ABD), Clinical Outcomes Solutions, Tucson, AZ, United States AIMS: Advanced basal cell carcinoma (aBCC) may influence patient quality of life (QoL) due to the aesthetic implications of the disease and its related, visible symptoms. This is the first critical psychometric assessment to identify and evaluate the emotional impact on aBCC patients’ lives. The Skindex-16 is a three-domain patient reported quality of life outcome measure developed for use in dermatologic indications. This study sought to evaluate the psychometric properties of the 7-item emotional domain in patients with advanced BCC (locally advanced and/or metastatic). METHODS: Data from the interim analysis of 501 patients in a single-arm, open-label, phase 2 multicenter clinical study of the Hedgehog inhibitor vismodegib in patients with locally advanced or metastatic BCC (STEVIE, ClinicalTrials.gov identifier NCT01367665) were utilized for this psychometric evaluation. Patients were administered vismodegib (150 mg/day) until disease progression or unacceptable toxicity. Psychometric evaluation included item level characteristics and performance (e.g., completion rate, floor and ceiling effects, Rasch analysis), reliability and anchor analyses. Anchor analyses were conducted statistically evaluating clinical response levels by mean change in the emotional domain scores using analysis of variance (p \ 0.05) and standardized effect size (greater than 0.80 implies large effect). Efficacy, in the form of investigator-assessed best overall response, was assessed using RECIST v1.1. RESULTS: 316 of 501 patients completed the Skindex-16 emotional domain items at baseline (97.5–99.1 % completion rate range). Floor effects were higher than expected in this patient population (range 11.4–31.3 %). Internal consistency reliability of the measure was acceptable (0.78) and the measure was unidimensional with 60.5 % of the variance measuring a single, latent concept. Test–retest reliability was slightly below the acceptable 0.70 criteria at 0.68. Patients who achieved complete or partial clinical response also reported statistically significant and meaningful improvement in emotional domain scores (mean change = -31.5 and -29.6 respectively; effect size = 1.06 and 1.01 respectively; p value \0.05 on each). CONCLUSIONS: The Skindex emotional domain demonstrated adequate performance in an locally advanced basal cell carcinoma (laBCC) patient population, though high floor effects were observed. Clinically meaningful improvement in the Skindex16 emotional domain suggests improved QoL for laBCC patients treated with vismodegib.

123

Qual Life Res (2015) 24:1–191 (103.5) Psychological distress in UK patients with idiopathic pulmonary fibrosis; use of emotion thermometers interpreted within a biopsychosocial constructionist framework Anne-Marie Doyle, D.Clin.Psy BSc MA, Royal Brompton hospital London, London, United Kingdom; Caroline Burdett, BA, ISoQol & Royal Brompton Hospital London, London, United Kingdom; Jennifer Gane, Patient Partner RN RHV, Royal Brompton hospital London, London, United Kingdom; Zahra Aden, MSc BSc, NHLI Royal Brompton and Imperial College London, London, United Kingdom; Anne-Marie Russell, MSc RN RHV LPE PGCE, Imperial College London & Royal Brompton Hospital, London, United Kingdom AIMS: There is little research about the nature of psychological distress in patients with idiopathic pulmonary fibrosis (IPF). The aim of the study was to explore dimensions of distress using the Emotion Thermometers; five visual-analogue scales that elicit self-report numeric data in relation to four constructs of distress and need for help. Constructionism is the position that knowledge and understanding is created by individuals/groups and sustained through social processes within societies. METHODS: Participants were invited to complete the Emotion Thermometers which involved marking on a thermometer scale 0–10, how much ‘emotional upset’ they had been experiencing in the past week, including that day. The first four thermometers were labelled ‘distress’, ‘anxiety’, ‘depression’ and ‘anger’, the fifth scale was labelled ‘need help’, with 0 rated as ‘can manage by myself’ and 10 rated as ‘desperately need help’. RESULTS: 74 IPF patients (mean age 69, range 41–88, 80 % male) completed the task. A significant proportion of patients selected ‘none’ in relation to the following constructs: distress (38 %), anxiety (39 %), depression (52 %) and anger (55 %). The remainder reported varying levels of ‘emotional upset’ with a significant minority reporting moderate to high levels (score \ 4) as follows: distress (27 %), anxiety (29 %), depression (18 %) and anger (18 %). In terms of need for help, 59 % reported that they could ‘manage by themselves’ and 17 % reported a moderate to high need for help. CONCLUSIONS: Psychological distress is conceptualised as a series of complex, functional, dynamic interactions between neurological, physiological, behavioural and socio-cultural systems. The Emotion Thermometers provide an effective way of facilitating communication between patients and healthcare professionals in the realm of psychological health. Further research could usefully examine strategies that facilitate well-being alongside on-going work to understand the nature of distress related to IPF symptoms, disability,

Fig. 1 .

Qual Life Res (2015) 24:1–191 loss of role function and end of life/palliative care issues. It is noted that stigma related both to living with a serious, progressive respiratory condition and of being witnessed to ‘not cope’ or to experience ‘psychological distress’ both warrant further study and may be relevant in relation to under-reporting of psychological distress and attempts to cope and maintain psychological integrity (Fig. 1).

(103.6) ‘‘First and foremost battle the virus.’’ Patient preferences in antiviral therapy for hepatitis C Axel Mu¨hlbacher, Professor Health Economics and Healthcare Management, Hochschule Neubrandenburg, Neubrandenburg, Germany; Susanne Bethge, Research Assistant, Hochschule Neubrandenburg, Neubrandenburg, Germany AIMS: Treatment options for patients with hepatitis C have made tremendous progress in the last 5 years. In the wake of the innovative triple therapies several innovative and even interferon-free regimen are in the authorization process. These innovations promise a huge reduction of burden of disease and side effects accompanied by an increase of patient’s benefit. It is unclear, however, what properties generate the highest benefit for patients. The aim of this study was to evaluate patient preferences in terms of innovative antiviral-therapies for hepatitis C. METHODS: A systematic literature search was conducted to identify patient-relevant outcomes for the success of therapy. Within N = 14 semi-structured interviews the decision model was generated. Eight patient relevant characteristics were identified and described by three or six levels. For the discrete choice experiment, an experimental design (3*3 + 5*6) was generated using Ngene-Software. The design consisted 72 choices, which were divided into 6 blocks. The survey was conducted in August 2014 with computer-assisted personal interviews. RESULTS: N = 561 hepatitis C-patients (58.1 % male) within different treatment states participated. The preference analysis [random parameter logit estimation (with 95 % CI)] showed a predominance for the attribute ‘‘sustained virological response 6 month after end of treatment’’ (coef.: 1108). A split second rank is occupied by the possible side effects ‘‘anemia’’(coef.: 0.396) and ‘‘rash’’(coef.: 0.372). The attributes that describe the administration of the treatment ‘‘number of interferon injections’’ (coef.: 0.224) and ‘‘duration of treatment’’(coef.: 0.183) share the third place. The attributes ‘‘fatigue/exhaustion’’(coef.: 0.088) and ‘‘nausea/diarrhea’’(coef.: 0.067) follow. Within the treatment decision of hepatitis C-patients the attribute ‘‘headache’’(coef.: 0.036) was not taken into account, as reflected by the lack of significance. CONCLUSIONS: In this study, the essential decision criteria for an optimal antiviral therapy of hepatitis C from the Hep C patients’ view were obtained by using a DCE. Thereby SVR is ranked highest against all patient relevant side effects and mode of administration characteristics. In the context of this study it could be demonstrated that with the help of the DCE patient-relevant properties can be weighted in terms of possible treatment options. The DCE stands out by a high degree of realism and good handling for the patient.

104: E-Health (104.1) Effect of electronic patient reporting of symptoms during cancer treatment: a randomized controlled trial Ethan Basch, MD, MSc, University of North Carolina, Chapel Hill, NC, United States; Allison Deal, Statistician, University of North Carolina, Chapel Hill, NC, United States; Amylou Dueck, Statistician, Mato Clinic, Scottsdale, AZ, United States; Antonia Bennett, University of North Carolina - Chapel Hill, Chapel Hill, NC, United States; Thomas M. Atkinson, PhD., Memorial Sloan Kettering

19 Cancer Center, New York, NY, United States; Lauren J. Rogak, MA, Memorial Sloan Kettering Cancer Center, New York, NY, United States; Jeff A. Sloan, PhD, Mayo Clinic, Rochester, MN, United States; Deborah Schrag, MD, MPH, Dana-Farber Cancer Institute, Boston, MA, United States AIMS: Prior studies suggest that patient self-reporting of symptoms enhances communication with clinicians and alters symptom management, but effects on clinical outcomes are unknown. METHODS: We randomly assigned patients receiving routine chemotherapy for advanced solid tumors at Memorial Sloan Kettering Cancer Center to report 12 common symptoms via tablet computers, or to usual care consisting of symptom monitoring at the discretion of clinicians. Those with home computers received weekly email prompts to report between visits. Treating physicians received symptom printouts at visits and nurses received email alerts when participants reported severe or worsening symptoms. The primary outcome was change in health-related quality of life (HRQL) at 6-months compared to baseline, measured by the EuroQol EQ-5D Index. Secondary endpoints included emergency room (ER) visits, hospitalizations, and 1-year survival. RESULTS: Among 766 patients randomized, HRQL improved among more participants in the intervention group than usual care (34 vs. 18 %) and worsened among fewer (38 vs. 53 %) (P \ 0.001). Intervention patients were less frequently admitted to the ER (34 vs. 41 %, P = 0.02) or hospitalized (45 vs. 49 %, P = 0.08). While 75 % of the intervention group was alive at 1 year, 69 % with usual care survived the year (P = 0.05), with differences also seen in quality-adjusted survival (mean of 8.7 vs. 8.0 months; P = 0.004). Benefits were greater for participants lacking prior computer experience. Most intervention patients (63 %) reported severe symptoms during the study. Nurses frequently initiated clinical actions in response to email alerts. CONCLUSIONS: Clinical benefits are associated with symptom self-reporting during chemotherapy for advanced cancer. (Clinicaltrials.gov: NCT00578006).

(104.2) The integration of a web-based outcome instrument and a graphic analysis tool for real-time, patient-centered outcome data collection and analysis during Justin E. Bird, MD, University of Texas MD Anderson Cancer Center, Houston, TX, United States, Theresa J. Nalty, PhD, PT, NCS, University of Texas MD Anderson Cancer Center, Houston, TX, United States; Jing He, University of Texas MD Anderson Cancer Center, Houston, TX, United States; Alexander N. Penny, PT, DPT, University of Texas MD Anderson Cancer Center, Houston, TX, United States; Kristi K. Posey, Physician Assistant, University of Texas MD Anderson Cancer Center, Houston, TX, United States; Valerae O. Lewis, MD, University of Texas MD Anderson Cancer Center, Houston, TX, United States AIMS: (1) Develop a web-based instrument for real-time collection of patient reported outcome measures, disease specific measures, patient-centered quality of life and satisfaction measures, and clinician reported functional measures; and (2) Determine the feasibility of providing patient-centered quality of life and functional outcome data to patients using interactive visual interfaces to enhance patient’s understanding of their disease and ease the decision-making process. METHODS: Patient, physician, and physical therapist reported outcome measures were selected for each anatomic site of disease treated surgically within the orthopaedic oncology department at our institution. An automated process was built to collect and transfer data into our surgical database. Study data was collected electronically with validated outcome measures created using a research electronic data capture (REDCap) tool. REDCap is a secure, web-based application designed to support data capture for research studies. Patient outcome data was then analyzed using customized visual analytic

123

20

Qual Life Res (2015) 24:1–191 feasibility of real-time data capture and the utility of visual analytics for communicating complex outcome data during routine clinical care workflows. Our experience further elucidated an added benefit of improved incentives for patients and providers to collect important outcome data due to this novel direct feedback mechanism. Future directions include performing qualitative analyses to determine the impact of visual analytics for the purpose of communication and shared decision making in cancer care (Fig. 1).

Fig. 1 Outcome measures (time points: baseline, 3 m, 6 m, 1y, 2y, 5y)

(104.3) Impact of an automated telehealth cancer symptom monitoring and management system on symptom burden and quality of life Kathi Mooney, RN, PhD, FAAN, University of Utah, Salt Lake City, UT, United States; Susan L. Beck, RN, PhD, FAAN, University of Utah, Salt Lake City, UT, United States; Debbie Wujcik, RN, PhD, FAAN, Vanderbilt-Ingram Cancer Center, Nashville, TN, United States; William Dunson Jr., MD, Huntman Cancer Institute, Salt Lake City, UT, United States; Gary Donaldson, PhD, University of Utah, Salt Lake City, UT, United States

Fig. 2 Clinician reviewing survival curve with a patient

Fig. 3 Visual analytic of 10 m walk test with timed 6 meter section based on surgical method, type of cancer, assistive device needed for ambulation, and the pot-operative day. The slope of each calculated linear regression line was used to generate a formula for predicting average outcome values with 95 % confidence intervals for each of the measures containing continuous summary scores

software. This enabled the clinicians to provide graphic representation of the most up-to-date data available at each clinic visit in order to illustrate the individual’s functional progress and health related quality of life status after surgery. Grouped regression formulas were used to project an individual’s functional return at any given postintervention date with 95 % confidence intervals. RESULTS: Communicating real-time data using an interactive visual interface enabled clinicians to more accurately communicate prognosis (Fig. 2), patient progress (Fig. 3), and project outcomes for a sample of patients who had pelvic sarcoma surgery. The visual interface provides clinicians, patients, and caregivers a means of understanding and communicating complex data sets. Figure 3 demonstrates a visual analytic profile of a timed walk test over the course of 3 years grouped by assistive device. CONCLUSIONS: This pilot study demonstrates both the

123

AIMS: To determine the impact of an automated telephone-delivered cancer symptom monitoring system on symptom burden and quality of life. METHODS: In an RCT, chemotherapy patients were randomized to the Symptom Care (SC) intervention or Usual Care (UC). All patients utilized the automated system daily to report symptom severity for 11 symptoms (0–10 scale) and monthly completed the SF-36. The SC intervention also received automated symptom specific coaching based on reported severity patterns. In addition those reporting moderate or greater severity received follow-up Nurse Practitioner calls for symptom care intensification utilizing national guidelines. The UC group was advised to contact providers for symptom concerns. RESULTS: 358 patients enrolled from two centers in Utah and Tennessee. Mean age was 56, 75 % were female, 44 % had breast cancer, 17 % had lung cancer. With no difference between groups, participants’ call compliance averaged 90 % of expected calls. The groups were equivalent for all socio-demographic and disease variables. Compared to UC, SC had significantly reduced symptom burden (treatment by time interaction, p = .008). The relative symptom burden reduction for intervention participants was 1.8 severity points (p \ .001). SF36 composite scores (39 for Physical Composite; 51 for Mental Health Composite) did not differ by group with physical function in the chronic illness range and normal range for mental health (MH); however, the SC intervention group had significantly better adjusted mental health (p = .003) using the Mental Health subscale alone rather than the composite. This improvement over UC was moderated by gender, with men receiving a direct benefit of 7.8 MH points (p = .016) plus a small mediated effect through decreased symptom burden (.54 net MH points; p = .035). SC men’s rate of change averaged 5.2 MH points higher each month than UC men over the 4 study months (21 total points). While SC women had substantial decreases in symptom burden, they did not benefit as SC men did for mental health. CONCLUSIONS: Symptom burden can be significantly reduced for both men and women through remote monitoring and provider follow-up. In addition, men appear to gain a significant mental health advantage from automated monitoring and support for emotional concerns during cancer treatment.

(104.4) Electronic patient reported outcomes and toxicities during radiotherapy for head and neck cancer Joshua R. Niska, MD, Mayo Clinic, Scottsdale, AZ, United States; Michele Y. Halyard, MD, Mayo Clinic, Scottsdale, AZ, United States;

Qual Life Res (2015) 24:1–191 Angelina D. Tan, Mayo Clinic, Rochester, MN, United States; Pamela J. Atherton, Mayo Clinic, Rochester, MN, United States; Samir H. Patel, MD, Mayo Clinic, Scottsdale, AZ, United States; Jeff A. Sloan, PhD, Mayo Clinic, Rochester, MN, United States AIMS: Providers strive to ameliorate toxicities and to maximize quality of life (QOL) during radiotherapy (RT) or chemoradiotherapy (CRT) for head and neck (H&N) cancer. However, many aspects of QOL may not be readily evident. Electronic patient reported outcomes (PROs) attempt to address this dilemma. Using electronic PROs, we characterized QOL and toxicities during H&N RT. METHODS: Sixty-five H&N RT patients completed electronic realtime 12-domain Linear Analogue Self-Assessments (LASA) at baseline and before biweekly appointments. LASA scores were scaled 0–100 (poor-excellent). Changes in QOL domains from baseline to biweekly appointments and last week of RT were calculated. Data on disease, CRT, RT, toxicities, and adverse events were retrospectively collected. Toxicities were retrospectively graded using Common Terminology Criteria for Adverse Events v4.0 and recorded at baseline, biweekly intervals, and last week of RT. RESULTS: The study population was primarily elderly (avg age 65.3), white (92.3 %), male (80.0 %) patients on adjuvant (63.1 %) CRT (61.5 %) for stage IV (59.9 %) squamous cell carcinoma (78.5 %) of the H&N. CRT was mostly cisplatin (55.0 %), cetuximab (17.5 %), or carboplatin (12.5 %). RT regimen averaged 63.7 Gy in 32 fractions. During RT, most patients had clinically significant decrease in all QOL domains except level of support, financial concerns, and legal concerns. QOL domains most impacted were fatigue (77.8 % with clinically significant decrease), social activity (75.4 %), and overall QOL (74.2 %). At last week of RT, average LASA score was worse than baseline in all QOL domains except level of support and financial concerns. All patients had grade 2 toxicity. Grade 3 toxicity affected 35.4 % overall, 50.0 % among CRT vs 12.0 % among RT alone. Grade 4 toxicity affected 3.1 % (both were CRT patients). Weight loss averaged 5.5 kg. Patients frequently needed IV hydration (52.3 %) and/or feeding tube (40.0 %). Emergency department (ED) visit without hospitalization occurred in 10.8 %. Emergent hospitalization at our institution occurred in 27.7 %. CONCLUSIONS: H&N RT significantly impacts QOL, particularly overall QOL and fatigue. Grade 3 toxicity is common, particularly for CRT. Many patients need IV hydration and/or feeding tube. ED visits and/or hospitalization are not uncommon. Real-time electronic PROs allow providers to monitor QOL at multiple time points during RT.

105: Women’s Health (105.1) Nausea, vomiting, fatigue and health-related quality of life of women in early pregnancy: the Generation R study Guannan Bai, MPH, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands; Esther H. Groen, PhD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands; Henriette A. Moll, PhD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland, Netherlands; Vincent W. Jaddoe, PhD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands; Ida Korfage, PhD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands; Hein Raat, PhD, MD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands AIMS: Nausea, vomiting and fatigue are the most common pregnancy-related symptoms in early pregnancy and they may affect women’s health-related quality of life (HRQOL). The aim of this study was to assess the independent associations between these symptoms and HRQOL of women during first trimester in a

21 community sample. METHODS: We included 5079 women in early pregnancy in the Generation R Study, a prospective mother and child cohort study. The frequency of nausea, vomiting and fatigue in the last 3 months was measured by questionnaire at enrollment. The 12-item Short Form Health Survey (SF-12) was used to measure HRQOL; physical and mental component summary (PCS/MCS) scores were calculated. We considered demographic characteristics (maternal/gestational age, ethnic background, education level, marital status, parity), life-style related factors (body mass index, tobacco and alcohol use), indicators of health status (chronic/infectious conditions, urogenital symptoms, sleep quality, headache, anxiety, and depression) as potential confounders, which were also measured by questionnaire. Independent associations of nausea, vomiting and fatigue with HRQOL were assessed using multivariate regression models adjusting for all potential confounders. Multiple imputation was applied to account for missing data in covariates. RESULTS: Mean age was 30.0 (SD 5.0) years; mean gestational age was 13.5 (SD 2.0) weeks; 44.1 % were non-Dutch. Percentages of women who reported nausea, vomiting and fatigue to be present daily in the last 3 months were 33.6, 9.6 and 44.4 %, respectively; 17.1, 56.6 and 2.0 % reported no nausea, vomiting and fatigue. Multivariate analyses demonstrated that compared to women who never reported these symptoms, those who reported nausea, vomiting and fatigue to be presented daily had decreased HRQOL in both PCS (respectively -3.0, 95 % CI -3.8, -2.2; -2.1, 95 % CI -3.1, -1.2; -5.2, 95 % CI -6.9, -3.5) and MCS (-1.8, 95 % CI -2.7, -0.96; -3.0, 95 % CI -4.0, -2.0; -3.0, 95 % CI -4.8, -1.1), after adjusting for confounders. CONCLUSIONS: Even after adjusting for a comprehensive set of potential confounders, presence of nausea, vomiting and fatigue are associated with lower HRQOL. These findings highlight the importance to take into account negative impacts of pregnancy-related symptoms on HRQOL by health professionals in daily practice. Further research should focus on evaluation of approaches to improve HRQOL during early pregnancy.

(105.2) Patient-reported neurocognitive deficits do not always agree with objectively-assessed neurocognitive test scores—why they differ as explained by a bayesian latent regression item response approach Yuelin Li, PhD., Memorial Sloan Kettering Cancer Center, New York, NY, United States; James C. Root, PhD., Memorial Sloan Kettering Cancer Center, New York, NY, United States; Thomas M. Atkinson, PhD., Memorial Sloan Kettering Cancer Center, New York, NY, United States; Tim A. Ahles, PhD., Memorial Sloan Kettering Cancer Center, New York, NY, United States AIMS: Many cancer survivors report memory lapses after cancer treatment, yet these deficits only weakly correlate with objectivelyassessed neurocognitive measures of memory. We explain this discordance using a Bayesian Latent Regression Item Response approach. METHODS: 132 breast cancer survivors and 45 healthy controls (mean age 54) were given a battery of 28 objective neurocognitive tests, including Memory (Verbal, Visual) and Executive Functioning (Attention, Processing Speed). Patient-reported cognitive deficits were assessed by the Multiple Ability Self-Report Questionnaire (MASQ), which consists of Language, Visual Perceptual, Verbal Memory, Visual Memory, and Attention subdomains. Assessments were made before treatment and 6, 12 and 18 months post-treatment. A Bayesian Latent Item Response Regression was fitted to model the extent to which patient-reported deficits at 6 months agree with objectively measured post-treatment deficits. RESULTS: Cancer survivors perceived their deficits to be primarily in Verbal Memory (e.g., ‘‘I forget to mention important issues during conversations’’) rather than Attention (e.g., ‘‘I can keep my mind on

123

22

Qual Life Res (2015) 24:1–191 (105.3) Validation of the measure of ovarian symptoms and treatment concerns (MOST), a patient-reported outcome measure (PROM) of symptom burden and subjective benefit with chemotherapy in recurrent ovarian cancer (ROC)

Fig. 1 .

Fig. 2 . more than one thing at a time’’) (p \ 0.01 by MASQ subscales). Yet the Bayesian model showed the opposite pattern; patient-reported MASQ symptoms at 6 months were most strongly associated with Executive Functioning (e.g., D-KEFS Trails Test Sequence 3, median effect size d = 0.65, 95 % HDR 0.25, 1.09); deficits in focused attention during memorizing (e.g., middle-region recall, median d = 0.33, 95 % HDR -0.03, 0.71), and were negligible in overall memory performance (e.g., CVLT memory test, median d = 0.03, 95 % HDR -0.43, 0.38). This pattern was observed in all five MASQ subdomains (Fig. 1) at a minimum of 90 % posterior confidence. The Bayesian model produced a 99 % posterior probability that patients attributed neurocognitive symptoms to memory rather than attention (also shown in model-predicted symptoms in Fig. 2). CONCLUSIONS: The Bayesian approach suggests that patient-reported memory problems may be caused by shortfalls in cognitive control and supervisory attentional systems; putatively, attentional control problems prevent the memory from registration/encoding in the first place, thus patients subjectively experience memory problems. Practitioners of patient-reported outcomes (PRO) research should strive to collect objective outcome assessments whenever available, so they can corroborate the PRO-version to achieve higher precision and accuracy in intervention design considerations.

123

Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia; Martin Stockler, NHMRC Clinical Trials Centre, Sydney, NSW, Australia; Rachel O’Connell, NHMRC Clinical Trials Centre, Sydney, Australia; Florence Joly, Centre Francois Baclesse, GINECO, Caen, France; Anne Lanceley, University College London, London, United Kingdom; Felix Hilpert, Klinik fur Gynakologie und Geburtshilfe, UKSH, Kiel, Germany; Aikou Okamoto, Jikei University School of Medicine, Tokyo, Japan; Eriko Aotani, Kitasato Academic Research Organization, Tokyo, Japan; Sandro Pignata, National Cancer Institute, Naples, Italy; Paul Donnellan, Galway University Hospital, Galway, Ireland; Amit Oza, Princess Margaret Hospital, Toronto, Canada; Elisabeth Avall-Lundqvist, Linkoping University, NSGO, Linkoping, Sweden; Jonathan Berek, Stanford Women’s Cancer Centre, Stanford, CA, United States; Luke Buizen, NHMRC Clinical Trials Centre, Sydney, Australia; Katrin Sjoquist, NHMRC Clinical Trials Centre, ANZGOG, Sydney, Australia; Kim Gillies, NHMRC Clinical Trials Centre, ANZGOG, Sydney, Australia; Phyllis Butow, Psycho-oncology Research Group (PoCoG), University of Sydney, Sydney, Australia; Michael Friedlander, Prince of Wales Hospital, ANZGOG, Sydney, Australia; on behalf of the GCIG Symptom Benefit Study Group, Gynecologic Cancer Intergroup (GCIG) AIMS: The MOST, developed in Stage 1 of the Gynecologic Cancer Intergroup Symptom Benefit Study (GCIG-SBS), was designed to provide statistically efficient assessment of patient-reported burden and benefit in trials of palliative chemotherapy for ROC by focusing on symptoms of ovarian cancer and adverse effects of chemotherapy. We aimed to assess the validity of the MOST Ovarian Symptom Index (MOST-OSI) relative to comparable scales from the most widely used ovarian-cancer-specific PROMS: EORTC Ovarian Cancer abdominal symptoms scale (QLQ-OV28-Abdo); Functional Assessment of Cancer Therapy-Ovarian Additional Concerns Scale (FACTO-ACS) and Trial Outcome Index (FACTO-TOI); FACTOvarian Symptom Index (FOSI). METHODS: Stage 2 of GCIG-SBS recruited 950 ROC patients internationally. Patients completed the QLQ-OV28, FACT-O/FOSI and MOST before each cycle of chemotherapy. At baseline, clinicians assessed patients’ performance status (PS), cancer-related symptoms, and presence of ascites. Concurrent validity of the MOST-OSI was assessed by correlation (Spearman) with comparable scales. Discriminative validity was assessed with Cohen’s D effect sizes (ES) for differences between groups classified by PS (0–1 vs C2), symptomatic ascites (yes, no), and cancer-related symptoms (yes, no). Relative efficiency (RE) was calculated as the squared ratio of t-statistics from 2-sample t-tests, with MOST-OSI as the numerator so RE [ 1 indicates that MOSTOSI is more efficient than the competing PROM. RE is proportional to the required sample size; e.g. RE = 2.0 means half as many subjects are required to detect a given effect. RESULTS: Data from 859 patients with baseline scores for all PROMs were analysed: 615 had cancer-related symptoms,190 had ascites, and 92 had PS C 2. Correlations of MOST-OSI with QLQ-OV28-Abdo, FACTO-ACS, FACTO-TOI and FOSI were 0.76, -0.64, -0.72 and -0.80, respectively. MOST-OSI ES were: for PS, ES = 0.77 (CI = 0.53–1.00): cancer-related symptoms, 0.96 (0.80–1.12); ascites, 0.86 (0.69–1.04). For PS, MOST-OSI was more sensitive than QLQ-OV28-Abdo (RE = 2.4, CI = 1.4–5.1) and FOSI (2.0, 1.2–3.3). For cancer-related symptoms, MOST-OSI was more sensitive than FACTO-ACS (2.1, 1.4–3.3) and FACTO-TOI (2.0, 1.4–2.9). For ascites, MOST-OSI was more sensitive than FACT-TOI (1.8, 1.2–2.8) and borderline for FACTO-ACS (1.5, 1.0–2.4). CONCLUSIONS: The MOST-OSI has good concurrent and discriminative validity relative to comparable

Qual Life Res (2015) 24:1–191

23 criteria, including the SIDI-F diagnostic criteria (B33), the FSDS-DAO cut-point for hypoactive sexual desire disorder (C11), and number of sexually satisfying events (\2; all p \ 0.001). CONCLUSIONS: The FSFI desire score demonstrated excellent reliability, validity, and ability to detect change in desire among women with FSD and is a suitable endpoint in clinical trials.

(105.5) Modeling overall quality of life in women with breast cancer Marie-Eve Letellier, McGill University, Montreal, QC, Canada; Susan C. Scott, McGill University Heath center, Montreal, QC, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada

Fig. 1 . scales from the most widely used PROMS for ovarian cancer, and comparable or superior statistical efficiency (Fig. 1).

(105.4) Measurement properties of the female sexual function index desire domain in women with female sexual dysfunction Hilary Wilson, Evidera, Bethesda, MD, United States; Dennis Revicki, PhD, Evidera, Bethesda, MD, United States; Carl Spana, Palatin Technologies Inc., Cranbury, NJ, United States; Robert Jordan, Palatin Technologies Inc., Cranbury, NJ, United States AIMS: Lack of sexual desire is a key symptom of female sexual dysfunction (FSD); however, no measure of desire currently exists that meets the Food and Drug Administration’s evidence requirements for use in support of a labeling claim. The Female Sexual Function Index (FSFI) was designed for use in clinical trials as a measure of sexual functioning in women. It assesses 6 domains: desire, arousal, lubrication, orgasm, satisfaction, and pain. The primary aim of this study was to evaluate measurement properties of the FSFI desire domain. METHODS: Subjects (n = 325; mean age 37.1 years; range 21–53) were participants in a Phase 2, randomized trial, evaluating the efficacy and safety bremelanotide versus placebo. Outcome measures included: FSFI (28-day recall version), Female Sexual Distress Scale-Desire/ Arousal/Orgasm (FSDS-DAO), the Sexual Interest and Desire Inventory-Female (SIDI-F), and the Female Sexual Encounter Profile (FSEPR). Analyses included: Cronbach’s a (internal consistency); intraclass correlation coefficients (ICCs) among stable patients (defined as ± 5 % change in SIDI-F; test–retest reliability); Spearman correlations (convergent validity); and, analysis of variance (known groups validity and ability to detect change). RESULTS: The internal consistency reliability of the FSFI desire score was excellent (Cronbach’s a C 0.81 at multiple time points). Test–retest reliability over 4-weeks in stable patients was supported with ICCs C 0.70. Correlations between the FSFI desire score and desire specific items of the FSDSDAO and SIDI-F were moderate or greater ([0.30), providing support for convergent validity. FSFI desire scores were significantly more severe (all p \ 0.0001) among participants that met the SIDI-F diagnostic cut-point for FSD (B33) and the FSDS-DAO cut-point for patients with hypoactive sexual desire disorder (C11), providing support for known groups validity. The ability of the FSFI desire score to detect change in participants known to have experienced a change in clinical status with treatment was supported by several responder

AIMS: Cancer was once considered as an acute disease with a binary outcome (dead or alive). The effectiveness of early diagnosis and treatment has now made breast cancer a chronic disease where impairments, activity limitations and participation restrictions, wellknown sequelae of (breast) cancer treatment, lead to poor health perception and sub-optimal quality of life (QOL). The Wilson-Cleary (W–C) model of health related QOL (HRQL), which suggests causal links between biological and physiological factors, symptoms status, functional status, general health perceptions and overall QOL, has not been tested in breast cancer. To improve our comprehension for how QOL emerges from the sequelae of breast cancer, the aim of this study was to empirically test a bio-psycho-social conceptual model of HRQL for breast cancer survivors. METHODS: Women completed five outcomes measures [SF-36, EORTC QLQ C30 and -BR23, WHODAS, and WHOQOL, all of which have been previously mapped to the International Classification of Functioning, Disability and Health (ICF)]; health care professionals completed the ICF Breast Cancer Core Set. Latent variables were created considering the variable’s/scale’s construct and their ICF mapping code, when applicable. All rubrics of the W–C model were fulfilled. A structural equation model (SEM) was fit using MPlus. RESULTS: A total of 245 women (mean age 57) participated. 193 (79 %) had breast cancer stage II or below and 44 % had two or more surgeries. The median time post-surgery for the evaluation was approximately 9 months, which is within the critical time (first 2 years) when most impairments, activity limitations and participation restrictions emerge. The hypothesized model consists of 24 latent variables covering all 5 rubrics of the W–C model: 9 for Biology; 7 for Symptoms; 6 for Function, and 1 each for Health Perception and overall QOL. An empirical model will be developed sequentially following the W–C framework. CONCLUSIONS: An empirical model of QOL, built from health and functional outcomes post-breast cancer, would be useful to inform a more integrated and person-centered approach to enhance QOL post breast cancer.

(105.6) Long-term prospective patient reported outcomes (PROs) after types of immediate pedicle back flap (latissimus dorsi) breast reconstruction and adjuvant treatment. Two- and three-year post-operative results from the Prospective Trial Management Group Zoe E. Winters, MB,ChB(Witw), DPhil(Oxon), FRCS, FCS, University of Bristol, Bristol, United Kingdom; Vasiliki Balta, Research Associate, University of Bristol, Brussels, Belgium; Maryam Afzal, Research Administrator, University of Bristol, Bristol, United Kingdom; Janet Freeman, Senior statistician, University of Bristol, Bristol, United Kingdom; Rebecca Llewellyn-Bennett, Specialist Registrar, University of Bristol, Bristol, United Kingdom; James Cook, Consultant Surgeon, North Bristol NHS Trust, Bristol, United Kingdom; Zenon Rayter, Consultant Surgeon, North Bristol

123

24 NHS Trust, Bristol, United Kingdom; Rosemary Greenwood, Senior RDS statistician, NIHR Research Design Support, Bristol, United Kingdom; Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia AIMS: To evaluate the impact at 2 and 3 years of implant-assisted latissimus dorsi (LDI) and autologous LD (ALD) flap breast reconstructions (BRRs) on patient-reported outcomes (PROs) adjusted for case-mix, and secondarily to describe the levels of change in PROs at 2 and 3 years. METHODS: Multi-centre prospective cohort study. The European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Questionnaires QLQ-C30 and QLQ-BR23; Functional Assessment of Cancer Therapy—Breast (FACT-B) and Hospital Anxiety and Depression Scale (HADS), were completed preoperatively and at 2 and 3 years after BRR. RESULTS: 206 patients (93 LDI and 113 ALD) were recruited (2007–2013); 66 % were lymph node negative; 34 % received radiotherapy (RT). Women with adverse clinico-pathology were more likely to have ALD. Each surgical group at 2 and 3 years showed clinically important (P \ 0.01) improvements over time in emotional scales, but worse physical functioning, social well-being, body image and anxiety. Multiple linear regressions were used to estimate the differences in PRO levels at 2 and 3 years, respectively for LDI and ALD by adjusting for clinico-demograpic factors at baseline. RT adversely affected social function at 2 years (P = 0.002). By 3 years, persistent effects were seen on arm symptoms (P = 0.005) after chemotherapy, and on physical well-being in younger women (P = 0.006). Women undergoing ALD BRR had significantly improved sexual functioning (P = 0.003) at 3 years relative to those who had LDI BRR. There were no other significant differences in long-term PRO levels comparing types of breast reconstruction. CONCLUSIONS: Clinically important changes occurred in core functioning domains, breast surgery symptoms and psychological distress. These are applicable PROs in future studies of BRR. Randomization designs remain the best option to compare surgical effects on PROs.

106: Health Systems and Quality of Life (106.1) A breath of fresh air: how eCOA technology integration boosts patient compliance in asthma trials Jessica Thilaganathan, CRF Health, London, United Kingdom; Antti Heikkila, CRF Health, London, United Kingdom; Juha Matero, CRF Health, London, United Kingdom; Paul O’Donohoe, CRF Health, London, United Kingdom; Karl McEvoy, CRF Health, London, United Kingdom; Katie Garner, CRF Health, London, United Kingdom AIMS: Over the last 10 years the prevalence of asthma has risen by 50 %, with over 235 million people suffering with the condition. For asthma patients, managing their disease often involves tracking both their peak expiratory flow rate (PEF) and forced expiratory volume (FEV1) values, which are common key endpoints in clinical trials. Patients are often required to fill in diaries twice a day, which increases patient burden and risks negatively impacting patient compliance; a historic weakness with paper diaries. Our aim was to develop an electronic solution to capture asthma patients’ symptoms which would lower patient burden while simultaneously increasing compliance and improving data quality. METHODS: We collated information from pharmaceutical sponsors and asthma trial protocols in order to understand the needs for an electronic daily diary to capture all home-based data in asthma clinical trials. A draft diary solution was developed and further refined with sponsor input to ensure it fit in with the typical experience of an asthma trial participant. RESULTS: Initial sponsor feedback highlighted concerns about patients needing to use both the PEF meter and diary device to

123

Qual Life Res (2015) 24:1–191 provide a large amount of data. To reduce this burden an asthma diary solution was developed that included a way for patients to measure PEF on a spirometer and transfer data via Bluetooth without having to manually enter values to the diary, improving usability and data quality. This integration drives a more complete symptom and physiological value data set with accurate date and time stamping. Use of the diary solution led to high compliance rates in asthma trials, with average daily compliance ranging between 88 and 92 % in multiple global studies. Feedback from these studies has led to a further iteration of the diary which utilizes a modular approach for collection of symptoms and PEF values. CONCLUSIONS: With an increasing amount of data being collected in asthma trials, an intuitive electronic diary can significantly reduce patient burden leading to high compliance, improved data quality and study efficiency, benefitting not only patients but all stakeholders involved.

(106.2) Development and use of patient reported outcomes during early drug development: an industry and academic collaboration Charles S. Cleeland, PhD, UT MD Anderson Cancer Center, Houston, TX, United States; Loretta Williams, PhD, UT MD Anderson Cancer Center, Houston, TX, United States; Tito Mendoza, PhD, UT MD Anderson Cancer Center, Houston, TX, United States; Ferran Prat, PhD, JD, UT MD Anderson Cancer Center, Houston, TX, United States; Brian Cuffel, PhD, Bayer Healthcare Pharmaceuticals, Whippany, NJ, United States; Barrett H. Childs, Bayer Healthcare Pharmaceuticals, Whippany, NJ, United States; Karen N. Keating, Bayer Healthcare Pharmaceuticals, Whippany, NJ, United States AIMS: There has been an exponential increase in the interest of the patient experience in response to oncology therapies. Gathering data about this experience is especially important with the introduction of new agents or combinations of agents. Standard toxicity ratings are an imprecise reflection of the patient’s feeling and function in response to new therapy, and provide little information that a clinician can provide to a patient about what they will experience while being treated. The Department of Symptom Research at MD Anderson Cancer Center and Bayer Healthcare recently formed a collaboration that assures that patient reported information contributes to the drug development process. This poster/presentation will describe how this collaboration will be implemented across the trajectory of the development of a new oncology drug/drug combination. METHODS: Early-phase clinical trials present an opportunity to measure symptoms early and often and to capture signs of toxicity and potential signs of symptom benefit, such as reduction of disease relatedsymptoms and improved function. This measurement can be done with a simple existing multi-symptom assessment instrument and through qualitative interviewing of patients to capture the emergence of additional treatment-related symptoms or symptom benefit. Even with the small numbers of patients, systematic use of symptom measurement tools administered at frequent intervals can give incremental information about treatment toxicities or benefits and allow for the identification or development of specific assessment instruments that can be incorporated into planning for registration trials. RESULTS: Early evidence of multiple treatment-related symptoms can serve as a warning sign to drug developers that treatment tolerability may be compromised, and the appropriateness of dose selection or plans for supportive care need to be considered. The next phase of instrument development is to employ the preliminary instrument in extension and later phase trials to develop evidence of the instrument’s psychometric properties and sensitivity to disease change, evidence that may help regulatory agencies understand the patient’s perspective on the impact of the new treatment. CONCLUSIONS: There are multiple advantages to begin the

Qual Life Res (2015) 24:1–191 development and refinement of patient reported outcomes very early in the trajectory of oncology drug development.

(106.3) Patients’ and health professionals’ understanding of and preferences for graphical presentation styles for individual quality of life scores obtained with the EORTC QLQ-C30 Wilma Kuijpers, The Netherlands Cancer Institute, Amsterdam, Netherlands; Johannes M. Giesinger, Medical University of Innsbruck, Innsbruck, Austria; August Zabernigg, Kufstein County Hospital, Kufstein, Austria; Teresa Young, Lynda Jackson Macmillan Centre, Mount Vernon Cancer Centre, Northwood, Middlesex, United Kingdom; Elizabeth Friend, Basingstoke & North Hampshire Hospital, Basingstoke, United Kingdom; Iwona M. Tomaszewska, Jagiellonian University Medical College, Krakow, Poland, Neil K. Aaronson, The Netherlands Cancer Institute, Amsterdam, Netherlands; Bernhard Holzner, Medical University of Innsbruck, Innsbru¨ck, Austria AIMS: To investigate patients’ and health professionals’ understanding of and preferences for different graphical presentation styles for individual EORTC QLQ-C30 scores. METHODS: We recruited cancer patients (any treatment and diagnosis) in Austria, the Netherlands, Poland and the UK, and health professionals from the Netherlands Cancer Institute. Using a questionnaire we assessed objective and self-rated understanding of QLQ-C30 scores using five presentation styles (bar charts and line charts, with or without color coding, and a heat map) and preferences for certain styles. RESULTS: In total, 548 patients (mean age 61 years; 54 % female; 26 % breast cancer, 13 % colorectal cancer) and 227 health professionals (mean age 45 years; 76 % female; 38 % medical specialists, 62 % nurses) completed the questionnaire. Eighty-three percent of patients and 85 % of health professionals indicated that the graphs were very or quite easy to understand. This self-rated understanding did not differ as a function of graphical presentation style or medical profession. The mean percentage of correct answers to questions assessing understanding was 61 % in patients, 78 % in medical specialist and 74 % in nurses. For patients, this objective understanding did not differ between graphical formats. Medical specialists were significantly better than nurses in understanding absolute scores. Within non-colored charts, patients’ preferences for bar and line charts were comparable (21 vs. 18 %, respectively), with 61 % without a preference. Within color-coded charts, bar charts (30 %) were preferred over heat maps (24 %) or line charts (16 %). We found a significant difference between medical specialist and nurses: medical specialists preferred heat maps (46 %), followed by non-colored bar charts (19 %). Nurses equally valued the heat map and non-colored bar chart (32 %). CONCLUSIONS: We found a substantial discrepancy between participants’ high self-rated and relatively low objectively measured understanding of graphical presentation of PRO results. This highlights the need to provide sufficient guidance when providing access to PRO results. Given that no presentation style was preferred by a clear majority of patients or health professionals, it may be appropriate, where possible, to adapt the presentation of PRO results to individual preferences. This could be facilitated when PROs are administered and fed back to patients and health professionals electronically.

(106.4) Misalignment between cancer patients’ values and their care experiences is associated with worse self-reported mental and physical health outcomes Kathleen J. Yost, Mayo Clinic, Rochester, MN, United States; Marlene Frost, Mayo Clinic, Rochester, MN, United States; Amy

25 Windham, American Institutes for Research, Washington, DC, United States; Sarah Jenkins, Mayo Clinic, Rochester, MN, United States; Kandace Lackore, Mayo Clinic, Rochester, MN, United States; Manshu Yang, PhD, American Institutes for Research, Chapel Hill, NC, United States; San Keller, PhD, American Institutes for Research, Chapel Hill, NC, United States AIMS: We developed and validated the Consumer-based Cancer Care Value Index (CCCVI; sponsored by National Patient Advocate Foundation). The CCCVI can be used to (1) measure the value patients place on different aspects of cancer care, (2) report the alignment between valued aspects of cancer care and the frequency with which that care is provided, and (3) identify services and resources patients value and would like to receive. Our objective in this study was to assess the impact on health outcomes of misalignment between what patients value in their cancer care and what they perceived experiencing. METHODS: We evaluated data for 721 patients (38.5 % male, 85.8 % White, 54.1 % age 65+) from four cancer centers in the U.S. We calculated two indicators of misaligned cancer care. A patient had Misaligned Aspects of Care (MAC) if one or more of the 24 aspects of cancer care included in the survey was rated as highly valued but was provided infrequently or not at all. Patients had Misaligned Services and Resources (MSR) if they indicated they wanted one or more of the 28 specified services or resources but did not receive them. Health outcomes were measured using the PROMIS Global Mental Health (GMH) and Global Physical Health (GPH) summary scales. We used multivariable linear regression to assess the impact of MAC and MSR on GMH and GPH after adjusting for covariates (cancer center, demographics, and clinical variables). RESULTS: MAC was experienced by 271 (37.6 %) patients and MSR by 291 (40.4 %). In the multivariable GMH model, mean T-scores were lower for patients who experienced MAC (-1.4, p \ .05) or MSR (-3.9, p \ .001). In the multivariable GPH model, mean T-scores were also lower for patients who experienced MAC (-2.0, p \ .01) or MSR (-2.7, p \ .001). Adding the two misaligned care independent variables increased the adjusted R2 for the GMH and GPH models by 5.1 and 3.2 %, respectively. CONCLUSIONS: Cancer care that was misaligned with patients’ values was significantly associated with worse self-reported mental and physical health. As this study was cross-sectional, future research should investigate whether providing cancer care that better aligns with patients’ values leads to better health outcomes.

(106.5) Management of patient-reported outcome alerts: a cross sectional survey Derek G. Kyte, PhD, University of Birmingham, Birmingham, United Kingdom; Jonathan Ives, PhD, University of Birmingham, Birmingham, United Kingdom; Heather Draper, PhD, University of Birmingham, Birmingham, United Kingdom; Melanie J. Calvert, PhD, University of Birmingham, Birmingham, United Kingdom AIMS: Assessment of patient-reported outcomes (PROs) in trials provides valuable information to inform patient-centered care, but can reveal ‘PRO alerts’: ‘concerning levels of psychological distress/physical symptoms that require an immediate response’. Ad hoc management of PRO alerts may result in suboptimal patient care or bias trial results. There is little data on how PRO alerts are currently handled in trials. To gain greater understanding of current practice we conducted a national survey of clinical trials personnel. METHODS: An anonymised online cross-sectional national survey of UK research nurses, data managers/coordinators, trial managers and chief/principal investigators involved in clinical trials using a primary/secondary PRO was undertaken. Respondents were self-selected volunteers from a purposive sample of eligible individuals identified through all UK

123

26 Clinical Research Collaboration Registered Clinical Trials Units and 19 NIHR Comprehensive Local Research Networks. Survey questions centred on the proportion of trial personnel encountering PRO alerts, how staff responded to alerts and whether current guidance was sufficient to support research personnel. Descriptive analysis was used to examine participant characteristics and survey responses. All analysis was conducted using SPSS (version 21, IBM). Free-text comments were analysed using directed content analysis. RESULTS: 767 participants completed the survey. 39–50 % of respondent group participants reported encountering PRO alerts. Of these, 83 % of research nurses and 54 % of data managers/trial coordinators reported taking action to assist the trial participant, but less than half were able to record the intervention in the trial documentation. 20 % of research nurses did not view completed PRO questionnaires and were not in a position to discover alerts. Research personnel reported current PRO alert guidance/training was insufficient. CONCLUSIONS: Research personnel are intermittently exposed to PRO alerts. Some intervene to help trial participants, but do not record this intervention in the trial documentation, risking co-intervention bias. Other staff do not routinely check PRO information, meaning alerts may remain undiscovered, or do not respond to alerts that are inadvertently encountered. Failure to monitor and react may impact on patient welfare and safety and raises issues around legal liability. Trial guidance is needed to support PRO alert management that protects the interests of trial participants whilst avoiding potential bias.

(106.6) Do HRQOL assessments impact the development of cancer treatments? Ivan Barofsky, PhD, The Quality of Life Institute, East Sandwich, MA, United States AIMS: First, to demonstrate that formal HRQOL assessments currently plays no meaningful role in the development of cancer treatments, and that new treatments have either a biological rationale, or clinical support. Second to shift the focus of developing HRQOLsensitive cancer treatment from patient reported outcomes (PRO’s) to inventor reported outcomes (IRO’S). METHODS: Examples from a variety of cancer treatment settings will demonstrate that most major changes in cancer treatment that impacts HRQOL occur quite independent of any formal HRQOL assessment. RESULTS: Examples of this include; Modern Hospice care procedures, for example, were first developed at St. Christopher, near London, in 1967 well before formal HRQOL assessments were developed. Second, the shift in the thinking of breast cancer as a localized to a systemic disease, paved the way for the abandonment of the Halstead radical mastectomy treatment, and the development and adoption of more conservative surgical procedures. Third, the shift from treating estrogen-receptor positive breast cancer patients with chemotherapy to such biologically more specific agents such as tamoxifen or aromatase inhitibors. Fourth, the history of changes to melanoma treatment has included a shift from common surgical procedures to the application of various forms of chemotherapy and immunotherapy. Each of these treatment changes, and others, have improved HRQOL outcomes without requiring formal HRQOL assessments. CONCLUSIONS: These examples suggest that the efforts to improve the quality of life outcome of cancer treatment should shift from the patient to the developers of new cancer treatments (IRO’s). To do this will require an institutional commitment (e.g., clinical trial groups, leaders in clinical and research oncology) to the objective of designing cancer treatments that are sensitive to HRQOL issues, and that this be viewed as just as necessary an end as extending cancer survival. Several examples will be given (e.g., the soft-tissue sarcoma trial, incorporating the cosmetic outcome of surgical during initial treatment, shielding during radiation to minimize the development of fibrotic

123

Qual Life Res (2015) 24:1–191 chest and bowel dysfunction, etc.) to demonstrate that this objective is feasible.

107: Psychometrics (107.1) Evaluating the dimensionality of complex pros using bifactor analysis within an exploratory structural equation modeling (ESEM) framework: an example using the patientreported scar evaluation questionnaire (PR-SEQ) Rohini Sen, Adelphi Values, Boston, MA, United States; Andreas M. Pleil, Pfizer, Inc., San Diego, CA, United States; Cheryl Coon, Adelphi Values, Boston, MA, United States; Alan L. Shields, Adelphi Values, Boston, MA, United States AIMS: The Patient-reported Scar Evaluation Questionnaire (PR-SEQ) has the following hypothesized conceptual framework (CF): scar appearance (five items), symptoms (three items), bother (five items), and impacts (fourteen items). Several factor analytic approaches were used to determine if more complex methods (e.g., exploratory structural equation modeling; ESEM) provided insight into the PR-SEQ’s multidimensional structure beyond traditional exploratory and confirmatory factor analysis (EFA and CFA). METHODS: As part of a psychometric study, the PR-SEQ was administered to 514 subjects (77 % female) with non-cosmetic (46 %) and cosmetic (33 %) related hypertrophic scars. In a confirmatory bifactor model, each item loads on a general factor (fg) and, at most, one noise factor (fn). An ESEM approach to bifactor analysis integrates EFA and CFA models, allowing for estimation of the fg via CFA while each item is allowed to load on multiple fn’s in a less restrictive EFA framework. A series of models compared model fit and factor structure of the PR-SEQ: (1) CFA to test the CF, (2) post hoc EFA to assess dimensionality, (3) bifactor CFA to test the revised CF with an fg, and (4) bifactor ESEM to evaluate the fn structure in a less restrictive model. RESULTS: Model (1) required complex model restrictions to achieve convergence and had poor model fit(RMSEA = 0.37, CFI = 0.38), thereby leading to Model (2) which showed strong evidence for a unidimensional structure with a first factor explaining 58 % of the variance and a second factor explaining 9 %. Model (3) did not converge due to identification issues. Model (4) converged with acceptable model fit (RMSEA = 0.086, CFI = 0.97), providing support for a unidimensional structure with three groups of items whose fn structure was consistent with the CF. CONCLUSIONS: CFA was too restrictive, EFA was too simplistic, and the bifactor ESEM provided the needed flexibility to identify a unidimensional scale containing groups of items with associations consistent with the PR-SEQ’s hypothesized CF. A traditional approach using EFA or CFA would have led to the rejection of the hypothesized CF, but rather than concluding that item reduction is necessary, bifactor ESEM supported the hypothesized CF and provided additional insights to questionnaire use and enhancements.

(107.2) Improving the measurement of HRQL in dementia: a Rasch measurement theory approach Sarah C. Smith, PhD, London School of Hygiene & Tropical Medicine, London, United Kingdom, A. A. Jolijn Hendriks, PhD, London School of Hygiene & Tropical Medicine, London, United Kingdom; Theopisti Chrysanthaki, PhD, London School of Hygiene & Tropical Medicine, London, United Kingdom; Stefan Cano, PhD, Modus Outcomes, Stotfold, United Kingdom; Nick Black, PhD, London School of Hygiene and Tropical Medicine, London, United Kingdom

Qual Life Res (2015) 24:1–191 AIMS: DEMQOL (28 items) and DEMQOL-Proxy (31 items) were developed to measure HRQL in people with dementia using psychometric methods based on classical test theory (CTT) (Smith et al. 2005, 2007). Both instruments are widely used in research and are currently being considered for routine use by the Department of Health (England). Given the limitations of CTT, we aimed to develop improved scoring algorithms for both DEMQOL and DEMQOLProxy based on Rasch Measurement Theory (RMT) and identify other areas in which the instruments could be improved. METHODS: In a sample of 498 patients and family carers we used RMT to investigate for both instruments: adequacy of scale to sample targeting, how well the items work together as a measuring instrument (response thresholds, pattern of item locations, item fit, item dependency, DIF by age, gender, severity or relationship) and how well the instrument measures the people in the sample (person separation index (PSI) and person fit). Subsequent analysis attempted to resolve any anomalies and to derive Rasch based scores which were compared with the raw, summed total scores from the original CTT analysis. RESULTS: Initial analyses indicated that targeting was better for DEMQOLProxy than for DEMQOL, 5 DEMQOL and 11 DEMQOL-Proxy items had response options that were not working as intended, 2 DEMQOL and 3 DEMQOL-Proxy items mis-fitted (fit residual [ 2.5; Chi square and/or F-probabilities \0.01; ICC suggesting misfit), 1 DEMQOL and 5 DEMQOL-Proxy item pairs (mainly positive emotions) had item dependency (r [ 0.4) and 5 DEMQOL-Proxy items showed DIF (4 for severity, 1 for relationship). In a post hoc analysis to explore ways forward, we resolved these anomalies for both DEMQOL and DEMQOL-Proxy by dichotomising response options, removing the positive emotion items and 1 additional item. PSI was 0.72 (DEMQOL) and 0.76 (DEMQOL-Proxy). Test characteristic curves showed a clear S-shaped relationship, indicating improvement in the interval nature of the scales. CONCLUSIONS: We have improved the scoring algorithms for both DEMQOL and DEMQOLProxy using RMT. Aspects of both instruments require further investigation, including identifying the most appropriate response options and developing new items to improve the content coverage. Funding: This research was commissioned and funded by the Department of Health Policy Research Programme (Using Patient Reported Outcome Measures to Assess Quality of Life in Dementia, 0700071). The views expressed in this publication are those of the authors and not necessarily those of the Department of Health.

(107.3) PRO development with small samples: an innovative use of repeated measures IRT R. J. Wirth, Vector Psychometric Group, LLC, Chapel Hill, NC, United States; Carrie Houts, Vector Psychometric Group, LLC, Chapel Hill, NC, United States AIMS: This presentation will present scale development methods that optimize repeated measures data in conjunction with the use of current gold-standard psychometric analyses. The item response theory (IRT) models discussed are specifically implemented to help the development of patient reported outcomes for rare or hard to reach populations. METHODS: The statistical foundations and practical benefits of conditional, longitudinal IRT models are briefly detailed. Two scale development examples are provided. The first example will use simulated data to demonstrate the model’s ability to recover the ‘‘population’’ measurement structure using a series of sample sizes (N = 50–500) and model complexities (number of items = 10–40). The second example uses previously published data (N = 255, 10 items) to demonstrate these methods in practice. RESULTS: The results of the simulation show that the ability to regain the population measurement structure rests largely on the interaction between sample size and the number of time points. Overall, the model performs well.

27 The results of the real-world data example demonstrate that using conditional, longitudinal IRT models researchers can obtain the same study results (N = 255) with half of the PRO development sample (N = 125). CONCLUSIONS: The ability to obtain stable measurement characteristics for a set of items with smaller samples than are typically recommended when employing complex statistical models allows researchers to optimize their study designs and have increased confidence in subsequent comparisons using scores obtained from such models. This innovative use of longitudinal IRT may be especially beneficial when attempting to develop scales for rare or hard to reach populations.

(107.4) Grooming a CAT: customizing CAT administration rules to increase response efficiencies in specific research and clinical settings Michael A. Kallen, PhD, MPH, Northwestern University, Feinberg School of Medicine, Missouri City, TX, United States; Karon F. Cook, PhD, Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Richard C. Gershon, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: We studied PROMIS item bank CAT administration assessment histories to determine: efficiencies of ‘‘standard’’ CAT administration rules in limiting the number of items administered per CAT to achieve established score precision levels; potential for refining CAT rules to increase item-administration efficiencies while maintaining score precision. In the PROMIS 1 Wave 2 Back Pain/Depression Study, CATs were administered to N = 417 cases assessed across 11 PROMIS domains (baseline, up to two follow-up occasions). CAT administration rules were: start with a pre-identified item of moderate difficulty; administer a minimum four items per case; stop when an estimated theta’s SE declines to \0.3 OR a maximum 12 items are administered. METHODS: Original CAT: 12,622 CAT administrations were analyzed. CATs ranged in number of items administered from 4 to 12 items; 72.5 % were 4-item CATs. The second and third most frequently occurring CATs were 5-item (n = 1102; 8.7 %) and 12-item CATs (n = 964; 7.6 %). 64,062 items total were administered, averaging 5.1 items per CAT. Customized CAT: Three new CAT stopping rules were introduced, each with potential to increase item-presentation efficiency and maintain required score precision: Stop if a case responds to the first two items administered using the ‘‘lowest’’ response category (indicating clinically-unimportant status); administer a minimum two items per case; stop if the change in SE estimate (previous to current item administration) is positive but \0.01. RESULTS: The three new stopping rules reduced the total number of items administered by 25,643–38,419 items (40.0 % reduction). After four items were administered, only n = 1824 CATs (14.5 %) were still in assessment mode (vs. n = 3477 (27.5 %) in the original CATs). On average, cases completed 3.0 items per CAT (vs. 5.1). CONCLUSIONS: Each new rule addressed specific inefficiencies in the original CAT administration process: Cases not having or possessing a low/clinically unimportant level of the assessed domain; allow the SE \ 0.3 stopping criterion to come into effect earlier in the CAT administration process; cases experiencing poor domain item bank measurement, (e.g., ‘‘floor,’’ ‘‘ceiling’’ cases). Customizing CAT administration rules to better suit specific research study or clinical practice needs can decrease response burden while maintaining score precision standards.

(107.5) A mixture partial credit model for identifying latent classes responsible for differential item functionning

123

28

Qual Life Res (2015) 24:1–191

Jean-Franc¸ois Hamel, Assistant professor, University hospital of Angers, University of Nantes, Angers, France; Veronique Sebille, Professor of Biostatistics, University of Nantes, Faculty of Pharmacy, France, Nantes, France; Yves Roquelaure, Professor, University of Angers, Angers, France; Jean-Benoit Hardouin, PhD, University of Nantes, Nantes, France AIMS: Ideally, the item response probability to a quality of life (QoL) questionnaire should only depend on the respondents’ QoL level. If such a probability also depends on other characteristics such as ethnicity, gender or socioeconomic status, differential item functioning (DIF) may be present. Identifying DIF plays a key role in verifying measurement invariance when validating questionnaires. Moreover, being able to take into account DIF allows limiting measurement biases when analyzing patient-reported ouctomes (PRO) data. Several methodologies have been proposed for dealing with DIF, one of the most flexible and powerful being the IRT-based likelihood ratio test. With such a method, the covariate suspected to be responsible for DIF on a given item can be identified. For this purpose, nested models (with and without DIF) are compared: one constrained to be DIF-free for the suspected item, and one considering DIF by including interactions between the item parameter and the considered covariate. The best model is then chosen for analyzing data, allowing taking into account DIF if necessary. However, some problems can occur when covariates responsible for DIF are not well identified. Multiple covariates can be wrongly suspected, leading in multiple comparisons thus in type I error rate inflation. Moreover, the covariate truly responsible for DIF might not be identified because it is not a directly observed covariate but a latent variable. METHODS: We propose an adaptation of the IRT likelihood ratio test based on mixture partial credit models (PCM). With these models, items parameters are considered as fixed effects and both the latent trait to be analyzed (for example QoL) and the covariate responsible for DIF are considered as continuous and categorical latent variables, respectively. Latent classes can finally be constructed based on such categorical latent variables using individual posterior probabilities, and then described using observed data. RESULTS: We illustrate the properties of such likelihood ratio test based on mixture PCM using both simulated data and observed data from the Pays-de-la-Loire Workers Surveillance Program (France), and provide a MPlus based macro-program working under Stata for performing such a procedure. CONCLUSIONS: We believe that such program may facilitate the use of these methods by researchers.

(107.6) Applying a common depression metric to independent samples: different approaches result in similar score estimates Gregor Liegl, Mag., Charite - Universitaetsmedizin Berlin, Berlin, Germany; Inka Wahl, Dr. phil., University Medical Center HamburgEppendorf, Hamburg, Germany; Anne Berghoefer, Dr. med., Charite Universitaetsmedizin Berlin, Berlin, Germany; Sandra Nolte, PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Christoph Pieh, Dr. med., Danube University Krems, Krems, Austria; Matthias Rose, MD PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Felix Fischer, ScD, Charite - Universitaetsmedizin Berlin, Berlin, Germany AIMS: There is still little experience with the practical application of common metrics combining different Patient-Reported Outcomes (PROs) on a common scale. Using the example of the well-established depression scale of the Patient Health Questionnaire (PHQ-9), the present study was aimed at examining the performance of a recently published depression metric in independent samples and at comparing different methods of linking scores. METHODS: Using secondary data analysis, we applied a common metric model based on Item-Response Theory (IRT) presented by Wahl et al. (2014) to four

123

Fig. 1 .

Fig. 2 . German-speaking samples (n = 3315) that completed the PHQ-9. A Confirmatory Factor Analysis (CFA) was conducted in each sample to establish unidimensionality and local independence. We then fitted a Generalized Partial Credit Model (GPCM) with item parameters fixed to the parameters reported for the common depression metric. For each sample as well as for the full sample, we compared resulting parameters to new parameter estimates derived from two different model estimation approaches with unknown item parameters (estimation with shifted prior and Stocking-Lord linking). We investigated the agreement between latent depression scores (theta) resulting from the different estimation methods by fitting a mixed-effects model and using Bland–Altman plots. RESULTS: We found slightly different IRT item parameters across samples and estimation methods. Fixed to a German general population mean of 50 and a standard deviation of 10, the estimated depression score differences between the methods were about D = 1.0 in the different samples. While these differences were statistically significant, differences between samples were 16–130 times larger compared to the observed effects of the linking method. CONCLUSIONS: Using published item parameters from the common depression metric for theta estimation is clinically equivalent to the use of sample-specific item parameter estimates. These findings provide evidence that expected-a posteriori (EAP) scoring with unconstrained normal prior using a common metric is an appropriate alternative to re-estimation methods. Common metric score estimation is available for researchers at http://www.common-

Qual Life Res (2015) 24:1–191

29 reliability for the different mode of administration. Logistic regression was also performed to explore patients’ characteristics associated with the reported moderate/big urinary, bowel or sexual problem. The sensitivity and specificity of the telephone mode in reference to gold standard were further measured for each of the health domains. RESULTS: From 221 men of Victorian Prostate Cancer Registry who agreed to participate in the study, 168 (76.0 %) returned completed surveys. Kappa-linear model resulted in a moderate agreement across the urinary/bowel/sexual bother scores for both modes of administration; with greatest concordance recorded for bowel bother (90 %). Patient’s age (\85 years), disease risk and active treatment type determined a moderate-to-good level of agreement between administration modalities with a Kappa varying between 0.44 and 0.73, Chi2: 8.18, p = 0.042. Sensitivity tests revealed, that 68 % of men with a moderate/big problem during the phone interviews would respond of suffering from a moderate/big sexual problem. CONCLUSIONS: This was a pilot study to determine the differences between self-administered and telephone interviews in men with prostate cancer. A health-related outcome initiative has been recently established to combine data from two clinical registries, where one collects telephone-administered health related quality of life data and the other collects self-administered data. Understanding the differences will assist in the analysis of pooled data. Methods and results of this study provide a good understanding for future health related quality of life studies to be conducted using the registry data.

(108.2) Reliability and validity of PROMIS measures administered by telephone interview in a prospective, populationbased study of men with localized prostate cancer

Fig. 3 . metrics.org. The application of common metrics is simple and offers a long-term perspective to improve the comparability of PRO measures (Figs. 1, 2, 3).

108: Prostate Cancer (108.1) Comparison of patient-reported quality-of-life and complications in men with prostate cancer, between two modes of administration Rasa Ruseckaite, Monash University, Melbourne, Australia; Fanny Sampurno, Monash University, Melbourne, Australia; Jeremy L. Millar, Alfred Health, Melbourne, Australia; Sue M. Evans, Monash University, Melbourne, Australia AIMS: To (1) assess the level of consistency between the quality-oflife scores of men with prostate cancer for urinary/bowel/sexual bother, collected via telephone versus self-administered survey in Victoria, Australia; (2) determine factors associated with variation in level of agreement; and (3) assess the efficacy of telephone interview as a mode of administration against the ‘‘gold standard’’ tool, EPIC26. METHODS: Descriptive analysis was conducted to describe demographic and treatment characteristics of the study participants, grouped according to the National Comprehensive Cancer Network (NCCN) risk of disease progression categories. To assess the agreement between self-administered paper and telephone scores, Cohen’s Kappa coefficients were calculated for each of the three ‘‘bother’’ items in the questionnaire in order to investigate the test–retest

Caroleen W. Quach, MSPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Michelle M. Langer, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Deborah S. Usinger, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Ronald C. Chen, MD, MPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States AIMS: To evaluate the reliability and validity of seven Patient-Reported Outcomes Measurement Information System (PROMIS) measures (anxiety, depression, fatigue, pain intensity, pain interference, physical function, sleep disturbance) administered by telephone interviews in a longitudinal, population-based sample of prostate cancer patients METHODS: The North Carolina Prostate cancer Comparative Effectiveness and Survivorship Study worked with the state cancer registry to enroll a cohort of newly-diagnosed localized prostate cancer patients before treatment. PROMIS measures were telephone administered at baseline (pre-treatment), and at 3-, 12-, and 24-months post-treatment initiation. Reliability and validity of seven PROMIS measures were evaluated using the 12-months data (N = 778). Internal consistency was assessed using Cronbach’s alpha. Bifactor models and explained common variance (ECV) were used to evaluate dimensionality. Convergent and discriminant validity were assessed with Pearson correlations of PROMIS anxiety, depression, and physical function measures with legacy measures, the Memorial Anxiety Scale for Prostate Cancer and the SF-12v2. Known-groups validity was examined with t-tests of mean PROMIS scores for age and race/ethnicity groups based on evidence from the published literature. RESULTS: Each PROMIS measure was sufficiently unidimensional (ECV [ 0.90), with high internal consistency (0.86–0.96), and low missing data (\1 %). Floor effects ranged from 15 (sleep disturbance) to 58 % (anxiety). No ceiling effects were detected (range 0–1 %). Convergent validity was established with strong correlations between PROMIS and legacy measures, ranging

123

30 from 0.54 to 0.77. Discriminant validity was demonstrated with weak correlations between measures of different domains, ranging from 0.29 to 0.33. PROMIS measures were able to detect health-related quality of life (HRQoL) differences across age and racial/ethnic groups: younger men aged \65 reported more anxiety, depression, and sleep disturbance than men aged C65; African Americans reported more pain interference and worse physical function than non-Hispanic Whites (each p \ .05). CONCLUSIONS: This study provides support for the reliability and validity of seven PROMIS domain measures in men with prostate cancer. Results also support the utility of telephone-administered PROMIS measures, which is critically important for assessing HRQoL in low literacy and underserved populations who may not have internet access. Future research is needed to assess responsiveness of these measures for prostate cancer patients.

(108.3) Comparing the responsiveness of two health-related quality of life instruments in a phase III randomized clinical trial of men with prostate cancer (NCIC CTG PR.3): The EORTC QLQ-C30 + 3 with PR17 trial specific checklist versus the FACT-P Michael Brundage, MD, MSc, Kingston General Hospital, Queen’s University, Kingston, ON, Canada; Harriet Richardson, PhD, Queen’s University, Kingston, ON, Canada; Padraig Warde, MD, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Wendy Parulekar, MD, NCIC Clinical Trials Group, Queen’s University, Kingston, ON, Canada; Bingshu Chen, PhD, NCIC Clinical Trials Group, Queen’s University, Kingston, ON, Canada; Andrea Bezjak, MD, Princess Margaret Cancer Centre, Universtiy Health Network, Toronto, ON, Canada, Yvonne Murray, MSc, NCIC Clinical Trials Group, Queen’s University, Kingston, ON, Canada AIMS: The EORTC QLQ-C30 and the FACT are two commonly used Health-Related Quality of Life (HRQL) instruments in cancer clinical trials, but there is limited data comparing them. The NCIC CTG PR.3 clinical trial compared Androgen Deprivation Therapy (ADT) alone with ADT plus radiation therapy (ADT + RT) in prostate cancer patients. In a PR.3 sub-study, we conducted a comparison of the EORTC QLQ-C30 + 3 and prostate module (PR17) to FACT-P by employing a cluster randomization of participating North American centers to HRQL instrument used on the PR.3 clinical trial. The purpose was to compare the responsiveness of two HRQL instruments to short-term radiation effects and long-term hormone effects in men treated for locally advanced prostate cancer. METHODS: 311 patients were randomized to the PR.3 sub-study in 29 participating centers. HRQL was assessed at baseline, 6 monthly (for 2 years), then annually; compliance exceeded 85 % to 3 years. The ability of each HRQL instrument to detect RT toxicity was determined by comparing mean change scores (ADT vs. ADT + RT arms) at 6 months by HRQL instrument (Wilcoxon rank-sum). The ability of each instrument to detect proportions changed (at 6 or 36 months) was determined by calculating proportions (clinically meaningful change defined as 10 % change from baseline) then comparing between instrument groups (Chi square). Finally, we compared instruments on time to clinically meaningful worsening of HRQL using Kaplan– Meier survival curves/Cox regression. RESULTS: The FACT-P detected significant between-treatment arm differences in urinary symptom change scores at 6 months. The EORTC QLQ-C30 + 3/ PR17 detected significant between-treatment arm differences in diarrhea and bowel/rectum symptom changes at 6 months. For functional domains and fatigue, no significant between-instrument differences were observed in proportions of patients improved/stable

123

Qual Life Res (2015) 24:1–191 and worsened at and up to 36 months. However, the FACT-P reported a faster rate of clinically meaningful HRQL decline for physical and role/functional domains. CONCLUSIONS: When randomly assigned to patients participating in a clinical trial, the FACT-P and EORTC QLQ-C30 + 3/PR17 instruments differed in responsiveness to changes in urinary and bowel symptoms attributable to radiotherapy. The FACT-P was more responsive to change in physical and role function over time.

(108.4) Five-year follow-up after active surveillance or curative treatment: preliminary quality of life outcomes of men with prostate cancer Lionne Venderbos, Erasmus University Medical Center, Rotterdam, Netherlands; Monique Roobol, Erasmus University Medical Center, Rotterdam, Netherlands; Chris Bangma, Erasmus University Medical Center, Rotterdam, Netherlands; Ida Korfage, PhD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands AIMS: A recent systematic review showed that men on active surveillance (AS) for prostate cancer (PCa) usually reported good levels of well-being and that they did not appear to suffer major negative psychological impacts of their disease. Whether AS causes anxiety and distress in the long-term is currently not known. With this study we aim to assess 5-year QoL of men on AS in the Prostate cancer International: Active Surveillance (PRIAS) study, which is an ongoing, prospective, observational cohort study, and compare the results to empirically collected data on QoL of two groups: (1) men who underwent radical prostatectomy (RP) and (2) men who did not have PCa. Here we will present some first preliminary outcomes. METHODS: Men with C4 years of follow-up in the PRIAS study were invited. In order to compare our results to men who have undergone curative treatment, we invited men who underwent RP between 2008 and 2010 in the context of the European Randomized study of Screening for Prostate Cancer (ERSPC) and a reference group of men without PCa whom where age matched to the AS and RP group. All men were sent a QoL questionnaire which included the SF-12, EQ-VAS, STAI-6 (general anxiety), EPIC (urinary, bowel and sexual function) and the MAX-PC (PCa-related anxiety). RESULTS: Response rates for the AS, RP and reference group amount to 62 (120/ 193), 52 (47/90) and 72 % (204/281) respectively. Median age for men in the AS, RP and control arm was 73, 76, and 74 respectively. The median SF-12 MCS and PCS scores amounted to 54.9 and 52.2 (AS); 49.9 and 45.3 (RP); 53.2 and 50.3 (control). On the EQ-VAS median health scores were 85 (range 55–100) for AS, 80 (range 30–100) for RP and 80 (range 30–100) for the control group. General anxiety scores (STAI-6), so far, seem comparable; all group show a median score of 30. CONCLUSIONS: So far, general health of men on AS versus men who underwent RP seems comparable after 5 years of follow-up. The majority of men in the AS and RP groups do not seem to be anxious.

(108.5) Defining quality of life outcomes for advanced prostate cancer trials: qualitative interview results compared with HRQL instruments in current use Michael J. Fabricius, MBBS BSc MRCS, Newcastle University, Newcastle upon Tyne, United Kingdom; Robert Pickard, MD FRCS (Urol), Newcastle University, Newcastle upon Tyne, United Kingdom; Elaine McColl, PhD, Newcastle University, Newcastle upon Tyne, United Kingdom

Qual Life Res (2015) 24:1–191 AIMS: Health related quality of life (HRQL) is an important outcome in the treatment of advanced prostate cancer (PC). PC-specific HRQL instruments are available, but these are not specific to advanced PC and assess lower urinary tract (LUTS), sexual function and bowel symptoms which are typically experienced following surgery or radiotherapy to treat localised PC and may not be relevant to patients with advanced disease. This study aims to compile a list HRQL factors that are important for patients with advanced PC, prior to refining this list using a Delphi consensus process and a large population of patients. The findings will provide a patient-centred, datadriven approach to selecting appropriate patient-reported outcome instruments for advanced PC trials. METHODS: We recruited participants with advanced (metastatic or castrate-resistant) PC from the oncology clinic and invited them to participate in a 1-h semi structured interview. Purposive sampling ensured that participants had experienced the full range of available treatments. Initial sample size was set at 15 with assessment for data saturation at that point. The participants’ experience of prostate cancer treatment was explored, including disease and treatment-related symptoms, psychological factors and the impact of their disease on all aspects of HRQL. Participants were also asked to discuss what defined HRQL for them. We performed thematic analysis of interview transcripts using inductive coding with Nvivo software. Current HRQL instruments were listed from a recent systematic review. RESULTS: The most commonly cited HRQL factor was pain, followed by factors in the role and social functioning domains including the ability to carry out normal activities (hobbies, work, responsibilities) and to continue relationships with family or friends. Physical function including the ability to walk or travel was also widely valued. Participants commonly described cancer-related symptoms including, lethargy, nausea and anorexia. Participants who described loss of libido and erectile dysfunction did not feel it impacted QoL. No patient described bothersome LUTS or bowel symptoms. CONCLUSIONS: Existing PC-specific HRQL instruments assess factors that have not been cited as important by this study sample, and may impose unnecessary burden on trial participants. This conclusion requires validation by the planned consensus process using a large cohort of patients.

109: Mental Health (109.1) Patient knows best: patient-reported functionality predicts course of clinical depression Carlos G G. Forero, Researcher, CIBER in Epidemiology y SAlud Pu´blica (CIBERESP), Barcelona, Spain; Elena Olariu, Predoc Student, Hospital del Mar Medical Research Insitute (IMIM), Barcelona, Spain; Jose Ignacio Castro-Rodrı´guez, Psychiatrist, Institute of Neuropsychiatry and Addictions (INAD), Parc de Salut Mar, Barcelona, Spain; Gemma Vilagut, Analyst, Hospital del Mar Medical Research Insitute (IMIM), Barcelona, Spain; Angels Pont, BSc, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain;CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP). Barcelona, Spain; Jordi Alonso, Head, Hospital del Mar Medical Research Institute, Barcelona, Spain AIMS: Functional disability (FD) is a compulsory criterion for psychiatric diagnosis in Major Depressive Episode (MDE). Recently issued DSM5 moved FD assessment from clinician-rated to patientreported, but whether any of these approximations is associated with disorder course remains uncertain. Here, we evaluate the ability of early (3-month) patient functioning change to predict 12-month depression episode course. METHODS: 139 patients seeking professional help for mood/anxiety symptoms were followed during 12 months with 4 assessments (baseline, and 1-, 3- and 12-months after baseline). At each moment, MDE status fulfilment of DSM5

31 symptom criteria), and FD was assessed clinician-rated with Global Assessment of Functioning (GAF: 1–100 scale, worst to best FD) and patient-reported with an Analogue Scale of Functioning (ASF: 1–100 scale, worst to best), and World Health Organization Disability Assessment Schedule (WHODAS 2.0: 12–60 scale, best to worst). We compared FD measures predictive ability during the first three assessments as time-varying predictors using GEE models with MDE status at 12-months as binary outcome (present/absent). Models were adjusted by sex, age, physical comorbidities, baseline severity (PHQ9) and MDE 3-month assessment (present/absent). FD effects and their interaction with time were tested with Wald’s chi2 test, and effect sizes with exponentiated parameters (OR-analogue). We determined predictive ability of FD measures using ROC analysis, using GEE predicted probability of 12-month MDE disorder status, computing sensitivity and specificity at the maximum Youden index value. RESULTS: Early FD measures were significant predictors of 12-month MDE status (GAF Chi2 = 7.33, p = 0.007, OR = 0.98; SAF Chi2 = 11.26, p = 0.001, OR = 0.97; WHODAS Chi2 = 12.65, p \ 0.001, OR = 1.05). Patient-reported disability assessed by the ASF was the only measure significantly interacting with time in predicting 12-month MDE (Chi2 = 5.65 p = 0.05). ROC analysis showed models with self-reported measures better predicted MDE disorder status (ASF: AUC = 0.79, Sens/Spec = 0.82/0.70; WHODAS: AUC = 0.81, Sens/Spec = 0.83/0.70), than clinician-rated GAF (AUC = 0.76, Sens/Spec = 0.93/0.61). CONCLUSIONS: Early change in self-reported FD predicts 12-month patient depression status as accurately as a general questionnaire. Results highlight the feasibility of using very few information based on simple patient-reported measures of functioning in clinical monitoring of MDE patients. Funding: funded by grants ISCII(PI13/00506) and ISCIII(PI10/ 00530) and DIUE of Generalitat de Catalunya (2014SGR748).

(109.2) Development of eight item banks to assess specific mental health problems Philip Batterham, PhD, The Australian National University, Acton, ACT, Australia; Matthew Sunderland, PhD, University of New South Wales, Randwick, Australia; Natacha Carragher, PhD, University of New South Wales, Randwick, Australia; Alison Calear, PhD, The Australian National University, Acton, Australia AIMS: This study aimed to develop item banks to assess eight specific mental health problems in the general population: social anxiety disorder, panic disorder, post-traumatic stress disorder, obsessive compulsive disorder, adult attention-deficit hyperactivity disorder, substance use disorder, suicide risk and psychosis. From these item banks, static and adaptive screeners were developed, with the aim of providing more brief and accurate assessment of mental health than existing screening measures. METHODS: A systematic process of item selection was conducted to develop each of the item banks. This process included: (1) collection of existing items through literature searches, (2) standardization of items to a common frame, (3) removal of duplicate items, and (4) collecting consumer and expert ratings of item relevance. After this process was complete, the remaining pool of items were calibrated in a community-based sample of 3175 Australian adults. From each item bank, short screeners assessing symptom severity were developed. RESULTS: Starting with between 464 and 1347 items in each pool, the item selection process reduced the pools to between 45 and 79 items that were then calibrated in the community-based sample. Development of final item banks from these data were based on modification indices and lack of significant differential item functioning by age, gender and educational attainment. Final item banks are currently being developed, with banks ranging from 19 to 25 items. Information

123

32 curves demonstrated that the new item banks provide more information than conventional measures in a range of severity from approximately -1 to +3 standard deviations (where 0 is the population mean). Short form screeners of 3–5 items provided greater information than existing measures containing more items, and were at least as accurate in assessing clinical caseness. CONCLUSIONS: This suite of item banks, when combined with existing item banks for depression, anxiety, anger and alcohol use, has great potential for rapid assessment of a broad range of mental health problems in a number of settings. The measures will also enable greater understanding of transdiagnostic patterns of mental health symptomatology, better accounting for the high rates of comorbidity between disorders.

(109.3) Impact of smoking, physical inactivity, heavy drinking, and obesity on health-related quality of life, life expectancy, and quality-adjusted life expectancy among adults with and without depression in the U.S. Haomiao Jia, Columbia University, New York, NY, United States; Matthew M. Zack, Centers for Disease Control and Prevention, Atlanta, GA, United States; William W. Thompson, PhD, U.S. Centers for Disease Control and Prevention, Atlanta, GA, United States AIMS: Depression is a leading cause of both poor health-related quality of life (HRQOL) and lower life expectancy. Depression is also associated with many risky behaviors. This study estimated the health impact of four modifiable risky behaviors—smoking, physical inactivity, heavy alcohol drinking, and obesity—on HRQOL, life expectancy, and quality-adjusted life expectancy (QALE) among adults with and without depression in the U.S. METHODS: We ascertained depressive symptoms using the 8-item Patient Health Questionnaire (PHQ 8) for data from the 2006, 2008, and 2010 Behavioral Risk Factor Surveillance System (BRFSS) respondents. HRQOL was estimated from the EQ 5D index scores based on BRFSS data, and constructed life tables from the U.S. Compressed Mortality Files were used to calculate life expectancy and QALE for individuals by depression status. These data were then cross-tabulated by each of the four risky behaviors (smoking, physical inactivity, heavy alcohol drinking, and obesity). RESULTS: Among depressed adults, smoking contributed a 0.040-point decrease (6.6 %) in the age-adjusted EQ-5D index, 12.9 years of life lost (24.3 %) from age 18, and 8.6 years of QALE loss (27 %). Physical inactivity contributed a 0.171-point decrease (25.6 %) in EQ-5D, 10.8 years of life lost (20.1 %), and 10.9 years of QALE loss (32 %). Heavy drinkers had a higher EQ-5D index (0.088-point higher, 15 %), a lower life expectancy (6.4-year lower, 13.3 %), and a non-statistically significant 0.7-year QALE loss. Obesity contributed a 0.053-point decrease (8.6 %) in EQ-5D, 0.4 years of life lost (0.9 %), and 1.7 years of QALE loss (5.8 %). At the population level, if the prevalence of physical inactivity among adults with depression (43.5 %) were reduced to the same level as adults without depression (20.6 %), depressed adults would gain 2.8 years of QALE. Similarly, if the prevalence of smoking among depressed adults (39.9 %) were reduced to that of adults without depression (19.5 %), depressed adults would gain 1.9 years of QALE (6.8 %). CONCLUSIONS: In this study, physical inactivity and smoking were shown to differentially impact the health of U.S. adults with depression relative to adults without depression. Obesity and heavy alcohol drinking had only a small or a non-statistically significant impact on respondents’ QALE.

(109.4) Development of Dutch computer adaptive tests for anxiety and depression using the PROMIS item banks

123

Qual Life Res (2015) 24:1–191 Gerard Flens, Stichting Benchmark GGZ (SBG) [Benchmark Foundation Mental Health Care], Bilthoven, Netherlands; Niels Smits, Department of Clinical Psychology and Department of Methodology, the EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, Netherlands; Caroline B. Terwee, VU University Medical Center, Amsterdam, Netherlands; Edwin de Beurs, Stichting Benchmark GGZ (SBG) [Benchmark Foundation Mental Health Care], Bilthoven, Netherlands AIMS: In the Netherlands, Routine Outcome Monitoring (ROM) has been implemented for mental health care patients nationwide. To limit the burden on patients, there is a need for measurement instruments that require less administration time without compromising their reliability. Computer Adaptive Testing (CAT) may address this need. We undertook a study to evaluate and calibrate the Dutch PROMIS item banks for Anxiety and Depression as input for CATs. METHODS: As item banks, we used the Dutch-Flemish translation of the Patient Reported Outcome Measurement Information System (PROMIS) adult item banks V1.0 for Anxiety (29 items) and V1.0 for Depression (28 items). In addition, we translated 26 US anxiety items and 28 US depression items that didn’t make it to the final PROMIS item bank for failing the Item Response Theory (IRT) assumptions. After a comprehensive psychometric evaluation of the item banks in a patient (n = 1008) and general population sample (n = 1002), we performed four item calibrations; two for each latent trait (anxiety and depression), and per trait, one for the original PROMIS item bank and one for the extended PROMIS item bank. Using the item calibrations, we performed four CAT simulations on patient data with various stopping rules to assess the efficiency for each item calibration. RESULTS: The psychometric evaluation of the item banks (unidimensionality, local independence, monotonicity, model and item parameter evaluation, Differential Item Functioning) resulted in eliminating 4 of the additional translated depression items and 1 additional translated anxiety item. With the final item banks, CAT simulations were performed. The results showed similar patterns for anxiety and depression. The extended PROMIS item banks were slightly more efficient than the original PROMIS item banks. In contrast, the CAT simulation scores of the extended PROMIS item banks showed somewhat lower correlations with the total scale score than the original PROMIS item banks. CONCLUSIONS: This study suggests that developing CATs for anxiety and for depression for Dutch adults can be useful as it leads to efficient measurement without compromising its reliability. We discuss the implications of using each item bank and stopping rule for a CAT application of anxiety and depression.

(109.5) The BELLA study: first results from the longitudinal mental health module of the German National Health Interview and Examination Survey among Children and Adolescents Anne-Catherine Haller, Dipl.-Psych., University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Fionna Klasen, PhD, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Christiane Otto, PhD, University Medical Center HamburgEppendorf, Hamburg, Germany; Sandra Nolte, PhD, Charite Universitaetsmedizin Berlin, Berlin, Germany; Annett Mierke, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Dana Barthel, Dipl.-Psych., University Medical Center HamburgEppendorf, Hamburg, Germany; Ute Ellert, Robert Koch-Institute, Berlin, Germany; Laura Schlepper, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Kathrin I. Fischer, M.Sc.med., Charite - Universitaetsmedizin Berlin, Berlin, Germany; Heike Hoelling, Robert Koch-Institute, Berlin, Germany; Otto Walter, PhD, MD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Matthias Rose, Charite - Universitaetsmedizin Berlin, Berlin,

Qual Life Res (2015) 24:1–191 Germany; Ulrike Ravens-Sieberer, PhD, MPH, University Medical Center Hamburg-Eppendorf, Hamburg, Germany AIMS: The longitudinal BELLA study is the mental health module of the German National Health Interview and Examination Survey among Children and Adolescents and collects comprehensive data on well-being, quality of life and mental health of children and adolescents in Germany since 2003. The BELLA study design allows the analyses of individual developmental trajectories of health-related quality of life and mental health from childhood into adulthood and the investigation of related risk and protective factors. METHODS: BELLA examined well-being, quality of life and mental health in a representative sample of families with children aged 7–17 years by means of computer-assisted telephone interviews and subsequent questionnaires through self-report and parent-report. Of the 2863 participants included in the BELLA study at baseline, 2423 (84.6 %) participated in the 1-year follow-up, 2190 (76.5 %) participated in the 2-year follow-up and 1429 (49.9 %) in the 6-years follow up. A total of 1255 (43.84 %) children and adolescents took part at all four BELLA measurement points. The latest follow-up examination (2015–2017) is designed as an online questionnaire and includes the first-time use of the Kids-CAT—a computer-adapted test for healthrelated quality of life in children and adolescents—in a population based sample. Data analyses include multivariate and longitudinal growth modelling methods. RESULTS: The overall findings demonstrate an overall high level of health-related quality of life. They also indicate a higher likelihood to develop mental health issues between the age of 7 and 12 years and after the age of 19 years. Among all children and adolescents participating at the first four measurement points, about 10 % showed clinically significant mental health problems at each measurement point, with a third of the cases at baseline also reporting mental health problems at the 6-year followup. Determinants of mental health and health-related quality of life of children and adolescents will be discussed and the latest follow-up examination will be presented. CONCLUSIONS: The BELLA study provides important information on young people’s well-being, quality of life and mental health in Germany. It highlights the importance of strengthening resources in the personal, familial and social domains as an integral part of prevention and treatment interventions.

110: Population-based Studies (110.1) Five year trajectories of health-related quality of life in Canada’s poorest postal code Skye P. Barbic, Post-Doctoral Fellow, Faculty of Medicine, Vancouver, BC, Canada; Ana Maria Rodrı´guez, PHD, McGill University, Montreal, QC, Canada; Tari Buchanan, Univeristy of British Columbia, Vancouver, BC, Canada; William G. MacEwan, St. Paul’s Hospital, Vancouver, Canada; William G. Honer, Univerisity of British Columbia, Vancouver, BC, Canada AIMS: Health-Related Quality of Life (HRQL) has been well studied across health conditions. Few studies have explored the construct from the perspective of individuals who are homeless or living in marginal housing conditions. The overall aim of this study was to contribute to the understanding of the marginally housed experience by evaluating the dynamics of HRQL over time. A secondary objective was to measure the extent to which the sociodemographic profile of marginally housed individuals impacts HRQL over time. METHODS: Between 2009 and 2014, we recruited participants from four single room occupancy hotels (SROs) in in Vancouver, Canada. We assessed participants monthly (for up to 60 months) including a detailed review of substance use, physical, psychological and social health. Using the SF-36 Physical and Mental Component Summary scores, we measured overall Physical and Mental Health. We used

33 group-based trajectory analysis (GBTA) to identify distinctive groups of individuals with similar trajectories and to estimate the extent to which age, sex, ethnicity, and the years spent marginally housed predict an individual’s membership in a trajectory. RESULTS: From the sample of 402 people (mean age 47 ± 11 years), six trajectories of Physical Health were identified, five of which were stable across time. None of the Physical Health trajectories scored higher than 60 (maximum score is 100). Similarly, five stable trajectories of Mental Health were identified. Age and aboriginal ethnicity had a detrimental effect on Physical Health group trajectory membership, as older people and being aboriginal were associated with a lower likelihood of being in a high Physical Health trajectory. Age also had a significant effect on Mental Health group trajectory membership, however the effect was protective, as older individuals showed a greater likelihood of being in a high Mental Health trajectory. CONCLUSIONS: Characterizing HRQL in this population would enable health agencies to legitimately address broader areas of health policy and initiate strategic collaborations with health and community partners. A greater understanding of the ‘‘positive’’ dimensions of health in this population may lend evidence towards how services can be orientated to meet the ‘‘health’’ needs of this group, rather than solely the illness management and survival requirements.

(110.2) Exploring the relationships between health-related quality of life and health conditions, costs, resource utilization and quality measures Eric Havens, MA, Humana, Knoxville, TN, United States; Jonathan Pena, MS, Humana, Louisville, KY, United States, Lane Slabaugh, PharmD,MBA, Humana, Louisville, KY, United States; Tristan Cordier, MPH, Humana, Atlanta, GA, United States; Andrew Renda, MD,MPH, Humana, Louisville, KY, United States; Vipin Gopal, PhD,MBA, Humana, Louisville, KY, United States AIMS: Healthy Days (HD) is a validated health-related quality of life instrument consisting of 4 survey questions developed by the U.S. Center for Disease Control and Prevention. This study of persons with commercial health insurance from a large national U.S. insurer examined the relationship between HD responses and a variety of health-related measures such as condition prevalence, medical and pharmacy costs, healthcare resource utilization (HRU), and compliance to Healthcare Effectiveness Data and Information Set (HEDIS) quality measures. METHODS: Surveys were completed by 47,910 adults between November and December 2014 who were insured by a commercial health plan. Those who completed a survey and were continuously enrolled in their health plan in 2014 were included in the analysis. HD data were merged with medical claims, pharmacy claims, and HEDIS data at the individual level. The primary outcome was an index of total unhealthy days, calculated from responses to two HD questions which ask people to recall the number of physically and mentally unhealthy days (UHD) within the past 30 days. Prevalence of six conditions (coronary artery disease [CAD], congestive heart failure [CHF], chronic obstructive pulmonary disease [COPD], diabetes, hypertension and depression) was examined among the study population for significant correlations to reported UHD using Wald Chi squares. T-tests were calculated to compare UHD to healthcare costs and utilization. HEDIS data were analyzed descriptively. RESULTS: A total of 22,667 people were eligible for the study (53 % female; mean age 47.5 years). A significant correlation was observed between prevalence of the examined conditions and UHD (p \ 0.001 for all six conditions). People with chronic conditions reported more unhealthy days than those without (Fig. 1). Medical and pharmacy costs, as well as HRU, significantly increased as the number of unhealthy days increased (p \ 0.001 for all cost and HRU variables). Among the HEDIS measures descriptively analyzed, no

123

34 clear relationships to UHD were observed. CONCLUSIONS: HD is well correlated to many important objective measures of health for commercial enrollees. Additional research is needed to determine if a causal relationship exists, and to determine if interventions resulting in improvements in HD result in corresponding improvements in these measures.

(110.3) Health-related quality of life in incident, established, and long-term cases of chronic illness: results from a populationbased survey Lucy Busija, Australian Catholic University, Melbourne, Victoria, Australia; Kerrie Sanders, Australian Catholic University, Melbourne, Victoria, Australia; Daniel O’Connor, Australian Catholic University, Melbourne, Victoria, Australia; Geoff Nicholson, Australian Catholic University, Melbourne, Victoria, Australia; Marita McCabe, Australian Catholic University, Melbourne, Victoria, Australia; Catherine Connaughton, Australian Catholic University, Melbourne, Victoria, Australia AIMS: The aim of this study was to describe loss of health-related quality of life (HRQoL) in incident, mid-term, and long-term mental and physical chronic health conditions, as compared with those without health conditions. METHODS: Data from the Australian National Health and Mental Wellbeing Survey 2007/08 were used to examine HRQoL in asthma, heart disease, arthritis, diabetes, bipolar disorder, depression, and anxiety. Physical conditions were ascertained through self-report of prior medical diagnosis. Mental health conditions are DSM-IV lifetime diagnoses, ascertained through Composite International Diagnostic Interview Schedule. HRQoL was captured by the Assessment of Quality of Life Scale 4D Global Utility score (0 = worst HRQoL, 1 = full HRQoL). Analysis of covariance with Sidak correction compared HRQoL of those with incident (1 year or less from initial diagnosis), established/mid-term ([1–5 years from diagnosis), and long-term ([5 years) conditions with those who had no condition, after controlling for age and gender. RESULTS: The 8841 respondents were aged 16–85 years (median 43 years, 50.3 % female). Anxiety was the most prevalent condition (n = 2727, 30.8 %), followed by heart disease (n = 1871, 21.2 %), arthritis (n = 1759, 19.9 %), and asthma (n = 1736, 19.6 %). Compared with those who did not have the selected health conditions, the largest initial losses in the mean HRQoL occurred in bipolar disorder, depression, and asthma, with 11, 10, and 7 % decline, respectively (Fig. 1). In heart disease and asthma, HRQoL declined for incident

Fig. 1 Mean health-related quality of life for individuals with incident (B1 year since diagnosis), mid-term ([1–5 years post diagnosis), and long-term ([5 years post diagnosis) health conditions, compared with those who do not have the selected health conditions (N = 8841)

123

Qual Life Res (2015) 24:1–191 cases (p \ 0.001) relative to those without these conditions, with an improvement for mid-term cases (p \ 0.05) and no further significant change for long-term cases. Compared with those without the condition, HRQoL in arthritis, diabetes, and bipolar disorder declined for incident cases (p \ 0.001), with no change for mid- and long-term cases. Anxiety showed a linear trend for lower HRQoL for cases with longer disease duration (p \ 0.001). In depression, HRQoL declined for incident cases (p \ 0.001), followed by an improvement for midterm cases (p \ 0.001), and a further decline long-term (p \ 0.001), compared with the ‘no condition’ group. CONCLUSIONS: Chronic illness has immediate and sustained impact on HRQoL, with only asthma and heart disease showing some recovery with longer disease duration. Mood disorders are associated with the largest initial loss of HRQoL.

(110.4) Quality of life independently predicts long-term vascular events and mortality: The Northern Manhattan Study Mandip S. Dhamoon, MD, MPH, Icahn School of Medicine at Mount Sinai, New York, NY, United States; Consuelo McLaughlin-Mora, MS, Columbia University, New York, NY, United States; Joshua Z. Willey, MD, MS, Columbia University, New York, NY, United States; Ying Kuen Cheung, PhD, Columbia University, New York, NY, United States; Yeseon P. Moon, MS, Columbia University, New York, NY, United States; Ralph L. Sacco, MD, MS, University of Miami, Miami, FL, United States; Mitchell S. Elkind, Columbia University, New York, NY, United States AIMS: Quality of life (QOL) has been previously associated with mortality, but with incomplete control for depression, cognitive status, and social support. Also, the relationship between QOL and subsequent vascular events is uncertain. We hypothesized that baseline QOL is independently associated with long-term mortality and vascular events in a large, multi-ethnic urban cohort. METHODS: In the prospective, population-based Northern Manhattan Study, participants had baseline assessment including the Spitzer QOL index (SQI, range 0–10, with 10 signifying the highest QOL). Participants were followed over a median of 11 years for stroke, myocardial infarction (MI), non-vascular and vascular death, and all events were dated and adjudicated. Cumulative risks were calculated for each outcome separately and for composite outcomes, overall and stratified by SQI (10 vs. \10), and difference was tested by the log-rank test. Cox proportional hazards regression was performed for each outcome, with SQI as the main predictor, analyzed as a continuous variable and dichotomized at 10 in separate models. Models were adjusted for baseline demographics, vascular risk factors, history of cancer, social support, cognitive status, and depression. RESULTS: Among 3298 participants, mean age was 69.7 (SD 10.3), 37 % were male, 52 % Hispanic, 32 % married, 74 % had hypertension, 22 % diabetes, 21 % coronary artery disease, and 4 % depression. Mean SQI was 9.1 (SD 1.3), and 1795 (54.4 %) had SQI of 10. Cumulative risk probabilities were greater among those with SQI \ 10 (compared to SQI = 10) for all-cause mortality, vascular death, non-vascular death, stroke or MI or death (all p \ 0.0001), and stroke or MI (p = 0.048). In fully adjusted Cox models, SQI of 10 (compared to SQI \ 10) was associated with reduced risk of all-cause mortality (HR 0.76, 95 % CI 0.69–0.85), vascular death (0.77, 0.65–0.91), nonvascular death (0.75, 0.64–0.86), stroke or MI (0.80, 0.65–0.98), and stroke or MI or death (0.74, 0.65–0.83). Results were consistent when SQI was analyzed continuously. CONCLUSIONS: In this large population-based cohort, baseline QOL was inversely associated with risk of vascular events and mortality over the long term, independently of medical conditions, social support, cognition, and depression. QOL, a patient-centered outcome, is a robust predictor of vascular events and mortality.

Qual Life Res (2015) 24:1–191

35

(110.5) 1000 Norms Project: health-related quality of life across the lifespan Jennifer N. Baldwin, PhD Student, The University of Sydney, Lidcombe, NSW, Australia; Marnee McKay, PhD Student, The University of Sydney, Lidcombe, NSW, Australia; Claire Hiller, PhD, The University of Sydney, Lidcombe, NSW, Australia; Elizabeth J. Nightingale, PhD, The University of Sydney, Lidcombe, NSW, Australia; Niamh Moloney, PhD, The University of Sydney, Macquarie University, Macquarie Park, NSW, Australia; Joshua Burns, PhD, The University of Sydney, Sydney Children’s Hospital Network, Lidcombe, NSW, Australia AIMS: Patient-reported outcomes (PROs) provide valuable information regarding health-related quality of life (HRQOL) in healthcare, clinical trials and economic evaluation of health services. The aims of this study were to generate HRQOL normative reference data and investigate the relationship between HRQOL, socio-demographic factors and physical function in a sample of healthy children and adults. METHODS: This study was conducted within the 1000 Norms Project, an observational study investigating objective and subjective measures of health and function in 1000 healthy Australians aged 3–101 years. Data for adult participants were analysed. Age, weight, height, waist circumference, highest educational level and marital status were collected. Physical activity level, workability and measures of physical function were also collected (Table 1). HRQOL was investigated using the AQoL-8D utility instrument assessing eight dimensions: independent living, pain, physical senses, self-worth, relationships, coping, happiness and mental health. Scores were converted to two super-dimensions (physical and mental) and a global ‘utility’ total score. RESULTS: To date data from 516 participants aged 18–91 years have been analysed. Adults aged 80+ years reported significantly lower scores for the independent living and senses dimensions and the physical super-dimension (p \ .001). There was no correlation between age and AQoL-8D utility, mental super-dimension, or relationships, coping, mental health or happiness scores (p [ .05) (Fig. 1). No significant associations were found between AQoL-8D utility scores and gender, marital status, education or

Fig. 1 Median scores (±SE) for the AQoL-8D utility (total) score, physical and mental super-dimensions for the different adult age groups. *Statistically significant difference (p \ .001)

Table 1 Physical activity, workability and measures of physical function collected in the study

Fig. 2 Scatterplot showing relationship between physical superdimension scores and 6-min walking test results

Item

Description

Physical activity

Physical activity in the past seven days assessed by the International Physical Activity Questionnaire (IPAQ) long version converted to a category score (low, moderate, high) [1]

Workability

Single item from the Workability Index measuring self-reported workability on a scale from 0 to 10 [2]

Six-minute walk test

Distance (in metres) covered in six minutes walking as quickly as possible

physical activity (p [ .05). Workability was associated with AQoL8D utility and both super-dimension scores (rho = .266, .354 and .159, p \ .001). Physical super-dimension scores were negatively correlated with BMI, waist circumference and weight (rho = -.290, -.293 and -.155, p \ .001), and positively with 6-min walking distance (Fig. 2), timed up and down stairs, chair stand, and BOT balance scores (rho = .404, .410, .332, .342, p \ .001). CONCLUSIONS: Physical super-dimension scores decreased in later life from 80 years. Mental super-dimension scores remained unchanged yet were surprisingly low across the sample. Low perceived workability, higher body mass, and poorer lower limb function were associated with lower physical super-dimension scores. By September 2015 data from all 1000 participants, including children and adolescents, will be available.

Timed up and down stairs test Time taken to quickly and safely ascend and descend a flight of stairs 30-second chair stand test

Number of full sit-to-stands from a chair performed in 30 s

Bruininks-Oseretsky test of 9-items assessing balance, summed to motor proficiency (BOT-2) give a total score from 0–37 points

201: Multiple Sclerosis (201.1) Development of a multidimensional computerized adaptive test for measuring quality of life for patients with multiple sclerosis: the MuSiQoL-MCAT Pierre Michel, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille,

123

36

Qual Life Res (2015) 24:1–191

France; Karine Baumstarck, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Pascal Auquier, EA 3279 SelfPerceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Jean Pelletier, Departments of Neurology and CRMBM CNRS6612, Timone University Hospital, APHM, Marseille, France; Anderson Loundou, 1. Aix-Marseille University, EA 3279 – Public Health, Chronic Diseases and Quality of Life - Research Unit, 13005, Marseille, France; Mohamed Boucekine, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Badih Ghattas, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Laurent Boyer, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France AIMS: Multidimensional computerized adaptive testing (MCAT) based on multidimensional item response theory (MIRT) models can improve the measurement precision of QoL instruments while reducing the length of the questionnaire and the completion time. The aim of this study is to develop the MusiQoL-MCAT for patients with multiple sclerosis (MS). METHODS: Patients with MS from 15 countries were enrolled in this international, multicenter, cross-sectional study. The development of the MusiQoLMCAT was based on the assessment of between-items MIRT model fit to our data, followed by real-data simulations. The MCAT algorithm was based on Bayesian maximum a posteriori estimation of latent traits and Kullback–Leibler information item selection. The starting item was the most informative. We examined four simulations based on a fixed number of items (5, 10, 15, 20). Accuracy was assessed using correlations between initial IRT scores and MCAT scores (r). Precision was assessed using the standard error of measurement (SEM) and the root mean square error (RMSE). Validity of the MusiQoLMCAT was examined according to the socio-demographic, clinical and QoL features. RESULTS: A total of 1992 patients were enrolled. The 9-dimension structure of the MuSiQoL was confirmed using multidimensional confirmatory factor analysis (RMSEA = 0.067, CFI = 0.950, SRMR = 0.042). The multidimensional graded response model (MRGM) yielded a better fit than the multidimensional generalized partial credit model (MGPCM) and was used to estimate item parameters and IRT scores. Among MCAT simulations, the 16-item version of the MusiQoL-MCAT was selected because the accuracy and precision became stable from 16 items with satisfactory levels (r = 0.9, SEM = 0.55 and RMSE = 0.3). External validity testing revealed that the MusiQoL-MCAT correlated significantly with QoL scores but also showed discriminant

Table 1 continued Item a1

a2

a3

a4

a5

a6

a7

a8

a9

d1

d2

-1.83 -0.77

d3

d4

10

–

3.47 –

–

–

–

–

–

–

11

–

2.25 –

–

–

–

–

–

–

-1.87 -0.55

0.58

1.61

12

–

2.21 –

–

–

–

–

–

–

-1.99 -0.89

0.19

1.10

13

–

–

2.50 –

–

–

–

–

–

-1.93 -1.15 -0.30

0.42

14

–

–

3.39 –

–

–

–

–

–

-1.85 -1.03 -0.19

0.44

15

–

–

1.17 –

–

–

–

–

–

-2.72 -1.63 -0.43

0.58

16

–

–

1.14 –

–

–

–

–

–

-2.93 -1.46 -0.27

0.72

17

–

–

–

4.75 –

–

–

–

–

-1.66 -1.09 -0.24

0.63

18

–

–

–

2.38 –

–

–

–

–

-1.75 -0.85

0.08

1.05

19

–

–

–

3.37 –

–

–

–

–

-1.75 -1.11 -0.28

0.60

20

–

–

–

–

2.95 –

–

–

–

-2.23 -1.56 -0.69

0.13

21

–

–

–

–

2.15 –

–

–

–

-2.48 -1.57 -0.65

0.33

22

–

–

–

–

2.69 –

–

–

–

-2.10 -1.54 -0.81

0.04

23

–

–

–

–

–

2.47 –

–

–

-2.91 -2.11 -1.06

0.06

24

–

–

–

–

–

2.41 –

–

–

-2.56 -1.81 -0.84

0.16

25

–

–

–

–

–

1.26 –

–

–

-3.03 -2.27 -1.12

0.24

26

–

–

–

–

–

–

3.90 –

–

-1.33 -0.84 -0.25

0.42

27

–

–

–

–

–

–

3.90 –

–

-1.13 -0.62

0.03

0.70

28

–

–

–

–

–

–

–

3.03 –

-1.30 -0.69

0.00

0.58

29

–

–

–

–

–

–

–

3.03 –

-1.65 -0.95 -0.22

0.43

30

–

–

–

–

–

–

–

–

2.64 -2.08 -1.49 -0.79 -0.18

31

–

–

–

–

–

–

–

–

2.64 -2.10 -1.41 -0.51

Table 2

0.12

0.96

0.15

MCAT simulations results

Number of items

ADL

PWB

SYMP

RFR

RFA

RHCS

SSL

COP

REJ

Score

61.61

59.77

68.33

63.73

58.59

58.28

57.22

58.88

59.18

SEM

0.43

0.75

0.80

0.42

0.80

0.90

0.37

0.88

0.81

RMSE

0.42

0.70

0.69

0.28

0.64

0.71

0.15

0.73

0.62

Score

50.70

52.37

59.31

56.71

64.98

50.99

55.04

57.42

50.73

SEM

0.36

0.45

0.48

0.36

0.74

0.89

0.37

0.44

0.75

RMSE

0.31

0.34

0.30

0.20

0.59

0.69

0.12

0.19

0.53

Score

50.27

50.73

56.82

56.87

61.85

53.91

55.55

56.64

57.54

SEM

0.31

0.43

0.46

0.35

0.52

0.86

0.37

0.43

0.56

RMSE

0.21

0.31

0.26

0.14

0.34

0.66

0.09

0.13

0.30

Score

49.94

50.64

58.14

57.08

63.42

56.63

55.58

55.21

57.48

SE

0.30

0.43

0.46

0.34

0.51

0.84

0.37

0.42

0.53

RMSE

0.19

0.30

0.26

0.13

0.32

0.64

0.09

0.11

0.26

Score

49.70

50.57

56.77

57.14

63.94

60.99

56.10

55.08

57.26

SE

0.30

0.42

0.45

0.34

0.50

0.81

0.36

0.42

0.52

RMSE

0.18

0.29

0.24

0.12

0.30

0.61

0.09

0.10

0.25

Score

49.95

50.85

57.01

57.13

63.80

64.80

55.98

55.19

57.08

SE

0.29

0.42

0.44

0.34

0.50

0.78

0.37

0.42

0.52

RMSE

0.16

0.28

0.22

0.11

0.29

0.58

0.07

0.10

0.24

5

10

15

16

Table 1 Item a1

Estimated item parameters a2

d1

d2

d3

d4

a3

a4

a5

a6

a7

a8

a9

1

3.39 –

–

–

–

–

–

–

–

-1.16 -0.55 -0.03

0.58

2

3.48 –

–

–

–

–

–

–

–

-1.21 -0.65 -0.02

0.46

3

2.92 –

–

–

–

–

–

–

–

-1.76 -1.17 -0.48

0.07

4

2.54 –

–

–

–

–

–

–

–

-0.92 -0.18

0.50

1.12

5

2.89 –

–

–

–

–

–

–

–

-1.28 -0.61

0.00

0.50

6

2.37 –

–

–

–

–

–

–

–

-1.38 -0.73 -0.05

0.54

7

2.12 –

–

–

–

–

–

–

–

-1.04

0.05

0.95

1.84

Score

49.67

50.81

54.85

57.28

63.68

64.92

55.84

55.56

57.15

8

1.98 –

–

–

–

–

–

–

–

-1.49 -0.30

0.78

1.74

SE

0.29

0.41

0.43

0.34

0.49

0.74

0.36

0.42

0.51

9

–

2.36 –

–

–

–

–

–

–

-1.82 -0.84

0.17

0.98

RMSE

0.15

0.26

0.21

0.09

0.28

0.54

0.07

0.09

0.23

123

17

18

19

Qual Life Res (2015) 24:1–191

37 validity for socio-demographic and clinical characteristics. CONCLUSIONS: The MusiQoL-MCAT presents satisfactory properties and can individually tailor QoL assessment to each patient, making it less burdensome to patients and more adapted for its use in clinical practice. The MusiQoL MCAT is an adaptive short-form that adaptively administer the 16 most informative items for each patient, with a high level of clinical validity (Tables 1, 2, 3).

Table 2 continued Number of items

ADL

PWB

SYMP

RFR

RFA

RHCS

SSL

COP

REJ

20 Score

49.67

52.52

54.89

57.37

63.34

63.21

55.70

55.89

57.15

SE

0.28

0.40

0.43

0.33

0.49

0.66

0.36

0.42

0.51

RMSE

0.14

0.25

0.20

0.07

0.27

0.46

0.06

0.08

0.22

Table 3

Validity of the 16-tem MCAT procedure ADL

PWB

SYMP

Age (N = 900)

-0.295

-0.041

-0.117

p-value

\0.001

0.219

\0.001

RFR

RFA

RHCS

SSL

COP

0.029

-0.007

-0.042

-0.086

-0.017

0.382

0.839

0.207

0.009

0.603

REJ

-0.12 \0.001

Sex (N = 913) Male (N = 288)

48.3 ± 20.4

52.6 ± 19.2

58.71 ± 19.6

53.3 ± 23.9

63.5 ± 19.1

56.1 ± 15.4

55.9 ± 25.0

56.9 ± 21.2

57.7 ± 16.9

Female (N = 625)

50.7 ± 21.1

49.9 ± 20.0

57.9 ± 21.9

58.8 ± 23.4

63.5 ± 20.2

56.9 ± 16.8

55.6 ± 25.5

54.7 ± 22.3

57.4 ± 18.2

p value

0.109

0.058

0.596

0.001

0.991

0.468

0.871

0.171

0.827

Educational level (N = 676) \12 years (N = 399)

46.2 ± 20.2

47.7 ± 20.2

56.1 ± 21.9

55.6 ± 22.5

62.4 ± 20.4

55.4 ± 16.2

53.3 ± 25.5

50.7 ± 22.4

54.9 ± 17.8

C12 years (N = 277)

52.8 ± 21.1

52.6 ± 19.6

58.8 ± 21.3

59.9 ± 23.0

63.1 ± 20.2

57.9 ± 16.5

56.1 ± 24.5

58.7 ± 20.3

60.4 ± 16.5

p value

\0.001

0.002

0.105

0.016

0.668

0.047

0.145

\0.001

\0.001

Marital status (N = 683) Not single (N = 564)

48.2 ± 20.5

49.5 ± 20.2

56.9 ± 21.7

57.7 ± 22.5

64.1 ± 19.1

57.1 ± 15.7

56.0 ± 23.9

53.8 ± 21.8

56.9 ± 17.4

Single (N = 119)

51.4 ± 22.1

51.5 ± 19.4

58.8 ± 21.4

56.9 ± 23.9

56.5 ± 23.9

53.8 ± 18.8

47.2 ± 28.4

56.2 ± 21.5

58.3 ± 18.2

p value

0.131

0.333

0.383

0.741

\0.001

0.047

\0.001

0.275

0.426

Labor (N = 669) Active (student/employee) (N = 469)

57.72 ± 18.29

54.06 ± 17.89

53.81 ± 16.91

52.99 ± 17.55

51.74 ± 16.96

53.28 ± 10.88

53.17 ± 17.29

54.33 ± 17.16

55.63 ± 15.77

Unemployed (N = 200)

44.42 ± 16.94

48.66 ± 18.00

48.19 ± 17.11

52.40 ± 17.85

52.96 ± 16.20

51.67 ± 9.48

51.07 ± 17.45

48.30 ± 17.11

47.41 ± 16.14

p value

\0.001

\0.001

\0.001

0.695

0.382

0.06

Disease duration (years) (N = 890)

0.154

\0.001

\0.001

-0.206

0.040

-0.026

-0.027

0.002

p value

\0.001

0.238

0.444

0.415

0.947

-0.029

-0.029

-0.029

-0.08

0.387

0.380

0.385

0.022

MS subtype (N = 901) RR (N = 668)

54.8 ± 19.3

50.7 ± 19.6

58.6 ± 21.6

PP (N = 68)

36.6 ± 17.5

52.4 ± 18.2

60.5 ± 18.6

59.1 ± 23.5

64.3 ± 31.9

56.3 ± 13.9

54.7 ± 24.2

56.4 ± 19.2

55.5 ± 15.0

SP (N = 151)

32.5 ± 15.7

48.8 ± 21.0

53.9 ± 20.1

54.5 ± 23.6

63.4 ± 20.2

53.0 ± 13.8

52.3 ± 25.9

50.2 ± 21.9

49.9 ± 17.1

69.7 ± 24.7

58.4 ± 22.4

72.3 ± 20.8

64.7 ± 19.2

66.5 ± 20.4

62.1 ± 23.6

64.1 ± 24.4

65.2 ± 25.7

CIS (N = 14) p value EDSS (N = 904) p value

\0.001

0.813

57.3 ± 23.8

63.3 ± 19.6

0.297

0.565

0.733

57.4 ± 16.8

0.028

56.3 ± 25.3

0.253

56.0 ± 21.9

0.062

59.2 ± 17.5

67.6 ± 22.9 \0.001

-0.644

-0.116

-0.164

-0.051

-0.025

-0.197

-0.185

-0.159

-0.365

\0.001

\0.001

\0.001

0.125

0.452

\0.001

\0.001

\0.001

\0.001

SF-36 PHYS (N = 914) p value SOC (N = 919) p value PBPHY (N = 914) p value PBPSY (N = 908)

0.784

0.264

0.339

0.138

0.116

0.305

0.286

0.305

0.522

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.295

0.416

0.337

0.408

0.525 \0.001

0.533

0.47

0.407

0.26

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.602

0.407

0.435

0.176

0.223

0.338

0.291

0.391

0.498

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.465

0.418

0.163

0.213

0.257

0.404

0.454

0.46

0.32

123

38

Qual Life Res (2015) 24:1–191

Table 3 continued

p value PSY (N = 917) p value VITAL (N = 917) p value DOUL (N = 915) p value SP (N = 910) p value PCS (N = 887) p value MCS (N = 887) p value

ADL

PWB

SYMP

RFR

RFA

RHCS

SSL

COP

REJ

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.376

0.689

0.447

0.305

0.415

0.478

0.411

0.508

0.529

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.541

0.567

0.515

0.279

0.336

0.457

0.388

0.419

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.53 \0.001

0.423

0.356

0.401

0.091

0.135

0.229

0.204

0.296

0.407

\0.001

\0.001

\0.001

0.006

\0.001

\0.001

\0.001

\0.001

\0.001

0.473

0.418

0.369

0.217

0.229

0.348

0.295

0.422

0.472

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.733

0.255

0.129

0.116

0.286

0.264

0.316

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

0.38 \0.001

0.5 \0.001

0.282

0.641

0.437

0.273

0.372

0.426

0.338

0.452

0.458

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

\0.001

External validity

% Fit

SF36: % of dimensions meeting three conditions

33

Age: % of dimensions with correlation coefficient \ 0.40

100

EDSS: % of dimensions meeting two conditions**

100

MS duration: % of dimensions with correlation coefficient \ 0.4

100

Gender: % of dimensions with ES \ 0.2 from ref

88.9

Educational level: % of dimensions with ES \ 0.2 from ref

100

Marital status: % of dimensions with ES \ 0.2 from ref

100

Occupational status: % of dimensions with ES \ 0.2 from ref

88.9

Total

88.85

(201.2) Distilling the essence of appraisal: a mixed methods study of people with multiple sclerosis Bruce D. Rapkin, PhD, Albert Einstein College of Medicine, Bronx, NY, United States; Carolyn Schwartz, ScD, Tufts University Medical School; DeltaQuest Foundation, Inc., Oslo and Akershus University College, Concord, MA, United States AIMS: The goal of this study is to identify the essential parts of the Quality of Life (QOL) Appraisal Profile that capture the most important differences in the ways that people with multiple sclerosis (MS) respond to patient-reported outcome (PRO) measures. This process will enable the eventual development of a more practical, less resource-intensive version of the QOL Appraisal Profile to facilitate its use in clinical research and practice. METHODS: This is a secondary analysis of longitudinal generic and disease-specific PRO data (n = 859) of participants in the North American Research Committee on Multiple Sclerosis (NARCOMS) registry. Following the Rapkin and Schwartz (2004) model, we computed a ‘‘standard QOL model’’, and then Multivariate Analysis of Variance (MANOVA) and discriminant function analysis to identify patterns of appraisal measures associated with group differences in response to each QOL outcome (Rand-12 physical and mental component scores; Performance Scales summary score). RESULTS: The ‘‘standard QOL model’’ explains a moderate amount of variance (i.e., 15–17 %) in physical functioning and disease-specific disability, and very little variance in mental health functioning. On the basis of the preliminary screen of the MANOVAs, we identified the appraisal variables that mattered by PRO, the discriminant function analysis included 10–16 of the 83 potential appraisal variables in two discriminant functions per outcome that distinguished groups with better, average and worse expected scores, as well as groups with scores that were better-thanexpected, as-expected, and worse-than-expected. The dominant

123

appraisal parameters were more similar between the generic and disease-specific measure of physical functioning and disability, respectively, than between the mental health measure and the former two measures. CONCLUSIONS: On the basis of relatively complex modeling and substantial ‘‘number-crunching’’, we have reduced the number of key appraisal variables by almost half. We have then further summarized and concluded that the essence of this subset of appraisal variables relates to evaluative, stability, and activity concepts. The implications this work are discussed in terms of interdisciplinary research on causal attribution, self-schema, and semantic differential; as well as future directions for the QOL Appraisal Profile.

(201.3) A Bayesian methodology to assess psychometric properties of quality of life measures in small sample sizes Mohamed Boucekine, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Pierre Michel, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Samy Clinchard, sante´ public, Marseille, France; Karine Baumstarck, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Laurent Boyer, EA 3279 SelfPerceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Anderson Loundou, 1. AixMarseille University, EA 3279 – Public Health, Chronic Diseases and Quality of Life - Research Unit, 13005, Marseille, France; Badih Ghattas, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Pascal Auquier, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France

Qual Life Res (2015) 24:1–191

39

AIMS: The sample size in validation studies of measurement scales can lead to erroneous conclusions being drawn if the sample is too small. The aim of this study was to explore the construct validity and reliability of a QoL instrument in small sample sizes of multiple sclerosis (MS) subjects. METHODS: The study relied on a crosssectional design and was performed in a neurology department (Marseille, France). The inclusion criteria were as follows: MS diagnosis according to the McDonald criteria, age C18 years. QoL was assessed using the MS specific MusiQoL instrument which included 9 domains (activity of daily living, psychological well-being, symptoms, and relationships with friends, relationships with family, and relationships with health care system, sentimental and sexual life, coping, and rejection). According to cognitive test assessment (executive functions, memory and attention performance), four groups of cognitive dysfunction were considered: (1) a no-impaired (no impairment), (2) low-impaired (1 impaired function), (3) medium-impaired (2 impaired function) and (4) highly-impaired groups (3 impaired functions). Statistical analyses were performed on the four samples defined above. Bayesian versions of each indicator of validity were computed using WinBUGS 1.4.3 software, considering 30,000 MCMC samples. The structure of the MusiQoL was explored using confirmatory factor analysis (CFA) to determine how it matched with the initial MusiQoL structure. The Partial Credit Model was fitted to the data. Unidimensionality of each domain was assessed using item goodness-of-fit statistics (INFIT). For each dimension scale, the internal consistency reliability was assessed by Cronbach’s alpha coefficient. RESULTS: A total of 124 MS patients participated in the study, among them 35 non-impaired, 40 low-impaired, 37 medium-impaired and 12 highly-impaired. The structure was confirmed by the Posterior Predictive P value ([0.05). The INFIT statistics were in the acceptable range (0.7–1.3) for non/mediumimpaired groups but less satisfactory for low/hightly-impaired groups. Cronbach’s alphas were globally satisfactory ([0.7) except for 3 domains (sentimental and sexual life, coping, and relationships with health care system). CONCLUSIONS: The study shows satisfactory validity of the MusiQoL in different small sample sizes (Tables 1, 3). Table 1 Infit Infit NI

LI

MI

HI

ADL

0.95–1.60

0.98–1.36

0.72–1.65

0.90–3.24

PWB

0.69–0.95

0.70–1.06

0.81–1.10

0.82–1.36

SPT

1.01–1.16

0.90–1.31

0.87–1.29

1.15–1.57

RFr

0.72–0.85

0.66–0.84

0.61–0.80

0.59–0.85

RFa

0.80–0.89

0.89–1.03

0.81–0.96

0.72–1.23

SSL

1.02–1.09

1.05–1.08

0.88–0.88

1.01–1.02

COP REJ

1.03–1.04 0.74–0.80

1.31–1.35 0.84–0.90

0.97–1.00 0.82–0.89

1.14–1.31 1.04–1.22

RHCS

0.83–0.91

0.78–1.14

0.83–1.02

0.82–0.94

Alpha: Cronbach’s alpha Infit: Infit MnSq statistics MusiQoL: ADL activity of daily living. PWB psychological wellbeing. SPT symptoms. RFr relationships with friends. RFa relationships with family. SSL sentimental and sexual life. COP coping. REJ rejection. RHCS relationships with health care system NI non-impaired (0 impaired composite). LI low-impaired (1 impaired composite). MI medium-impaired (2 impaired composites). HI highly-impaired (3 impaired composites) Bold values: unsatisfactory values

Table 3 Cronbach’s alpha Alpha NI

S.E.

LI

S.E.

MI

S.E.

HI

S.E.

ADL

0.87

0.03

0.86

0.03

0.86

0.03

0.80

0.08

PWB

0.77

0.05

0.79

0.05

0.76

0.06

0.79

0.09

SPT

0.75

0.06

0.73

0.07

0.77

0.06

0.70

0.14

RFr

0.68

0.10

0.71

0.08

0.72

0.08

0.65

0.19

RFa

0.73

0.08

0.73

0.07

0.67

0.09

0.61

0.23

SSL

0.64

0.11

0.69

0.10

0.66

0.12

0.58

0.27

COP REJ

0.62 0.67

0.14 0.12

0.63 0.71

0.12 0.10

0.63 0.71

0.12 0.10

0.66 0.68

0.20 0.20

RHCS

0.57

0.12

0.65

0.10

0.64

0.10

0.52

0.26

MusiQoL: ADL activity of daily living. PWB psychological wellbeing. SPT symptoms. RFr relationships with friends. RFa relationships with family. SSL sentimental and sexual life. COP coping. REJ rejection. RHCS relationships with health care system NI non-impaired (0 impaired composite). LI low-impaired (1 impaired composite). MI medium-impaired (2 impaired composites). HI highly-impaired (3 impaired composites) S.E. standard errors

(201.4) Association between bookmarking-derived cut scores and self-classified cut scores for neuro-QOL measures of sleep disturbance, lower & upper extremity function, and fatigue in a sample of persons with multiple sclerosis: validating both approaches David Victorson, Northwestern University, Evanston, IL, United States; Karon F. Cook, PhD, Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Deborah Miller, PhD, Cleveland Clinic, Cleveland, OH, United States; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: In previous qualitative work, cut scores for levels of 4 NeuroQOL measures were generated using ‘‘Bookmarking’’, a method borrowed and adapted from educational standard setting. The current study evaluated the validity of these thresholds in a clinical sample (N = 827–951). METHODS: Participants completed the Neuro-QOL measures and then rated their levels of the outcome (none, mild, moderate, severe). Mean Neuro-QOL scores were calculated by selfclassified severity level. We operationalized self-classification cutpoints as the mid-point between score means of adjacent categories. For example, the mean fatigue scores of those classifying themselves as ‘‘mild’’ and ‘‘moderate’’ were 46.35 and 55.52, respectively. The self-classification threshold between mild and moderate, therefore, was 50.69. This approach allowed us to compare two sets of thresholds for each Neuro-QOL measure: those generated using Bookmarking and those generated based on self-classifications. The associations between sets of thresholds were estimated using Tau-B and Spearman correlation coefficients. Values [0.6 were considered indicative of good agreement. RESULTS: Self-classified cut points for the 4 measures resulted in the following score ranges: fatigue (none \ 40.74, mild C 40.74, moderate C 50.69, severe C 60.30); sleep (none \ 47.49, mild C 47.49, moderate C 56.37, severe C 64.15); lower extremity (none \ 51.32, mild C 51.32, moderate C 40.39, severe C 31.50); upper extremity (none \ 45.95, mild: C 45.95, moderate C 36.52, severe C 29.84). See Table 1 for Tau-b and Spearman values and confidence intervals. All Tau-b values and Spearman correlations were in the ‘‘high’’ range for

123

40

Qual Life Res (2015) 24:1–191

Table 1 Association between self-classified cut scores and standard setting derived cut scores Tau-b

Spearman

Value CI

Value CI

Sleep Patients Clinicians Split random half derived classifications

0.652 0.628–0.676 0.704 0.680–0.728 No data available 0.696 0.667–0.727 0.757 0.727–0.787

Lower Ext Patients

0.800 0.783–0.817 0.855 0.838–0.870

Clinicians

No data available

Random half class

0.801 0.775–0.823 0.858 0.835–0.877

Fatigue Patients Clinicians Split random half derived classifications

0.749 0.732–0.763 0.810 0.793–0.824 0.752 0.733–0.771 0.806 0.787–0.823 0.753 0.724–0.778 0.807 0.779–0.830

Upper Patients

0.674 0.649–0.698 0.713 0.688–0.737

Clinicians

0.680 0.656–0.704 0.735 0.711–0.756

Split random half class 0.691 0.656–0.720 0.750 0.713–0.777

agreement (between .65–.80 and .70–.85, respectively). There were no statistically significant differences among the methods for classifying individuals in either approach. CONCLUSIONS: Results support the validity of the cut scores derived from the Bookmarking exercises. For fatigue and upper extremity, there were slight differences in the associations between Bookmarking and self-classified cut scores, however they were negligible. While Bookmarking engages respondents in a unique way, it can be expensive and time consuming. Our results indicate that, when an adequate sample size is available, cut points for levels of outcome can be effectively operationalized as the distance between score means of adjacent self-classification categories.

(201.5) Interdigitating performance-based tasks and patient reported outcomes to produce a linear hierarchical unidimensional measure of global physical disability in MS Stanley Hum, PhD (candidate), McGill University, Montreal, QC, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada AIMS: Multiple Sclerosis (MS) disability can only be measured indirectly by how it is manifested. The goal is to develop a linear hierarchical unidimensional measure of global physical disability measure including the patient perspective and relevant to a multidisciplinary team. METHODS: The conceptual framework for this study was the International Classification of Functioning, Disability, and Health. Several ICF MS core sets provided content validity for this analysis. Rasch analysis was used to develop the global disability measure and provided evidence of construct validity. This is a secondary analysis of information from 189 MS patients comprehensively assessed on health indices. Performance-based tasks (PBTs) and patient reported measures (PROs) were assessed at baseline and questionnaires were assessed twice (6 months apart). A total of 136 items were selected from domains of body function impairment and activity limitation as a starting point. RUMM2020 software was used

123

to fit a partial credit Rasch model. Items with continuous scores were categorized by their frequency distribution. Measurement fit to the Rasch model was tested for threshold order, model fit, response dependency, unidimensionality, DIF, and item/person targeting. All p values \0.05 after Bonferroni adjustment were considered to be significant. RESULTS: Disordered threshold was observed in 55.1 % of the items. Residual correlation C0.3 were observed in 266 item pairs. The final 22-item global physical disability measure fit the expectations of the Rasch measurement model. There was some mistargeting with mean person location of 1.595 (SD = 2.188). Four PRO measures contributed items to the final measure; 5 from the SF36-PFI, 1 item from the PBSI, 1 item from the EQUI-SCALE, 7 items from the DASH. A total of 8 performance-based tasks (6-min walk, 9-hole peg test dominant hand, push-ups, partial-curl-ups, gait speed (comfortable and fast), VO2 max, and vertical jump) were also included in the measure. Consistently, PROs were left of the continuum and performance based tasked located on the right. CONCLUSIONS: PBTs and PRO health indices can co-exist in a linear hierarchical unidimensional measure of global physical disability. However, these items were not evenly distributed along the linear continuum. PROs were easier than PBTs.

(201.6) Developing the Guy’s Neurological Scale (GNDS) in 28 languages Catherine Acquadro, Mapi Research Trust, Lyon, France; Bing Zhu, Biogen, Cambridge, MA, United States; Alain Nguyen, Mapi, Lyon, France; Basil Sharrack, Professor, Sheffield Teaching Hospitals NHS Foundation Trust & University of Sheffield, Sheffield, United Kingdom; Nazilabee Toraubully, Mapi, Lyon, France AIMS: The Guy’s Neurological Disability Scale (GNDS) is an interviewer-administered measure developed in British English to measure disability in multiple sclerosis (MS). Twelve domains are assessed: cognition, mood, vision, speech, swallowing, upper-limb function, lower-limb function, bladder function, bowel function, sexual function, fatigue and ‘‘others.’’ The domains are graded according to their severity and impact on patients. The objective of this abstract is to present the rigorous methodology in translating GNDS to 28 languages. METHODS: The GNDS was translated following a methodology in compliance with the ISPOR guidelines. For most of the languages, the process consisted of: (1) definition of each concept behind each item, (2) two forward translations by native translators, reconciliation of the translations, (3) one back-translation by a native English-speaking translator, (4) final reconciliation, (5) clinician review, (6) cognitive interviews with five native-speaking patients with MS, and (7) proofreading. RESULTS: Most of the concepts assessed were cross-culturally relevant, but the word ‘‘rollator’’ (a walking frame with wheels) was added in the scoring section of the lower-limb disability domain in Switzerland where rollators are more commonly used than simple walking frames. Clinicians were essential in ensuring that the term ‘‘disability’’ was carefully translated, and that the clinician scoring section of each domain was faithfully rendered by reviewing the translations of medical terms (e.g., diplopia, urethral/suprapubic, etc.). The interviews with patients revealed some difficulties in understanding ‘‘cognitive’’ (in the heading ‘‘cognitive disability,’’), but this difficulty did not affect the understanding of the corresponding items since most of the respondents inferred the meaning from the content of the items (i.e., memory and concentration). Some respondents had difficulties in discerning the difference between usual activities and physical activities in the fatigue domain. CONCLUSIONS: The rigorous methodology used, involving the participation of the developer, 28 clinicians and 140 MS patients, was essential in producing translations of the GNDS conceptually equivalent to the British English original. This methodology

Qual Life Res (2015) 24:1–191

41

may be adopted to ensure the successful translation of health-related questionnaires.

and determine the minimal clinically important difference score (MCID).

202: Measure Development

(202.2) Determining the psychometric properties of the CLEFT-Q using two different sets of response options

(202.1) Development of the preschool cystic fibrosis questionnaire-revised (CFQ-R) Alexandra L. Quittner, PhD, University of Miami, Miami, FL, United States; Anjana M. Morris, PhD, MPH, University of Miami, Miami, FL, United States; Claire Wainwright, MD, Lady Cilento Children’s Hospital, Brisbane, Australia; Joyce Cheney, Lady Cilento Children’s Hospital, Brisbane, Australia AIMS: Patient-reported outcomes (PROs) are critical to evaluating the effects of chronic disease. The Cystic Fibrosis QuestionnaireRevised (CFQ-R; Quittner et al., 2005) is the best validated PRO for cystic fibrosis (CF), with strong reliability and validity in patients with CF ages 6 through 70 (Quittner et al., 2014). However, currently there is no validated PRO to measure HRQoL in patients under age 6. As new treatment guidelines for young children with CF are implemented, it will be important to assess their effects. Thus, this study aimed to develop a pictorial adaptation of the CFQ-R for children ages 3–6 as an outcome measure for routine care, disease progression and efficacy of newly introduced treatments. METHODS: Children (n = 12, mean age = 4.51 years, SD = .84) completed open-ended, qualitative interviews to assess how CF affects their daily functioning. Recorded interviews were transcribed and coded in Atlas.ti to verify content saturation and generate preliminary items. Cognitive testing with forced-choice responses measured the clarity and comprehensiveness of the measure and rating scales. Psychometric validation was then conducted with preschool-age children in the United States and Australia (n = 248, mean age = 4.21 years, SD = .93) to evaluate item and scale level reliability and convergent validity. RESULTS: Saturation of content was achieved. Most children could answer questions about their physical and emotional functioning, respiratory and digestive symptoms, energy level, body image and treatments. The preliminary child instrument consisted of 35 items; however we deleted items assessing social functioning (a = .53), which were not well understood, and revised pictures for the Treatment Burden Scale (a = .21; see Fig. 1). Other items were modified or deleted based on internal consistencies and item-to-total correlations for each scale. The revised instrument consists of 28 items. Adequate to strong reliabilities were obtained for all scales in this revised version (a = .68-.80). CONCLUSIONS: Young children were able to reliably report on their daily symptoms and functioning. With further validation, this PRO may be used in clinical trials of medications for young children with CF ages 3–6. Future research will identify other psychometric properties of the Preschool CFQ-R

Elena Tsangaris, MSc, McMaster University, Hamilton, ON, Canada; Karen Wong, MD MSc FRCSC, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; Christopher Forrest, MD MSc FRCSC, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; Karen Harman, MD, McMaster University, Hamilton, ON, Canada; Andrea Pusic, MD MPH FACS, Memorial Sloan-Kettering Cancer Center, New York City, NY, United States; Stefan Cano, PhD, Modus Outcomes, Stotfold, United Kingdom; Timothy Goodacre, MBBS BSc FRCS, University of Oxford, Oxford, United Kingdom; Anne Klassen, DPhil, McMaster University, Hamilton, ON, Canada AIMS: Few scientifically sound PRO instruments currently exist to measure satisfaction with appearance in patients with facial differences. The CLEFT-Q is the only PRO instrument designed to measure outcomes that matter to patients with cleft lip and/or palate (CLP) aged 8–29 years. The CLEFT-Q is composed of 13 independently functioning scales that measure appearance, QOL and facial function. Limited research was available to choose the most appropriate response option set for measuring appearance in children and young adults. The aim of this study was to decide between two different sets of response options for the CLEFT-Q appearance scales: (1) dislike a lot/dislike a little/like a little/like a lot; and (2) not at all/a little bit/ quite a bit/very much. METHODS: Participants with CLP were recruited from the Hospital for Sick Children and McMaster Children’s Hospital between November 2014 and April 2015. Participants completed two versions of the Facial Appearance scale in a randomized order, with the remainder of the CLEFT-Q scales administered in between. Rasch Measurement Theory (RMT) analysis was used to identify the best response option set. RESULTS: A total of 271 individuals, i.e., 145 males and 126 females, participated. Average time to complete the entire CLEFT-Q was 20 min. All items for both versions of the scale had ordered thresholds. The person estimates ranged from -2.25 to +5.11 (like) and -5.32 to +5.11 (frequency), and the item thresholds ranged from -3.70 to +3.52 (like) and -4.00 to +3.38 (frequency). The proportion of extreme scores were similar, i.e., 10 % (27 of 264, like) and 12 % (31 of 251, frequency). The primary difference between the two options was in the best point of measurement, located in the centre of person estimate distribution for the frequency-based options and skewed to the left for the like/dislike options. CONCLUSIONS: Response options for scales may change the measurement properties of an instrument. Testing the two different response options showed that the frequencybased response options resulted in marginally better measurement of the population compared to the like/dislike options. The CLEFT-Q scales will employ frequency-based response options as a result of this study.

(202.3) Psychometric development of three new condition-specific questionnaires to measure quality of life (Aneurysm-DQoL), symptoms (Aneurysm-SRQ) and treatment satisfaction (Aneurysm-TSQ) of individuals with abdominal aortic aneurysms

Fig. 1 Revised items for Treatment Burden Scale

Jacquelyn Romaine, Health Psychology Research Unit, Royal Holloway, University of London, Egham, Surrey, United Kingdom; George Peach, St George’s Vascular Institute, London, United Kingdom; Matt Thompson, St George’s Vascular Institute, London, United Kingdom; Robert Hinchliffe, St George’s Vascular Institute,

123

42 London, United Kingdom; Clare Bradley, Health Psychology Research Unit, Royal Holloway, University of London; Health Psychology Research Ltd, Royal Holloway, University of London, Egham, Surrey, United Kingdom AIMS: With increasing interest in patient reported outcomes for people with (or following repair of) abdominal aortic aneurysms (AAA), this study reports on the psychometric evaluation of three new condition-specific measures: the Aneurysm-Dependent Quality of Life (Aneurysm-DQoL) questionnaire, the Aneurysm Symptom Rating Questionnaire (Aneurysm-SRQ) and the Aneurysm Treatment Satisfaction Questionnaire (Aneurysm-TSQ). METHODS: Two hundred and eighty six participants from five UK NHS Hospital Trusts were invited to complete the new questionnaires. Participants were recruited and completed questionnaires returned by post. Retest questionnaires were administered to 65 participants approximately 4 months after initial completion. Test–retest reliability was assessed using a two-way random effects single measure intraclass correlation coefficient (ICC 2.1, Index range 0–1). Exploratory Factor Analysis (EFA) was employed to examine the structure of the data. RESULTS: Responses were returned by 197 participants (69 %), including 18 women. Useable data were received from 19 participants under preoperative AAA surveillance and 174 who had already undergone aneurysm repair using open (n = 70) or endovascular (n = 104) surgery. Participants’ mean age was 75 years (SD = 8.09, range 60–95). EFA of the Aneurysm-DQoL revealed a one-factor structure that included 20 of the total 23 items (a = 0.96, ICC = 0.77) and explained 55.54 % of the variance. EFA of the Aneurysm-SRQ demonstrated a six-factor structure comprising 24 of the total 44 items. The six factors were: Emotional (n = 5, a = 0.88, ICC = 0.69); Weight (n = 3, a = 0.72, ICC = 0.47); Lower Limb (n = 5, a = 0.76, ICC = 0.69); Cognitive (n = 3, a = 0.84, ICC = 0.82); Malaise (n = 4, a = 0.69, ICC = 0.60); and Gastrointestinal (n = 4, a = 0.73, ICC = 0.80), explaining 52.62 % of the variance. EFA of the Aneurysm-TSQ identified a subscale for preoperative treatment satisfaction (Items = 7, a = 0.87) that explained 52.83 percent of the variance in the data and an 11-item scale for postop patients only (a = 0.90, ICC = 0.40–0.88) explaining 49.69 % of the variance. CONCLUSIONS: The findings from the present study suggest that the Aneurysm-DQoL, Aneurysm-SRQ and AneurysmTSQ have clear structure, good internal consistency reliability and strong test–retest reliability. The questionnaires are now ready for use in clinical trials and routine care. Their use will clarify the precise impact of AAA and its treatment, highlight the issues most relevant to AAA pre- and post-repair and facilitate targeted improvements in care.

(202.4) Development and psychometric validation of a PRO instrument for obese, bariatric and cosmetic body contouring patients: the BODY-Q Anne Klassen, DPhil, McMaster University, Hamilton, ON, Canada; Amy Alderman, MD, Swan Center for Plastic Surgery, Alpharetta, GA, United States; Mark Soldin, MD, St Georges Healthcare NHS Trust, London, United Kingdom; Sam Robson, MD, Temple Medical, Aberdeen, United Kingdom; Manraj Kaur, MSc, McMaster University, Hamilton, ON, Canada; Stefan Cano, PhD, Modus Outcomes, Stotfold, United Kingdom; Achilles Thoma, MD, McMaster University, Hamilton, ON, Canada; Andrea Pusic, MD MPH FACS, Memorial Sloan-Kettering Cancer Center, New York City, NY, United States AIMS: Increasing rates of morbid obesity requiring bariatric surgery and the consequent need to get rid of the resulting excess skin have recently increased the demand for body contouring. The aim of our study was to develop a new PRO instrument (the BODY-Q) to

123

Qual Life Res (2015) 24:1–191 address the lack of scales designed to measure HR-QOL and satisfaction with appearance for reconstructive and cosmetic patients. METHODS: We followed recommended guidelines for PRO instrument development to ensure that the BODY-Q meets requirements of regulatory bodies. In Phase I, 63 patient interviews led to the development of a conceptual framework and set of independently functioning PRO scales that measure satisfaction with appearance (body, abdomen, arms, back, buttocks, thighs, hips, skin, scars) and HR-QOL (physical, psychological, social and sexual function and body image). Scales were refined through cognitive interviews (N = 22 patients) and expert input (N = 9 surgeons). Phase II involved recruitment of 977 participants (obese, bariatric and body contouring) from Canada, USA, and UK. Rasch Measurement Theory (RMT) analysis was used to refine the scales. RESULTS: The final scales are short (4–10 items) with a low grade reading level (3.8, range 0–5.3 by scale). In the RMT analysis, all items had ordered thresholds providing support for the hypothesis that the successive integer scores worked as a continuum. Item fit was good, i.e., 76/97 items were within -2.5 to +2.5 and none were significant in terms of the Chi square p-values. PSI and Cronbach alpha values ranged from 0.82 to 0.94 and 0.80 to 0.98 respectively. BODY-Q scores worked as hypothesized, with significantly worse scores for all appearance and HR-QOL scale reported by participants in the highest versus lowest BMI groups and participants with a little versus a lot of excess skin. Finally, appearance was strongly correlated with HR-QOL (e.g., Satisfaction with Body correlated 0.50, 0.42, 0.50, 0.57, 0.81 with Psychological, Social, Sexual and Physical Function and Body Image respectively). CONCLUSIONS: The BODY-Q is a comprehensive PRO instrument that can be used in clinical practice and research to measure outcomes with cosmetic patients and with weight loss patients across the entire weight loss journey starting with obesity and ending after body contouring surgery.

(202.5) Validation of a crosswalk between measures of work role limitations Milena D. Anatchkova, PhD, University of Massachusetts Medical School, Evidera, Lexington, MA, United States; Nisha Kini, University of Massachusetts Medical School, Worcester, MA, United States; John Ware, PhD, University of Massachusetts Medical School, Worcester, MA, United States; Jakob B. Bjorner, MD, PhD, QualityMetric; University of Copenhagen, Copenhagen, Denmark AIMS: To validate in an independent sample a previously developed crosswalk between the short form of the Work Limitations Questionnaire (WLQ, 8 items) and the Role Physical (RP, 4 items) and Role Emotional scales (RE, 3 items) of the SF-36 Health survey. These measures assess limitations in role performance attributed to health (emotional, physical, or both) and different breadth of impact (work vs. work and other activities). We have previously presented crosswalk tables allowing back and forth comparisons between WLQ and SF-36 role scales. Such crosswalk can be used to expand the information from studies that have used only one of the measures, provide a broader basis for interpretation and can inform future decisions on data collection strategies. METHODS: We used data from employed participants in a cross-sectional panel sample n = 301, 45 % female, 90 % Caucasian, 47 % with preselected chronic conditions, 21 % with fair/poor health. We used previously developed IRT based calibration tables to estimate WLQ scores from the SF-36 RE and RP scale. We evaluated the agreement of observed and estimated WLQ scale scores and the ability of estimated and observed WLQ scores to discriminate among known groups of patients. RESULTS: Estimated and observed scale scores were highly correlated (r = .69 for SF36 RE estimates and r = .74 for RP estimates) and the correlations were only slightly lower than the ones

Qual Life Res (2015) 24:1–191 observed in the original sample of the crosswalk development. Both observed and estimated WLQ successfully differentiated between levels of self reported general health and between patients with and without chronic conditions. The estimated WLQ means based on the SF36 RP score were not statistically different than the observed WLQ means, while estimated WLQ scores from the SF36 RE scale were lower than the observe WLQ group means. CONCLUSIONS: Our results demonstrate that the crosswalk of scores from the SF-36 role scales and the WLQ achieved using IRT methods is valid in an independent validation sample.

(202.6) The computer adaptive test (CAT) version of the EORTC QLQ-C30 Morten A. Petersen, The Research Unit, Department of Palliative Medicine, Bispebjerg Hospital, University of Copenhagen, Copenhagen, Denmark; Neil K. Aaronson, The Netherlands Cancer Institute, Amsterdam, Netherlands; Juan I. Arraras, Oncology Departments, Complejo Hospitalario de Navarra, Pamplona, Spain; Wei-Chu Chie, Institute of Epidemiology and Preventive Medicine and Department of Public Health, College of Public Health, National Taiwan University, Taipei, Republic of Taiwan; Thierry Conroy, Medical Oncology Department, Institut de cance´rologie de Lorraine, Vandoeuvre-le`s-Nancy, France; Anna Costantini, Psychoncology Unit, Department of Oncological Sciences, Rome, Italy; Eva-Maria Gamper, Department for Nuclear Medicine, Medical University Innsbruck, Innsbruck, Austria; Johannes M. Giesinger, Medical University of Innsbruck, Innsbruck, Austria; Esther Habets, Medical Center Haaglanden, Department of Neurology, The Hague, Netherlands; Eva Hammerlid, Dept of Otolaryngology Head and Neck Surgery, Sahlgrenska University Hospital, Go¨teborg University, Gothenburg, Sweden; Marianne J. Hjermstad, Regional Centre for Excellence in Palliative Care, Department of Oncology, Oslo University Hospital, Oslo, Norway; Bernhard Holzner, Medical University of Innsbruck, Innsbru¨ck, Austria; Colin Johnson, Surgical Unit, University of Southampton, Southampton, United Kingdom; Stein Kaasa, Palliative Medicine Unit, University Hospital of Trondheim, Trondheim, Norway; Jon H. Loge, European Palliative Care Research Centre, Faculty of Medicine, Norwegian University of Science and Technology (NTNU), Trondheim, Norway; Susanne Singer, Institute of Medical Biostatistics, Epidemiology and Informatics (IMBEI), University Medical Centre Johannes Gutenberg University, Mainz, Germany; Martin Taphorn, VU University Medical Center, Department of Neurology, Amsterdam, Netherlands; Krzysztof Tomaszewski, Department of Anatomy, Jagiellonian University Medical College, Krakow, Poland; Galina Velikova, MD PhD, Leeds Institute of Cancer Studies and Pathology University of Leeds, Leeds, United Kingdom; Irma Verdonck-de Leeuw, Clinical Psychology, VU University, Amsterdam, Netherlands; Teresa Young, Lynda Jackson Macmillan Centre, Mount Vernon Cancer Centre, Northwood, Middlesex, United Kingdom; Mogens Groenvold, The Research Unit, Department of Palliative Medicine, Bispebjerg Hospital, University of Copenhagen, Copenhagen, Denmark AIMS: Computerized adaptive test (CAT) methods allow one to tailor questionnaires based on individual’s responses, at the same time yield scores that are directly comparable across individuals. CAT instruments are generally more efficient, relevant, and flexible than standard, fixed length questionnaires. The EORTC Quality of Life Group (QLG) has developed CAT item banks for all 14 functional and symptom domains of the widely used EORTC QLQ-C30 questionnaire. The CAT version facilitates measurement within the established health-related quality of life (HRQOL) framework of the QLQC30, and aims at being backward-compatible with the many studies which have used the QLQ-C30. METHODS: The EORTC QLG’s

43 strategy for developing CAT includes: literature search and conceptualization, formulation of new items in the ‘‘QLQ-C30 item style’’, expert and patient evaluations, international field-testing (N C 1000), and psychometric analysis including evaluation of dimensionality, calibration and testing of IRT models and DIF analysis. RESULTS: There was substantial variation across the QLQ-C30 domains in the number of items that could be developed, while fulfilling our requirements regarding item style, unidimensionality, international applicability, etc. The final item banks include between seven and 34 items. Generally, we observed close agreement between scores based on the full item banks and CAT scores (also for short CATs with just 2–4 items). The CATs generally improved measurement precision compared to the original QLQ-C30 scales. Evaluation based on available data indicated that the CATs typically can reduce sample size requirements by 20–40 % without loss of statistical power. CONCLUSIONS: The challenges encountered in developing unidimensional CAT item banks, as well as the potential benefit of CAT varied across domains. Nevertheless, development of CAT was achievable across all 14 HRQOL domains and the new CAT instrument appears to have clearly superior measurement properties compared to the original QLQ-C30. The measurement properties of the CAT instrument should be validated in independent samples. An international clinical validation study is in progress. Meanwhile, the CAT-versions of the QLQ-C30 scales are available for ‘‘experimental use’’ as short-forms or CATs using our web-based platform or by integrating it in other existing platforms for collecting electronic questionnaire data.

203: Social Support, Coping and Resilience (203.1) What keeps us healthy in the face of adversity?— Development of a new Stress Resilience Questionnaire (StResQ) Nina Tamm, Charite´ - Universita¨tsmedizin Berlin, Berlin, Germany; Alexander Obbarius, MD, Charite´ - Universita¨tsmedizin Berlin, Berlin, Germany; Sandra Nolte, PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Felix Fischer, ScD, Charite Universitaetsmedizin Berlin, Berlin, Germany; Kim-Marie Engster, Charite´ - Universita¨tsmedizin Berlin, Berlin, Germany; Matthias Rose, MD PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany AIMS: Today, a plethora of well validated questionnaires are available measuring perceived stress as well as external stressors. In contrast only few instruments assessing stress resilience exist. A review of available questionnaires suggests that they have not been developed following guidelines, resulting in unsatisfactory content validity or instable factor structure. In summary, to date no gold standard to assess stress resilience has been identified. A new content valid Stress Resilience Questionnaire (StResQ) is being developed to capture conditions that keep persons healthy in the face of adversity and allow an adaptive response to stress. METHODS: Following the guidelines of the Food and Drug Administration (FDA) on questionnaire development, focus groups including clinical experts and psychosomatic patients were performed. Items were generated on the basis of the Conceptual Framework developed from focus group results. The item pool was revised iteratively incorporating results from cognitive testing. As a first result we generated a pilot version of the StResQ which contained 131 items that are measured on a 4-point Likert scale. Data collection was performed at the psychosomatic outpatient clinic (N = 500) using smartphones and tablet computers. RESULTS: Preliminary results (N = 372) of an explorative factor analyses with Varimax rotation suggest an eight factor solution with (1) positive introjection, (2) self-efficacy, (3) ability to resist pressure/positive attitude towards demands, (4) confidence, (5) social integration skills, (6) self-structuring ability, (7) sense of self/self-

123

44 care, (8) safe relationship in childhood/development conditions for basic trust. The Pearson-Correlation of the StResQ (Sum Score) with the Connor-Davidson Resilience Scale (Connor and Davidson, 2003) was .84. Data collection is still in progress, but final results of the complete data set will be available by October 2015. Different factor solutions will be compared, items will be reduced (targeted number of items: 30–40) and reliability analyses will be conducted. CONCLUSIONS: A new Stress Resilience Questionnaire was developed. First evaluation shows promising results for content validity. The questionnaire will be validated further. Once finalized the StResQ will be used to identify patients in need of psychological support to allow an adaptive response to stress and therefore increase their health-related quality of life.

(203.2) Age and general self-efficacy are directly associated with physical health related quality of life in Norwegian patients with neuroendocrine tumors Trude Haugland, PhD, Diakonova University College, Oslo, Norway; Astrid K. Wahl, PhD, University of Oslo, Oslo, Norway; Dag Hofoss, PhD, University of Oslo, Oslo, Norway; Holli DeVon, PhD, University of Illinois Chicago, College of Nursing, Chicago, IL, United States AIMS: Being diagnosed with neuroendocrine tumors (NET) may provide challenges due to physical side effects of and mental response to treatment resulting in increased perceived stress and poorer health related quality of life (HRQoL). General self-efficacy, social support and cancer-related stress has been considered a key factor in the coping process in that, problems-solving strategies and social support may influence adjustment to cancer and HRQoL. The aim of the study was to examine the interrelationship between general self-efficacy, social support, cancer-related stress and HRQoL in patients with NET. METHODS: A cross sectional design was chosen to evaluate 196 Norwegian patients. Descriptive analyses were used to assess background characteristics, mental and physical HRQoL, general selfefficacy, social support and cancer-related stress. The interrelationship between general self-efficacy, social support, cancer-related stress and mental and physical components scores was tested using Path Analysis with AMOS21 with maximum standard likelihood estimation. RESULTS: The patients demonstrated decreased physical HRQoL compared to the norm population. The majority of patients demonstrated mild to moderate experience of stress and 56 % of the patients reported low levels (B30) of general self-efficacy. The mean score for social support was 3.1 and 44 % of the patients had a support level above 3 points. Acceptable model fit was obtained for reciprocal causation between general self-efficacy, social support, cancer-related stress and HRQoL. The model demonstrated that age was directly and negatively correlated with physical HRQoL, general self-efficacy and social support. General self-efficacy mediated the effects of age on physical HRQoL. The model demonstrated that those with higher social support experienced less cancer-related stress compared to those with less support and that cancer-related stress was not associated with physical or mental HRQoL. CONCLUSIONS: This analysis demonstrates that general self-efficacy mitigates the effect of aging on physical HRQoL. Furthermore, patients having high social support experienced less cancer-related stress compared to those with less support. Thus, addressing available resources of general self-efficacy and social support in order to support the patients self beliefs and assist establishing social network may be of importance in improving physical HRQoL. Further research should test interventions based on self-efficacy to improve HRQoL.

123

Qual Life Res (2015) 24:1–191 (203.3) Resilience in the context of chronic disease and disability Dagmar Amtmann, PhD, University of Washington, Seattle, WA, United States; Aimee M. Verrall, University of Washington, Seattle, WA, United States; Alexandra L. Terrill, University of Utah, Salt Lake City, UT, United States; Amanda Smith, University of Washington, Seattle, WA, United States; Arielle Silverman, University of Washington, Seattle, WA, United States; Dawn M. Ehde, University of Washington, Seattle, WA, United States; Kevin Alschuler, University of Washington, Seattle, WA, United States; Mark Jensen, PhD, University of Washington, Seattle, WA, United States; Rozanne Wilson, PhD, University of Washington, Seattle, WA, United States; Kurt Johnson, University of Washington, Seattle, WA, United States; Ivan Molton, University of Washington, Seattle, WA, United States AIMS: Resilience is predictive of health and quality of life, however no instruments are freely available that were developed using Item Response Theory (IRT) and thus can be dynamically administered. The purpose of this abstract is to describe the results of the qualitative stage of the development of an IRT-based item bank to measure resilience in adults. Once completed the instrument will be publicly available, psychometrically sound, sensitive to change, flexible, customizable and brief to administer, and appropriate for people with chronic health conditions and disabilities. METHODS: An expert panel of psychologists, clinicians and outcomes measurement experts defined the construct of resilience and developed a set of items measuring resilience. Focus groups of people living with multiple sclerosis (MS), their care partners, and MS advocates were convened to discuss resilience. Cognitive interviews will be conducted with people with chronic pain. Items will be administered to 1600 people with disabilities and calibrated to a two parameter IRT model. RESULTS: Resilience was defined as the capacity to (1) maintain or regain emotional stability and to (2) maintain or regain progress towards valued goals in the face of adversity. Discussions differentiated resilience from self-efficacy, coping strategies, post-traumatic growth, and optimism. A total of 32 people participated in four focus groups. Transcripts from the focus groups were coded by 2 researchers and analyzed using Dedoose on-line software. Participants described resilience as bouncing back, rolling with the disease, or living well. Factors that support resilience included psychological adaptation, social connection, life meaning, planning, and physical wellness. Stakeholders described resilience as problem-solving and stressed the importance of ‘‘flexibility in thinking’’. CONCLUSIONS: Expert panel and focus group feedback was used to define resilience. People with high resilience maintain emotional stability and normal activities when faced with daily hassles and relatively quickly regain emotional stability and normal activities after adverse events. People with MS and their care partners conceptualized resilience differently from the stakeholders who work with or advocate for people with MS.

(203.4) Identifying the relationship between social support and quality of life in a biobehavioral cancer survivorship trial Kathryn Osann, PhD, University of California - Irvine, Irvine, CA, United States, Justin Wilford, PhD, University of California - Irvine, Irvine, CA, United States; Edward Nelson, MD, University of California, Irvine, Irvine, CA, United States; Susie Hsieh, PhD, University of California, Irvine, Irvine, CA, United States; Jo Tucker, PhD, University of California, Irvine, Irvine, CA, United States; Bradley J. Monk, MD, University of Arizona Cancer Center-Phoenix,

Qual Life Res (2015) 24:1–191 Phoenix, AZ, United States; Lari Wenzel, PhD, University of California-Irvine, Irvine, CA, United States AIMS: Benefits of social support (SS) during chronic illness and cancer survivorship are complex. This study examines changes in SS over time to determine its relationship to health-related quality of life (HRQOL) and T-helper type 2 (Th2) cytokine changes. METHODS: We conducted a randomized trial of 204 cervical cancer survivors who had completed definitive treatment and were 9–30 months from diagnosis to test if psychosocial telephone counseling (PTC) could improve HRQOL and other outcomes compared to usual care (UC). The MOS Social Support 19-item survey and Functional Assessment of Cancer Therapy-Cervix (FACT-Cx) were collected at baseline, 4 months and 9 months post enrollment. Biospecimens were collected at each time point to investigate associations with change in patient reported outcomes. Data were analyzed using multivariate general linear models and stepwise linear regression methods. RESULTS: Participants mean age was 43, and 41 % were of Hispanic ethnicity. Significantly more PTC participants experienced increasing SS (54 %) compared to UC (37 %) at 4 (p = 0.016) and at 9 months (PTC: 47 vs. UC 33 %, p = 0.034). Both higher baseline SS and increase in SS over time were significantly associated with improved QOL at 4 (p B 0.001 for each) and 9 months (p \ 0.05 for each) independent of study arm and patient characteristics. Improved SS contributed to significant improvement in each HRQOL domain: physical, functional, social, emotional and additional concerns subscales. Emotional/informational support and tangible support were the most important contributors to improved QOL in stepwise models whereas affection and positive interaction did not contribute independently to prediction of change in QOL. Participants with increasing SS had significant longitudinal decreases in IL-4 and nonsignificant decreases in IL-5 and IL-13, at 4 months. CONCLUSIONS: Participants in PTC were significantly more likely to experience improved SS compared to UC. Improved SS was also significantly associated with improved QOL independent of study arm after adjusting for patient characteristics. Decreasing Th2 cytokines were observed for those with increasing SS, which provides support for an improved capacity to mount an anti-tumor response.

(203.5) Determinants of health utility among caregivers of individuals with multiple sclerosis (MS) Namita Joshi, MS, PhD Candidate, University of Mississippi, San Mateo, CA, United States; Rahul Khanna, PhD, University of Mississippi, Oxford, MS, United States; John P. Bentley, PhD, University of Mississippi, Oxford, MS, United States; Donna WestStrum, PhD, University of Mississippi, Oxford, MS, United States; Glenn A. Phillips, PhD, Biogen, Cambridge, MA, United States; Matthew Strum, Pharm D, University of Mississippi, Oxford, MS, United States; Marie Barnard, University of Mississippi, Oxford, MS, United States AIMS: Adequate information about health state preferences (utility values) among caregivers of individuals with multiple sclerosis (MS) is unavailable. This study assessed health utility among MS caregivers using the EuroQoL-5-dimensions questionnaire (EQ-5D-3L) and compared with population norms in the United States (US). Determinants of health utility were also identified. METHODS: The study utilized a cross–sectional, non–experimental research design. Caregivers were recruited via a MS patient registry—North American Research Committee on Multiple Sclerosis (NARCOMS). A web– based, self-administered survey was distributed to caregivers using Qualtrics survey software. A revised stress, coping, and appraisal model was used as a guiding framework to examine the predictors (care recipient disability status, objective and subjective caregiver burden, caregiver social support, and coping) of health utility among

45 MS caregivers. Descriptive analyses were performed on all variables. Health utility values were compared with US population norms using one-sample t-tests. Path analysis was employed to identify determinants of caregiver health utility and quantify direct and indirect relationships among predictors. Data analyses were conducted using SPSS v22.0 and Mplus v7.11. RESULTS: A total of 215 usable responses were received. The majority of caregivers were male (60.8 %), Caucasian (87.9 %), with a mean age of 56.5 years. Health utility scores among MS caregivers aged 18–44 years were significantly lower than those in the general population (0.83 vs. 0.91). Male caregivers had lower health utility scores compared to their counterparts in the US population (0.83 vs. 0.88). Caregivers reporting greater subjective burden were found to have lower health utility scores (Path coefficient = -0.069). Care recipient disability status had an indirect negative influence on caregiver health utility through its impact on subjective caregiver burden. The relationship between objective caregiver burden and health utility was mediated via ‘avoidance and critical’ coping and subjective burden. CONCLUSIONS: Caregiver burden (objective and subjective), care recipient disability status, and ‘avoidance and critical’ coping were found to influence health utility among MS caregivers. In order to alleviate the burden among MS caregivers, greater resources should be allocated towards provision of support/respite services. Educational efforts are warranted to assist MS caregivers in developing positive coping strategies.

(203.6) Does resilience rest on relationships? Exploring social context and personal resources as moderators of the effect of disability on quality of life in older adults Eithne Sexton, Trinity College Dublin, Dublin, Ireland; Bellinda King-Kallimanis, PhD, Pharmerit, Boston, United States; Richard Layte, Trinity College Dublin, Dublin, Ireland; Anne Hickey, Royal College of Surgeons in Ireland, Dublin, Ireland AIMS: It has been suggested that social relationships are key drivers of resilience throughout the life-course. This paper aims to examine the extent to which social context (supportive relationships, social participation and church attendance) and personal characteristics (personality and personal religiosity) modify the negative effects of disability on quality of life (QoL) in older people. METHODS: We used data from the Irish Longitudinal Study of Ageing (TILDA), a general-population sample of adults aged 50+ (n = 2701). QoL was measured with the two dimensions of the CASP-R12—control/autonomy and self-realisation/pleasure (meaning and enjoyment in life). Disability was measured as number of body function and activity deficits. We used conditional change regression analysis to test the effect of change in disability on change in QoL over 2 years, with interaction terms to examine moderating effects of supportive relationships (partner, relatives, and friends), social participation, church attendance, personality (neuroticism and extraversion) and personal religiosity (importance of religion to the person). Effects of age, sex, education, personality, retirement status and having a low income were adjusted for. RESULTS: Having a supportive partner significantly reduced the negative effect of increased disability on control/ autonomy (interaction effect = 0.11, p = 0.01), as did having a supportive relative (0.08, p = 0.04). Personal religiosity reduced the negative effect of increased disability on meaning and enjoyment in life (0.14, p \ 0.01). Social participation, church attendance and personality did not independently modify effects of increased disability on either QoL dimension. CONCLUSIONS: Social context matters for maintaining QoL in declining physical health. Family relationships may provide key instrumental support, allowing older adults to maintain independence as health declines. Social participation may be directly beneficial for QoL but not enhance resilience in

123

46

Qual Life Res (2015) 24:1–191

the face of declining health. Resilience also depends on personal resources: among those for whom religion is very important, physical health may be less important for living a fulfilling and enjoyable life. Further waves of TILDA data will facilitate analysis of resilience over a longer time period.

United States; Milena D. Anatchkova, PhD, University of Massachusetts Medical School, Evidera, Lexington, MA, United States; Arlene Ash, University of Massachusetts Worcester, Worcester, MA, United States; Catarina Kiefe, University of Massachusetts Worcester, Worcester, MA, United States; Jeroan Allison, University of Massachusetts Worcester, Worcester, MA, United States

204: Cardiac Conditions

AIMS: We identified longitudinal trajectories of health-related quality of life (HRQoL) and examined predictors of these trajectories during 6 months following hospitalization for acute coronary syndrome (ACS). METHODS: We used data from the Transitions, Risks, and Actions in Coronary Events—Center for Cardiovascular Outcomes and Education (TRACE-CORE) prospective cohort of patients hospitalized with ACS. We measured health-related quality of life using the physical components (PCS) and mental component summaries (MCS) of the SF-36 during hospitalization and at 1-, 3-, and 6-months post-discharge. We identified MCS and PCS trajectories and their predictors using latent growth curve models and only individuals who completed surveys for all time points. RESULTS: Participants (N = 658) had mean age 63.7 (SD 10.7) years, 33.0 % female, and 83.6 % non-Hispanic white. We identified 2 MCS trajectories: AVERAGE and IMPAIRED HRQoL. The majority of participants (81.0 %) had AVERAGE MCS at baseline (mean MCS 53.6) and slight improvement in scores over time. A minority (19.0 %) had IMPAIRED HRQoL at baseline (mean MCS 36.7) and slight improvement in scores over time (Fig. 1a). We identified 2 similar

(204.1) Change in individual trajectories of change in patientreported outcomes after implantable cardioverter-defibrillator Sandra B. Lauck, PhD, St. Paul’s Hospital, Vancouver, BC, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; Joy L. Johnson, PhD, University of British Columbia, Vancouver, Canada; Karin Humphries, DSc, University of British Columbia, Vancouver, Canada; Pamela A. Ratner, PhD, Universtiy of British Columbia, Vancouver, Canada AIMS: Some people, because they have severe heart disease or a genetic predisposition, are at high risk for cardiac arrhythmias that could cause their hearts to stop. The implantable cardioverter-defibrillator (ICD) is an effective therapy that recognises abnormal heart beats, can administer an electrical shock to stop a potentially lethal heart rhythm, and affords protection from the devastating consequences of sudden cardiac arrest. We sought to study the change in patient-reported outcomes (PROs) after ICD implantation to identify people’s patterns of change, explore individual trajectories of change, and identify predictors of differences in individuals’ trajectories. METHODS: The study was grounded in the Wilson and Cleary conceptual framework of quality of life and informed by the PatientReported Outcomes Measurement Information System (PROMIS) domain framework. Using a prospective, longitudinal study design, data were obtained from 171 people undergoing ICD implantation at quaternary centres in British Columbia, Canada. PRO assessments were obtained immediately before implantation and at 1, 2, and 6 months following implantation. We employed individual growth modelling to analyse change within and between people. RESULTS: The participants had different physical, mental, and social health status PROs at baseline and, on average, demonstrated improvement. At most of the measurement occasions, the participants’ PROs remained poorer than those of average adult, urbandwelling Canadians. There was significant individual variability in most of the trajectories, especially in the social functioning domains. Relative to men, women reported worse PROs initially; the relative mean difference in men’s and women’s scores ranged from 4.5 to 24.7 % for 6 of the 12 indicators. Yet, the women’s rates of improvement were significantly faster than those of men. Women equalled or exceeded the men’s PROs at the 6-month assessment; the relative mean difference ranged from 4.5 to 10.4 %, depending on the PRO. CONCLUSIONS: Men and women differed in their trajectories of change in self-reported health status, both in terms of their starting points and their rates of change. Further research is needed to explore the individual change trajectories identified in this study, fully test the conceptual model that framed the research, and evaluate interventions aimed at improving PROs after ICD implantation.

(204.2) Trajectories of health-related quality of life following acute coronary syndrome in TRACE-CORE Lisa Nobel, University of Massachusetts Worcester, Worcester, MA, United States; Jennifer Tjia, University of Massachusetts Worcester, Worcester, MA, United States; Jane Saczynski, University of Massachusetts Worcester, Worcester, MA, United States; Molly Waring, University of Massachusetts Worcester, Worcester, MA,

123

Fig. 1 a Trajectories of physical component subscale over time following discharge from an acute coronary syndrome hospitalization. b Trajectories of mental component subscale over time following discharge from an acute coronary syndrome hospitalization

Qual Life Res (2015) 24:1–191 PCS trajectories with similar patterns of scores over time: AVERAGE (71.1 %) and IMPAIRED (28.9 %) HRQoL at baseline (Fig. 1b). With AVERAGE HRQoL as the referent, we found that older age predicted membership in the IMPAIRED MCS and PCS trajectories [PCS Odds Ratio (OR) per 10-years 0.70 (95 % CI 0.51–0.96); MCS OR 0.56 (95 % CI 0.39–0.80)]. Males and non-Hispanic white patients, respectively, were less likely to have IMPAIRED PCS [males PCS OR 0.51 (95 % CI 0.33–0.78); non-Hispanic white race/ethnicity PCS OR 0.56 (95 % CI 0.33–0.96)] but not IMPAIRED MCS. Individuals with myocardial infarction were less likely than those with unstable angina to belong to the IMPAIRED trajectory for MCS but not PCS [MCS NSTEMI OR 0.40 (95 % CI 0.24–0.67) and STEMI OR 0.35 (95 % CI 0.17–0.75)]. CONCLUSIONS: 10 % of ACS patients experience impaired mental HRQoL during the 6 months following hospital discharge, and 29 % experience impaired physical HRQoL. Predictors of these impairments differ from those of mental impairments. Identifying patients with poorer expected trajectories may help target early interventions and improve post-hospitalization quality of life.

(204.3) Patient-centered outcomes in treatment of small abdominal aortic aneurysms—a qualitative interview study Stephanie M. Tomee, PhD candidate, Leiden University Medical Centre, Leiden, Netherlands; Winifred A. Gebhardt, Associate Professor of Health Psychology, Leiden University, Leiden, Netherlands; Jan H. Lindeman, Associate professor of Surgery, Leiden University Medical Centre, Leiden, Netherlands; Jaap F. Hamming, Professor of Surgery, Leiden University Medical Centre, Leiden, Netherlands AIMS: An abdominal aortic aneurysm (AAA) is a potentially lethal disease, which is relatively common in elderly men. As repair has only proven benefit above 5.5 cm, quite a large group of patients with smaller aneurysms do not receive specific treatment. They are in a watchful-waiting period in which aorta diameter is assessed by ultrasound every 6–12 months. Therefore, these patients have little contact with their vascular specialist or other hospital care personnel. Qualitative data from patients with small AAA’s is scarce and little is known about the psychological impact of living with a small AAA. More insights into patients’ view on diagnosis and disease burden of small AAA is important to provide patients with additional support in coping this disease. METHODS: For this purpose we perform ten indepth semi-structured qualitative interviews with AAA patients who are in a watchful-waiting period. After the interview patients are asked to fill in four questionnaires assessing: health related quality of life (SF-36), illness perceptions (IPQ), anxiety and depression (HADS) and general demographic information. All interviews are audio taped and transcribed verbatim. Interview data will be coded and analyzed according to the constant comparative method (Glaser & Strauss, 1967). After ten completed interviews, data saturation will be evaluated. Additional interviews will be performed until data saturation has been reached. RESULTS: Preliminary analyses of data derived from six interviews shows that patients often have little knowledge on AAA-disease and treatment options. However, they accept the conservative watchful-waiting plan proposed by the surgeon. In addition, a response shift phenomena appeared to be exist, as patients rated their own quality of life as highly positive, despite severe comorbidity and sometimes uncertain life expectancy. Patients don’t experience an AAA as a disease due to the absence of symptoms and active intervention. Aneurysm patients feel they have little control over their AAA and largely rely on the surgeon for judgment. CONCLUSIONS: Patients with small AAA are generally content with a conservative treatment plan and consider themselves to enjoy a good health. In addition, patients depend on their vascular specialist to take control in AAA treatment.

47 (204.4) What impacts recovery after cardiac surgery? A mixed method approach Lisa M. Fornaresio, PhD, Inova Heart and Vascular Institute, Falls Church, VA, United States; Sari D. Holmes, PhD, Inova Heart and Vascular Institute, Falls Church, VA, United States; Niv Ad, MD, Inova Heart and Vascular Institute, Falls Church, VA, United States AIMS: Cardiac surgery represents a major challenge for patients physically and emotionally. Improving clinical outcomes postoperatively remains a primary focus. As adverse events have been minimized, there are opportunities to improve the subjective patient experience after cardiac surgery. The purpose of this study was to use a mixed method approach to understand the recovery experience of cardiac surgery patients. METHODS: This study consisted of two study phases. In Phase I (qualitative), focus groups were conducted with patients who had cardiac surgery (N = 132). In Phase II (quantitative), surveys were mailed to all cardiac surgery patients with complete baseline HRQL data, and 712 patients returned the survey. Based on themes from Phase I, regressions were conducted for selfreported ‘‘recovery’’ from Phase II including: time since surgery, age, gender, euroSCORE II risk score, cardiac rehabilitation, and social support (ISEL). RESULTS: Qualitative analyses from focus groups revealed gender differences in recovery duration; men described *6 months and women *1 year. Social support was noted as important in recovery, especially for women. Patients found cardiac rehabilitation valuable, noting comfort in supervised exercise as well as social benefits from other patients. Phase II mean age was 65 ± 11 years and 24 % were female. For patients who reported feeling recovered (n = 454), mean time to recovery was 7 ± 9 months. Longer time to self-reported recovery was independently associated with female gender (B = 3.41, P = 0.002) and unexpectedly by participation in cardiac rehabilitation (B = 2.16, P = 0.014). Factors independently associated with recovery by 12 months were lower operative risk via euroSCORE II (OR = 0.93, P = 0.019) and greater social support (OR = 1.04, P = 0.019). Patients who self-reported feeling fully recovered by 12 months had greater improvement in the SF-12 physical composite score (F = 6.69, P = 0.003) and the SF-12 mental composite score (F = 4.90, P = 0.010) from before surgery to 12 months after surgery (Fig. 1). CONCLUSIONS: Results indicated the importance of

Fig. 1 . social support, cardiac rehabilitation, and gender in the recovery experience of cardiac surgery patients. Considering oneself recovered was shown to impact HRQL up to 12 months after surgery. These results can be used to educate and prepare patients and clinicians for the experience of patients after cardiac surgery.

(204.5) Not so transient: fatigue, psychological and cognitive impairment following transient ischaemic attack (TIA) Grace M. Moran, University of Birmingham, Birmingham, United Kingdom; Tom Marshall, University of Birmingham, Birmingham, United Kingdom; Melanie J. Calvert, PhD, University of

123

48 Birmingham, Birmingham, United Kingdom; Max Feltham, University of Birmingham, Birmingham, United Kingdom; Ronan Ryan, University of Birmingham, Birmingham, United Kingdom AIMS: Transient ischemic attack (TIA) is defined by short-lasting, stroke-like symptoms and treatment is focused on secondary stroke prevention. However, a recent systematic review suggested that TIA patients may experience ongoing residual impairments. TIA patients are followed up in primary care; therefore, it is important for family practitioners to understand the holistic consequences of TIA. We aimed to investigate whether TIA is associated with subsequent consultation for fatigue, psychological or cognitive impairment in primary care. METHODS: A retrospective matched cohort study of first-ever TIA patients using anonymised electronic primary care medical records from The Health Improvement Network (THIN), a primary care database which covers approximately 6 % of the UK population. Stroke-free patients who experienced a first-ever TIA between 2009 and 2013 were matched (approximately 1:5) to strokefree controls by age (±2 years), sex and family practice. Follow up was censored at date of death, stroke or leaving the database. Outcomes were the first consultation for fatigue, psychological or cognitive impairments. Kaplan–Meier (K-M) survivor functions estimated time to first consultation with log-rank tests to compare TIA and control patients. RESULTS: 9419 TIA patients were matched to 46,511 controls. There were a statistically significant differences (P \ 0.0001) in K-M curves between TIA and control patients for all three impairments. The time taken for a quarter of the sample to consult for each of the impairments was significantly shorter in TIA patients than controls: fatigue 18 versus 46 months; psychological impairment 3 versus 13 months; and cognitive impairment 25 versus 51 months. CONCLUSIONS: The findings suggest that TIA may be more than a transient event and that these patients may experience long-term impairments. Residual impairments are likely to present in primary care; therefore, family practitioners may need to consider more than just secondary stroke prevention when treating TIA patients. Future research should investigate if integration of electronic patient reported outcomes into routine primary care software could facilitate the identification of residual impairments post-TIA and improve patient centred care.

205: Meaningful Change (205.1) Determining meaningful cutpoints for the interference items of the MD Anderson Symptom Inventory Qiuling Shi, MD, PhD, MD Anderson Cancer center, Houston, TX, United States; Tito Mendoza, PhD, UT MD Anderson Cancer Center, Houston, TX, United States; Xin Shelley Wang, MD, MPH, MD Anderson Cancer Center, Houston, TX, United States; Charles S. Cleeland, PhD, UT MD Anderson Cancer Center, Houston, TX, United States AIMS: The MD Anderson Symptom Inventory (MDASI)-interference items measure symptom impact on functional status. However, the interpretation of the MDASI-interference in practice is challenged by lack of a clear delineation of meaningful threshold for function levels. We conducted a secondary analysis to statistically establish good/ poor cutpoints for MDASI interference scale-measured functional status in cancer patients. METHODS: Data were extracted from an aggregated cancer symptom database of the Department of Symptom Research at MD Anderson Cancer Center. Total 3809 cancer patients completed MDASI and ECOG Performance status (PS) measures. Using ECOG-PS (good vs. poor) as the anchor, receiver operating characteristics (ROC) curves were used to identify cutpoints for 6 MDASI-interference scales. The samples were equally divided and randomized into two datasets: development and validation. External

123

Qual Life Res (2015) 24:1–191 validation was conducted by comparing SF12 physical and mental component scores between good and poor MDASI-interference scales in 1102 patients. RESULTS: ROC analyses showed that all MDASI interference items had area under the curve (AUC) over 0.7 (P \ .0001), except for relations with others (AUC = 0.65, P \ .0001). The optimal cut point was 4 for general activity, 3 for walking, 4 for work, 3 for mood, 2 for relations with others, and 3 for enjoyment of life. Using the external validation dataset, SF12-physical component (PCS) and mental component scores (MCS) were significantly higher in patients with good scores than those reporting poor scores of all 6 MDASI-interference items. Large Cohen’s d effect sizes were found for PCS between good and poor general activity (1.45), walking (1.43) and work (1.35) and for MCS between good and poor mood (1.29), relations with others (1.12), and enjoyment of life (1.14). CONCLUSIONS: The current study statistically defined functional impairment categories. These cutpoints of good/ poor MDASI-interference scores offer clinically interpretable parameters for reporting symptom-induced functional impairment in cancer patients and may be useful in clinical trial design and treatment decision making.

(205.2) A comparison of effect size measures for differential item functioning analysis with ordinal logistic regression techniques in a PROMIS anxiety measure Michelle M. Langer, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Caroleen W. Quach, MSPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Deborah S. Usinger, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Ronald C. Chen, MD, MPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States AIMS: The literature offers few empirical applications and little guidance for which effect size measure is appropriate for differential item functioning (DIF) detection using ordinal logistic regression (OLR) methods. Assessing DIF in health-related quality of life (HRQoL) instruments often offer additional challenges due to few items and skewed score distributions. Effect size measures offer investigators a sense of magnitude of DIF impact. This study will compare 9 effect size measures for detecting DIF with OLR. METHODS: The 5-item PROMIS Anxiety measure was administered 1 year after treatment initiation to prostate cancer patients. OLR models were fit for 527 men with more than high school education ([HS) versus 246 men with less education (BHS). Uniform DIF was evaluated by comparing a model with summed score to one adding an education level variable. Non-uniform DIF was evaluated comparing the latter model with one including a summed score*education interaction. Effect size was measured by: change in R-square, change in summed score regression parameter, significance of education level regression parameter, the conditional log odds ratio for education level, comparison of mean predicted summed score by education level, clinical relevance criteria, change in predicted item and summed score compared to minimally important difference, graphical approaches. RESULTS: The PROMIS summed score distribution was positively skewed. Mean score for [HS was 7.03 (SD = 3.28) versus 7.19 (SD = 3.60) for BHS. Item 4 ‘‘I felt uneasy’’ was flagged for significant uniform DIF (p \ .005). Item mean for [HS was 1.54 (SD = 0.87) versus 1.47 (SD = 0.86) for BHS. Figure 1 depicts the expected score given the OLR coefficients. The potential DIF in item 4 was considered negligible according to 7 of 8 effect size measures. However, a 1.81 conditional log odds ratio for education level suggested large DIF ([1.5). CONCLUSIONS: Despite a short test and skewed score distribution, OLR methods detected significant DIF by

Qual Life Res (2015) 24:1–191

49 PGTCS thresholds were consistently higher than POS on Seizure Worry and Daily Activities across both methods. CONCLUSIONS: These results support the importance of calculating separate thresholds for populations of differing severity. Additionally, the high anchor-based thresholds for PGTCS patients on Seizure Worry and Daily Activities suggest that these may be the most responsive domains in PGTCS clinical responders.

(205.4) Interpretation of verbal descriptors for response options commonly used in verbal rating scales in patient-reported outcome instruments Fig. 1 Item 4 expected score for OLR model accounting for DIF education level education level in one item. However, the magnitude of DIF was unclear given conflicting effect size measures. The practical impact of DIF is critically important for evaluating HRQoL in multiple populations. Future research should incorporate both simulation studies and empirical applications implementing a wide variety of effect size measures.

(205.3) Minimal important difference (MID) thresholds for the QOLIE-31-P in epilepsy patients with primary generalized tonic–clonic seizures (PGTCS) Stacie Hudgens, MA (ABD), Clinical Outcomes Solutions, Tucson, AZ, United States; Wan Tsong, M.S., Eisai, Inc., Woodcliff Lake, NJ, United States; Anna Forsythe, PharmD, MSc, MBA, Eisai, Inc., Woodcliff Lake, NJ, United States; Francesco Bibbiani, MD, Eisai, Inc., Woodcliff Lake, NJ, United States; Antonio Laurenza, MD, Eisai, Inc., Woodcliff Lake, NJ, United States AIMS: Our objective is to estimate minimal important difference (MID) thresholds for the Quality of Life in Epilepsy Inventory (QOLIE-31-P) in epilepsy patients diagnosed with primary generalized tonic–clonic seizures (PGTCS). We utilized data from a multinational, phase III, double-blind, randomized, controlled trial designed to evaluate the efficacy of adding perampanel to the standard of care in a population of PGTCS patients. While MID values have been previously calculated for partial onset seizure (POS) patients, PGTCS patients experience seizures which are less frequent but more severe suggesting that a meaningful change for a POS patient may be different than that for a PGTCS patient. METHODS: QOLIE-31-P data were collected for 96 patients in which a valid translation was available out of a total 182 patients with efficacy data. The QOLIE31-P subset did not appear to be biased as demographic and efficacy values were similar to the overall population. MID thresholds were derived using distribution- and anchor-based methods. Distributionbased methods included the calculation of the standard error of measurement (SEM) for multi-item domains [SEM = standard deviation (SD) 9 (1 - reliability)] and ‘ SD for the single item domain. Clinical responders (defined as a 50 % or greater reduction in PGTCS) were used to estimate the anchor-based MID thresholds. RESULTS: Distribution-based thresholds ranged between 6.11 and 15.30 as measured by the SEM at baseline for the multi-item domains with the greatest values for Medication Effects (15.3), Seizure Worry (14.04), and Daily Activities (14.17). For the single item ‘‘health status’’, ‘ SD was calculated as 10.79. Anchor-based thresholds ranged between 3.1 and 12.1 with the greatest values for Seizure Worry (12.1) and Daily Activities (7.8). The greatest differences between our calculated PGTCS MID threshold and published POS MID thresholds were on Seizure Worry, Energy, and Daily Activities.

Alex Mutebi, PhD, Amgen, Thousand Oaks, CA, United States; Marion Slack, PhD, University of Arizona, Tucson, AZ, United States; Terri L. Warholak, PhD, University of Arizona, Tucson, AZ, United States; Stacie Hudgens, MA (ABD), Clinical Outcomes Solutions, Tucson, AZ, United States; Stephen J. Coons, PhD, Critical Path Institute, Tucson, AZ, United States AIMS: Assess the interpretation of verbal descriptors commonly used in response scales of patient-reported outcome (PRO) instruments; identify verbal descriptors whose interpretation yields closest to interval level data; and examine the association between demographic factors and the interpretation of verbal descriptors. METHODS: Subjects were recruited and screened through an online drug–drug interaction service (MediGuard.org). Using an 11-point scale (0 = lowest possible, 10 = highest possible), subjects assigned scores to the interpretations of verbal descriptors in 5 commonly used sets: Set 1 (none, mild, moderate, severe, very severe); set 2 (never, rarely, sometimes, often, always); set 3 (poor, fair, good, very good, excellent); set 4 (not at all, a little bit, moderately, quite a bit, extremely); and set 5 (not at all, a little bit, somewhat, quite a bit, very much). Paired t-tests (adjusting for multiple comparisons) were used to test for equality between mean scores assigned to the 1st, 2nd, 3rd, 4th, and 5th verbal descriptors in each set and the fixed intervals of 0.0, 2.5, 5.0, 7.5, 10.0 respectively. Ordinal regression assessed adjusted associations between demographic factors and scores assigned to verbal descriptors. RESULTS: This Study included 350 subjects, 64 % females, with mean (SD) age of 56.9 (12.1) years, and median of 5 health conditions per subject. Set 1 had 3 verbal descriptors with mean scores not significantly different than their corresponding fixed intervals: ‘‘none’’ = 0.34 (p = 0.10), ‘‘mild’’ = 2.46 (p = 0.55), and ‘‘moderate’’ = 5.01 (p = 0.90). Other notable mean scores included: ‘‘often’’ = 7.50 (p = 0.95); ‘‘quite a bit’’ = 7.42 (p = 0.26) [set 4] and 7.68 (p = 0.04) [set 5]. Scores assigned to all other verbal descriptors were significantly different than their corresponding fixed intervals (p B 0.01). Significant associations (p B 0.05) included: Age and scores for ‘‘sometimes,’’ ‘‘poor,’’ and ‘‘somewhat’’; sex and scores for ‘‘very severe’’; and level of education and scores for ‘‘sometimes.’’ CONCLUSIONS: Response scales using ‘‘none,’’ ‘‘mild,’’ ‘‘moderate,’’ ‘‘severe,’’ ‘‘very severe’’ yield closest to interval level data. Analyses of non-randomized data collected using verbal descriptors should consider adjusting for age, sex, and levels of education.

(205.5) What is a meaningful change? Ask the patient! Josephine M. Norquist, Merck Sharp & Dohme Corp., North Wales, PA, United States; Tjeerd Korver, Merck Sharp & Dohme Corp., North Wales, PA, United States; Fang Chen, Merck Sharp & Dohme Corp., North Wales, PA, United States; Rob Arbuckle, Adelphi Values, Cheshire, United Kingdom; Alice Turbull, Adelphi Values, Cheshire, United Kingdom; Allison M. Nguyen, Merck Sharp & Dohme Corp., North Wales, PA, United States

123

50 AIMS: Methods for interpretation of objective and subjective clinical trial endpoints have been developed and debated over decades. However, challenges remain in understanding the degree of change considered important to patients, physicians, payers, and regulators. This abstract summarizes methodologies to determine meaningful change in pain scores based on direct patient input. METHODS: Blinded data from a multi-national, RCT among adult women with primary dysmenorrhea were used to qualitatively compare minimally important difference (MID) (anchor [A] and distribution [D]-based), responsiveness, receiver operating characteristic (ROC) curves and minimal detectable change (MDC) 90 and 95 % confidence levels, to identify a peak pelvic pain change score considered important to patients. Pelvic pain was assessed on a 0–10 NRS in the Dysmenorrhea Daily Diary (DysDD). A Global Assessment of Change (GAC) questionnaire, administered at end of study (Treatment Cycle 2), asked patients to rate their change since the start of the study on a 7-point scale (‘much worse’ to ‘much better’) and to report whether the change was meaningful (‘‘Was this increase/decrease in pelvic pain an important change for you?’’). Including a question on the significance of change is a new approach which provided insight into the patients’ ratings of change. RESULTS: Data from 335 patients were analyzed. MID-A analyses based on GAC of ‘a little better’ suggested an MID in DysDD peak pelvic pain of -2.9. Based on ROC analyses, a reduction of C3 in peak pelvic pain was the best predictor of GAC ratings of ‘‘a little better’’, ‘‘better’’ or ‘‘much better’’. Among subjects achieving a C3 reduction, 95.6 % reported that the reduction was important. Supportive analyses from MID-D estimation suggested MIDs of 1.6 (0.5 SD), 1.7 (SEM), 2.8 (MDC90) and 3.4 (MDC95). CONCLUSIONS: Directly asking subjects to rate the significance of their change, combined with MID-A, responsiveness and ROC curve analyses, provided a novel approach to determine and substantiate a meaningful change in peak pelvic pain. Results were supported by the distribution-based methods. Taken together, a C3-point reduction in peak pelvic pain, as assessed in the DysDD, is considered a meaningful reduction that can be used in future clinical trials.

206: Pulmonary Conditions (206.1) Update of the EORTC questionnaire for assessing quality of life in lung cancer patients (EORTC QLQ-LC13) Michael Koller, University Hospital Regensburg, Regensburg, Germany; Marianne J. Hjermstad, Regional Centre for Excellence in Palliative Care, Department of Oncology, Oslo University Hospital, Oslo, Norway; Krzysztof Tomaszewski, Department of Anatomy, Jagiellonian University Medical College, Krakow, Poland; Amelie Harle, The Christie NHS Foundation Trust, Manchester, United Kingdom; Juan I. Arraras, Oncology Departments, Complejo Hospitalario de Navarra, Pamplona, Spain; Ofir Morag, Chaim Sheba Medical Center, Ramat-Gan, Israel; Cecilia Pompili, St.James’s University Hospital, Leeds, United Kingdom; Georgios Ioannidis, Nicosia General Hospital Cyprus, Nicosia, Cyprus; Loukia Georgiou, Bank of Cyprus Oncology Centre, Nicosia, Cyprus; Morfo Georgiou, Bank of Cyprus Oncology Centre, Nicosia, Cyprus; Chiara Navarra, Rome, Italy; Wei-Chu Chie, Institute of Epidemiology and Preventive Medicine and Department of Public Health, College of Public Health, National Taiwan University, Taipei, Republic of Taiwan; Colin Johnson, Surgical Unit, University of Southampton, Southampton, United Kingdom; Ayje Himpel, University Hospital Regensburg, Regensburg, Germany; Christian Schulz, University Hospital Regensburg, Regensburg, Germany; Hans-Stefan Hofmann, University Hospital Regensburg, Regensburg, Germany; Dagmara Kulis, EORTC, Brussels, Belgium; Andrew Bottomley, EORTC Headquarters, Brussels, Belgium

123

Qual Life Res (2015) 24:1–191 AIMS: The EORTC QLQ-LC13 was the first module developed to be used in conjunction with the core questionnaire C30. Since the publication of the LC13 in 1994, major advances in the treatment of lung cancer as well as in the assessment of quality of life have been made. Therefore, the EORTC decided to support a project to update the LC13. METHODS: In accordance with the EORTC Module Development Manual a classical Phase I-III project was initiated. Phase I generated a list of 110 quality of life issues that are potentially relevant for lung cancer. Phase II resulted in a 78 item provisional lung cancer module. The provisional module was translated into Chinese (Taiwan), German, Greek, Hebrew, Italian, Norwegian, Polish, and Spanish. An international multicentre Phase III study was conducted to test the provisional module with regard to relevance, comprehensibility, and acceptance. RESULTS: A total of 12 centers from Cyprus, Germany, Italy, Israel Spain, Norway, Poland, Taiwan, and UK participated in this study. 200 patients with histologically confirmed lung cancer were enrolled. Mean age was 64 years (range 39–91). 59 % of the patients were male, 82 % had NSCLC, 56 % of the patients were treated with a palliative approach. The items yielding the highest importance ratings ([90 % of the patients checking ‘‘yes, include’’) tapped into worry about the future, shortness of breath when walking, fear of tumor progression, loss of energy, fear of pain, and decrease in physical capabilities. None of these high ranking items had been included in the original LC13. CONCLUSIONS: Further analyses will be performed to investigate other item criteria (distribution, missing values, intrusiveness). The final decision for an updated Phase III module will be consented in a group of experts. The first results make clear that it was necessary and timely to come up with a revised module that better fits the needs of patients and keeps up with the opportunities and side effects new therapeutic approaches.

(206.2) The longitudinal relationship between asthma control and quality of life in children: a report from the PROMIS Pediatric Asthma Study Zheng Li, University of Florida, Gainesville, FL, United States; Heather Gross, University of North Carolina at Chapel Hill, Chapel Hill, United States; Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Darren DeWalt, MD MPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; I-Chan Huang, PhD, St. Jude Children’s Research Hospital, Memphis, TN, United States AIMS: This study described the trajectories of health-related quality of life (HRQOL) over a 2-year period, and examined the effect of asthma control status on HRQOL by taking baseline socio-demographic factors into consideration. METHODS: A total of 238 pairs of asthmatic children and their parents were recruited for assessing asthma control status and HRQOL from baseline to 3 separate followups over 2 years. Asthma control status was reported by parents using the Asthma Control and Communication Instrument. HRQOL that captures functional limitations of asthma was reported by children using the Patient-Reported Outcomes Measurement Information System (PROMIS) Pediatric Asthma Impact Scale. Bivariate analyses were applied to examine the relationship among asthma control status, HRQOL, and participants’ socio-demographic characteristics. Unconditional latent growth modeling (LGM) was performed to examine the trajectories of HRQOL during the observational period, and conditional LGM was conducted to examine the contribution of asthma control and participants’ socio-demographic characteristics to the changes of HRQOL. RESULTS: Bivariate analyses revealed that children who were black, had more chronic health conditions, and had parents with lower levels of education (high school or below) were associated to poorly-controlled asthma status (p’s \ 0.5). Children

Qual Life Res (2015) 24:1–191 with poor asthma control and being female, and parents with lower levels of education (high school or below) (p’s \ 0.1) were associated with worse pediatric HRQOL. Unconditional LGM showed that HRQOL improved over time. Conditional LGM suggested that taking into account asthma control and participants’ socio-demographic characteristics, the variation in the baseline level of HRQOL was significant, yet the rate of change over time was not. Conditional LGM also revealed that poorly-controlled asthma was associated with impaired HRQOL at each time point of assessment (p’s \ 0.05). Low parental education was associated with lower baseline HRQOL (p’s \ 0.05). Hispanic children had slower rates of change of HRQOL (p’s \ 0.01) than non-Hispanic White children. CONCLUSIONS: Asthma control status is associated with pediatric HRQOL; parental education and children’s race/ethnicity influence the baseline status and rate of change of HRQOL, respectively. This study provides a foundation for identifying vulnerable children at risk for poorly controlled asthma and impaired HRQOL, which can inform how interventions are developed and implemented.

(206.3) Use of concept elicitation interviews to determine potential differences in disease-related symptom concepts between earlyversus advanced-stage non-small cell lung cancer (NSCLC) patients Thomas M. Atkinson, PhD., Memorial Sloan Kettering Cancer Center, New York, NY, United States; Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States; Kelly P. McCarrier, PhD, Health Research Associates, Inc, Seattle, WA, United States; Michael Scanlon, Health Research Associates, Inc., Seattle, WA, United States; Lisa M. Peterson, Memorial Sloan Kettering Cancer Center, New York, NY, United States; Jean Marie Arduino, Merck & Co., Inc., Kenilworth, NJ, United States; Vasudha Bal, Novartis Pharmaceuticals Corporation, East Hanover, NJ, United States; Kendra DeBusk, Genentech, Inc., South San Francisco, CA, United States; Dagmar Kaschinski, Boehringer Ingelheim GmbH, Ingelheim, Germany; Astra M. Liepa, Eli Lilly and Company, Indianapolis, IN, United States; Stephen J. Coons, PhD, Critical Path Institute, Tucson, AZ, United States, On behalf of the PatientReported Outcome (PRO) Consortium’s Non-Small Cell Lung Cancer Working Group AIMS: Prior research has demonstrated that early-stage (i.e., Stage I-II NSCLC) patients are relatively asymptomatic compared to advanced-stage (i.e., Stage III-IV) patients. Given the regulatory expectations outlined in the U.S. Food and Drug Administration’s 2009 guidance on the use of patient-reported outcome (PRO) measures to support product labeling claims, it is important to empirically establish patient experiences using qualitative methodology. This report describes symptom concepts for early-stage NSCLC patients, relative to advanced-stage patients. METHODS: As part of an effort to develop a new PRO measure meeting FDA requirements to assess NSCLC symptoms as an endpoint in clinical trials, concept elicitation interviews were conducted in a non-random purposive patient sample. Patients were eligible if they were age 18 years or older, English speaking, had a diagnosis of Stage I-IV NSCLC, were treatment naı¨ve or had recovered from prior treatment-related adverse events, and had an ECOG Performance Status of 0–2. Patients were asked to consider both pre- and post-NSCLC diagnosis and (1) spontaneously report NSCLC disease symptoms and (2) state whether they recognized having experienced specific symptoms from an expert-derived disease model. RESULTS: A total of 51 patients (mean age = 64.8 years; 51 % female) diagnosed with NSCLC (12 % early-stage, 88 % advanced-stage) were recruited from six U.S. clinical sites. Patients identified concepts within the following sub-domains: fatigue/tiredness, pain/discomfort, respiratory symptoms, digestive symptoms,

51 and other symptoms. Early-stage patients experienced 29 % of the total concepts identified prior to NSCLC diagnosis, but only reported having experienced 7 % of the total concepts identified post-NSCLC diagnosis. No unique symptom concepts were identified by earlystage patients. Concepts common across stages included tiredness, chest pain, coughing, and dyspnea, whereas concepts such as weakness, dizziness, difficulty swallowing, and taste change were only raised by advanced-stage patients. CONCLUSIONS: Patients with early-stage NSCLC reported fewer symptom concepts compared to the advanced-stage patients. However, there is a great deal of overlap of symptoms in both groups, such as tiredness, chest pain and coughing. Despite the small number of early-stage patients in our sample (n = 6), these findings support the notion that regardless of severity, key disease-related symptom concepts are consistent across the full range of NSCLC disease stages.

(206.4) A matched-samples approach to evaluating the impact of COPD on quality of life, healthcare utilization, and asthma burden Brian D. Stucky, PhD, RAND Corporation, Santa Monica, CA, United States; Cathy Sherbourne, The RAND Corporation, Santa Monica, CA, United States; Nicole Eberhart, PhD, RAND Corporation, Santa Monica, CA, United States; Maria O. Edelen, PhD, RAND Corp, Boston, MA, United States AIMS: There has been recent attention to understanding different subtypes of asthma and COPD, including an asthma-COPD phenotype in which individuals have both disorders. However, in describing asthma-COPD, it is unclear whether outcomes are specific to this particular comorbidity, or just a function of having any comorbidity. To disentangle this issue, we used a matched samples design to evaluate the role asthma, COPD, or other comorbid conditions on various domains of QoL, healthcare utilization, and asthma factors. METHODS: Using data from an initial sample of 2032 people with asthma, we derived three subsamples that were matched according to gender and age. The three groups were demographically equivalent and include patients with (1) ‘‘Asthma-only’’ (i.e., no other comorbid conditions; N = 97), (2) asthma and other non-COPD comorbid conditions (e.g., diabetes, heart disease, sinusitis, etc.; ‘‘Asthma-comorbid’’; N = 96), and (3) asthma and COPD (‘‘Asthma-COPD’’; N = 97). ANOVA and Cohen’s d effects size estimates were used to estimate the differences between these groups on various QoL asthma-specific QoL, PROMIS generic QoL, healthcare utilization, and asthma symptoms and control. RESULTS: Differences between the Asthma-only and the Asthma-comorbid conditions matched groups were small; however, the differences between matched Asthma-COPD group and both the Asthma-only and Asthma-comorbid groups were significant, large, and indicated worse outcomes for Asthma-COPD. For example, across PROMIS QoL domains including Sleep Disturbance, Social Roles and Activities, Anxiety, Physical Function, and Global Mental Health and Global Physical Health, the Asthma-COPD group had lower quality of life relative to the Asthma-comorbid and Asthma-only groups, with effect sizes ranging from 0.69 to 1.54. Across scales measuring asthma symptoms, control, and asthma-specific QoL differences between these groups were also significant with effect sizes ranging from 1.02 to 1.39. However, the differences across health care utilization variables (e.g., hospital visits, nights spent at hospital, etc.), though statistically significant, were smaller in magnitude (d range = 0.51–0.58). CONCLUSIONS: Research evaluating the impact of asthma on QoL outcomes is often confounded by other chronic medical conditions such as COPD. In this study, we found that COPD, above and beyond age, gender, asthma and other comorbid conditions, strongly impacts QoL and other asthma outcomes.

123

52 (206.5) Quality of life trajectory during surgical treatment for NSCLC: preliminary analysis on baseline evaluation on 108 patients Cecilia Pompili, St.James’s University Hospital, Leeds, United Kingdom; Alberto Sandri, St James University Hospital, Leeds, United Kingdom; Jacintha Lee Mei Ying, University of Leeds, Leeds, United Kingdom; Tadg O’Connor, University of Leeds, Leeds, United Kingdom; John White, St James University Hospital, Leeds, United Kingdom; Sandra Dixon, St James University Hospital, Leeds, United Kingdom; Alessandro Brunelli, St James University Hospital, Leeds, United Kingdom; Jonathan Robson, MD, St James’s University Hospital, Leeds, United Kingdom AIMS: Patient reported outcomes are becoming increasingly important in the evaluation of the thoracic surgical candidate.Lung resection for cancer should aim at improving survival and symptoms without compromising the dignity of an acceptable quality of life. The objective of this investigation was to assess whether age or gender were associated with differences in baseline quality of life (QOL) in lung resection candidates. The study may serve as a preliminary analysis to assist future investigations on residual QOL after lung resection. METHODS: This is a retrospective analysis on prospectively collected data on 108 consecutive patients submitted to curative anatomic lung resection for non-small cell lung carcinoma during 6 months in a single centre. All patients were administered an EORTC QLQC30 and LC13 module within 2 weeks before operation during their outpatient clinic consultation. Patients were grouped by age (older or younger than 70 years) and gender and the scores of each quality of life domain were compared between groups by using Mann–Whitney test. To assess the clinical importance of the difference we used the Cohen’s effect size method (an effect size of 0.5 represent a medium difference, an effect size of 0.8 represents a large difference). RESULTS: 59 (55 %) patients were older than 70 years of age. Compared to younger patients, the elderly had a significantly higher emotional functioning score (effect size 0.6, p \ 0.001), lower nausea and vomiting symptoms (effect size 0.55, p \ 0.001), lower insomnia (effect size 0.51, p = 0.007), lower appetite loss (effect size 0.9, p \ 0.001), lower financial difficulties (effect size 0.7, p \ 0.001). The elderly group had also lower symptoms of sore mouth (effect size 0.51, p = 0.008). No significant differences were observed in the other QOL domains. 62 (57 %) patients were female. Compared to male patients, females had higher insomnia score (effect size 0.63, p = 0.001), while no differences were observed in the other QOL domains. CONCLUSIONS: We were able to find that most of the QOL domains were similar in older and younger patients and in both genders. The elderly reported less symptoms and a better emotional functioning than younger patients. These findings may be used as baseline in future studies evaluating the perioperative changes of QOL.

207: Health Utilities (207.1) Estimating health utility scores for children with autism using a mapping algorithm: secondary analysis using NDAR Nalin Payakachat, PhD, University of Arkansas for Medical Sciences, Little Rock, AR, United States; J Mick Tilford, PhD, University of Arkansas for Medical Sciences, Little Rock, AR, United States AIMS: Comparative effectiveness of interventions for children with autism that incorporates cost is lacking due to the scarcity of information on health utility scores (HU) typically used for calculating quality-adjusted life years. We developed algorithms for mapping clinical and behavioral measures for children with autism to HU. This study estimates HU using data shared in the National Database for Autism Research (NDAR). NDAR is an NIH-funded research data

123

Qual Life Res (2015) 24:1–191 repository created to accelerate autism research by integrating heterogeneous datasets through data sharing. METHODS: We reviewed projects that shared their data in NDAR and decided to use a project from the University of Washington-Autism Center of Excellence, Early Connections because it contained longitudinal clinical and behavioral outcomes. The Early Connections Study is designed to enhance social motivation and promote early social engagement and reciprocity for improving behavioral outcomes for infants who may be at risk for developing autism. All needed data files were then linked using a global unique identifier (GUID) to create an analytical file. The data file contained GUID, child age of interview, phenotypes, Mullen IQ scores, and Vineland-II Adaptive Behavioral Scale scores. We excluded subjects that did not have follow-up data from the final analytic file. A mapping algorithm was generated using data from our previous study: HU = (-0.4359) + (0.009316 * VinelandII composite score) + (0.01933 * child age) + (-0.000261 * child age^2) + (0.072534 * log(IQ)). RESULTS: 74 subjects were included in this study with 3 follow-up at age 12, 18, and 24 months old. The majority were boys (61 %) with an average age of 6.5 months old at baseline. Children in this study had normal cognitive functioning with Mullen IQ scores averaging 101.2 ± 15.2. The estimated HU increased over the 18-month follow-up period from 0.796 ± 0.082 at baseline to 0.902 ± 0.064. CONCLUSIONS: We demonstrated the feasibility of a mapping algorithm to generate health utility estimates in children where direct elicitation is not possible. In the absence of direct elicitation of HU in clinical trials, a mapping approach can be a useful alternative. As the collections contained in NDAR continue to grow, researchers will have ample opportunities to identify interventions that warrant dissemination and implementation based on economic value.

(207.2) Validity of the new EQ-5D-5L in a cohort of asthmatic patients in Europe Gimena Herna´ndez, MD,MPH, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Spain, Universitat Auto`noma de Barcelona, Bellaterra, Barcelona, Spain, Monica Avila, BPharm, MPH, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Spain, Universitat Pompeu Fabra, Barcelona, Spain; Olatz Garin, PhD, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Spain, Universitat Pompeu Fabra, Barcelona, Spain; Angels Pont, BSc, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Barcelona, Spain; Alexandra Dima, PhD, Dept of Communication Science, ASCoR, University of Amsterdam, Amsterdam, The Netherlands, Amsterdam, Netherlands; Laurent Laforest, MD, PhD, Unite´ de Pharmaco e´pide´miologie, Faculte´ d’Odontologie, UMR 5558 CNRS - Universite´ Claude-Bernard Lyon, CHU, Lyon, France; Montse Ferrer, MD,PhD, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salud Pu´blica (CIBERESP), Spain;Universitat Auto`noma de Barcelona, Bellaterra, Barcelona, Spain AIMS: A previous study evaluating the psychometric properties of the traditional EQ-5D-3L in asthmatic patients showed a high ceiling effect (59 % of patients with perfect health) questioning its usefulness in these patients. Our objective was to examine construct validity and score distribution of the new EQ-5D-5L in an European sample of asthmatic patients. METHODS: A subgroup of 313 patients between

Qual Life Res (2015) 24:1–191 12 and 40 years included in the AstroLab cohort completed the EQ5D-5L via online questionnaire. The EQ-5D-5L is a brief, multiattribute, generic, preference-based health status measure consisting of five dimensions with five response levels. The index value ranges from 1 (best health possible) to -0.594 (negative values indicate health states worse than death). Index values were calculated using the crosswalk value sets for the EQ-5D-5L developed from the EQ5D-3L for France. To examine the distribution of the index measures, we calculated measures of central tendency, dispersion, ceiling and floor effects, and observed range. Construct validity was assessed by comparing means of known groups defined by the Asthma Control Questionnaire (ACQ-5) that measures the presence of asthma symptoms during the previous week in 7 Likert scale response options, with a score ranging from 0 to 6 (lower score better asthma control). Three groups of asthma control were defined according to tertiles: well-controlled (ACQ-5 \ 0.4), intermediate (ACQ-5 0.4–1.2) and not well-controlled (ACQ-5 [ 1.2). ANOVA tests and effect sizes (ES) between groups were calculated. RESULTS: Mean EQ-5D-5L index was 0.834 (0.184). The observed range was -0.07 to 1. Floor and ceiling effects were 0 % and 28.1 %, respectively. Mean EQ-5D5L index for patients with well-controlled asthma was 0.913 (95 % CI 0.890–0.937); 0.841 (95 % CI 0.810–0.873) for those with intermediate control and 0.734 (95 % CI 0.691–0.777) for patients not wellcontrolled. These indexes were significantly different (p \ 0.001), and the effect size between the extreme groups was large (ES = 1.32). CONCLUSIONS: The new EQ-5D-5L provides an adequate index distribution in patients with asthma, with an acceptable ceiling effect. Statistical and substantive differences across known asthma control groups indicates good construct validity for assessing health related quality of life in these patients.

(207.3) Comparison of PROMIS and EQ-5D quality-adjusted Life years John D. Hartman, MA, Moffitt Cancer Center and University of South Florida, Tampa, FL, United States, Benjamin M. Craig, PhD, Moffitt Cancer Center and University of South Florida, Tampa, FL, United States AIMS: Using quality-adjusted life years (QALYs) based on patient reported outcomes (PROs) as a summary measure of health-related quality of life (HRQoL) is an essential component to any economic evaluation comparing alternative treatments. While multiple psychometric papers have compared PRO instruments, no paper has compared QALYs based on EQ-5D and Patient Reported Outcomes Measurement Information System (PROMIS) items. METHODS: As part of a nationally representative survey, 2623 US adults completed the 29-item PROMIS health profile instrument (PROMIS-29) and the 3 level EQ-5D (EQ-5D-3L). Their responses were summarized on a QALY scale using published estimates. Using regression analysis, PROMIS and EQ-5D QALYs were compared with each other as well as with general self-reported health (i.e., In general, would you say your health is: Excellent, Very good, Good, Fair, Poor; Move the slider on the scale to indicate how your health is today: 0 [The worst health you can imagine] to 100 [The best health you can imagine]). RESULTS: Respondents reported 1670 and 74 distinct QALY values for the PROMIS-29 and EQ-5D-3L, respectively. This difference is largely attributable to the greater number of questions and levels in the PROMIS-29 (29 5-level questions vs. 5 3-level questions). While losses in PROMIS QALYs are much greater than the losses in EQ-5D QALYs, likely due to a difference in scale, PROMIS QALYs map well to EQ-5D QALYs (correlation = .782) and exhibit greater concordance with general self-reported health. Mapping from EQ-5D

53 QALYs to PROMIS QALYs proved more difficult owing to the large ceiling effect of the EQ-5D (43 % of respondents are reported in perfect health) and the greater number of questions in the PROMIS29. CONCLUSIONS: Differences in QALY scaling between these instruments are likely attributable to differences in preference elicitation and econometric methods (e.g., rescaling of worse than dead time trade-off responses). Compared to the EQ-5D-3L, PROMIS-29 QALYs are more sensitive and can be mapped to EQ-5D QALYs. Future research is needed to compare the PROMIS QALYs to QALYs from the 5 level EQ-5D and to assess whether adding questions to the EQ-5D (i.e., bolt-on) might sufficiently improve its sensitivity to allow for mapping to PROMIS QALYs.

(207.4) A time trade-off-derived value set of the 5-level EQ-5D for Canada Feng Xie, PhD, McMaster Univerity, Father Sean O’Sullivan Research Centre, St. Joseph’s Healthcare, Hamilton, ON, Canada; Eleanor Pullenayegum, PhD, Toronto Hospital for Sick Children, Toronto, ON, Canada; Kathryn Gaebel, McMaster University, Hamilton, Canada; Nick Bansback, PhD, University of British Columbia and Centre for Health Evaluation and Outcomes Sciences, Vancouver, BC, Canada; Stirling Bryan, PhD, University of British Columbia, Vancouver, BC, Canada; Arto Ohinmaa, PhD, University of Alberta, Edmonton, AB, Canada; Lise Poissant, PhD, University of Montreal, Montreal, QC, Canada, Jeffrey A. Johnson, PhD, University of Alberta, Edmonton, AB, Canada AIMS: The 5-level version of the EQ-5D (EQ-5D-5L) was recently developed by the EuroQol Group. A number of preference-based scoring systems are being developed for several countries around the world. Our objective was to develop a value set for the EQ-5D-5L based on societal preferences for the EQ-5D health states in Canada. METHODS: We used age, sex and education quota sampling from the general population from 4 cities across Canada (Vancouver, Edmonton, Hamilton and Montreal). Composite time trade-off (cTTO) and traditional TTO (tTTO) were used as the main elicitation technique. A total of 86 EQ-5D-5L health states grouped into 10 blocks were valued using cTTO, while a subset of 18 severe states was also valued using tTTO. Participants meeting predefined inconsistency criteria were excluded from the analyses. For the value set development, we used tTTO and positive cTTO values, while censoring negative and zero cTTO values at zero. Models with the main effects presented using linear terms combined with various additional terms were estimated. The preferred model was selected based primarily on logically ordered coefficients, and secondly model fit. RESULTS: Out of 1209 participants who completed interview, 136 met criteria for inconsistent responses which excluded them from modelling. The demographics and socio-economic status of the remaining 1073 participants were similar to the Canadian general population. The preferred model has five linear terms for the main effects, a term for level 4 or 5 for each dimension, and a term for the squared total number of level 4 or level 5 beyond the first. For this preferred model, health utilities ranged from -0.148 for the worst (55,555) to 0.949 for the best (11,111) EQ-5D-5L states. Sensitivity analyses suggested that exclusions due to inconsistent responses and censoring cTTO values at or below zero did not substantially alter the final model. CONCLUSIONS: This is the first TTO-based value set of the EQ-5D-5L for Canada. It can be used to derive an overall index score for the EQ-5D-5L, in support of various applications in patient-reported outcomes, population health and economic evaluations to inform reimbursement decision-making in Canada (Tables 1, A2).

123

54

Qual Life Res (2015) 24:1–191

Table 1

Table A2 continued

Demographics of the participants in the Canadian Valuation Study Canadian general population*

Mean age (SD), years 18–39

35.5 %

Full sample (n = 1209)

Included sample (n = 1073)

Excluded sample (n = 136)

p value

47.6 (17.4)

47.1 (17.6)

51.9 (15.5)

0.002

425 (35.2)

392 (36.5)

33 (24.3)

40–59

38.0 %

434 (35.9)

382 (35.6)

52 (38.2)

60–89

25.4 %

350 (28.9)

299 (27.9)

51 (37.5)

Female

51.5 %

663 (54.8)

595 (55.5)

68 (50.0)

Marital status

0.229 0.009

Married/common law partner Single

53.4 %

646 (53.4)

574 (53.5)

72 (52.9) 26 (19.1)

28.0 %

345 (28.5)

319 (29.7)

Separated

3.0 %

33 (2.7)

28 (2.6)

5 (3.7)

Divorced

8.5 %

129 (10.7)

108 (10.1)

21 (15.4)

Widowed

6.0 %

50 (4.1)

40 (3.7)

10 (7.4)

Education 437 (36.2)

374 (34.9)

63 (46.3)

29.1 %

386 (31.9)

337 (31.4)

49 (36.0)

25.3 %

380 (31.4)

358 (33.4)

22 (16.2)

405 (33.5)

359 (33.5)

46 (33.8)

3.5 %

139 (11.5)

107 (10.0)

32 (23.5)

$15,000–$45,000

19.7 %

321 (26.6)

278 (25.9)

43 (31.6)

$45,000–$75,000

23.5 %

260 (21.5)

237 (22.1)

23 (16.9)

[$75,000

53.3 %

353 (29.2)

334 (31.1)

19 (14.0) 104 (76.5)

0.622

0.332

University or higher Full-time employment

Annual household income \$15,000

Born in Canada

75.1 %

910 (75.3)

806 (75.1) 174 (14.4)

815 (67.4)

729 (67.9)

86 (63.2)

None

530 (43.8)

471 (43.9)

59 (43.4)

1

301 (24.9)

264 (24.6)

37 (27.2)

2

191 (15.8)

168 (15.7)

23 (16.9)

C3

187 (15.5)

170 (15.8)

17 (12.5)

82.3 (14.2)

82.7 (13.6)

78.7 (18.0)

Language spoken at home-English

64.8 %

# Chronic conditions

Self-reported EQVAS

0.932

0.723

0.014

P values were for the comparison between the included and excluded samples * Source: The distribution on education came from Statistics Canada 2006 Census, while all other demographics from 2011 census

Table A2 Variable

Intercept MO

Sensitivity analyses results

Preferred model (censored at zero, with consistent participants)

Censored below zero, with consistent participants

Censored at zero with all participants

Censored below zero, with all participants

1.1351 (0.01712)

1.1163 (0.01503)

1.0836 (0.01740)

1.0703 (0.01525)

-0.0389 (0.00478)

-0.0341 (0.00417)

-0.0347 (0.00457)

-0.0299 (0.00400)

SC

-0.0458 (0.00453)

-0.0394 (0.00402)

-0.0396 (0.00438)

-0.0337 (0.00388)

UA

-0.0195 (0.00435)

-0.0180 (0.00389)

-0.0145 (0.00420)

-0.0137 (0.00375)

PD

-0.0444 (0.00442)

-0.0390 (0.00392)

-0.0391 (0.00423)

-0.0343 (0.00374)

AD

-0.0376 (0.00492)

-0.0373 (0.00433)

-0.0355 (0.00480)

-0.0352 (0.00423)

MO45

-0.0510 (0.01597)

-0.0698 (0.01394)

-0.0479 (0.01518)

-0.0660 (0.01327)

SC45

-0.0584 (0.01513)

-0.0809 (0.01345)

-0.0562 (0.01434)

-0.0771 (0.01273)

123

Censored at zero with all participants

Censored below zero, with all participants

UA45

-0.1103 (0.01433)

-0.1134 (0.01274)

-0.1079 (0.01372)

-0.1087 (0.01218)

PD45

-0.1409 (0.01455)

-0.1465 (0.01264)

-0.1311 (0.01384)

-0.1349 (0.01203)

AD45

-0.1060 (0.01289)

-0.1277 (0.01539)

-0.1238 (0.01343)

-0.1090 (0.01477)

Num45sq

0.0085 (0.00156)

0.0158 (0.00124)

0.0047 (0.00152)

0.0114 (0.00123)

MAE*

0.0443

0.0779

0.0557

0.0989

(207.5) A comparison of EQ-5D-5L valuation data in England and the United Arab Emirates Koonal K. Shah, Office of Health Economics, London, United Kingdom

\0.001

194 (16.1)

Rural dwellers

Censored below zero, with consistent participants

MAE = mean absolute error which was calculated without using leave-a-state-out cross validation approach

45.7 %

College diploma

Preferred model (censored at zero, with consistent participants)

Standard errors are given in parentheses. MO, SC, UA, PD, and AD were linear variables for the 5 domains; MO45, SC45, UA45, PD45, and AD45 were dummy variables for whether or not there exists any level 4 or level 5 within a domain; Num45 was defined as the number of level 4 or level 5 beyond one across the five domains; Num45sq was defined as Num45 squared

\0.001

High school or lower

Variable

AIMS: The EuroQol Research Foundation recently established a new international protocol for valuing EQ-5D-5L health states. In 2012/2013, data were elicited from 996 English individuals in order to develop an EQ-5D-5L value set for England. In 2013, corresponding data were elicited from 200 Emirati individuals in order to test the feasibility of valuing EQ-5D-5L in the United Arab Emirates (UAE). The aim of this presentation is to compare the findings of these two studies, both of which followed the standardised EuroQol protocol. METHODS: Values were elicited in face-to-face computer-assisted personal interviews. Respondents completed 10 time trade-off tasks and seven discrete choice experiment tasks, followed by debriefing questions. Respondents in the UAE study were asked additional questions examining cultural and religious influences on their responses. Descriptive analyses were used to assess the face validity of the data and to make comparisons between the studies. RESULTS: The range of mean values observed was wider in the England study. Respondents in the UAE study were less likely to give ‘worse-than-dead’ valuations. Interviewer effects were observed in both studies. There appear to be important differences between the countries—pain/discomfort and anxiety/depression were the least important dimensions in UAE study and the most important dimensions in the England study. The majority of Emirati respondents stated that their religious beliefs influenced their responses to the valuation tasks. CONCLUSIONS: The results highlight the importance of developing country-specific value sets, and suggest that some adaptation of the methods is required to improve their acceptability in the UAE.

(207.6) Investigating the influence of individual valuation block severity on TTO estimates Andre´a Libo´rio Monteiro, HTA Ms MSc, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil; Marisa Santos, MD, PhD, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil; Monica Cintra, Medical coordinator, HC-FMRP- USP, Ribeira˜o Preto, Brazil; Bra´ulio Santos, MD, MSc, Instituto Nacional de Cardiologia, Rio de Janeiro, Brazil; Moˆnica V. Andrade, PhD, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil; Kenya Noronha, PhD, Universidade Federal de Minas Gerais, Belo Horizonte, Brazil; Luciane N. Cruz, MD PhD, Universidade Federal de Minas Gerais, Porto Alegre, RS, Brazil; Suzi Camey, Universidade

Qual Life Res (2015) 24:1–191 Federal de Minas Gerais, Porto Alegre, Brazil; Bernardo Tura, MD PhD, Instituto Nacional de Cardiologia, Rio de Janeiro, Brazil AIMS: One of the main assumptions when selecting EQ-5D health states for health valuation studies is to seek a balance between the severities of health states in order to maximize the representation of the full valuation space both on global and on individual level. This analysis aims to assess the influence of individual valuation blocks severity on TTO response patterns and on utility estimates. METHODS: The dataset includes 5785 respondents with data collected in four urban areas in Brazil (Rio de Janeiro, Minas Gerais, Recife and Porto Alegre). A geographic-based probabilistic sample of the general population, aged 18–64, was recruited. Respondents were asked to value seven health states (six variable and the states 11,111, 33,333 and ‘‘death’’) using TTO. While defining the blocking strategy we chose not to restrict health states combinations that resulted on ‘‘mild’’ or on ‘‘severe’’ individual valuation blocks (IVB). As a result, both balanced (34 %) and unbalanced (66 %) blocks were obtained. Unbalanced blocks are those that do not have at least two mild, two moderate and two severe health states. Within imbalanced IVBs we also defined ‘‘mild’’, ‘‘moderate’’ and ‘‘severe’’ groups. All 243 health states were valued by at least 135 subjects. TTO data were modelled at both individual and aggregate levels by using Ordinary least squares and Random Effects methods. RESULTS: The varied blocking strategies had very limited impact both on health states mean values and dispersion. The proportion of inconsistent valuations did not differ significantly between balanced and unbalanced IVB. Other differences on response patterns (clustering at certain values, proportion of worse than health valuations), were also non-significant. We found very small differences on models yield by both balanced and unbalanced IVBs data (unbalanced IVBs yield estimates slightly higher). However, within unbalanced IVBs, both ‘‘mild’’ and ‘‘severe’’ groups yielded significantly higher utility estimates than ‘‘moderate’’group. CONCLUSIONS: Further research in needed to understand the implications of individual valuation blocks severity on health valuations, however based on our results it seems that the composition of IVB has an non-negligible impact on TTO valuation.

208: Concordance Between Patients and Others (208.1) Individual, linguistic and method-related impact on discrepancies between self- and proxy-reports on healthrelated quality of life of children and adolescents with chronic conditions Holger Muehlan, PhD, Ernst-Moritz-Arndt University, Greifswald, Germany; Sivan Edelstein, University of Greifswald, Greifswald, Germany; Monika Bullinger, Vice Chief of Department, University of Hamburg, Hamburg, Germany; Silke Schmidt, Prof., Ernst-MoritzArndt University Greifswald, Greifswald, Germany AIMS: This study evaluated sources of differences between self- and proxy-report of the health-related quality of life of children and adolescents with a chronic condition. It aimed to analyse discrepancies as well as agreement and to clarify the underlying mechanisms. METHODS: A sample of 225 children and adolescents with bronchial asthma, juvenile arthritis and diabetes mellitus and their parents took part in the study, being assessed on self- and proxy-report versions of DISABKIDS’ chronic-generic and disease-specific modules. Approaches to analyse discrepancies included graphical, descriptive and statistical testing techniques. Agreement was measured by means of intraclass correlations, percentage of ratings within clinically specified limits and Bland–Altman-Plots. T-tests proved discrepancies on general, scale and item level. To evaluate factors of influence an explicit distinction was made between individual, linguistic and and method-related attributes being analysed by t-test, analysis of

55 variance, regression and non-parametric techniques. RESULTS: Though high correlations were shown, parents tend to underestimate their children’s quality of life indicating the importance of differentiating between consistency and agreement. Individual-related factors of influence were the parents’ mental condition, children’s age and coping style. Non-individual factors with substantial impact included the explicit linguistic reference and attribution to the disease and differences between the children’s and parents’ response styles. CONCLUSIONS: This work identifies various factors of influence and hints to important aspects in the development of instruments using self-and proxy-report, however not being able to elu-ci:date the underlying mechanisms entirely.

(208.2) PROMs and PREMs for elderly patients in acute care and their families: an EMPRO evaluation Eric K. H. Chan, PhD, Trinity Western University, Langley, BC, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; Anne M. Gadermann, PhD, University of British Columbia, Vancouver, BC, Canada; S. Robin Cohen, PhD, McGill University and Lady Davis Institute, Montreal, QC, Canada; PROMs PREMs Knowledge Synthesis Team AIMS: Patient-reported outcome measures (PROMs) and patient-reported experience measures (PREMs) are increasingly used with seriously ill elderly patients in acute care and their families. In this knowledge synthesis, we sought to identify and evaluate the psychometric properties of PROMs and PREMs that had evidence of being used for research or practice in acute care settings, and to provide recommendations for the selection and utilization of PROMs and PREMs for this patient population and their families. In this paper, we present the results of the evaluation of the quality of the identified PROMs and PREMs. METHODS: In the early stage of this project we identified a list of 88 PROMs and PREMs that had been developed or used within the past 5 years for elderly patients in acute care and their families. In the current stage, we evaluated the quality of the PROMs and PREMs that had published psychometric validation studies that focused on elderly patients in acute care and/or their families. We used the Consensus-based Standards for the Selection of health Measurement Instruments (COSMIN) search filter to search for validation studies on PubMed. Quality appraisal on each of the identified instruments was conducted using the Evaluating the Measurement of PatientReported Outcomes (EMPRO) standardized tool for appraising the quality of self-report health instruments. RESULTS: Two PREMs, the Canadian Health Care Evaluation Project Questionnaire (CANHELP), and its shorter version, CANHELP LITE, scored over 90 (out of 100) on the EMPRO. The instruments received excellent evaluation on the validity and respondent burden domains, as well as very high scores on the conceptual and measurement model, interpretability, and administrative burden domains. No information was provided in the publications to evaluate the responsiveness (an instrument’s ability to detect change) and language versions of the two instruments. CONCLUSIONS: The CANHELP and CANHELP LITE are strongly recommended for use with elderly patients in acute care and/or their family caregivers. Although there are many PROMs and PREMs used in acute care for seriously ill elderly patients and their families, many of them have not been validated for this patient population. More validation studies are urgently needed.

(208.3) Patient perspective of treatment benefit in clinical trials: it never hurts to ask—or does it? Alison Greene, MPH, Genentech, South San Francisco, CA, United States; Christopher Cabanski, PhD, Genentech, South San Francisco,

123

56 CA, United States, Alison Berryhill-Matsui, MPH, Genentech, South San Francisco, CA, United States AIMS: Having patients complete PRO assessments independently, prior to other study procedures, is the standard for collecting these data. Yet, there is little published evidence that this administration practice impacts patients’ responses. To determine the effect of how PRO data are reported, we included both ‘‘interview’’ and ‘‘independent’’ global impression items as exploratory endpoints in a Ph2 asthma clinical trial and compared the level of agreement between clinician and patient global item ratings, based on how the data are collected. METHODS: Patients completed two global impression items at the last study visit: patient global impression of change (PGIC) and patient global evaluation of treatment effectiveness (GETEp) items. Clinicians also completed two items: clinician global impression of change (CGIC) and clinician’s global evaluation of treatment effectiveness (GETEc) items. These items collect the rater’s assessment of treatment benefit compared to study start; PGIC and CGIC items measure symptom change on a 7-point response scale (very much improved to very much worse), while GETEp and GETEc items measure treatment effectiveness on a 5-point response scale (excellent to worsening). Patients and clinicians completed their respective GIC items independently. The GETEp and GETEc items were interviewer-administered during the physical exam. Contingency tables were created to compare the GETEp and GETEc, as well as the PCIG and CGIC. All items were categorized by responder status (excellent/good for GETE; improved/very much improved for GIC). Comparisons were also made for ACQ5 responders (achieved .5 MID improvement from baseline) and non-responders. Cohen’s kappa statistic was calculated to measure the amount of agreement between the groups. Missing data were removed from the analysis. RESULTS: There was stronger agreement between GETEp and GETEc responses (kappa = .73) than PGIC and GCIC responses (kappa = .35). The substantial agreement between GETEp and GETEc held regardless of ACQ5 responder or non-responder status (kappa = .65 and .88, respectively). CONCLUSIONS: Patients may rate themselves differently, based on whether they complete assessments independently or via interview. Unless required for a special study population, independent completion of patient-reported outcomes assessments in clinical trials should remain the standard.

(208.4) Cancer and the family: a study of Canadian Chinesespeaking and Anglophone patients and family caregivers Joyce W. Lee, University of British Columbia; Trinity Western University, Vancouver, BC, Canada; Arminee Kazanjian, University of British Columbia, Vancouver, BC, Canada; Carolyn Gotay, University of British Columbia, Vancouver, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada AIMS: This study examined the association between patient distress, family caregiver distress and patient quality of life (QoL) in a Chinese-speaking and Anglophone cancer population in British Columbia, Canada. METHODS: A cross-sectional survey of Chinesespeaking and Anglophone patients diagnosed with stage 1–3 cancer in the last 6 months to 2 years (N = 55) and their family caregivers (N = 40) was used to examine patient QoL and correlates: specifically, patient and family caregiver characteristics (age, sex, education, years in Canada, English fluency), cancer diagnosis (site, stage, treatment), and type and duration of family caregiving. Multiple imputation of missing data for incomplete dyads resulted in data for 29 Chinese-speaking and 28 Anglophone dyads. Multiple linear regression and mediation analyses examined predictors of QoL and QoL domains and mediation effects of patient and family caregiver

123

Qual Life Res (2015) 24:1–191 distress. The FACT-G was used to measure QoL and its physical, social/family, emotional and functional well-being domains. RESULTS: Patient age was a significant predictor (p \ .05) of patient emotional well-being in the combined sample (b = 0.03), such that older patients had better emotional functioning. Patients with high school education (relative to having more schooling) had lower social well-being (bs = -.31 to -.46). Among Chinese-speaking patients, distress had the greatest impact on functional well-being (b = -.69) and the smallest impact on physical well-being (b = -.46). For Anglophone patients, distress had the greatest impact on physical well-being (b = -.76) and the smallest impact on social well-being (b = -.45). Family caregiver distress was not significantly related to patient QoL or the QoL domains. The effects of patient age on patient emotional well-being in the combined sample were mediated by patient distress, such that lower distress in older patients explained better emotional functioning. CONCLUSIONS: This study highlights the importance of examining QoL domains, specifically in the context of exploring the cancer’s impact on patient QoL in Canadian Chinesespeaking and Anglophone cancer patients and family caregivers. Results indicate that the effect of age on patient emotional well-being QoL is explained by the mediation role of patient distress.

(208.5) Developing a conceptual framework of paediatric CFS/ ME: the value of card ranking within qualitative interviews Roxanne M. Parslow, BSc (HONS), University of Bristol, Bristol, United Kingdom; Alison Shaw, BA, MSc, PhD, University of Bristol, Bristol, United Kingdom; Kirstie Haywood, Senior Research Fellow, Warwick Medical School, Warwick University, Coventry, United Kingdom; Esther Crawley, BA(Hons), BM BCh, MRCP, FRCPCH, PhD, University of Bristol, Bristol, United Kingdom AIMS: Paediatric Chronic Fatigue Syndrome/Myalgic Encephalomyelitis (CFS/ME) is common (1–2.34 %) and disabling. Child-derived measures specific to CFS/ME do not exist; widely used generic and domain-specific measures lack relevance. We aim to develop a childderived patient-reported outcome measure (PROM) which captures what really matters to children. We describe a child-friendly process for enabling them to influence which outcomes must be included in a new PROM. METHODS: Semi structured interviews were carried out with children with CFS/ME and separately with their parents. Participants were presented with a list of printed cards depicting 15 areas of life described as important by children with CFS/ME in previous work. Participants were asked to select the most important areas and arrange them in terms of what they would most like to improve. Once completed, the areas and suggested order were further explored in a discussion with the interviewer. Data was analysed using framework analysis using techniques of constant comparison. RESULTS: 21 children with CFS/ ME were interviewed (16 females, 5 males, mean age 14.4 years, range 12–17 years old; 9 (12–13), 6 (14–15) and 6 (16–17). 20 mothers and two fathers were interviewed. Three areas were consistently ranked as important by children and their parents: ‘‘symptoms’’, ‘‘tiredness’’ and ‘‘payback and crashing’’. Children ranked ‘‘school’’ higher than parents (ranked 2nd by children; 9th by parents); and parents ranked ‘‘mood’’ problems higher than children (ranked 9th by children; 5th by parents). Girls ranked ‘‘symptoms’’ higher than boys (ranked 1st by girls; 8th by boys); and boys ranked ‘‘activities and hobbies’’ as more important (ranked 7th by girls; 4th by boys). Mood was ranked higher by older children (14–15 years of age) and their parents. CONCLUSIONS: Card ranking is an interactive technique that has facilitated an understanding of children’s priorities and preferences, and those of their parents, whilst providing a framework for deeper discussion within an interview. The study produced quantitative and qualitative data which allowed us to differentiate between subgroups. The results will inform a developing conceptualisation that includes outcomes viewed as important to children

Qual Life Res (2015) 24:1–191 Table 1

57

Ranking order of most important areas of life in paediatric CFS/ME to improve as ranked by children with CFS/ME and their parents and according to different sub groups

Important areas of life to improve in paediatric CFS/ME: card ranking order All children (12–17 No. years) of Pts

Parents

No. of Pts

Children (females)

No. of Pts

Children (males)

No. of Pts

Children (12–13 years)

No. of Pts

Children (14–15 years)

No. of Pts

Children (16–17 years)

No. of Pts

Symptoms

15

Symptoms

13

Symptoms

14

School

5

School

7

School

4

Symptoms

6

School

15

Tiredness

12

School

10

Tiredness

4

Payback & crashing 6

Your future

4

Tiredness

5

Tiredness

13

Payback & crashing

11

Tiredness

5

Symptoms

4

School

3

9

Activities & hobbies

5

Tiredness

3

Payback & crashing 2

Payback & crashing Your future

9

Payback & crashing

9

Payback & crashing 2

Tiredness

7

Activities & hobbies

2

Symptoms

9

Mood

8

Your future

7

Family

2

Family

4

Mood

3

Fluctuation

2

Activities & hobbies

7

How your child feels about him/ herself

8

Friends

6

Mood

2

Friends

3

Friends

2

Activities & hobbies

2

Friends

7

Sleep problems

7

Activities & hobbies

5

Your future

2

Your future

3

Sleep problems

1

Friends

2

Family

6

Fluctuation

7

Fluctuation

4

Symptoms

1

Daily activities

2

Daily activities 1

Family

2

Mood

6

School

7

Family

4

Sleep problems

1

Fluctuation

2

Payback & crashing

1

Your future

2

Fluctuation

5

Family

5

Mood

4

Daily activities

1

Activities & hobbies

2

Fluctuation

1

Sleep problems

1

Daily activities

4

Daily activities

3

Daily activities

3

Fluctuation

1

Mood

2

How you feel about yourself

1

Daily activities

1

Sleep problems

3

Friends

2

How you feel about yourself

3

Friends

1

Sleep problems

1

What people think about you

1

Mood

1

How you feel about yourself

3

Your child’s future

2

Sleep problems

2

Independence

1

How you feel about 1 yourself

Activities & hobbies

0

How you feel about 1 yourself

Independence

1

Independence

2

Independence

0

How you feel about 0 yourself

Independence

Family

0

Independence

0

Boyfriend/girlfriend

0

Boyfriend/girlfriend

0

Boyfriend/girlfriend 0

Boyfriend/girlfriend 0

0

Boyfriend/girlfriend 0

Boyfriend/girlfriend

with CFS/ME and their parents, providing an important basis for the new PROM (Table 1).

1

Independence

0

quality of your home meet your needs?’’), A1 (‘‘To what degree are you satisfied with your isomerism relationships?’’) were selected to be added into ‘‘Social Relationships’’ domain. In Item Response Theory, the mean information of the whole revised scale was 19.8 when

209: Cross-cultural Research (209.1) The revision and psychometric evaluation of WHOQOLBREF Jialing Li, Master of Medicine, Sun Yat-Sen University; Guangdong Provincial Center for Disease Control and Prevention, Guangzhou, Guangdong, China; Yuantao Hao, PhD; Professor, Sun Yat-sen University, Guangzhou, Guangdong, China; Jing Gu, PhD; Associate Professor, Sun Yat-sen University, Guangzhou, Guangdong, China; Yanxia Zhao, Sun Yat-Sen University, Guangzhou, China AIMS: Add superior items into the ‘‘Social Relationships’’ domain of WHOQOL-BREF by using Item Response Theory, Classical Test Theory and Delphi method. Evaluate the psychometric properties of the revised scale WHOQOL-BREF-R. METHODS: Nominal group worked interactively with focus group to build the item pool. In the first phase of investigation, a self-administered questionnaire survey was conducted among 700 healthy people and patients. Superior items in the item pool were selected using Item Response Theory (discrimination a, difficulty b1–b4, item information function and differential item functioning), Classical Test Theory and Delphi method (experts’ importance rating) to add into the ‘‘Social Relationships’’ domain. WHOQOL-BREF-R was administered in the second phase of investigation which 562 people were recruited. The psychometric properties of the WHOQOL-BREF-R were then evaluated by using Item Response Theory and Classical Test Theory. RESULTS: According to the selection criterias, three items D36 (‘‘Do you feel that other people respect you?’’), F14.3 (‘‘To what degree does the

Table 1

The results of parameter estimation using IRT (N = 700) Items

a

Personal relationship

F13.1

1.23 -3.09 -2.04 -0.27 1.46 0.428

0.9

169.6

F13.2

1.25 -5.02 -3.43 -1.26 0.92 0.398

143.4

14.7

F13.4

1.41 -3.20 -1.92

0.00 2.24 0.521

134.9

23.5

D34

1.48 -3.75 -2.21

0.23 2.16 0.541

7.5

150.4

D35

1.47 -4.10 -2.49 -0.31 2.20 0.524

5.2

11.2

D36#

1.68 -3.68 -2.81 -0.43 1.89 0.626

0.2

6.3

O23

0.82 -2.61 -1.55

0.26 2.04 0.197

134.3

13.1

F14.1

1.29 -3.10 -1.78

0.06 2.44 0.448

131.8

18.9

F14.2

1.17 -3.30 -1.84

0.02 2.45 0.378

3.2

16.0

F14.3# 1.42 -3.73 -2.50 -0.67 1.50 0.596

6.0

2.4

Support

Sex/companion

b1

h [ (-3,3) DIF for DIF gender for Average information G1 (df = 5) age G2 (df = 5)

Facets

b2

b3

b4

F15.1

1.55 -2.35 -1.40

0.48 2.00 0.625

131.4

36.7

F15.2

1.37 -2.39 -1.46

0.56 2.27 0.504

4.0

36.9

F15.4

0.87 -4.88 -3.37 -0.74 1.14 0.203

15.1

3.0

O21

0.86 -3.26 -2.18

0.20 2.36 0.217

4.8

3.9

A1#

1.82 -3.28 -2.22 -0.02 2.03 0.762

8.7

2.2

‘‘#’’: The items meet all the selection criterions of DIF.

v2(0.01)

(df = 5) = 15.1

123

58 Table 2

Qual Life Res (2015) 24:1–191 The results of traditional item selection methods and experts’ importance rating

(N = 700) Standard Corre- Factor Cronbachs’ deviation lation analysis a (a = coeffi- (3) (1) 0.635) (4) cient (2)

Facets

Items

Personal relationship

F13.1 0.98

0.36

0.52

0.645

–

–

7.9

F13.2 0.75

0.38

0.59

0.659

–

–

9.5

F13.4 0.82

0.48

0.59

0.696

0.544* 1

7.5

D34

0.77

0.50

0.64

0.706

0.399

–

8.4

D35

0.71

0.49

0.66*

0.695

–

1

7.5

D36

0.68

0.51

0.69*

0.707*

0.276

2

7.7

O23

1.24*

Support

Multi- Selected Experts linear times evaluation regression (5)

0.20

0.43

0.520

0.163

1

6.3

F14.1 0.87

0.36

0.60

0.650

–

–

8.5

F14.2 0.90

0.42

0.57

0.675

0.313

–

8.0

F14.3 0.79

0.43

0.63

0.689

0.282

–

9.4

0.55*

0.57

0.717*

0.613* 3

7.1

0.52*

0.62

0.703

0.427

1

7.3

F15.4 0.94

0.29

0.39

0.616

–

–

5.7

O21

1.05*

0.28

0.46

0.578

0.144

1

7.0

A1

0.72

0.59*

0.66*

0.737*

0.819* 3

7.1

F15.1 0.95 Sex/ companF15.2 0.95 ion

(1) The standard deviation of items’ scores. (2) The correlation coefficients between items and ‘‘social relationship’’ domain. (3) The factor loadings of the items on the common factor. (4) The Cronbach’ a coefficient of the domain after adding the items (The original Cronbach’ a coefficient is 0.635). (5)The parameters of selected items in multiple linear regression. ‘‘*’’: The items are selected by the corresponding methods. Experts evaluation: The average scores of 11 experts’ importance rating (Ten-point system)

Fig. 1 The item characteristic curve of each item in WHOQOLBREF-R h [ (-3,3), and the marginal reliability of whole revised scale was 0.95. In Classical Test Theory, the Cronbach’s a coefficients of the four domains ranged from 0.84 to 0.91, while the ICCs ranged from 0.85 to 0.91. The spearman correlation coefficients between items with their hypothesized domain were higher than those with other domains. The main indexes of goodness of fit in Structural Equation Modeling such as x2/df, RMSEA, CFI reached the adequate standards, while NFI and NNFI did not reach but very closed to the standard of 0.90. The marginal reliability, Cronbach’s a coefficient and ICC of the original ‘‘Social Relationships’’ domain(only 3 items) in WHOQOL-BREF were all lower than the revised ‘‘Social Relationships’’ domain. CONCLUSIONS: Three superior items were added into the ‘‘Social Relationships’’ domain to form the revised

123

Fig. 2 The Information Function Curve of WHOQOL-BREF-R scale WHOQOL-BREF-R. The WHOQOL-BREF-R has adequate psychometric properties and the revised ‘‘Social Relationships’’ domain has better reliability than the original one (Tables 1, 2; Figs. 1, 2).

(209.2) Can the original factor structure of WHOQOL-BREF (English) be replicated in Singapore? Kok Joon Chong, National University of Singapore, Singapore, Singapore; Yin Bun Cheung, National University of Singapore, Singapore, Singapore; Eric Yin Hao Khoo, National University of Singapore, Singapore, Singapore; Khung Keong Yeo, National Heart Centre Singapore, Singapore, Singapore, Hwee Lin WEE, National University of Singapore, Singapore, Singapore AIMS: World Health Organization Quality of Life—BREF (WHOQOL-BREF) is a widely used generic health-related quality of life instrument, which consists of 26 items organised into four factors, namely: Physical Health, Psychological, Social Relationships and Environment. This instrument has shown excellent psychometric properties in several countries worldwide. However, it has also been reported that the original factor structure of WHOQOL-BREF cannot be replicated in some countries. The aim of this study is to explore the factor structure of WHOQOL-BREF (English) in the Singapore general population. METHODS: A total of 913 community-dwelling participants residing in public housing were recruited via door-to-door surveys using quota sampling. The interviews were administered in English, Chinese or Malay language. In this study, only data from those completing the English surveys were included. Confirmatory factor analysis (CFA) was performed to evaluate the model fit of the original factor structure. Exploratory factor analysis (EFA) using principal component analyses with varimax rotation will be performed if the model fit for CFA was poor. RESULTS: In CFA, the model fit was poor (Root Mean Square Error of Approximation = 0.0932, Tucker-Lewis Index = 0.7263 and Comparative Fit Index = 0.7516). In EFA, one item regarding negative feeling did not load onto any of the factors. The remaining 25 items loaded onto four factors. Of which, three factors are largely similar to the original. Instead of Social Relationships, the fourth factor (named Satisfaction with Life) comprised of items that have the same question stem of ‘‘how satisfied are you with…’’. This suggests a framing effect, a form of cognitive bias where the participant’s response is influenced by how the question was presented. This framing effect was also observed in Nigeria where 3 of the 8 scales revealed in EFA were Satisfaction with Self, Satisfaction with Others and Satisfaction with Life. CONCLUSIONS: A different interpretation of the four factor scores of WHOQOLBREF is warranted when it is used in Singapore. In multi-national clinical trials or research studies, WHOQOL-BREF scores obtained from Singapore would require further evaluation or adjustments when being compared with other countries or used in pooled analysis.

Qual Life Res (2015) 24:1–191 (209.3) Inuit version of the aboriginal children’s health and well-being measure: qanuippit? Karen Baker-Anderson, Executive Director, Ottawa Inuit Children’s Centre, Ottawa, ON, Canada, Nancy L. Young, PhD, Laurentian University, Sudbury, ON, Canada; Mary Jo Wabano, MHK, Wikwemikong Health Centre, Wikwemikong, ON, Canada; Lynda Brown, BA, Ottawa Inuit Children’s Centre, Ottawa, ON, Canada; Courtney Henderson, BA, Ottawa Inuit Children’s Centre, Ottawa, ON, Canada; Rachel Quinn, BEd, Ottawa Inuit Children’s Centre, Ottawa, ON, Canada; Natlie Lloyd, Ottawa Inuit Children’s Centre, Ottawa, ON, Canada; Christine Kudluk, Ottawa Inuit Children’s Centre, Ottawa, ON, Canada; Janice Messam, ECE, Ottawa Inuit Children’s Centre, Ottawa, ON, Canada AIMS: The Ottawa Inuit Children’s Centre (OICC) supports urban Inuit children to become strong, healthy and proud community members with knowledge of their culture, connection to the local Inuit community, equitable access to social services and supported by their families. This is achieved through culturally strength-based programs and individualized services for children and their families that promote a good life. This presentation reports on the adaptation of the Aboriginal Children’s Health and Well-Being Measure (ACHWM) for urban Inuit children. METHODS: The ACHWM was reviewed by staff at the Ottawa Inuit Children’s Centre to identify modifications that were required to fit the cultural context of their community. Then Inuit children (8–18 years) and their caregivers were recruited from the OICC to complete the ACHWM with an interviewer who probed the children’s understanding of each question and relevance to their culture. Caregivers completed the same process with a second interviewer, separate from the child interviews. After each pair of interviews the research team had a brief discussion to determine common concerns and discuss potential modifications. RESULTS: OICC staff identified a need to replace ‘‘mother earth’’ to reflect a connection to ‘‘the land’’. They also had concerns with ‘‘traditional medicines’’ but a clear solution was not evident. They translated the name of the measure to Qanuippit in Inuktitut. The revised version was reviewed in detail by 7 children and 6 caregivers (one with 2 children) via interviews. The 3 girls and 4 boys ranged in age from 9.2 to 15.0 years (mean of 10.6). Caregivers included 4 mothers, 1 father, and 1 grandmother. Based on their responses, minor modifications were made to 9 of the 59 questions. Three new questions were identified: 1 related to food insecurity; and 2 related to hope. The revised version was then reviewed and confirmed. CONCLUSIONS: The changes were important to ensure the measure was appropriate within this distinct culture. The revisions to the questions have subsequently been reviewed and accepted by our First Nations partners. We now have a 62-item version of the ACHWM that is consistent for the Aboriginal children in Canada: First Nations, Inuit and Me´tis.

(209.4) What traditional chinese medicine told modern medicine about health status and quality of life: content analysis of traditional Chinese medicine instruments for quality of life comparing with the WHOQOL and SF-36 Chang-He Yu, MD, China Academy of Chinese Medical Sciences, Beijing, China; Ya-Nan Sun, PhD, Xuanwu Hospital Capital Medical University, Beijing, China; Li-Yun He, PhD, China Academy of Chinese Medical Sciences, Beijing, China; Wen-Jing Bai, PhD, China Academy of Chinese Medical Sciences, Beijing, China; Bao-Yan Liu, China Academy of Chinese Medical Sciences, Beijing, China AIMS: Traditional Chinese Medicine (TCM) was regarded as a complementary and alternative medicine besides the Modern Medicine (MM). TCM theory and its concept determine different views of

59 health status, which may lead to a different conceptual frame of quality of life. Thus, the aim of this study was to identify the concepts of TCM-QOL that underpin the same and different structure of WHOQOL-100 and SF-36 by a content analysis. METHODS: Based on the previous systematic review of TCM instruments for the quality of life, Chinese Quality of Life Instrument (ChQOL), whose psychometric property evidences showed good and solid and which reflected the TCM theory, was chosen as the TCM-QOL instrument. The comparators were WHOQOL-100 and SF-36, because (1) they were used as criteria instrument in the criterion validity of ChQOL; (2) they were generally applied to measure people’s QOL. A content analysis process was used to code and categorized all items from ChQOL, WHOQOL and SF-36. The methods of coding contained the main content and property of each items, and then the phrases of the content and property and the structures (domain or subdomain) that items underpinnings were compared between the ChQOL and the combination of the other two instruments. RESULTS: One hundred and eighty six items were coded and categorized. Among 50 items from ChQOL, 32 items, holding different contents, were totally different from, and the other eight items, holding same contents but different properties, were partly different from those from WHOQOL-100 or the SF-36. Based on the results of such differences, the subdomains of Complexion, Appetite & Digestion, Climate Adaptation, Spirit of the Eye, Verbal Expression and Fear & Anxiety from ChQOL were totally different, and the subdomains of Stamina, Consciousness, Thinking, Joy, Anger and Depress were partly different. But the sleep subdomain was the same as the other MM instruments. CONCLUSIONS: The TCM-QOL instruments could complement some domain and items that may be importent to the Qaulity of life, even though the development of TCM-QOL was not mature. A comprehensive structure of quality of life and health should be considerd and established in future to capture all the characteristics of health.

(209.5) Assessing the performance of an item bank derived from the World Health Organisation Quality of Life-100 measure for computer adaptive testing across diverse cultures Chris Gibbons, PhD CPsychol AFBPsS, NIHR Fellow, University of Manchester, Greater Manchester, United Kingdom AIMS: To develop an item bank suitable for computer adaptive administration of quality of life (QOL) measures using the WHOQOL-100 instrument across diverse cultures. Simulated computer adaptive tests were conducted to assess item bank performance. METHODS: Data from the international WHOQOL-100 field trial were psychometrically assessed using a combination of Mokken and Rasch analyses. The original 100-item bank was divided into four domains, representing the structure of the WHOQOL-BREF instrument. Differential item functioning (DIF) was assessed using ANOVA with post hoc Tukey tests. Where DIF was evident between countries, a strategy of ‘splitting’ items was employed to allow item ‘difficulties’ to vary across diverse cultures. Computer adaptive testing was simulated using the FIRESTAR application to assess item bank performance at different standard error thresholds using maximum posterior weighted information estimation. RESULTS: Mokken analysis confirmed the suitability of a four-factor structure consisting of domains measuring Physical QoL, a Psychological QoL, Social QoL and Environmental QoL. However not all items within each domain scaled uniformly (Loevinger’s Ho \ .30) and 22 items were removed prior to Rasch analysis. Rasch analysis further reduced 55 items across all scales. All resultant scales showed excellent fit to the Rasch model (p [ 0.5), were reliable (Cronbach’s a [ .85), unidimensional and free from local dependency. Whilst the items of the WHOQOL-100 have been shown to be conceptually equivalent, tests

123

60 of differential item functioning suggested that a number of items provided different measurement information across different cultures. A complex solution which allowed item ‘difficulties’ to vary by country was used to allow valid pooling of data. Where standard errors was set to .55 (equivalent to Cronbach’s a = .70) the four item banks could be successfully administered using an average of 4 items each. CONCLUSIONS: The current research demonstrates that brief QoL item banks can support fundamental measurement whether administered using a computer adaptive testing protocol or a standard pencil-and-paper test. Approaching the issues of DIF by country in the manner demonstrated here may help to strengthen cross-cultural research.

210: Musculoskeletal Conditions (210.1) Cutting edge solutions to improving the efficiency of PRO measurement: from real-data simulations to pilot testing before and after total joint replacement in a national registry John Ware, PhD, University of Massachusetts Medical School, Worcester, MA, United States; Barbara Gandek, PhD, University of Massachusetts Medical School, Worcester, MA, United States; Patricia Franklin, MD, University of Massachusetts Medical School, Worcester, MA, United States; Celeste Lemay, University of Massachusetts Medical School, Worcester, MA, United States AIMS: Widespread inclusion of PROs in the health care database requires much more practical yet comprehensive estimates of the outcomes that matter most to patients. This presentation reviews the logic and evidence behind a radical new approach to automatically adapting PRO measurement to be more efficient and useful and discusses issues in transitioning from legacy to new methods. It also presents results from a real-world longitudinal pilot test comparing new with legacy approaches. METHODS: NIH-sponsored projects progressively evaluated and improved generic (QGEN) and diseasespecific (QDIS) measures using cross-sectional and longitudinal realdata simulations for 5418 adults in five disease groups including osteoarthritis. QDIS item banks and a disease checklist enabled standardized estimates of QOL impact attributed to 35 diseases and overall comorbidity impact. QGEN item banks estimated profiles and physical and emotional summary scores. Both used new adaptive survey logic (ASLX) to substantially improve measurement efficiency over both fixed-length legacy forms and routine computerized adaptive tests (CAT). After successful real-data simulations, a pilot test (N = 25) evaluated the feasibility of clinic/home administrations and the usefulness of QDIS and QGEN before and after total joint replacement (TJR) in comparison with legacy tools (e.g., SF-36, WOMAC) used in the national AHRQ-sponsored FORCE-TJR registry. RESULTS: Simulation methods achieved reliable and valid scores in five disease groups with 75 % reductions in survey length compared to legacy measures and 50 % reductions in relation to routine CAT. Reductions were greatest for patients with fewer comorbid conditions and/or less severe disease. The pilot test confirmed large gains in efficiency and very large ([1 SD) pre- vs. postTJR differences in both QDIS and QGEN, replicating the pattern observed for legacy measures. Despite the small pilot test sample, convergent validity correlations were very high for joint-specific QDIS and legacy (r = 0.96, p \ 0.01) and generic QGEN and legacy (r = 0.80–0.89, p \ 0.01) measures. The QDIS overall comorbidity impact score predicted generic physical summary outcomes (r = -0.55, p \ 0.03). CONCLUSIONS: Large gains in efficiency with adaptive measurement achieved in real-data simulations were replicated in a national registry pilot test. Use of standardized QDIS disease impact measures to interpret variations in generic PROs warrants further study in comparative effectiveness research.

Qual Life Res (2015) 24:1–191 (210.2) Cross-cultural validity of the Animated Activity Questionnaire (AAQ) for assessing activity limitations in patients with hip or knee osteoarthritis Wilfred F. Peter, VU University Medical Center, Amsterdam, Netherlands; Henrica C. de Vet, VU University Medical Center, Amsterdam, Netherlands; Maarten Boers, VU University Medical Center, Amsterdam, Netherlands; Jaap Harlaar, VU University Medical Center, Amsterdam, Netherlands; Leo D. Roorda, Amsterdam Rehabilitation Research Center | Reade, Amsterdam, Netherlands; Rudolf W. Poolman, Onze Lieve Vrouwe Gasthuis, Amsterdam, Netherlands; Vanessa A. Scholtes, Onze Lieve Vrouwe Gasthuis, Amsterdam, Netherlands; Martijn Steultjens, Glasgow Caledonian University, Glascow, United Kingdom; Ewa M. Roos, Institute of Sports Science and Clinical Biomechanics, Odense, Denmark; Francis Guillemin, University of Lorrain, Nancy, France; Maria G. Benedetti, Istituto Ortopedico Rizzoli, Bologna, Italy; Antonio Escobar, Basurto University Hospital, Bilbao, Spain; Hanne S. Dagfinrud, National Advisory Unit for rehabilitation in rheumatology, Oslo, Norway; Caroline B. Terwee, VU University Medical Center, Amsterdam, Netherlands AIMS: The Animated Activity Questionnaire (AAQ) measures activity limitations of patients with hip and knee osteoarthritis (HKOA). Good validity and reliability results were found [1]. The AAQ uses videos in which an animation of a basic daily activity is shown with different levels of difficulty. Patients are asked to choose which video best matches their own performance. By showing standardized animations the AAQ minimizes the influence of the patient’s own reference frame and no translation is needed. Therefore, the AAQ should show minimal Differential Item Functioning (DIF) across countries, which means that patients from different countries with the same level of activity limitations should have the same score on each item. The aim of this study is to evaluate DIF as a parameter of cross-cultural validity of the AAQ. METHODS: In 7 European countries patients were asked to complete the AAQ. An example of an item of the AAQ is shown on: https://www. youtube.com/v/ELw9S3X0z5M?hl=nl_NL&version=3&rel=0 Ordinal logistic regression was used to evaluate DIF across languages. As a criterion for uniform DIF we used an odds ratio for translation (Dutch versus one of the six other languages) outside the interval 0.53–1.89. A significant interaction (p value \ 0.05) between translation and total score was considered as non-uniform DIF. Analyses were adjusted for sex, age, and affected joint. RESULTS: Data of 1183 patients were available from: Netherlands (N = 279), Italy (N = 203), United Kingdom (N = 200), Denmark (N = 188), France (N = 173), Spain (N = 78), and Norway (N = 62). Compared to Dutch, none of the 17 items showed DIF in French; uniform DIF occurred in 1 item for Norwegian versus Dutch, Danish versus Dutch, and English versus

Fig. 1 .

123

Qual Life Res (2015) 24:1–191

61 Srabani Banerjee, PhD, CADTH, Ottawa, ON, Canada; Michelle Mujoomdar, PhD, CADTH, Ottawa, ON, Canada

Fig. 2 .

Dutch; for Spanish versus Dutch 2 items showed uniform DIF, and 1 item showed non-uniform DIF; for Italian versus Dutch 2 items showed uniform DIF, and 4 items showed non-uniform DIF. CONCLUSIONS: Cross-cultural validity of the AAQ looks promising. Further analyses using Item Response Theory will be conducted to evaluate the quality of the individual items. The AAQ seems to have great potential for use in international research and daily clinical practice (Figs. 1, 2).

(210.3) Pooling outcome data across multi-attribute utility measures: feasibility in rheumatoid arthritis and hip arthroplasty Sheri L. Pohar, BScPharm, MSc, PhD, CADTH, Ottawa, ON, Canada; Chris Cameron, PhD, CADTH, Ottawa, ON, Canada;

AIMS: The importance of evaluating treatment efficacy from the patient perspective is increasingly recognized in the evaluation of health technologies. Pooling patient-reported outcome data across trials can be challenging when the construct is measured using instruments with underlying differences that contribute to heterogeneity in the estimated treatment effect. However, having a single pooled estimate of effect can be attractive for decision-making. The aim of this study was to evaluate feasibility of pooling HRQL data derived from different multi-attribute utility (MAU) measures using alternate methods of treatment effect estimation. METHODS: RCTs that compared standard dose biologic DMARD with methotrexate (MTX) to MTX alone in patients with rheumatoid arthritis (RA) were identified by searching the literature. Data for hip arthroplasty were obtained from a previously published scoping review. For each condition, data were pooled using random effects models. Treatment effects were estimated using the standardized mean difference (SMD), minimally importance difference units (MIDs) and by converting all measures to natural units of the EQ-5D by simple rescaling or using cross-walk formulas. RESULTS: In the RA studies, HRQL was measured with the HUI3 (n = 2), SF-6D (n = 2) and EQ-5D (n = 1). The SMDs between biologic DMARDs combined with MTX versus MTX alone ranged from 0.36 (SF-6D) to 0.57 (HUI3), with a pooled estimated SMD of 0.41 (95 % CI 0.30–0.52). Differences expressed in MID units ranged from 1.5 (SF-6D) to 3.0 (HUI3) in the individual studies, with a pooled estimate of 1.91 (95 % CI 1.40–2.42) MIDs. The pooled estimates of mean differences were 0.09 (95 % CI 0.06–0.12), and 0.10 (95 % CI 0.07–0.13) crosswalked and rescaled to the EQ-5D respectively. Statistical heterogeneity ranged from 0 to 61 % depending on the method of pooling.

Table 1 Characteristics of Generic Utility Measures of Health-Related Quality of Life Instrument Number of attributes

Attributes

Range of scores

Minimally important difference

EQ-5D

5

Mobility, self-care, usual activities, pain/discomfort, anxiety/depression

-0.59 to 1.00

0.05

HUI3

8

Vision, hearing, speech, ambulation, dexterity, emotion, cognition, pain -0.36 to 1.00

0.06

SF-6D

6

Physical functioning, role participation, social functioning, bodily pain, mental health, vitality

0.03

0.0 to 1.00

EQ-5D = European Quality of Life-5 Dimensions; HUI3 = Health Utilities Index Mark 3; SF-6D = Short Form 6D Table 2 Characteristics of included studies Study

Instrument

NCT008533853

EQ-5D

Group

Duration of follow-up

Placebo + MTX

3 months

From Torrance et al. (2004)

HUI3 4

DE019

99

0.05 (0.32) 0.16 (0.27)

24 weeks

64

0.04 (0.31)

57

0.22 (0.31)

52 weeks

187

0.07 (0.35)

191

0.21 (0.35)

ADA 40 mg + MTX HUI3

From Torrance et al. (2004)4 Strand et al. (2012)5

Placebo + MTX Placebo + MTX ADA 40 mg + MTX

SF-6D

Placebo + MTX

24 weeks

TCZ 4 mg/kg + MTX Emery et al. (2006)6

SF-6D

Placebo + MTX ABA 10 mg/kg + MTX

Mean (SD) change from baseline

189

ADA 40 mg + MTX ARMADA

n

12 months

158

0.08 (0.13)

161

0.13 (0.13)

119 115

0.06 (0.11) 0.11 (0.11)

ABA = abatacept; ADA = adalimumab; MTX = methotrexate; SD = standard deviation; TCZ = tocilizumab

123

62

Fig. 1 Treatment effect as standardized mean difference (SMD)

Fig. 2 Treatment effect in minimally important difference units

Qual Life Res (2015) 24:1–191 after TKR. METHODS: A prospective cohort study identified 247 patients who had primary elective TKR. Patients completed baseline interviews before surgery and were then followed at 6-months and 2-years post-operatively. The Quality of Life Inventory(QOLI) is a multiple-item, domain format which uses a life satisfaction construct. This 32 item measure rates satisfaction of certain aspects of wellbeing but it also rates the importance of those items. Satisfaction of meaningful aspects of life has a greater contribution to well-being. The QOLI generates 16 domains and a summative global QOL score with higher scores indicating higher life satisfaction. The WOMAC is a disease-specific measure of knee pain and function with lower scores representing less pain and greater function. Changes in QOLI scores between baseline and follow-up were assessed using Wilcoxon signed rank sum test. Pearson correlations were calculated between QOLI-raw score and WOMAC score at each time-point. RESULTS: The mean age was 64.9 (sd 8.7) years with 162 (65.6 %) female. The mean global raw score of QOLI improved from 2.88 (sd 1.53) preoperatively, to 3.27 (sd 1.31) at 6 months (p \ 0.001) and to 3.08 (sd 1.46) at 2 years (p = 0.11). Pre-operative QOLI-health was ranked as one of the top important aspects to overall happiness (1.7 sd 0.4) yet with the lowest satisfaction (0.3 sd 2.2). At 6 months, patients reported significant improvements in the QOLI subdomains of health, self-esteem and play. WOMAC pain and function significantly improved after surgery and were associated with overall-QOLI at follow-up, although the magnitude of the association decreased overtime (6-mon: pain r = -0.27, p \ 0.001; function r = -0.25, p \ 0.001; 2 years: pain r = -0.17, p = 0.02; function r = -0.15, p = 0.05). CONCLUSIONS: Initial improvements in health result in significant increases in overall happiness, but over time happiness regresses to its baseline state.

(210.5) The performance and association between patientreported and performance-based measures of physical functioning in research on individuals with arthritis

Fig. 3 Mean difference in EQ-5D units (crosswalked) Significant heterogeneity was observed in the pooled analysis for hip arthroplasty, which suggested that pooling across MAU measures in this indication may not be reasonable. CONCLUSIONS: Preliminary results suggest that it may be feasible to pool outcome data across MAU measures in RA, but not hip arthroplasty. Expressing differences in MID units or natural units of the EQ-5D may facilitate interpretation (Tables 1, 2; Figs. 1, 2, 3).

(210.4) Total knee replacement provides joint pain and function improvement but are patients overall happy? Allyson Jones, PhD, University of Alberta, Edmonton, AB, Canada; Andrea Barbo, MD Anderson Center, Houston, TX, United States; Glenn Landon, St Luke’s Episcopal Hospital, Houston, TX, United States; Sherwin Siff, St Luke’s Episcopal Hospital, Houston, TX, United States; Maria Suarez-Almazor, MD PhD, MD Anderson Center, Houston, TX, United States AIMS: While large improvements with health-related quality of life are seen with total knee replacement (TKR), up to 30 % of patients are not satisfied with outcomes. The concept of QOL also includes satisfaction with life or overall happiness which has typically not been evaluated with this elective surgery. AIM: to examine the associations of joint pain and function with life satisfaction over 2 years recovery

123

Laura C. Pinheiro, MPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Leigh F. Callahan, PhD, School of Medicine, Thurston Arthritis Research Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Lloyd J. Edwards, PhD, Gillings School of Global Public Health, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Rebecca J. Cleveland, PhD, School of Medicine, Thurston Arthritis Research Center, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States AIMS: Researchers conducting evaluations of behavioral interventions such as exercise often include physical functioning as an outcome to determine effectiveness. Physical functioning is a unique endpoint as it can be assessed by a variety of methods including observable performance-based (PB) tasks and patient-reported outcome (PRO) assessments using measures such as the NIH’s PatientReported Outcomes Measurement Information System (PROMIS) or Health Assessment Questionnaire (HAQ). The objective of our study was to evaluate the association between PRO and PB measures of physical functioning to inform decisions of which to use when evaluating the effectiveness of a physical activity intervention. METHODS: We conducted a secondary data analysis of a non-randomized two-arm pre-post community trial of 462 individuals who self-identified as having arthritis and participated in the Walk with Ease (WWE) intervention. Two PRO (PROMIS and HAQ) and eight PB assessments were collected at baseline (pre-intervention) and 6-weeks follow-up. We calculated correlations between PB and PRO measures, assessed how each type of measure captured changes in physical functioning from baseline to follow-up, and compared PRO

Qual Life Res (2015) 24:1–191 and PB measures to self-reported arthritis symptoms of pain, stiffness, and fatigue. RESULTS: Strength of correlations between PB and PRO measures varied depending on the PB measure, ranging from 0.21 to 0.54. PRO and PB measures captured significant physical functioning improvements from baseline to follow-up, as expected none showed significant differences between the WWE modalities (instructor-led or self-directed groups). Correlations with arthritis symptoms were stronger for the PROs (0.30–0.46) than the PB measures (0.03–0.31). CONCLUSIONS: The parent study evaluation of the effectiveness of two WWE modalities would have yielded similar findings had they used eight PB measures or one PRO measure (PROMIS or HAQ). PRO measures may provide us with insights into aspects of physical functioning that are not captured by PB measures alone. Use of PRO measures allows patients to communicate their own perceptions of physical functioning, which may provide a more accurate representation of overall physical functioning. Our study does not suggest abandoning the use of PB measures to capture physical functioning, but recommends that PRO measures may serve as complementary or surrogate endpoints for some studies.

301: Response Shift (301.1) Minimal impact of response shift for SF-12 mental and physical health status in the homeless and vulnerably housed: an item-level multi-group analysis Anne M. Gadermann, PhD, University of British Columbia, Vancouver, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; Anita Palepu, Professor, University of British Columbia, Vancouver, BC, Canada; Anita M. Hubley, PhD, University of British Columbia, Vancouver, Canada; Bruno D. Zumbo, PhD, University of British Columbia, Vancouver, Canada; Tim Aubry, Professor, University of Ottawa, Ottawa, Canada; Susan Farrell, PhD, Royal Ottawa Mental Health Centre & University of Ottawa, Ottawa, ON, Canada; Stephen Hwang, Professor, University of Toronto, Toronto, Canada AIMS: Failure to adjust for response shift can result in an under- or over-estimation of change, thereby confounding the measurement of longitudinal outcomes. Homeless and vulnerably housed individuals (HVHIs) may experience response shift in their perceived health outcomes as they adapt to or experience changes in their living circumstances, which are often transient in nature. The aim of the present study was to evaluate whether HVHIs experienced response shift over a 12-month time period in their physical and mental health status. METHODS: The data for our analyses were obtained from the Health and Housing in Transition (HHiT) study, a longitudinal multisite cohort study in three Canadian cities. HVHIs were recruited using a two-stage sampling method (N = 1190) and were re-interviewed 12 months later (N = 968). Multi-group confirmatory factor analysis (MG-CFA) and methods for response shift detection at the item level, based on the approach by Oort, were used to test for reconceptualization, reprioritization, and recalibration response shift on the SF-12 in four groups of individuals who were homeless (n = 170), housed (n = 437), or who reported a difference in their housing status [from homeless to housed (n = 285) or housed to homeless (n = 73)] at the two measurement occasions. A probit link function with mean and variance adjusted weighted-least squares estimation was used to accommodate the ordinal and binary distributions of the SF-12 variables using the statistical software Mplus. RESULTS: Using MGCFA, a strict invariance model showed that the measurement model was equivalent for the four groups at baseline. Although we found small but statistically significant response shift for several measurement model parameters, the impact on the predicted average mental and physical health scores within each of the groups was very small. Specifically, differences in effect sizes (standardized mean

63 differences) between the model accommodating response shift and the model ignoring response shift were all smaller than .03 for mental and physical health within each of the four groups. CONCLUSIONS: Response shift does not appear to be a significant concern when using the SF-12 to obtain change scores over a 12-month period in this population.

(301.2) Response shift in patients receiving a pre-emptive kidney transplant compared to similar recipients having experienced a short dialysis period: study protocol for a prospective, observational, multicenter controlled study Veronique Sebille, Professor of Biostatistics, University of Nantes, Faculty of Pharmacy, France, Nantes, France; Jean-Benoit Hardouin, PhD, University of Nantes, Nantes, France; Angelique BonnaudAntignac, Professor of Psychology, University of Nantes, Nantes, France; Philippe Tessier, Associate Professor of Health Ecoomics, University of Nantes, Nantes, France; Magali Giral, Professor of Nephrology, ITUN, Nantes, France; Yohann Foucher, Associate Professor of Biostatistics, University of Nantes, Nantes, France; Aurelie Meurette, Hospital Practionner in Nephrology, ITUN CHU de Nantes, Nantes, France AIMS: Treatment of end stage renal disease has an impact on the physical and psychological health of patients, including quality of life (QoL). Nowadays, it is known that reducing the dialysis period has many advantages regarding QoL and medical outcomes. Although preemptive transplantation is the preferred strategy to prevent patients undergoing dialysis, its psychological impact and the influence of response shift on outcome assessment (e.g. QoL) is unknown compared to recipients who were treated by dialysis before transplantation. Whether response shift is experienced in a different way among patients who were on dialysis and those who still had a functioning kidney at time of surgery is unknown and will be studied. Understanding the psychological impact of preemptive transplantation is an important issue since it can be associated with long-term patient and graft survival. METHODS: Adult patients with a pre-emptive transplantation (PPT group, n = 130) from deceased donors will be prospectively included along with a comparable control group of patients with a short pre-transplant dialysis period \12 months (PDT group, n = 260; matching on recipient gender and center). Only first and single kidney transplantation will be considered. Primary and secondary endpoints include: comparison of change between PPT and PDT groups in specific and generic QoL (ReTransQol, SF-36), anxiety and depressive disorders (HAD), perceived stress (PSS), taking into account response shift. These criteria will be evaluated every 6 months prior to surgery, at hospital discharge, at 3 and 6 months, 1 and 2 years after transplantation. RESULTS: The main assumption is that pre-emptive transplanted patients, who generally have a better health level than patients undergoing dialysis, might although have a lower post-transplantation QoL and experience response shift differently and not to the same extent. CONCLUSIONS: Our study will give more insight on the response shift experience of pre-emptive patients compared to patients with a short period under dialysis. It will help organizing therapeutic educational programs and psychological support specifically adapted to pre-emptive patients before and following surgery. This point is also of major importance to prevent poor adherence to treatment.

(301.3) Response shift and perceptions of quality of life following neurogenic bowel and bladder in persons with spinal cord injury Denise G. Tate, Professor, University of Michigan Medical School, Ann Arbor, MI, United States; Edward Rohn, Research Associate,

123

64 University of Michigan, Ann Arbor, MI, United States; Colette Duggan, Senior Research Associate, Detroit Medical Center/ Rehabilitation Institute of Michigan, Detroit, MI, United States; Roxane Madrid, Research Assistant, University of Michigan, Ann Arbor, MI, United States; Martin Forchheimer, Senior Research Associate, University of Michigan, Ann Arbor, MI, United States; Emily Zafiroff, Research Assistant, University of Michigan, Ann Arbor, MI, United States AIMS: Defining QOL after disability has implications for long-term adaptation and treatment readiness. The concept is highly subjective varying for same person over the life trajectory. ‘‘Response shift’’ captures individual changes over time, as influenced by patterns of coping, reframing expectations, and meaning-making. Understanding patients’ changing perceptions of their own QOL following loss of bladder and bowel functions is important for clinicians in assessing and implementing treatment plans and assisting patients in coping with their injury long-term. This study reviews data from persons with recent and long-term spinal cord injury both from civilian and military backgrounds. It compares these experiences while investigating the associations between response shift and QOL perceptions. METHODS: A mixed method approach was used to study perceptions. Using qualitative interviews, data were collected on the impact of neurogenic bowel and bladder on quality of life for veterans and non-veterans with SCI. Participants also completed a series of questionnaires within 2 weeks of their interview (BBTI, PROMIS, SCIQOL) to assess bladder and bowel management, complications and QOL. Data was analyzed using NVivo software, Chi square and ANOVA. Participants (n = 40) were recruited from registries at a large medical center and the local VA clinic. RESULTS: Preliminary data based on a sample of 29 persons indicates that the average time since injury was 21.3 years with 10 participants with injuries of less than a year. Conceptualizations of ‘‘physical independence’’ emerged as a salient theme across participants. Comparing group responses revealed significant conceptual and meaning-driven differences in the nature and place of independence in participants’ QOL. Newly injured participants characterized loss of independence as a negative influence on their QOL. Those injured more than 10 years described independence as something once challenged and now regained, having an overall positive influence on their QOL. References of bladder and bowel effects on QOL were presented among 45 % of those newly injured. Differences were noted with respect to QOL ratings (p \ .04) and with social role activities (p \ .001). There were no differences between military and civilian participants. Preliminary data suggests an inverse relationship between QOL and response shift, with those rating their QOL lower presenting greater indication of response shift. CONCLUSIONS: Notions of ‘‘independence’’ appear to change over time, reflecting a response shift as those living longer with SCI learn to adapt and redefine their independence as it relates to QOL. Recognizing these patterns can assist clinicians by signaling patient readiness to treatment that can best improve QOL. Changes in reconceptualization of self, greater acceptance, and awareness of a new self with realistic expectations of life goals suggest adjustments to QOL and a greater sense of control over life. Differences in perception and response shift were observed between recent and chronic SCI.

(301.4) Evaluation of the Oort’s procedure for response shift detection at item level—a simulation study Alice Guilleux, University of Nantes, Nantes, France; Myriam Blanchin, PhD, University of Nantes, Nantes, France; Jean-Benoit Hardouin, PhD, University of Nantes, Nantes, France; Veronique Sebille, Professor of Biostatistics, University of Nantes, Faculty of Pharmacy, France, NANTES, France

123

Qual Life Res (2015) 24:1–191 AIMS: Issues regarding the detection of the response shift (RS) at item-level remain seldom investigated. To our knowledge no SEM based method has been proposed for the detection of the RS at itemlevel. In this context, a method based on the Oort’s procedure which is often used for RS detection at the dimension level was proposed. This procedure is composed of 4 steps (1-Establishing a measurement model 2-Overall test of absence RS 3-RS detection 4-True change estimation). These steps allow detecting and taking into account the different types of RS: recalibration, reprioritization and reconceptualization. However, several questions arise regarding the procedure and its use at the item level: (1) the adaption of the procedure for item-level analyses. (2) the different methodological choices at each step of the procedure. METHODS: A simulation study has been carried out in order to find the appropriate method to detect RS at item-level based on structural equation modeling. Eight different versions of the Oort’s procedure were tested in order to compare the performance of each method. For this purpose, the type I error and the power for detecting the proper type of RS on the correct items were computed. Data were simulated with several parameters (for example, the number of items or categories and the type of RS) to constitute different scenarios. Each version of the Oort’s procedure was realized at item-level and different methodological choices during the operationalization of the procedure were assessed. Namely, the need to include a global of RS at step 2, the need to hierarchically test for the different forms of RS during step 3, the requirement to adjust p-values for multiple comparison in the different tests of step 3. RESULTS: The first results of this simulation study show that only uniform recalibration is properly detected. Moreover, regarding the methodological choices, the test of global occurrence of RS and adjustment of p-values are required. CONCLUSIONS: Oort’s procedure needs to be adapted to be used as a SEM-based method to detect RS at item level. This simulation study allows proposing paths of research for the implementation of the procedure at item-level.

(301.5) Identification of response shift in the Manitoba follow-up study: at the individual and item level Maryam A. Alshammari, University of Manitoba, Winnipeg, Canada; Robert Tate, PhD, University of Manitoba, Winnipeg, Canada; Donna Collins, MSc, University of Manitoba, Winnipeg, Canada; Ruth Barclay, PhD, University of Manitoba, Winnipeg, Canada AIMS: The objective was to provide an individual level identification of response shift (RS) in community-dwelling older men over a 1-year period (2010–2011). METHODS: Since 1948, a cohort of 3983 male World War P Royal Canadian Air Force aircrew have been followed by the Manitoba Follow-up Study (MFUS) to identify incidence of cardiovascular diseases, and recently, predictors of successful aging. Data from the MFUS, Successful Aging questionnaire was used. In 2 years of data, participants identified the importance of 15 items, which reflect 2 physical, 5 mental, and 8 social domains of quality of life. Items were scored as ‘not important, moderately important or very important’. Types and direction of RS in relation to domains were identified at an individual level. RESULTS: 360 men had a mean age of 89.7 (2.9) years in 2011. Overall, the least amount of RS was identified in the 5 mental domain items ranging from 9.3 to 24.3 % of individuals (median 20.2 %), and the most RS was identified in the 8 social domain items ranging from 14.8 to 39.3 % (median 30.5 %). In detecting specific types of RS, reprioritization was identified in all of the items with a minimum 6.3 % and a maximum 27.2 % of individuals. Less than 3 % of older men demonstrated reconceptualization in 9 items, and 5.6–19 % demonstrated it in other 7 items. Direction of RS varied among items; e.g., of those respondents who demonstrated reprioritization in the item ‘Helping family/friends’, 11.6 % showed an increase in

Qual Life Res (2015) 24:1–191

65

Table 1 Concordance of patient-reported outcomes based performance measures assessed using different patient-reported outcome measures PRO-PM

PROM

% Depression at baseline visit

% Depression remission

Group-level Concordance with PRO-PM assessed using PHQ-9

Patient-level Concordance with PRO-PM assessed using PHQ-9

7* [3.12]

PHQ-9 [ 9

35.1 % (2012/ 5736)

n/a

n/a

n/a

PROMIS SF

52.7 (11.2)

PROMIS [ 59.9**

26.5 % (1518/ 5736)

n/a

91.4 %

82.5 % (4734/5736)

PHQ-9PROMIS

55.3 (10.5)

PROMIS [ 59.9**

32.1 % (1844/ 5736)

n/a

97.0 %

95.6 % (5486/5736)

Depression diagnosis PHQ-9 (N = 5736)

Depression remission (n = 701)

PROM Score Depression [IQR], (SD) threshold

PHQ-9

8* [4.14]

PHQ9 [ 9

41.5 % (291/701)

6.5 % (19/291) n/a

n/a

PROMIS SF

55.7 (14.4)

PROMIS [ 59.9**

35.2 % (247/701)

5.7 % (14/247) 99.6 % 

92.9 % (184/198) 

PHQ9PROMIS

57.5 (10.3)

PROMIS [ 59.9**

38.1 % (267/701)

6.7 % (18/267) 97.9 % 

98.5 % (256/260) 

PRO-PM = patient-reported outcomes based performance measure, PROM = patient-reported outcome measure PHQ-9 = patient health questionnaire 9, PROMIS SF = PROMIS depression short-form, PHQ-9PROMIS = PHQ-9 calibrated to PROMIS metric IQR = interquartile range, SD = standard deviation * Median, ** based on thresholds used by Choi et al  

Includes only patients who had diagnosis of depression by both PHQ-9 and relevant PROM

importance and 15.6 % showed a decrease in importance. In reconceptualization, more older men dropped items from their list of importance. For example, in the ‘Being socially active’ item, 12 % dropped it from their list, whereas, 7 % added it. CONCLUSIONS: This secondary data analysis from the MFUS provides the opportunity to identify RS at the individual level and item level. Using individual level methods will improve the understanding of RS. Also, it will be used for identification of factors that predict response shift in older men.

302: New Developments in PROMIS (302.1) Use of scale linkage to determine performance of PROPMs assessed with different patient-reported outcome measures Irene Katzan, Cleveland Clinic, Cleveland, OH, United States; Fan Youran, Cleveland Clinic, Cleveland, OH, United States; Sandra D. Griffith, Cleveland Clinic, Cleveland, OH, United States; Paul Crane, University of Washington, Seattle, WA, United States; Nicolas Thompson, Cleveland Clinic, Cleveland, OH, United States; Deborah Miller, PhD, Cleveland Clinic, Cleveland, OH, United States; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: Patient-reported outcome based performance measures (PROPMs), which incorporate outcomes from the patient’s perspective into quality of care (‘‘performance’’) measures, has greater potential to impact healthcare than measures focusing on processes of care. One challenge to PRO-PM implementation is the variety of PRO measures (PROMs) used in clinical practice to assess the same outcome. A potential solution is to establish linkages between scores of different scales that would provide equivalent scores for different PROMs. To evaluate this potential solution, we determined the group-level and patient-level concordance in performance of 2 depression-related PRO-PMs assessed using different depression PROMs: Patient Health Questionnaire-9 (PHQ-9), PROMIS depression short-form (PROMIS SF), and the PHQ-9 co-calibrated on the PROMIS metric (PHQ9(PROMIS)). METHODS: The study population consisted of patients seen in 8 ambulatory neurologic clinics from January 23 to June 15, 2012 who had completed both a PHQ-9 and PROMIS SF at the same visit. PHQ-9 was co-calibrated on the PROMIS metric as described

by Choi. The depression PRO-PMs assessed were: (1) Depression Diagnosis: The percentage of patients with depression (PHQ-9 [ 9) at the time of initial assessment (based on NQF #0418). (2) Depression Remission: The percentage of patients with initial PHQ-9 [ 9 who have a follow-up PHQ-9 \ 5 (based on NQF #0710). RESULTS: There were 5736 patients in the cohort with mean age 46.8 years (SD 15.1), 701 patients had PROMs from [2 visits. Group-level concordance of the Depression Diagnosis PRO-PM assessed with different PROMs was 91.4 % between PHQ-9 and PROMIS SF and was 97.0 % between PHQ-9 and PHQ-9(PROMIS). Patient-level concordance was 82.5 % between PHQ9 and PROMIS SF and 95.6 % between PHQ-9 and PHQ-9(PROMIS) (see Table 1). Group-level concordance of the Depression Remission PRO-PM was 99.6 % between PHQ-9 and PROMIS SF and 97.9 % between PHQ-9 and PHQ-9(PROMIS). Patient-level of concordance was 92.9 % between PHQ-9 and PROMIS SF and was 98.5 % between PHQ-9 and PHQ9(PROMIS). CONCLUSIONS: There was a high level of concordance in performance at the group level for depression PRO-PMs measured using PHQ-9, PROMIS SF and PHQ-9(PROMIS). These findings provide support for the ability to use linkage of scale scores to assess performance of PRO-PMs using different PROMs.

(302.2) Development of the PROMIS Nociceptive Pain Scale Cindy J. Nowinski, MD, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Dennis Revicki, PhD, Evidera, Bethesda, MD, United States; Dagmar Amtmann, PhD, University of Washington, Seattle, WA, United States; Kaleb Michaud, PhD, University of Nebraska Medical Center, Omaha, NE, United States; Michael A. Kallen, PhD, MPH, Northwestern University, Feinberg School of Medicine, Missouri City, TX, United States; Robert Askew, PhD, MPH, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: The experience of chronic pain is comprised of multiple components, and identifying specific mechanisms responsible for pain is critical for selecting an appropriate treatment. Analysis of the original PROMIS Pain Quality item pool suggested potentially

123

66 clinically meaningful factors were associated with different types of pain (nociceptive, neuropathic); these factors could be of value in determining underlying pain mechanisms. The overall goal was to extend the measurement use of the PROMIS Pain Quality item pool by developing additional measures of distinct components of pain quality. We report on the new PROMIS Nociceptive Pain Scale. METHODS: The original PROMIS adult pain quality item pool was expanded and harmonized with pediatric pain quality items. The resulting 69 items were administered to 735 patients diagnosed with a nociceptive (osteoarthritis, rheumatoid arthritis) or neuropathic (diabetic neuropathy, cancer chemotherapy-induced neuropathy) pain condition. Participants were recruited from the National Data Bank for Rheumatic Diseases, Northwestern University, and the University of Washington. ANOVA and confirmatory factor analysis (CFA) were used to identify candidate nociceptive pain characteristic items. Subsequent IRT model and item-fit analyses were conducted to insure the final set of items met established criteria for inclusion in a calibrated scale. RESULTS: ANOVAs conducted on the 39 pain quality items using a 5-point response set (‘‘not at all’’ to ‘‘very much’’) identified seven items showing nociceptive pain patients with statistically significantly higher levels of a pain characteristic than neuropathic pain patients. A preliminary CFA provided evidence of unidimensionality. However, in subsequent IRT analyses two items were eliminated (local dependence; misfit), resulting in a 5-item (‘‘sore’’, ‘‘achy’’, ‘‘tender’’, ‘‘deep’’, ‘‘steady’’) nociceptive scale meeting psychometric criteria. Internal consistency reliability was estimated to be 0.83. No nociceptive vs. neuropathic pain patient differential item functioning was identified. Nociceptive pain patients scored significantly higher (ANOVA, p \ .001) on the Nociceptive Pain Scale (mean T-score = 52.7, SD = 8.4) than did neuropathic pain patients (mean T-score = 46.4, SD = 10.9). CONCLUSIONS: The 5-item PROMIS Nociceptive Pain Scale is a valid, reliable, IRT-calibrated measure that can be used to assess nociceptive pain in chronic pain conditions. Harmonization of the pediatric and adult pain quality items enables nociceptive pain assessment across the life span.

(302.3) Extending PROMIS and neuro-QOL to Huntington disease: new measures of chorea, speech and swallowing difficulties, and end of life concerns Noelle Carlozzi, University of Michigan, Ann Arbor, MI, United States; Jin-Shei Lai, Northwestern University, Chicago, IL, United States; Stephen Schilling, University of Michigan, Ann Arbor, MI, United States; Elizabeth A. Hahn, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Anna Kratz, University of Michigan, Ann Arbor, MI, United States; Nancy Downing, University of Iowa, Cedar Rapids, IA, United States; Jane Paulsen, University of Iowa, Cedar Rapids, IA, United States; Martha Nance, Park Nicollet, Golden Valley, MN, United States; Praveen Dayalu, University of Michigan, Ann Arbor, MI, United States; Joel Perlmutter, Washington University, St. Louis, MO, United States; Michael McCormack, Rowan University, Glassboro, NJ, United States; Kimberly Quaid, Indiana University, Indianapolis, IN, United States; Christopher Ross, Johns Hopkins University, Baltimore, MD, United States; Susan Perlman, UCLA, Los Angeles, CA, United States; Jennifer Miner, University of Michigan, Ann Arbor, MI, United States; Stacey Barton, Washington University, St. Louis, MO, United States; Humberto Marin, Rutgers University, Piscataway, NJ, United States; Michael Geschwind, UCSF, San Fransisco, CA, United States; Stephen Rao, Cleveland Clinic, Cleveland, OH, United States; Richard C. Gershon, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States

123

Qual Life Res (2015) 24:1–191 AIMS: Existing patient-reported outcomes measures (PROs) used in Huntington’s disease (HD), a debilitating genetic disease that impacts physical, behavioral, emotional, cognitive, and social health, are typically not comprehensive and do not fully account for clinically meaningful changes in function. While new PROs examining healthrelated quality of life (HRQOL) such as Neuro-QOL (www. neuroqol.org) and PROMIS (www.nihpromis.org) offer solutions to these shortcomings, they do not include HD-specific content. The HDQLIFE study addresses this shortcoming by extending PROMIS/ Neuro-QOL to include HD-specific content. Specifically, the purpose of this study was to develop and calibrate new HD-specific item banks and computer-adaptive tests (CATs) according to established PROMIS methodology. METHODS: Data were collected from 539 individuals with prodromal, early-stage or late-stage HD. Participants completed newly developed HD-specific item pools (159 items) examining chorea, speech and swallowing difficulties, and end of life concerns. Item bank and CAT development included: establishing unidimensionality (exploratory and confirmatory factor analysis [EFA and CFA]), establishing psychometric properties with graded response model (GRM), and excluding items exhibiting differential item functioning (DIF). RESULTS: Chorea. EFA identified 36 unidimensional items; CFA indicated satisfactory fit indices; 2 items were deleted due to poor fit; no items exhibited DIF. The final bank includes 34 items. Speech. EFA identified 27 unidimensional items; CFA indicated good fit indices; all items fit GRM model; no items exhibited DIF. The final bank includes 27 items. Swallowing. EFA identified 17 unidimensional items; CFA indicated good fit indices for 16 items; all items fit GRM model; no items exhibited DIF. The final bank includes 16 items. Death and Dying. EFA identified 15 unidimensional items; CFA indicated good fit indices for 12 items; all items fit GRM model; no items exhibited DIF. The final bank includes 12 items. CONCLUSIONS: Four new item banks and CATs were developed as part of HDQLIFE: Chorea, Difficulties with Speech, Difficulties with Swallowing, and Concern with Death and Dying. Six-item short forms were selected for each of these measures; a 4-item short form was also developed to assess Meaning and Purpose. While these new tests have been developed and calibrated in HD, they have potential utility in other neurological diseases with overlapping symptomatology.

(302.4) Validation of PROMIS scales in adolescents and young adults with special health care needs: a report from the PROMIS linking study Carrie R. Howell, St. Jude Children’s Research Hospital, Memphis, TN, United States; Heather Gross, University of North Carolina at Chapel Hill, Chapel Hill, United States; Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Darren DeWalt, MD MPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; I-Chan Huang, PhD, St. Jude Children’s Research Hospital, Memphis, TN, United States AIMS: Using valid and reliable instruments to measure health-related quality of life (HRQoL) for people with special health care needs (SHCN) is important for research and clinical practice. We sought to validate the NIH Patient-Reported Outcomes Measurement Information System (PROMIS) scales for adolescents and young adults with SHCN by examining HRQoL for different categories and severity of SHCN, and comparing HRQoL between the SHCN and the norm of PROMIS calibration sample. METHODS: 292 adolescents (14–17.9 years of age) and 300 young adults (18–20) with SHCN were recruited from public insurance programs and on-line survey panels. Presence of SHCN was classified into three domains (dependency, service use, functional limitations) based on the SHCN Screener, and severity of SHCN was categorized by the number of

Qual Life Res (2015) 24:1–191 impaired domains (one, two, three). HRQoL was measured with the NIH’s PROMIS Pediatric and Adult Short Forms, including depressive symptoms, anxiety, anger, fatigue, pain interference, peer relationships, and physical functioning. Differences in HRQoL domain scores between SHCN groups and the PROMIS calibration sample (mean = 50; SD = 10) were tested using t-tests. Associations of HRQoL domain scores with impairment in individual SHCN domains and with severity levels of SHCN were tested using multivariable linear regression with and without adjusting for covariates (age, gender, and chronic conditions). RESULTS: Mean scores of all PROMIS domains were statistically worse than the norms in adolescents and young adults. Adolescents and young adults classified as SHCN on the domains of service use and functional limitations reported worse HRQoL than their counterparts without SHCN (p’s \ 0.01). Adolescents and young adults with more severe SHCN reported worse HRQoL on all domains than those with less severe SHCN (p’s \ 0.001), except for the social health domain in young adults. SHCN alone accounted for the most variance on the mobility, depressive symptoms, and anxiety domains for adolescents (14–18 %) and on the pain interference and physical functioning domains for young adults (21–24 %). CONCLUSIONS: Adolescents and adults with SHCN have worse HRQoL compared to the norms, and HRQoL varies across different categories and severity of SHCN. This study suggests PROMIS Pediatric and Adult Short Forms possess satisfactory validity related to SHCN status.

(302.5) Refinement and validation of a general self-efficacy item bank and short form for the NIH PROMIS John M. Salsman, Wake Forest School of Medicine, Winston-Salem, NC, United States; Thomas V. Merluzzi, University of Notre Dame, Notre Dame, IN, United States; Benjamin D. Schalet, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Crystal L. Park, University of Connecticut, Storrs, CT, United States; Elizabeth A. Hahn, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Mallory A. Snyder, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: General self-efficacy is associated with adaptive coping strategies and more positive health outcomes. The Patient-Reported Outcomes Measurement Information System (PROMIS) has developed self-efficacy item banks and short forms for managing symptoms, daily activities, social interactions, emotions, and medications and treatment. However, there is no measure of general self-efficacy within PROMIS to link to its specific domains of self-efficacy. We sought to address this gap by refining and validating an item response theory (IRT)-based patient-reported outcomes assessment tool of general self-efficacy for inclusion in PROMIS. METHODS: Informed by theory and qualitative input from patients and content experts, 10 items were modified from the NIH Toolbox Self-Efficacy Survey by creating new ‘‘confidence’’ response options. For calibration and validation purposes, the newly modified items were administered to a general population sample (n = 1000) along with the NIH Toolbox Self-Efficacy Survey and legacy measures (Life Orientation Test-Revised, LOT-R; Generalized Expectancy for Success-short form, GES-SF). We randomly split the sample into halves and conducted exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). IRT analyses included an assessment of differential item functioning (DIF). RESULTS: Participants had a mean age of 47.8 years old and were primarily male (50.3 %), white (68.3 %), and had some college education or more (50.2 %). In the EFA scree plot (n = 500), all 10 items loaded onto one dominant factor and CFA (n = 500) yielded a good fitting model for a general self-efficacy bank with confidence response options (CFI = 0.987,

67 TLI = 0.984, RMSEA = 0.090). All items were free of gender, age, education and race DIF. Item calibrations and content considerations informed the selection of a 4-item short form. Internal consistency estimates were alpha = .94 and .88 for the new 10-item PROMIS General Self-Efficacy bank and 4-item PROMIS General Self-Efficacy short form, respectively. The 10-item PROMIS General SelfEfficacy Bank was significantly correlated with the LOT-R (r = .58), the GES-SF (r = .55), and the NIH Toolbox Self-Efficacy Survey (r = .87). CONCLUSIONS: The PROMIS General Self-Efficacy measure demonstrated sufficient unidimensionality and displayed good internal consistency, model fit, and convergent validity. Further psychometric testing of the PROMIS General Self-Efficacy item bank and short form will help evaluate the utility of this measurement tool in patients with chronic health conditions.

303: Behavioral Interventions (303.1) Health behavior change counseling improves rehabilitation engagement and leads to better functional outcomes following lumbar spine surgery Richard L. Skolasky, ScD, Johns Hopkins University, Baltimore, MD, United States; Anica M. Maggard, Johns Hopkins University, Baltimore, MD, United States; David Li, Johns Hopkins University, Baltimore, MD, United States; Lee H. Riley, III, MD, Johns Hopkins University, Baltimore, MD, United States; Stephen T. Wegener, PhD, Johns Hopkins University, Baltimore, MD, United States AIMS: Health Behavior Change Counseling (HBCC) is a motivational-interviewing-based intervention designed to improve engagement in rehabilitation and lead to better functional outcomes after lumbar spine surgery. Our goal was to examine whether the brief intervention had an effect on therapist and patient-reported participation in physical therapy (PT) and/or home exercise (HEP) and lead to reduction of disability and improvement in health status after lumbar spine surgery. METHODS: From December 2009 through August 2012, 122 consecutive patients underwent lumbar spine surgery at our institution. Based on date of enrollment date, participants were assigned either to a control (n = 59) or HBCC intervention (n = 63, HBCC) group in a prospective lagged-control design. Following surgery, patients underwent rehabilitation consisting of PT and/or HEP. Therapists assessed engagement using the Hopkins Rehabilitation Engagement Rating Scale (HRERS). Patients reported attendance in PT and/or HEP. At 3 and 6 months following surgery, disability and health status were assessed using the Oswestry Disability Index (ODI) and the Short Form 12, version 2 (SF12). Hierarchical linear models (HLM) assessed the impact of HBCC on participation in rehabilitation and subsequent functional recovery. Significance was set at p \ 0.05. RESULTS: Participation in rehabilitation was significantly higher among those in the HBCC group compared to those in the control group: HRERS, 23.6 (standard deviation (SD) 2.7) vs. 21.2 (SD 4.6), p \ .001); PT, 0.82 (SD 0.16) vs. 0.67 (SD 0.22), p \ .001; and HEP, 0.75 (SD 0.22) vs. 0.65 (SD 0.23), p = .019. Similarly, there was a significantly better functional recovery among those in the HBCC group compared to those in the control group at 3 months (difference: ODI -10.7 (SD 4.4), p = .015 and SF12 6.2 (SD 2.2), p = .004) and 6 months (difference: ODI -12.7 (SD 4.8), p = .008 and SF12 8.9 (SD 2.4), p \ .001). HLM demonstrated that improvements in functional recovery were mediated by participation in rehabilitation. CONCLUSIONS: Patients in the HBCC group more fully participated in rehabilitation and experienced greater functional recovery following surgery compared to those patients in the control group. This improvement in functional recovery was mediated by their improved participation in rehabilitation.

123

68 (303.2) Effects of lifestyle interventions on quality of life in breast cancer patients Haleh Ghavami, PhD., Urmia University of Medical Sciences, Urmia, Azarbayjane Gharbi, Iran; Neriman Akyolcu, PhD., Istanbul University, Istanbul, Turkey AIMS: To compare a 24-week lifestyle intervention program (doing aerobic exercises 45–60 min three times per week for 24 weeks with dietary energy restriction training) to usual care on quality of life (QOL) in women with breast cancer. METHODS: This is a randomized clinical trial study.The Study samples were 80 women with stage I, II, or III breast cancer, that operated for breast cancer and their chemotherapy or radiation therapy completed 3–18 months ago. They are divided randomly into two groups; control group and lifestyle interventions group. Those in the lifestyle intervention group were instructed to practice aerobic exercises 45–60 min three times per week for 24 weeks with dietary energy restriction training. Those in the control group were instructed to continue normal activities and their routine health care. Data were obtained from the patient information form and EORTC QLQ-BR23 questionnaire that completed before and after the lifestyle intervention in both groups. RESULTS: Controlling for baseline scores, change over time between groups was significantly different for the women who practiced lifestyle intervention; those in the intervention group reported significantly improvement in quality of life as compared to control group (p [ 0.05).No baseline differences existed between the two groups for the mean of QLQ-BR23 sub scale scores (p [ 0.05) before the study; but the differences in ‘‘body image’’, ‘‘sexual function’’, ‘‘sexual enjoyment’’, ‘‘future perspective’’ ‘‘Systemic therapy side effects’’, ‘‘breast symptoms,’’ ‘‘Arm symptoms’’ and ‘‘Upset by hair loss’’ sub scale scores between the two groups of study were statistically significant (p \ 0.001)after the intervention. CONCLUSIONS: Patients with breast cancer characterize a group worthy of attention and oncology nurses are in a unique position to offer suggestions to help manage physical and psychological problems, and lifestyle intervention could be considered as part of a cancer survivorship program.Breast cancer patients may benefit mentally from participating in lifestyle intervention. For women with breast cancer, lifestyle intervention can improve QOL. Additional research in lifestyle intervention along with cognitive behavioral therapy also may be beneficial.

(303.3) Exploring the relationship of psychosocial determinants in the context of a weight loss intervention Jessica L. Lawson, MSc, M.A., Yeshiva University, Hoboken, NJ, United States; Elizabeth Seng, Yeshiva University, New York, NY, United States; Charles Swencionis, Yeshiva University, New York, NY, United States; Judith Wylie-Rosett, Einstein School of Medicine, New York, NY, United States AIMS: This study aimed to explore psychosocial determinants that contribute to weight loss success. Specifically, the effects of gender, psychological well-being and marital status were examined on change in BMI. Given the contrasting research in this area, the study aimed to further examine the relationship between variables that are central to both quality-of-life and weight loss dimensions. METHODS: This study analyzed data from a 12 month cognitivebehavioral computer-guided weight loss intervention RCT with three incremental levels of weight loss intensity. Secondary analyses were conducted using longitudinal mixed modeling (N = 588). RESULTS: Mixed models for repeated measures analysis revealed a significant 4-way, Time 9 Group 9 Gender 9 Psychological Well-being interaction, F(2, 572) = 3.20, p \ .05 that qualified the 3-way Group 9 Gender 9 Time

123

Qual Life Res (2015) 24:1–191 Table 1 Patient demographics and characteristics Characteristic

Participants Mean ± SD

Participants (n = 588) %

Gender Female

(484) 82.3

Male

(104) 17.7

Age (years)

52.2 ± 11.7

Baseline BMI

34.9 ± 6.6

25–29.9 Overweight

(112) 19.1

30+ Obese Ethnicity

(476) 80.9

White

(488) 83.0

Black

(65) 11.1

Hispanic

(16) 2.7

Asian/Pacific Islander

(7) 1.2

American Indian

(1) 0.2

Other

(11) 1.8

Education Grade 10–11 H.S.

(3) 0.1

H.S. degree

(91) 15.5

1+ yr H.S.

(143) 24.3

College degree

(152) 25.9

Graduate degree

(199) 33.8

Marital status Single Married

(91) 15.5 (397) 67.5

Divorced

(64) 10.9

Widowed

(36) 6.1

Religion Catholic

(125) 26.7

Jewish

(152) 32.4

Protestant

(36) 7.7

None

(14) 2.5

Other

(18) 3.8

No response

(124) 26.4

interaction, F(2, 575) = 4.04, p \ .05, which qualified the significant main effect of Psychological Well-being, F(1, 580) = 4.58, p \ .05. Post hoc analysis revealed that weight loss was significant across all arms of the intervention, however, compared to the rest of the sample, women with lower psychological well-being lost the most weight if they were randomized to the most intense arm of the intervention and received professional support in addition to the computer-guided intervention, t(574.88) = 2.52, p = 0.01. CONCLUSIONS: Compared to men, the level of weight loss intervention intensity was significantly more meaningful for women, particularly those with lower baseline psychological well-being. This finding has clinical implications regarding weight loss, which extend to gender and socio-economic disparities. Further, appropriate allocation of professional resources may improve individuals’ weight loss success and overall health related quality-of-life (Tables 1, 2; Fig. 1).

Qual Life Res (2015) 24:1–191

69

Table 2 Fixed effects Variable

Numerator df

Denominator df

F

Sig.

Time

1

589.75

Treatment Group [Tx]

2

589.31

.50 .61

Psych Well-Being [Psych]

1

579.60

4.58 .03*

.87 .34

Marital Status [Marital]

1

573.76

.05 .82

Gender

1

577.27

.08 .78

Tx * Time

2

576.56

.15 .86

Time * Psych Marital * Time

1 1

583.65 605.55

.04 .85 .12 .73

Gender * Time

1

576.09

.82 .37

Tx * Psych

2

589.15

.91 .40

Tx * Marital

2

565.54

1.22 .30

Tx * Gender

2

594.29

.51 .60

Marital * Psych

1

572.39

.00 .94

Gender * Psych

1

576.14

.05 .83

Marital * Gender

1

563.19

.21 .65

Tx * Time * Psych

2

571.68

.36 .70

Tx * Marital * Time

2

588.01

1.48 .23 4.04 .02*

Tx * Gender * Time

2

575.47

Marital * Time * Psych

1

601.25

.00 .96

Gender * Time * Psych

1

569.98

1.44 .23

Marital * Gender * Time

1

576.68

.30 .58

Tx * Marital * Psych Tx * Gender * Psych

2 2

564.76 594.24

1.46 .23 .35 .70

Tx * Marital * Gender

2

562.43

.39 .67

Marital * Gender * Psych

1

562.26

.15 .70

Tx * Gender * Time * Psych 2

571.86

3.20 .04*

* p \ .05

(303.4) Patient-reported outcomes and the use of a diabetes diary mobile application to attain lifestyle changes for persons with type 2 diabetes Heidi Holmen, MSc, Oslo and Akershus University College of Applied Sciences, Oslo, Norway; Astrid Torbjørnsen, MSc, Oslo and Akershus University College of Applied Sciences, Oslo, Norway; Lis Ribu, PhD, Oslo and Akershus University College of Applied Sciences, Oslo, Norway AIMS: To discuss how the use of a mobile phone–based self-management system for persons with diabetes, with and without health counseling by a diabetes specialist nurse, could improve patient reported outcomes in persons with type 2 diabetes. METHODS: We conducted a low-intensity 1-year randomized controlled trial with two intervention groups and one control group provided outside the ordinary care. The patients monitored their blood-glucose, eating habits, physical activity and personal goals with a mobile diabetes application, and with health counseling as a booster for the first 4 months in one of the groups. Inclusion criteria were age C18, HbA1c C 7.1 %, being able to use the technology and answer the questionnaires in Norwegian. The primary outcome was HbA1c collected through the GPs. Secondary outcomes included: self-management (heiQ), HRQL (SF-36), depressive symptoms (CES-D), and lifestyle changes. Data were analyzed using univariate methods (t test, ANOVA) and multivariate linear and logistic regression. RESULTS: A total of 151 adults were included, mean age 57 years, HbA1c 8.2 % (SD1.1), 66 mmol/mol (SD12.3), mean BMI 31.7 kg/m2 (SD 6.03); multimorbidity C2.48 %; mean diabetes duration 10 years. We found no effect on HbA1c after the 1-year intervention. Secondary outcomes showed an increase in the self-management domain (heiQ) of skill and technique acquisition. CONCLUSIONS: We found no significant differences between the groups in the primary outcome HbA1c after 1-year of intervention. The lack of findings may relate to the subjects disease burden with overweight and comorbidities, and somewhat outdated mobile phone used, due to the rapid development of mobile phones and the relative long study period. In addition, some difficulties was experienced with the Bluetooth pairing required for automatic transmission of data from the glucometer to the app in the mobile phone. Further, the intervention might have been too weak, due to the participants’ disease burden. More research within this area is needed, with longer follow-ups, and with more intensive HCP support. The use of diabetes mobile apps in health care based on our findings will be discussed more thorough at the ISOQOL conference.

(303.5) Sense of coherence (SOC) among Japanese white collar workers participating in ‘‘Salutogenic Cafe´’’. A 6 months followup study investigating development of SOC and the use of new skills Shinichiro Sasahara, Occupational Psychiatry, University of Tsukuba, Tsukuba-City, Ibaraki pref., Japan; Junko Sakano, Health and Welfare Science, Okayama Prefectural University, Soja, Okayama, Japan; Eva Langeland, University college of Bergen, Bergen, Norway; Ragnhild Sollesnes, University college of Bergen, Bergen, Norway; Yuichi Oi, University of Tsukuba, Tsukuba, Japan; Kazuya Usami, University of Tsukuba, Tsukuba-City, Japan; Yoshihiko Yamazaki, Nihon Fukushi Univerisity, Nagoya, Japan; Ichiyo Matsuzaki, University of Tsukuba, Tsukuba, Japan

Fig. 1 Change in BMI by intervention group, gender and psychological well-being

AIMS: The aim of the present study was to investigate the development of sense of coherence (the main concept in salutogenesis) and the extent of implementation of new skills after participating in the interaction program ‘‘Salutogenic Cafe´’’ (SC). Salutogenic interaction training programs are considered important for promoting stress coping power in a work place. METHODS: A prospective design

123

70

Qual Life Res (2015) 24:1–191 (303.6) Examining the association between modifiable health behaviors and health-related quality of life among U.S. adults William W. Thompson, PhD, U.S. Centers for Disease Control and Prevention, Atlanta, GA, United States; Vincent Campbell, PhD, Center for Disease Control & Prevention, Atlanta, GA, United States; John P. Barile, PhD, University of Hawaii, Honolulu, HI, United States; Gloria L. Krahn, PhD, Oregon State University, Corvallis, OR, United States

Fig. 1 Profile of mean SOC-13 score time by the level of practice after the intervention of Salutgenic cafe´ (SC) Table 1 Estimated regression coefficients from a mixed model to show the effect of Saljtoiienic Cafe by practice level on outcome (SOC-13) over time after adjustment for gender and age Outcome

SOC (baseline)

b

Semirobust SE(b)

z/t

95 % CI P Lower Upper value

97.25 6.63

14.68 0.00

85.09

111.14

Time (before, after, 6 M)

102.43 6.92

14.80 0.00

89.73

116.93

Practice level*

12.03 2.11

5.70 0.00

7.89

16.17

1.81 0.73

2.49 0.01

0.39

3.24

Gender Age Constant

0.01 0.02

0.63 0.53

0.00

0.25

48.03 1.63

29.55 0.00

44.83

51.24

* Practice level; ‘‘To what extent do you have used the new skills learned from SC in the daily life?’’ (4 points likert scale; Always, sometimes, seldom, never) included a baseline assessment and two follow-ups. A total of 718 business Japanese white collar workers in a certain company participated in our program SC which is composed of 2-h group discussion at one time. The program is built on the Salutogenic Talk Therapy Group program (Langeland E et al. 2006). The discussion was focused on how to cope with various difficulties in their work place. Sense of Coherence was measured by SOC13 in Japanese version. The use of new skills level (practice level) was measured by the following question: ‘‘To what extent do you have used the new skills learned from SC in the daily life?’’ The main effect of SC and practice level was analyzed in the linear mixed model adjusted by gender and age. RESULTS: A total of 409 persons (response rate; 57.1 %, men/women; 334/75 persons, mean age; 39.7 ± 7.6 years) completed self-administered questionnaires of SOC-13 scale 2 weeks before, 2 weeks after and 6 months after the program including the questionnaire of practice level after the SC. SOC was increased over time by the level of practice (Fig. 1). When controlling for gender and age, it was revealed that SOC significantly increased in all levels of practice. In addition SOC increasing power was significantly higher with increasing level of practice (Table 1) CONCLUSIONS: This study indicates that The Salutogenic Cafe´ could promote Japanese white collar workers’ SOC in their workplace and promote coping. However, in lack of a control group we have to be cautious with our interpretation. The next step is planned to investigate the effect of Salutogenic Cafe´ in a randomized controlled trial.

123

AIMS: Previous studies have shown that most chronic conditions are associated with health-related quality of life (HRQOL), but few studies have attempted to compare and contrast the impact and effect size of different modifiable health behaviors on HRQOL. The aim of this study is to characterize the relationship between two modifiable health behaviors (smoking and walking) and HRQOL. METHODS: We used the 2010 National Health Interview Survey (NHIS), a crosssectional household interview survey of the non-institutionalized civilians in the U.S., to examine the potential impact of smoking and walking on HRQOL. There were four smoking categories (everyday, some days, former and never). The walking variable was based on a question that asked whether you had difficulty walking a quarter mile and this question had 5 response categories (not at all, only a little, somewhat, very, and can’t do at all). The Patient Reported Outcomes Measurement Information System (PROMIS) Global Health scale was used to assess physical and mental health. The physical and mental scales comprise four items each and are in t-score units. A total of 24,043 adults were included in the analyses. All analyses were weighted to be nationally representative and were adjusted for age, sex, marital status, education, and income. RESULTS: We found dose response relationships for the effects of smoking and difficulties walking a quarter mile on both physical and mental health. For physical health, the difference between current smokers and those who never smoked was 2.6 (95 % CI 2.31–2.91) t-score points while the difference between those who had no difficulty walking a quarter mile and those who could not do this activity was 17.6 (95 % CI 17.0–18.1) t-score points. For mental health, the difference between current smokers and those who never smoked was 3.6 (95 % CI 3.2–3.9) t-score points while the difference between those who had no difficulty walking a quarter mile and those who could not do this activity was 9.4 (95 % CI 8.6–10.1) t-score points. CONCLUSIONS: Determining effect sizes and potential benefits of different modifiable health behaviors will be valuable for designing effective interventions to reduce smoking and increase physical activity among all adults.

304: Quality of Life in Children (304.1) Health related quality of life in children and adolescents born with esophageal atresia—use of focus groups in the development of a condition-specific questionnaire Michaela Dellenmark-Blom, Quality Manager, The Queen Silvia Children’s Hospital, Sahlgrenska University Hospital, Gothenburg, Sweden; John Chaplin, PhD, AFBPsS, CPsychol., Inst. Clinical Sciences, Sahlgrenska Academy at the University Gothenburg, Sweden, Gothenburg, Sweden; Linus Jo¨nsson, Pediatric Surgeon, The Queen Silvia Children’s Hospital, Sahlgrenska University Hospital, Gothenburg, Sweden; Vladimir Gatzinsky, Pediatric Surgeon, The Queen Silvia Children’s Hospital, Sahlgrenska University Hospital, Gothenburg, Sweden; Kate Abrahamsson, Pediatric Surgeon, The Queen Siliva Children’s Hospital, Sahlgrenska University Hospital, Gothenburg, Sweden AIMS: To identify the framework for the development of a healthrelated quality-of-life (HRQOL) instrument for esophageal atresia (EA). In line with medical advances in treatment of rare conditions individuals born with EA, are living longer with their chronic medical

Qual Life Res (2015) 24:1–191 condition. METHODS: This study followed the two stage focus group methodology outlined by the DISABKIDS project. Firstly, moderated focus group discussions were used to explore subjective HRQOL as the concepts of children and their parents. Secondly, participants were asked to reflect upon the discussion and write their own questions. The discussions were digitally recorded, transcribed and extracted statements content-analyzed using a card sorting procedure. RESULTS: Ten focus groups were stratified for severity of EA, age (0–7, 8–12, 13–17 years) and gender. Altogether 30 families (18 children, 32 parents) were invited and all of them participated as planned (100 %). Focus group discussions were led by one moderator; a research assistant was also present during the child focus groups. A total of 1371 HRQOL statements were identified in the discussions and categorized into 32 dimensions. A total of 104 selfwritten questions were contributed by focus group participants and categorized into 18 dimensions. All participants contributed, however the majority of HRQOL statements in focus group discussions (69 %) and self-constructed questions (78 %) were generated from the severe EA group. The most frequently HRQOL statements regarded physical activity such as sport and play (n = 116), social exclusion and bullying (n = 88) and food limitations, restrictions, special food preferences or needs (n = 82). The most common self-constructed questions regarded; self-confidence and self-perception due to EA (n = 19); food and eating experiences (n = 13); impact of medical treatment (n = 13). Two age related versions of the pilot questionnaire were subsequently constructed: a 50 item proxy questionnaire for parents of children under 8 years and a 100 item self-completion & proxy version for children and adolescents 8–17 years. CONCLUSIONS: A multi-method approach utilizing focus group discussions and reflective question writing by participants has identified HRQOL experiences not yet reported in previous literature. This demonstrates the utility of the method and will contribute to the development of the first condition-specific HRQOL questionnaire for patients with EA.

(304.2) Quality of life in children is associated with severity of atopic dermatitis and with depressed mood: a cross sectional study Andrea Na´jera, General Practitioner, Centro de la Piel, Ecuador., Quito, Ecuador; Andrea C. Cueva, Attending Physician/Researcher, Centro de la Piel, Quito, Ecuador; Julia E. Cabezas, Dermatology Resident, Centro de la Piel, Quito, Ecuador; Santiago A. Palacios, Dermatologist, Centro de la Piel. Universidad Cato´lica. Universidad Central, Quito, Ecuador; Marcos Serrano, Director-Department of Biostatistics, Hospital Carlos Andrade Marin, Centro de la Piel-Quito, Quito, Ecuador AIMS: To determine the correlation of the Scoring Atopic Dermatitis scale (SCORAD) and the Center for Epidemiological Studies Depression Scale for Children (CES-DC) with the Children’s Dermatology Life Quality Index (CDLQI). METHODS: The methodology for this cross-sectional study followed STROBE guidelines. The study was carried out at a single dermatology center (CEPI) in Ecuador from October 2013 to January 2014. Patients with a diagnosis of atopic dermatitis (AD) according to Williams’s criteria were invited to participate. After informed consent, two trained physicians performed a standardized clinical history, physical examination and SCORAD; and they guided patients or their guardians in the completion of the CDLQI and CES-DC questionnaires. Patients had a minimum of two follow-up visits. For consistency, one investigator extracted and analyzed data using SPSS v18.0 using Spearman’s Rank Correlation and Multivariate Linear Regression. RESULTS: 196 patients ages 5–18 (mean ± SD, 8.52 ± 3.83) were included.

71 The mean SCORAD was 25.90 ± 15.08; the score band distribution was mild AD (n = 50, 25.5 %), moderate AD (112, 57.1 %) and severe AD (34, 17.3 %). The mean CDLQI was 5.92 ± 4.86. The score band distribution showed no effect on quality of life (QoL) (n = 32, 16.3 %), small effect (95, 48.5 %), moderate effect (46, 23.5 %), severe effect (20, 10.2 %), and very severe effect (3, 1.5 %). The mean CES-DC was 7.56 ± 8.78; 31 subjects (16 %) had a depressed mood and 165 (84 %) had a non-depressed mood CDLQI correlated with SCORAD (r = 0.66, p = 0.0001) and with CES-DC (r = 0.43). The SCORAD also correlated with CES-DC (r = 0.27). Multiple regression analysis demonstrated an increase in the CDLQI score of 0.21 for each unit change in SCORAD, and an increase in the CDLQI score of 0.14 for every unit change in CES-DC (p \ 0.0001). The coefficient of determination was 0.742. CONCLUSIONS: Both the severity of AD and depressed mood have a strong impact in the QoL of affected children. These results correlate with previous work that demonstrate how illness severity modifies QoL; they also reveal the importance of a holistic assessment with a focus in the psychological realm of AD and emphasize the importance of identifying and treating depression in children affected by severe AD.

(304.3) Quality of life in childhood epilepsy: a KIDSCREEN comparison with typical children and children with cerebral palsy Meron Mezgebe, BHSc, Mercer University, Atlanta, GA, United States; Gileh-Gol Akhtar-Danesh, BHSc, McMaster University, Hamilton, ON, Canada; David Streiner, PhD, McMaster University, Hamilton, ON, Canada; Nora Fayed, PhD, McMaster University, Hamilton, ON, Canada; Peter Rosenbaum, MD, McMaster University, Hamilton, ON, Canada, Gabriel Ronen, MD, MSc, McMaster University, Hamilton, ON, Canada AIMS: Little is known about how the quality of life (QOL) of children with epilepsy compares to that of children with other neurological impairments or typically developing children. The objective of this study was to compare the QOL of children with epilepsy to children with cerebral palsy (CP) and children from the general population. We measured self- and proxy-reported QOL of children with epilepsy, and contrasted that with data from children with CP and from the general population from the SPARCLE group and European KIDSCREEN project, respectively. METHODS: Participants: Children ages 8–12 years with epilepsy were recruited from 6 sites across Canada (N = 345). Same-aged children with CP (N = 489) came from 6 European countries, while typical children aged 8–11 (N = 5950) came from 13 European countries. All children completed the self- and proxy-reported KIDSCREEN-52 questionnaire. This transcultural measure of QOL was developed based on typical children’s qualitative methodological evaluations. RESULTS: There were no clinically important differences (i.e., [0.5 standard deviation) between self-reported quality of life in children with epilepsy compared to children with CP or children in the general population. In contrast, proxy-reported outcomes revealed clinically important differences among the three groups in five of ten KIDSCREEN-52 domains. Compared to the children with CP and the typical children, parents of children with epilepsy reported better QOL in physical wellbeing, autonomy and social support, but poorer QOL in the domains of bullying and mood and emotions (all at p \ 0.001). Among the epilepsy group, children scored better ([0.5 SD) than their parents in the domains of psychological wellbeing and social support. CONCLUSIONS: Families should find comfort in the results of this study as they indicate that children with epilepsy do not perceive any important differences in QOL in comparison to their typical peers as measured by the KIDSCREEN-52.

123

72 (304.4) Administration of the first German-speaking computer adaptive test for assessing health-related quality of life in children and adolescents in a clinical sample: study design, methods and first validation results of the Kids-CAT study Dana Barthel, Dipl.-Psych., University Medical Center HamburgEppendorf, Hamburg, Germany; Christiane Otto, PhD, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Sandra Nolte, PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Kathrin I. Fischer, M.Sc.med., Charite Universitaetsmedizin Berlin, Berlin, Germany; Ute Thyen, MD, University Medical Center Schleswig–Holstein, Lu¨beck, Germany; Marcus Klein, MD, University Medical Center Schleswig– Holstein, Kiel, Germany; Otto Walter, PhD, MD, Charite Universitaetsmedizin Berlin, Berlin, Germany; Felix Fischer, ScD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Silke Schmidt, Prof., Ernst-Moritz-Arndt University Greifswald, Greifswald, Germany; Holger Muehlan, PhD, Ernst-Moritz-Arndt University, Greifswald, Germany; Matthias Rose, MD PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Ulrike Ravens-Sieberer, PhD, MPH, University Medical Center HamburgEppendorf, Hamburg, Germany AIMS: Recently, the Kids-CAT study group developed the first German-speaking computer adaptive test (CAT) for the assessment of health-related quality of life (HRQoL) in children and adolescents. The Kids-CAT assesses generic HRQoL by five item banks (‘‘physical well-being’’, ‘‘psychological well-being’’, ‘‘parent relations’’, ‘‘social support & peers’’, and ‘‘school well-being’’) and chronic generic HRQoL by one additional item bank. This talk aims at presenting the study design of the Kids-CAT study and at evaluating the validity of this newly developed tool under real application conditions. Furthermore, time needed for completion was investigated stratified by age of patients. METHODS: The KidsCAT study has a longitudinal prospective study design with eight measurement points over the course of 1 year. N = 306 consecutive patients aged 7–17 with asthma, diabetes or rheumatoid arthritis filled out the Kids-CAT at two University Medical Centers in northern Germany. The Kids-CAT was completed by the patients at respective clinic or at home. In order to validate the Kids-CAT, several legacy instruments assessing HRQoL (e.g. Pediatric Quality of Life InventoryTM, Kidscreen-27, KINDL-R) and associated constructs (e.g. social support, self-efficacy, life satisfaction) were administered via an electronic questionnaire. RESULTS: We found moderate to strong correlations of the Kids-CAT dimensions with scales from well-established generic HRQoL instruments (e.g. Pediatric Quality of Life InventoryTM, Kidscreen-27, KINDL). Furthermore, both correlations between the Kids-CAT chronicgeneric dimension with the DISABKIDS and correlations with instruments assessing associated constructs (e.g. ‘‘Family Assessment Device’’ for the dimension family well-being, ‘‘Social Support Scale’’ for the dimension social support and peers) were moderate as well. Children (aged 7–11 years) had a mean completion duration of 10:49 min (SD = 3:48; range: 3:46–25:27 min), while adolescents (aged 12–17 years) needed 6:02 min on average (SD = 2:11; range 3:06–15:26 min) for the Kids-CAT at the first measurement point. Children needed significantly more time for completing the KidsCAT than adolescents. CONCLUSIONS: The Kids-CAT is a feasible and valid tool that assesses HRQoL highly efficiently and less burdensome for children and adolescents compared to traditional paper–pencil instruments.

123

Qual Life Res (2015) 24:1–191 305: Aging Worldwide (305.1) General health perception indicates service needs posthospitalization among vulnerable seniors Sabrina Figueiredo, PhD Candidate, McGill University, Montreal, Canada; Alicia Rosenzveig, McGill University, Montreal, Canada; Jose Morais, MD, PhD, McGill University, McGill University Health Centre, Montreal, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada AIMS: Across Canada, people over 75 years of age, represent 16 % of all hospital admissions. At the McGill University Health Center (MUHC), this translates to over 6400 seniors annually. Compared with rates for younger adults, the rate of hospital discharges for seniors is higher for all care types examined. For inpatients, the discharge rate per 100,000 seniors is 4.9 times higher for acute care and 22 times higher for complex continuing care. For those with newly arising disabilities such as from stroke or myocardial infarction, post-discharge services are in place. However, for the vast majority of discharged seniors, no provisions are made to manage disabilities that are well known to lead to social disengagement and the onset of further functional decline. This study aimed to estimate the extent to which self-reported health can be used as an indicator of service needs among vulnerable seniors recently discharged from a large teaching Hospital (MUHC). METHODS: 940 seniors met contact criteria, 543 were contacted, and 148 seniors agreed to answer the survey, which included questions related to symptoms, physical and social activities, feelings and emotions, and health status. Those not participating included a group who were well and a group who were too ill to be interviewed. Age and sex adjusted logistic regression was used to estimate the link between functional status indicators and fair or poor self-reported health. Results are reported as odds ratio (OR) and 95 % confidence intervals (CI). RESULTS: Among the 103 respondents (mean age 79 ± 7; 60 % women), 40 % rated their health as fair or poor. As shown in the accompanying table, there was a strong relationship between fair or poor health and even moderate problems with pain, fatigue, mobility, climbing stairs, household tasks, and social participation. CONCLUSIONS: Fair or poor selfrated health, asked as a single item, indicates persistent impairments as well as key functional limitations. In a vulnerable population such as recently discharged seniors, this simple screening question could be used to identify a high degree of need for further follow-ups and services to promote recovery.

(305.2) Quality of life among senior citizens in Bhutan: associations with adverse childhood experiences and recent stressful life events Nidup Dorji, PhD Candidate, Khesar Gyalpo University of Medical Sciences of Bhutan, Thimphu, Bhutan, Brisbane, Queensland, Australia; Michael Dunne, Professor, Queensland University of Technology, Brisbane, Queensland, Australia; Charrlotte Seib, Lecturer, Queensland University of Technology, Brisbane, Queensland, Australia; Sibnath Deb, Professor, Pondicherry University, Puducherry, Tamil Nadu, India AIMS: To explore links between adverse life experiences, healthrelated quality of life among people aged 60 years and over in Bhutan. METHODS: In-depth one-to-one interviews were completed in homes and community settings. A total of 337 people aged

Qual Life Res (2015) 24:1–191 60–101 years were recruited from stupas, temples and local markets in four geographic regions. Measures included the WHO Adverse Childhood Experiences (ACE) International Questionnaire (with additional questions for Bhutanese context), checklist of recent stressful life events (SLE) and chronic health conditions, the WHOQOL-BREF, WHO Wellbeing Index, the K-10 Psychological Distress scale, and the scale measuring social connectedness. RESULTS: Males reported fewer physical and mental health problems, and better quality of life (QOL) than females. Most common childhood adversities were: Forced to do compulsory labour, required to take on adult roles in the household, and extended periods with insufficient basic needs (food, water, clothing or shelter). During adult years, deaths of close family members and friends, loss of property, and change in their relationship with their children were common SLEs. Univariate regression indicated current physical and mental health problems explained 29.6 and 15.5 % variance in QOL. ACEs had significant though modest associations with health, explaining 7.3 and 2.8 % of variance in physical and psychological problems respectively. Distal and proximal (past year) SLEs explained 17.7, and 10.9 % of variance in physical health and 2.2 and 6.9 % of variance in psychological distress. Overall, they reported high levels of social connectedness with their community. CONCLUSIONS: Elderly people in Bhutan have had a complex variety of stressful life events that may impact on QOL. Prominent adversities were loss of family and friends, though most reported close social integration. During childhood, many had lived in conditions of severe deprivation, and this was associated with health problems more than four decades later.

(305.3) Psychological risk factors and resources for early frailty detection in older adults Simone Freitag, Ernst-Moritz-Arndt University Greifswald, Greifswald, Germany; Silke Schmidt, Prof., Ernst-Moritz-Arndt University Greifswald, Greifswald, Germany AIMS: Frailty is a geriatric syndrome which describes physical deterioration and adverse outcomes. Psychological factors and quality of life are contributing to frailty, but results are still heterogonous and need further exploration. This study aims to explore psychological factors which predict frailty status in older adults. METHODS: A sample of 210 people mean age M = 75.3 years was recruited in Germany. Measurements comprised the Tilburg Frailty Indicator (TFI) as frailty measure and psychological variables: resilience (RS11), social support (FSozU), self-efficacy, quality of life (Eurohis-8), depression (PHQ-9), trauma (PCL-C) and anxiety (GAI). Logistic regression analyses were performed. RESULTS: Frailty status was significantly predicted by depression, resilience and quality of life in the total sample (R-squared = 53.8). Depression was associated to an increased risk for frailty, whereas resilience and quality of life are significantly associated with lower risk for frailty. Gender differences were evident. For women depression was significantly associated with an increased risk and quality of life with a decreased risk for frailty (R-squared = 54.3). For men frailty was significantly predicted by higher trauma scores and low levels of resilience and quality of life (R-squared = 61.9). CONCLUSIONS: Psychological contributors of frailty were investigated. In general, health-related quality of life, resilience and depression showed to be predictors for frailty. One risk factor is depression which confirms common results. Gender differences were evident and showed that men are more at risk through accumulation of traumatic life events, whereas for women depression is the main risk factor for frailty. Quality of life was the main protective factor for men and women. Additionally men were at lower risk for frailty when resilience was high. According to the findings, it

73 is suggested that health care and frailty prevention is gender specific in older adults. For women early detection of depression for men screening for accumulated traumatization is important. In general, protective factors quality of life and resilience should be promoted in men and women to delay early onset of frailty.

(305.4) Health-related quality of life during progression of Alzheimer’s disease based on latent growth curve model Hongmei Yu, Shanxi Medical University, Taiyuan, China; Yanbo Zhang, Shanxi Medical University, Taiyuan, China AIMS: To explore the dynamic changes of health-related quality of life (HRQOL) during progression of Alzheimer’s disease (AD) by making quality of life combined with disease status. METHODS: Cognitive normally aging, mild cognitive impairment (MCI), moderate or severe cognitive impairment and AD were defined as state 1, 2, 3 and 4, respectively. HRQOL was assessed by Quality of LifeAlzheimer’s Disease (QOL-AD) Chinese version. Based on collected three wave’s longitudinal data (every half year), latent growth curve model (LGCM) was used to focus on the dynamic changes of HRQOL with the state transitions during AD progression. RESULTS: LGCM analysis showed that the average QOL were 29.9 at baseline, 30.8 half year later, 29.1 1 year later and there were significant differences between individuals (t = 4.917, P \ 0.001). During followup periods, there were inverted-V shapes about QOL, but change speeds did not have significant differences between individuals (t = 1.012, P = 0.312). Besides, the initial level of QOL was not significantly correlated with the following change speed (t = -1.425, P = 0.154). For each transition group, groups of state 2 to state 3 (t = 4.080, P \ 0.05) and state 3 to state 4 (t = 2.450, P \ 0.05) had significant inverted-V trend about QOL. Compared with the latter, the former had higher increased pace and had significant differences between individuals. However, groups of state 1 to state 2 (t = 1.726, P [ 0.05) and state 3 to state 2 (t = 1.854, P [ 0.05) did not reach significant levels. CONCLUSIONS: HRQOL in older adults with cognitive impairment may not change much during 1–2 year’s observation period. The initial levels of HRQOL and trends of development are different among different state transition groups (Tables 1, 2). Table 1 Parameter estimates based on latent growth curve model Parameter

Estimate SE

t

P

Standard estimate

Mean Intercept Slope

29.865* 0.392 0.969* 0.455

76.123 \0.001 2.129

7.917

0.033

0.490

4.917 \0.001

1.000

1.012

1.000

Variance Intercept Slope

14.230* 2.894 3.907

3.862

0.312

Covariance Intercept and slope -3.786 Factor loading QOL1

0.000

QOL2

1.000

QOL3

-0.819

2.656 -1.425

0.154 -0.508

0.000

–

–

0.000

0.000

–

–

0.365

0.789 -1.038

0.299 -0.248

* P \ 0.05

123

74

Qual Life Res (2015) 24:1–191

Table 2

Parameter estimates for state transitions based on latent growth curve model

Parameter

Cognitive normally aging ? MCI

MCI ? moderate or severe cognitive

Moderate or severe cognitiveimpairment impairment ? AD

Moderate or severe cognitive impairment ? MCI

Estimate SE

Estimate SE

Estimate SE

Estimate SE

t

Standard estimate

t

Standard estimate

t

Standard estimate

t

Standard estimate

Mean Intercept

31.447

0.494 63.669*

9.027*

30.859

0.245 125.883* 9.104*

28.547

0.291 98.016*

7.954*

28.547

0.291 98.016*

8.194*

Slope

0.418

0.242 1.726

0.297

0.992

0.243 4.080*

0.686

0.584

0.238 2.450*

0.320*

0.485

0.262 1.854

0.294

Intercept

12.137

1.761 6.891*

1.000

11.489

1.286 8.932*

1.000

12.881

2.265 5.687*

1.000

12.137

1.761 6.891*

1.000

Slope

1.981

1.259 1.574

1.000

2.091

0.942 2.219*

1.000

3.336

1.441 2.315*

1.000

2.729

1.565 1.744

1.000

0.048

0.748 0.064

0.010

-1.785

0.693 -2.575*

-3.64*

-1.284

1.029 -1.248

-0.196

0.048

0.748 0.064

0.008

QOL1

0.000

0.000 –

0.000

0.000

0.000 –

0.000

0.000

0.000 –

0.000

0.000

0.000 –

0.000

QOL2

1.000

0.000 –

1.407*

1.000

0.000 –

1.446*

1.000

0.000 –

1.826*

1.000

0.000 –

1.652*

QOL3

-1.437

0.447 -3.213* -2.022*

-1.437

0.447 -3.213*

-2.077*

-1.437

0.447 -3.213* -2.624*

-1.437

0.447 -3.213* -2.373*

Variance

Covariance Intercept and slope Factor loading

Estimates of QOL for three waves

V1 = 31.447

V1 = 30.859

V1 = 28.547

V1 = 28.547

V2 = 31.447 + 1 * 0.418 = 31.865

V2 = 30.859 + 1 * 0.992 = 31.851

V2 = 28.547 + 1 * 0.584 = 29.131

V2 = 28.547 + 1 * 0.485 = 29.032

V3 = 31.447 + (-1.437) * 0.418 = 30.846 V3 = 30.859 + (-1.437) * 0.992 = 29.433 V3 = 28.547 + (-1.437) * 0.584 = 27.708 V3 = 28.547 + (-1.437) * 0.485 = 27.851

(305.5) How can we interpret proxy reports of HRQL when it is no longer possible to obtain a self-report from people with dementia? Sarah C. Smith, PhD, London School of Hygiene & Tropical Medicine, London, United Kingdom; A. A. Jolijn Hendriks, PhD, London School of Hygiene & Tropical Medicine, London, United Kingdom; Theopisti Chrysanthaki, PhD, London School of Hygiene & Tropical Medicine, London, United Kingdom; Stefan Cano, PhD, Modus Outcomes, Stotfold, United Kingdom; Nick Black, PhD, London School of Hygiene and Tropical Medicine, London, United Kingdom AIMS: Rising prevalence of dementia worldwide has created a growing need to robustly evaluate treatments and care for people with dementia, in both research and routine contexts. Due to the deteriorating nature of dementia self-reports are not always possible and it is sometimes necessary to rely on a proxy-report; yet to appropriately interpret proxy reports they need to be on the same metric as the selfreports. We present an equating analysis of DEMQOL (self-reported) and DEMQOL-Proxy (proxy-reported) using Rasch Measurement Theory (RMT) to determine if they can be placed on a common metric and, if so, to determine how DEMQOL-Proxy scores differ from DEMQOL scores. METHODS: We combined item-level data from DEMQOL and DEMQOL-Proxy (23 and 25 items, respectively; N = 498) and examined the extent to which the two sets of items were appropriate for equating (i.e., targeting, item threshold ordering, fit residuals, dependency, and differential item functioning). We then examined person estimates generated from the equated scales (2-way ANOVA) in mild, moderate and severe dementia, determined by MMSE. RESULTS: The findings from the RMT analysis supported the equating of the 48 items (e.g., no disordered thresholds or misfitting items, 1 pair of dependent items and 1 item with DIF (i.e., age). Comparisons of both scales on this common metric revealed the following patient-reported mean scores for DEMQOL: mild = 2.06; moderate = 2.07; severe 2.21 and the following proxy-reported mean scores for DEMQOL-Proxy: mild = 1.39; moderate = 1.18; severe = 1.61. There was a significant difference for reporter (i.e. DEMQOL or DEMQOL-Proxy; Wald Chi square = 31.8; p \ 0.01), but we found no main effect for severity and no interaction between reporter and severity. CONCLUSIONS: We recommend that the

123

DEMQOL and DEMQOL-Proxy continue to be used together where possible, and this equating study provides a potential common metric for both scale scores to be appropriately compared. Our preliminary analysis also suggests that, like many other health conditions, proxies tend to report poorer HRQL than patients. But notably this appears not to differ with severity, thus providing an initial basis for interpreting proxy-reports when self-reports are no longer possible. Funding: This research was commissioned and funded by the Department of Health Policy Research Programme (Using Patient Reported Outcome Measures to Assess Quality of Life in Dementia, 0700071). The views expressed in this publication are those of the authors and not necessarily those of the Department of Health.

(305.6) Exploring the impact of the built environment on older adults’ quality of life Lidia Engel, MSc, Simon Fraser University, Vancouver, BC, Canada; Maureen C. Ashe, PhD, The University of British Columbia, Vancouver, BC, Canada; Anna Chudyk, MSc, The University of British Columbia, Vancouver, BC, Canada; Heather A. McKay, PhD, The University of British Columbia, Vancouver, BC, Canada; David G. Whitehurst, PhD, Simon Fraser University, Vancouver, BC, Canada; Stirling Bryan, PhD, University of British Columbia, Vancouver, BC, Canada AIMS: The built environment is increasingly recognized as being associated with older adults’ quality of life (QoL). However, this relationship has remained unexplored in the area of Metro Vancouver. The aim of this study is to investigate the association between aspects of the built environment (perceived and objective) as well as the social environment and QoL in community-dwelling older adults from Metro Vancouver. METHODS: Cross-sectional data acquired from the Walk-the-Talk study were used. Health-related QoL (HRQoL) and capability wellbeing were assessed using the EQ-5D5L and the ICECAP-O, respectively. Measures of the environment comprised the NEWS-A (perceived built environment measure), the Street Smart Walk Score (objective built environment measure), and the SC-5PT (social environment measure). Statistical analyses

Qual Life Res (2015) 24:1–191 included correlation analyses and Tobit regression analyses (stepwise backward Tobit regression analysis as well as Tobit regression models for each built environment feature separately). RESULTS: We included 160 individuals (male n = 58, female n = 102) aged C65 years. The mean (SD) scores for the EQ-5D-5L and the ICECAP-O were 0.794 (0.154) and 0.843 (0.112), respectively. Correlation analyses showed numerous weak associations (r \ 0.3) between NEWS-A subscales and the EQ-5D-5L, while the ICECAP-O was weakly correlated with the SC-5PT (r = 0.293). Both sets of regression analyses indicated that after controlling for covariates, street connectivity and the social environment surfaced as potentially important predictors for the ICECAP-O, i.e., better social cohesion and poor street connectivity were associated with higher ICECAP-O index scores. No similar associations were found for the EQ-5D-5L. CONCLUSIONS: Our exploratory analyses demonstrated that features of the BE may impact older adults’ QoL and should be considered in policy decisions. However, results should be interpreted with caution due to study limitations. More longitudinal research is needed to investigate this association further.

Outstanding Poster Award Finalist Abstracts (1001) Health-related quality of life in CLARINET, a phase III trial of lanreotide autogel 120 mg in patients with non-functioning entero-pancreatic neuroendocrine tumour: analytical challenges and statistical solutions Antoine Regnault, Mapi, Lyon, France; Loic Ferrer, INSERM, Universite´ Bordeaux, Bordeaux, France; Je´rome Dinet, IPSEN Pharma, Boulogne, France; Sylvie Gabriel, IPSEN Pharma, Boulogne, France; Marianne E. Pavel, Universita¨tsmedizin Berlin, Berlin, Germany; Philippe B. Ruszniewski, Beaujon Hospital, Clichy, France; Martyn E. Caplin, Royal Free Hospital, London, United Kingdom AIMS: A clinical trial was conducted to investigate the anti-tumoral effect and safety of Lanreotide in patients with non-functioning entero-pancreatic neuroendocrine tumour, in which HRQL was a secondary endpoint. These data raised analytical challenges, such as censoring due to lack of post-progression data or non-normal distribution of scores. Our objective was to explore how sophisticated statistical methods, such as joint models, would allow improving HRQL data analysis in this context. METHODS: 204 patients were included in a randomised placebo-controlled study of deep subcutaneous injections of Lanreotide Autogel 120 mg administered every 28 days, with progression-free survival as primary efficacy endpoint. HRQL was measured by the EORTC QLQ-C30 and its gastrointestinal neuroendocrine tumour specific module QLQ-GI.NET21. Four HRQL scores of main clinical interest were selected for analysis: Overall quality of life (QL); Cognitive Functioning (CF); Sleep/Insomnia (SL); and Disease-Related Worries (DRW). Joint modelling with shared random effects or with latent class was applied to the QL and DRW scores transformed to a Gaussian distribution. Joint modelling could not be used for CF and SL because they could not be transformed to a Gaussian distribution. Cumulative probit mixed models were performed on the longitudinal data for these scores to reflect their ordinal properties. RESULTS: Joint modelling showed that progression was not associated with dynamics of the QL and DRW scores. The time-to-event submodel of the joint models showed a significantly lower risk of progression in Lanreotide arm (HR = 0.44, 95 % CI [0.28,0.68]). The longitudinal analyses of the four scores (i.e. longitudinal submodels of the joint models and cumulative probit mixed models) showed non-significant differences between Lanreotide and placebo in the change in scores over the course of the trial. CONCLUSIONS: The sophisticated statistical

75 methods confirmed that HRQL was maintained for patients on Lanreotide Autogel 120 mg as compared to placebo, reinforcing the credibility of the initial descriptive results (no deleterious effect of the drug). Our approach showed that the analysis of HRQL data collected in clinical trials should be carefully designed for all the potential issues to be taken into account and that a variety of statistical solutions exists to address the challenges associated with these data. (1002) Quality-adjusted life years (QALY) loss throughout remainder of expected life due to selected chronic conditions for U.S. adults aged 65 years and older Haomiao Jia, Columbia University, New York, NY, United States AIMS: Quality-adjusted life years (QALY) is a health outcome measurement that combines both years of life lived and health-related quality of life (HRQOL). This study estimates QALY loss throughout remainder of expected life due to each of the following chronic conditions—depression, diabetes mellitus, hypertension, heart diseases, stroke, emphysema, asthma, arthritis, and cancer. METHODS: We ascertained respondents’ HRQOL scores and mortality status from the 2004–2010 National Health and Nutrition Examination Survey (NHANES) with mortality follow-up data through December 31, 2011. We included respondents aged 65 years and older (n = 3680). QALY was estimated using a hybrid estimator that calculated QALY from two parts: QALY during the follow-up period and QALY beyond the follow-up period. The first part was estimated from a Kaplan–Meier method based estimator. The second part was estimated by extrapolating survival time beyond December, 31, 2011 using parametric survival models with Weibull distributions. We calculated QALY by each of the nine selected chronic conditions. RESULTS: For those aged 65 and older, QALY throughout reminder of expected life was 12.3 years. After accounting for age and sexrelated differences, depression contributed 8.2 years of QALY loss; diabetes, 5.6 years; hypertension, 2.5 years; heart diseases, 5.4 years; stroke, 6.4 years; emphysema, 8.0 years; asthma, 4.8 years; arthritis, 0.3 years; and cancer, 2.5 years. Compared to persons without any chronic conditions, persons with one condition were associated with 4.7 years of QALY loss; persons with two conditions were associated with 7.9 years of QALY loss; and those with three or more conditions were associated with 10.8 years of QALY loss. CONCLUSIONS: This study presents a hybrid QALY estimator for respondents from a follow-up study. The analyses show good precision and relatively small bias in estimating QALY even with relatively small sample sizes. In this study, depression contributed the biggest QALY loss for those aged 65 years and older, followed by emphysema, stroke, diabetes, heart diseases, asthma, cancer, and hypertension. Arthritis had a small and non-statistically significant impact on respondents’ QALY. These results indicate that mental health in the elderly warrants more attention.

(1003) Use of PROs in clinical guidance for the management of chronic conditions in primary care in the United Kingdom Ian Porter, Dr, University of Exeter Medical School, Exeter, Devon, United Kingdom; Jose M. Valderas, MD, PhD, MD, University of Exeter, Exeter, United Kingdom AIMS: The use of patient Reported Outcomes (PROs) in clinical practice represents a promising potential application of PROs for improving health systems and delivering patient centred care. Current clinical practice in the UK is typically informed by evidence based standards, such as the National Institute for Health and Care Excellence (NICE) and quality of care clinical indicators in incentive schemes. This project investigates what clinical guidance is available linking PROs to the management of selected chronic conditions in

123

76 Primary Care. METHODS: A systematic review was undertaken of available clinical guidance for the management of 6 chronic conditions (asthma, COPD, diabetes, heart failure, depression, and osteoarthritis of the hip and/or knee) in relation to use of PRO measures in UK Primary Care. NICE, Quality and Outcomes Framework (QOF) and other UK clinical practice guidelines (where available) were identified for each condition, and we extracted information on: (a) whether PROs (conceptualized according to an established conceptual model) were explicitly mentioned; (b) if favoured measures were specified, and, if so, (c) whether guidance explicitly linked scores to specific management options. RESULTS: PROs were explicitly mentioned in guidance for all conditions, with reference to symptoms, functional status, and/or quality of life. Generic advice on the management of conditions as informed by PROs (but not linked to specific scores) was provided for all conditions except for diabetes. The clinical guidance reviewed recommended the use of specific PROs for: asthma (Royal College of Practitioners 3 Asthma questions), COPD (Clinical COPD Questionnaire), and depression (Patient Health Questionnaire-9; Hospital Anxiety and Depression Scale; Beck Depression Inventory-II) but not for diabetes, osteoarthritis, or heart failure. Specific management advice tailored to PRO scores was identified only for COPD in relation to pulmonary rehabilitation. CONCLUSIONS: Although clinical guidance reference PROs in the management of conditions, there is limited information on how to measure PROs and act upon scores. This represents a significant barrier to PROs becoming a routine part of Primary Care in the UK, which needs to be addressed if the explicit policy aim of increased use of these measures is to become a reality.

(1004) How can Rasch measurement methods inform a measurement paradigm shift in mental health care? Skye P. Barbic, Post-Doctoral Fellow, Faculty of Medicine, Vancouver, BC, Canada; Sean A. Kidd, Centre for Addiction and Mental Health, Toronto, ON, Canada; William G. MacEwan, St. Paul’s Hospital, Vancouver, Canada; William G. Honer, Univerisity of British Columbia, Vancouver, BC, Canada; Kwame McKenzie, Centre for Addiction and Mental Health, Toronto, ON, Canada AIMS: Across the globe, there is an increasing commitment to ‘‘Recovery’’ as the expectation for people with mental illness. Despite ‘‘Recovery’’ being labeled as the guiding vision for assessment and service provision, there is little agreement about how to define or measure it. The aim of this study was to develop a conceptual and measurement model for personal recovery of people with mental illness. METHODS: We performed this study in three phases. In Phase 1, we asked individuals with serious mental illness (SMI) to complete 4 recovery rating scales. We used Rasch Measurement (RMT) methods to understand how well the items in each scale captured the full range (±4 logits) of the concept of interest. In Phase 2, we asked a focus group of people with SMI to review the item hierarchy of each rating scale and propose a set of items that could cover the full range of recovery. In Phase 3, a second sample of adults with SMI completed the item pool and we used RMT methods to test the extent to which these items collectively met the criteria for robust measurement of recovery and reflected the full clinical story hypothesized in Phase 2. RESULTS: In Phase 1, 228 individuals (mean age 46 ± 12 years) completed the rating scales. None of the scales met the criteria for robust measurement outlined by RMT. In Phase 2, 40 items were generated. In Phase 3, a total of 110 individuals (mean age 40 ± 13 years) completed the items. Analysis of the new item set showed overall fit to the Rasch model (v2 = 76.3, df = 62, p = .11), high reliability (rp = 0.94), ordered response scale structure, and no item bias (gender, age). Items covered a range of 7.8 logits with no gaps along the theoretical continuum. CONCLUSIONS: This study provides preliminary evidence for a set of

123

Qual Life Res (2015) 24:1–191 items that can be used to measure recovery for people with SMI. The conceptual and measurement model underpinning this set of items will be essential to establish the clinical meaning of this scale’s scores and informing an evidence-based approach to mental health practice.

(1005) Advancing patient-reported outcome measures (PROMs) in Canada Ellis Chow, MBA, Canadian Institute for Health Information, Toronto, ON, Canada; Maria Zaccaria Cho, Canadian Institute for Health Information, Toronto, ON, Canada; Laura L. Faye, Canadian Institute for Health Information, Toronto, ON, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; Eric K. H. Chan, Ph.D., Trinity Western University, Langley, BC, Canada; Greg Webster, Canadian Institute for Health Information, Toronto, ON, Canada AIMS: Routine use of PROMs for health services quality improvement and monitoring is in very early stages of development and consideration in Canada. The Canadian Institute for Health Information (CIHI) conducted several activities to explore opportunities for standardizing PROMs data collection and reporting across Canada. METHODS: In 2014, CIHI conducted an environmental scan of the Canadian and international PROMs landscape, which included a literature review. CIHI subsequently hosted a PROMs Forum in February 2015 to provide an opportunity for Canadian health leaders to share their perspectives on opportunities for standardizing PROMs data collection and reporting. Information from these sources were analyzed to arrive at preliminary recommendations. RESULTS: National programs for routine collection of PROMs have already been implemented in several countries. In Canada, although there are some regional-level PROMs initiatives, a standardized program for routine PROMs collection and reporting does not exist. A common approach to data collection is required for comparability of information. The SF-family of instruments (including the VR12) and EQ-5D were identified as the generic tools most suitable for use in Canada. The importance of ensuring clinician and administrative stakeholder engagement in national PROMs initiatives was emphasized. There is potential for leveraging existing projects, infrastructure and data holdings, including clinical registries. A PROMs Advisory Committee and PROMs demonstration projects are recommended for examining how PROMs can be used to inform Canadian healthcare. Hip and knee replacements and renal care were identified as promising areas where a demonstration project can showcase the value of PROMs as PROMs data can be integrated with information from existing CIHI registries. CONCLUSIONS: PROMs have potential for widespread use across the health care system, including health system management, policy-making, and decision-making. There is significant benefit to having a common approach to PROMs data collection across the country. CIHI is currently assessing options to continue supporting PROMs in Canada. Initiatives may include forming a leadership group to further discuss PROMs, collaborating with stakeholders to implement PROMs demonstration projects, and assessing the feasibility of submitting PROMs using CIHI’s existing clinical registries or acute and ambulatory care data holdings.

(1006) Disease-specific calibrations or disease-specific content: which is responsible for the enhanced sensitivity of conditionspecific measures compared to generic measures? San Keller, PhD, American Institutes for Research, Chapel Hill, NC, United States; Manshu Yang, PhD, American Institutes for Research, Chapel Hill, NC, United States AIMS: The Adult Sickle Cell Quality of Life Measurement Information System—ASCQ-Me—is a condition-specific measure for

Qual Life Res (2015) 24:1–191

77

Table 1 Differences in severity groups (F-statistics) for a conditionspecific versus generic measure: raw scores and scores based on sample-specific calibrations ASCQ-Me/PROMIS measure title Pain impact/pain interference Social impact/ satisfaction with social roles Sleep impact/sleep disturbance

Raw scores

IRT-calibrated scores

ASCQ-Me PROMIS ASCQ-Me PROMIS 8.41**

5.49*

10.78**

3.15

10.19**

5.36*

8.17**

5.63*

7.07**

0.44

7.84**

0.25

** p \ 0.01; * p \ 0.05 adults with sickle cell disease (SCD); whereas the Patient-Reported Outcome Measurement Information System (PROMIS) is a generic measure designed to apply across chronic diseases. We found that ASCQ-Me measures were more sensitive than PROMIS measures of similar health concepts, to differences in SCD severity. Because ASCQ-Me differs from PROMIS both in content and in the calibration sample, the greater sensitivity of ASCQ-Me could be due to either. METHODS: PROMIS questions were generated based on the review of legacy measures and focus group (FG) data provided by patients with a variety of chronic diseases. While the ASCQ-Me measures comprise questions which do not refer to SCD, the content was generated based on FG data from adults with SCD. Following PROMIS, ASCQ-Me measures were then constructed and scored using Item Response Theory (IRT) calibrations with Graded Response Model (GRM). While the ASCQ-Me calibration sample was 550 adults with SCD, the PROMIS calibration sample was from the general population and numerous respondents with a variety of chronic diseases recruited from clinics throughout the US. We evaluated the ability of ASCQ-Me (five-item) and PROMIS (seven or eight-item) short form scores for similar health concepts to discriminate among groups of SCD patients defined by SCD severity according to tertiles on the SCD Medical History Checklist (SCD-MHC). To evaluate whether the greater sensitivity of ASCQMe was truly due to the disease-specific content, we conducted these analyses using two types of scoring for ASCQ-Me and PROMIS: IRT scoring (with weights determined by the GRM) and raw scoring (with unit weights). RESULTS: In analysis of variance models controlling for age, sex and genotype, F-statistics associated with differences in mean scores across tertiles of SCD severity were greater for ASCQ-Me scores than PROMIS scores for similar health concepts. This difference between ASCQ-Me and PROMIS was found for both raw- and IRT-based scoring algorithms. CONCLUSIONS: These results suggest that compared to generic measures, the greater sensitivity of condition-specific measures, is due to the origin of the content rather than to the calibration sample (Table 1).

Poster Session 1001: Thursday Poster Session 1 Reviews (1007) Selecting patient reported outcome measures for patients with Abdominal Aortic Aneurysm (AAA): integrating systematic review and qualitative evidence synthesis methods Rosie Duncan, University of Sheffield, Sheffield, United Kingdom; Georgina Jones, University of Sheffield, Sheffield, United Kingdom;

Munira Essat, University of Sheffield, Sheffield, United Kingdom; Edith Poku, University of Sheffield, Sheffield, United Kingdom; Patrick Phillips, University of Sheffield, Sheffield, United Kingdom; Helen Buckley Woods, University of Sheffield, Sheffield, United Kingdom; Andrew Booth, University of Sheffield, Sheffield, United Kingdom; Simon Palfreyman, University of Sheffield, Sheffield, United Kingdom; Jonathan Michaels, University of Sheffield, Sheffield, United Kingdom AIMS: Abdominal Aortic Aneurysm (AAA) is a dilatation of the abdominal aorta. When the diameter exceeds 5.5 cm there is a significant risk of rupture, which is usually fatal. Depending on the size of the aneurysm treatment options include either undergoing surgical intervention or ongoing monitoring. The aim was to identify and evaluate existing patient reported outcome measures (PROMs) for use in patients with AAA to inform the selection of appropriate outcomes for use in vascular services. METHODS: Two systematic reviews were undertaken using standard review methods, the first to identify existing PROMs validated in the AAA population and the second, a qualitative evidence synthesis to explore patients’ experiences of AAA on health and quality of life. PROMs studies were evaluated for their psychometric properties using established assessment criteria, their methodological quality using the COSMIN checklist, and PROM domains were mapped. Identified qualitative studies were synthesised using Framework Analysis informed by the review domain map. RESULTS: Only two PROMs from two studies were identified; the SF-36 and Australian Vascular Quality of Life Index. The methodology of both the studies was rated as poor using the COSMIN checklist. The SF-36 was found to have good reliability and responsiveness, validity was not reported. The qualitative synthesis found only 3 studies that met the inclusion/exclusion criteria. These reported on a total of 23 patients who had either undergone surgery or were being treated conservatively. All patients reported that they did not feel any physical symptoms as a result of the AAA. Psychological effects were reported for those being treated conservatively. Psychological and physical outcomes were reported following surgical intervention, although the long-term patient reported outcomes were not reported qualitatively. CONCLUSIONS: There has been little research into PROMs for AAA. The qualitative evidence suggests that the current measures may not capture the outcomes important to patients with AAA. There is a need for high quality qualitative research to be undertaken to understand the outcomes of those either treated conservatively or who undergo surgical intervention. As a result, the development and validation of a new PROM for patients with an AAA may be warranted.

(1009) What do health-related quality of life questionnaires designed for adolescents and young adults (AYA) with cancer measure? A systematic review Natasha Wickert, McMaster University, Hamilton, ON, Canada; Elena Tsangaris, MSc, McMaster University, Hamilton, ON, Canada; Samantha Anthony, Hospital for Sick Children, Toronto, ON, Canada; Robert J. Klaassen, University of Ottawa, Children’s Hospital of Eastern ON, Ottawa, ON, Canada; Anne Klassen, DPhil, McMaster University, Hamilton, ON, Canada AIMS: The adolescent and young adult (AYA) period is unique in that individuals are faced with personal and developmental challenges, which are amplified by having a diagnosis of cancer. In order to include the voice of AYA in the assessment of treatment outcomes, well-defined, valid, reliable and responsive PRO instruments are needed. The aim of this study was to identify patient-reported outcome (PRO) instruments used with AYA cancer patients and survivors, and examine the development process and content. The

123

78 overall goal was to develop a preliminary conceptual framework of HRQL concerns that could be used to guide future research. METHODS: Findings from two previous systematic reviews, and a new literature search were used to identify self-report cancer-specific PRO instruments developed for AYA cancer patients and survivors. The updated search involved Medline, PsycINFO, EMBASE and CINAHL from January 2008 to December 2014. The development process was examined in relation to published guidelines for PRO instruments, and the content was analysed by multiple reviewers, with each item assigned a primary domain, subdomain and identifying concept. RESULTS: Twelve instruments were identified of which 4 were developed from patient interviews. The target population included cancer patients and/or survivors aged between 8 and 35 years. The content (418 items) was categorized to reveal 6 domains as follows: psychological (n = 207), social (n = 88), physical (n = 76), general (n = 23), sexual (n = 18) and spiritual (n = 6) health. A total of 21 sub-domains, and 51 unique health concepts were also identified. Only 3 instruments had content that spanned all 6 domains. Important differences were noted between the pediatric versus young adult-specific instruments, e.g., the pediatric tools do not measure spirituality, goal setting/future plans, and sexual and reproductive health, which are important concerns for AYA. CONCLUSIONS: Current PRO instruments used with AYA have content limitations. Using the most appropriate PRO instrument is crucial, i.e., if an instrument is used with AYA that does not ask the right questions, it may appear that an effective treatment has little or no benefit (Fig. 1). To select from the currently available AYA instruments, it advised to closely inspect the content to ensure it measures AYA concerns.

Qual Life Res (2015) 24:1–191 (1011) Systematic review of patient-reported outcome measures used to assess symptoms associated with heart failure Lauren A. Braam, MIH, Health Research Associates, Inc., Seattle, WA, United States; Donald M. Bushnell, MA, Health Research Associates, Inc., Seattle, WA, United States; Mona L. Martin, RN MPA, Health Research Associates, Inc., Seattle, WA, United States; Renee F. Pierson, MBA, Janssen Global Services, Raritan, NJ, United States AIMS: Background: Although a number of questionnaires assessing heart failure (HF) exist, their developmental history and ability to comprehensively measure symptoms and impacts for patients may vary significantly. This study sought to identify existing HF symptom measures in order to explore the breadth of concepts measured and assess their suitability relative to the US Food and Drug Administration’s (FDA) guidance regarding patient-reported outcomes (PRO) measures used to support product labeling claims (e.g., content validity (patient input), reliability, construct validity, sensitivity to change, and low burden). Aims of the Study: To identify existing selfreported measures for HF symptoms and impacts and assess their developmental history, concepts measured, measurement properties, and use in clinical studies. METHODS: Methods: A systematic review of HF measures was conducted by both (1) further describing development and measurement properties of measures identified from a previous review conducted in 2009, and (2) identifying new outcome measures developed since 2009 using PubMed, PRO Measurement Group and the Cochrane Library. The search was limited to those articles and instruments in English for which information on both their development process and psychometric properties were available. RESULTS: Results: Twelve HF-specific measures were

Table 1 Heart failure measures included in review Documented Documented Documented Prior use content reliability validity in HF validity 1 CHFQ

x

x

x

x

2 KCCQ

x

x

x

x

x

x

x

x

x

x

x

x x

x

x

x

x

x

x

9 HFFSI

x

x

x

10 MILQ x 11 QLQ-CHF x

x x

x x

x x

3 MacNew 4 MLHFQ 5 CHAT

x

6 SDHFQ 7 CDS 8 DASI

12 QLI 13 LCADL 14 HeartQoL

Fig. 1 .

123

x

x

x

x

x

x

x

x

x

1 Chronic Heart Failure Questionnaire, 2 Kansas City Cardiomyopathy Questionnaire, 3 MacNew Heart Disease Heart-Related Quality of Life, 4 Minnesota Living with Heart Failure Questionnaire, 5 Chronic Heart Failure Assessment Tool, 6 San Diego Heart Failure Questionnaire, 7 Cardiac Depression Scale, 8 Duke Activity Status Index, 9 Heart Failure Functional Status Inventory, 10 Multidimensional Index of Life Quality, 11 Quality of Life in Heart Failure Questionnaire, 12 Quality of Life Index–Cardiac Version, 13 London Chest Activity of Daily Living Scale, 14 HeartQoL, derived from MacNew and MLHFQ

Qual Life Res (2015) 24:1–191

79

included from the earlier (2009) search (CHFQ, KCCQ, MacNew, MLHFQ, CHAT, SDHFQ, CDS, DASI, HFFSI, MILQ, QLQ-CHF, QLI). As a result of the new (post 2009) literature search, 490 articles were identified from which 11 were selected for full-text review. Fulltext review identified two additional HF measures of interest (LCADL, HeartQoL). Reviewed instruments (see Table 1) varied with respect to the concepts measured, content validity, response options, anchoring, reliability, and recall period. The Kansas City Cardiomyopathy Questionnaire (KCCQ) and the Minnesota Living with Heart Failure Questionnaire (MLHFQ) are most commonly used in clinical studies. CONCLUSIONS: Conclusion: A wide array of instruments exists to assess patient-reported aspects of heart failure and its treatment. While a few have been used extensively in HF studies and many have documented reliability and validity statistics, most lack sufficient evidence of direct patient input and content validity. A well-developed PRO is prudent to measure aspects of heart failure and its treatments and impacts.

TCM-HSS, ChQOL, ChPRO and HSTCM had better quality. CONCLUSIONS: Based on available information, TCMQOLs mainly demonstrated TCM theory, but they are still on the first stage. And with the available psychometric evidence, ChQOL and TCM-HSS were recommended in the comparative and longitudinal studies, but HSTCM had better properties of reproducibility and discrimination for assessing quality of life. Further evidence from rigorous and high quality studies was warranted.

(1013) Traditional Chinese Medicine Instruments for Quality of Life: a systematic review of psychometric evidence

AIMS: Informal carers face many challenges when caring for patients with palliative care needs. Selecting valid and reliable measures to assess carer outcomes is important. This study aimed to identify outcome measures used for carers of patients with palliative care needs, and to evaluate the measures’ psychometric properties. METHODS: A systematic review was conducted in the following electronic databases: ASSIA, CINAHL, the Cochrane Library, Embase, Pubmed, PsycINFO, Social sciences citation index, and Sociological Abstracts. The studies identified were evaluated independently by three of the authors (CM, MB, MP) and any uncertainties were resolved by consensus with the other two authors (AA, BW). Data were extracted and evaluated regarding study characteristics and psychometric properties of the measures. RESULTS: The search yielded 4505 studies of which 112 met the inclusion criteria. A total of 38 measures had been used with the most commonly used generic measure being the SF-36 (n = 16, 14 %) and the most frequently used carer-specific measures the Caregiver Reaction Assessment (CRA) (n = 21, 19 %). Psychometric properties were reported in only 46 % (n = 52) of the studies, in relation to 24 measures. Where psychometric data were reported, the focus was mainly on internal consistency (n = 45, 87 %), construct validity (n = 27, 52 %), and/or reliability (n = 14, 27 %). Of the 24 measures, only four had been formally validated in informal carers in palliative care. These were the Quality of Life in Life Threatening Illness-Family Carer Version (QOLLTI-F), Family Appraisal of Caregiving Questionnaire for Palliative Care (FACQ-PC), Caregiver Burden Scale at End-of-life-care (CBS-EOLC), and Caregiver Quality of Life Index (CQLI). CONCLUSIONS: Despite a broad range of outcome measures having been used for informal carers in the palliative care context, little formal psychometric testing has been undertaken and therefore only limited information is available when research teams select an outcome measure for their study. Measures validated in this population group were not used as frequently as those that had not been validated. Due to their extensive validation, it is likely that generic measures are suitable for use in carers in palliative care, however further validation of carer-specific measures is required.

Changhe Yu, China Academy of Chinese Medical Sciences, Beijing, China; Ya-Nan Sun, PhD, Xuanwu Hospital Capital Medical University, Beijing, China; Li-Yun He, PhD, China Academy of Chinese Medical Sciences, Beijing, China; Bao-Yan Liu, China Academy of Chinese Medical Sciences, Beijing, China; Wen-Jing Bai, PhD, China Academy of Chinese Medical Sciences, Beijing, China AIMS: To evaluate the psychometric properties of Traditional Chinese Medicine Instruments for Quality of Life. METHODS: A search was conducted in CNKI, VIP, WangFang, PUBMED, EMBASE and OVID. Studies assessing the quality of life instruments with TCM theory were included. The quality of the instruments was evaluated by the psychometric properties of the outcome measures. The quality of studies investigating the instruments was assessing by the COSMIN list. RESULTS: Twenty-six articles including 20 validity studies identified 8 instruments. The studies were mainly conducted in Guangzhou, Xiamen and Xianggang Cities during 2004–2013. The sample population, ranging from 39 to 1298, came from community, university, health center, nursing home and hospital. All the included instruments had the common aim to assess people’s health condition based on the TCM theory. Other characteristics included the 30–60 items with Likert scales, easy to moderate degrees of scoring, completion time ranging from 10 to 30 min, and the completion rate more than 90 %. The current evidence showed that ChQOL scored comprehensively and best from fair to excellent; TCM-HSS, HSTCM and ChPRO were better (Table 1). The articles and studies describing the

Table 1 Quality of psychometric properties of Traditional Chinese Medicine Instruments for Quality of Life

(1015) Carer outcomes in palliative care: a systematic review of the measures and their psychometric properties Michele Peters, PhD, University of Oxford, Oxford, United Kingdom; Charlotte Michels, MSc, Radboud University, Nijmegen, Netherlands; Astrid Adams, MD, Oxford University Hospitals Trust, Oxford, United Kingdom; Bee Wee, MD PhD, Oxford University Hospitals Trust, Oxford, United Kingdom; Mary Boulton, PhD, Oxford Brookes University, Oxford, United Kingdom

(1017) Using preference-based measures to improve care: a synthesis David Feeny, PhD, Department of Economics, McMaster University, Hamilton, ON, Canada AIMS: Preference-based measures were developed to support economic evaluation, health technology assessment, and to provide

123

80 summary measures of population health. Recently these measures have been applied in clinical care setting. I summarize evidence on such applications. METHODS: A non-systematic review of the application of preference-based measures to improve the quality of care and/or support case management at the individual level was conducted. RESULTS: A number of inductive generalizations emerge. First is the identification of under recognized problems at the cohort level. In a clinic serving brain tumour patients, clinicians were surprised by the burden of pain reported. As a result the clinic began to ask about pain routinely and altered its pain management program. Second, the confirmation and more precise quantification of already recognized problems. Clinicians serving high-risk primary-care patients had been well aware that problems with mobility, emotional health, and chronic pain were common but had lacked precise quantification of the relative severity of these problems. Again changes in management followed as a result of having more information. Third, the inclusion of preference-based measures in registries has provided valuable information on health outcomes and, when multiple measures of health-related quality of life have been included, has provided valuable information for methodological studies about the relationships among measures. Fourth, in longitudinal population health surveys, baseline overall and single-attribute scores have been shown to predict mortality. Fifth, similarly, baseline utility scores predict subsequent utilization of healthcare services. Sixth, declines in utility scores have in some clinical studies been shown to be predictive of major exacerbations and even mortality. Seventh, the use of preference-based measures in longitudinal studies provides the basis for computing quality-adjusted survival. Such evidence can be used to counsel new patients about what to expect. Eighth, the inclusion of preference-based measures both in clinical studies and population health surveys allows investigators to compare their clinical cohorts to population norms, quantifying the burden of illness. CONCLUSIONS: The value-added of preference-based measures goes well beyond their contributions to cost-utility analyses.

(1019) Development of a database of published regulatory recommendations concerning the use of Clinical Outcomes Assessments (COAs) in drug development: outcomes InSite Cecile Perret, Mapi Research Trust, Lyon, France, Catherine Acquadro, Mapi Research Trust, Lyon, France AIMS: As defined by the FDA, a Clinical Outcome Assessment (COA) directly or indirectly measures how patients feel or function and can be used to determine whether or not a drug has demonstrated to provide a treatment benefit. There are four types of COAs: patientreported (PRO), observer-reported (ObsRO), clinician-reported (ClinRO), and performance outcomes (PerfO). The aim of this research is to create an online database of published regulatory recommendations concerning the use of COAs to determine treatment benefit and to inform drug approval and health technology assessment (HTA) review. METHODS: All clinical guidelines published from 1977 to 2014 by North American regulatory authorities (FDA and Health Canada), the European Union (EMA), and NICE, were retrieved on the corresponding websites and reviewed. The information was categorized as follows: guideline title, date of publication, guideline status (‘‘draft’’ or ‘‘final/adopted’’), therapeutic area and indication based on Mesh classification, guideline purpose(s), and for each COA, the concept of interest and context of use (e.g., targeted patient population, endpoint positioning and study design). RESULTS: For the EMA, we reviewed 110 scientific guidelines representing 62 therapeutic indications, of which approximately more than half recommend COA endpoints. For the FDA, we analyzed 86 guidance representing 56 therapeutic indications, of which almost

123

Qual Life Res (2015) 24:1–191 60 % recommend COA endpoints. For Health Canada, five guidance representing five therapeutic indications were retrieved. The 96 recommendations published by the NICE correspond to 75 therapeutic indications, and are currently under review. For example, we retrieved eight guidance in respiratory diseases (five EMA, one FDA, one NICE, one Health Canada). The EMA guidance for COPD recommends as co-primary endpoints three different PROs (COPD symptoms, exacerbations, and health status), and two PerfOs (lung function and physical capacity). Detailed information will be provided. CONCLUSIONS: Outcomes InSite will be a unique source of centralized information about how COAs are recommended by regulatory authorities in drug development and HTA review. It will be useful to observe the current trends in COAs recommended, and to integrate them as early as possible in the product development. The database will be available online in an independent website, with access by subscription.

(1021) Patient experience with hereditary angioedema (HAE): a review of the literature Larissa M. Stassek, MPH, Health Research Associates, Inc., Seattle, WA, United States; Donald M. Bushnell, MA, Health Research Associates, Inc., Seattle, WA, United States; Kelly P. McCarrier, PhD, Health Research Associates, Inc, Seattle, WA, United States; Dylan Supina, PhD, Shire Pharmaceuticals, Wayne, PA, United States AIMS: Hereditary angioedema (HAE) is a rare but potentially fatal condition associated with recurrent episodes of swelling in different parts of the body. The severity and unpredictable nature of these attacks can lead to major adverse impacts on patient lives. In order to gain a better understanding of the full range of impacts HAE has on patients’ lives, we conducted a comprehensive review of the literature since 1990. METHODS: A Boolean search was performed using PubMed to identify peer-reviewed articles examining the patient experience and disease burden among HAE patients. Papers were selected if they were published in 1990 or later, available in English, and explored topics including patient experience, burden, outcomes, symptoms, impacts, and quality of life. Atlas.ti (v7.5.3) was used to code and organize articles. RESULTS: Seventy-four articles were included in the final review. Among the concepts identified, approximately 17 were primarily associated with acute HAE attacks, and approximately 20 were long-term impacts. Impacts associated with acute attacks have been measured frequently, and include limitations on physical functioning and daily activities, acute treatment side effects, and immense economic burden resulting mainly from emergency healthcare utilization and missed work. We found fewer explorations of long-term HAE impacts, including inhibited social functioning, impaired mental health (including depression, anxiety, and fear), economic burden due to disease maintenance and lost productivity, and reduced overall quality of life. The literature falls short in describing many of these impacts from the patient perspective, including particularly large gaps surrounding social and emotional impacts, which makes it difficult to identify the impacts most important to patients. Much of the existing work used instruments not specific to HAE, and general ‘‘impairment’’ is usually reported without explaining the meaning to patients or importance relative to other HAE impacts. CONCLUSIONS: Since 1990, 74 publications have explored patient experience among HAE patients. However, few rigorous qualitative studies exist examining the social and emotional impacts of HAE, and none to our knowledge explored how patients rank their HAE impacts. In order to understand the full burden of disease on HAE patients, more qualitative work is needed to explore these areas.

Qual Life Res (2015) 24:1–191

81

(1023) Design, implementation and reporting strategies to reduce the instance and impact of missing patient-reported outcome data: a systematic review Rebecca Mercieca-Bebber, University of Sydney, Sydney, NSW, Australia; Michael J. Palmer, Queen’s University, Kingston, ON, Canada; Michael Brundage, MD, MSc, Kingston General Hospital, Queen’s University, Kingston, ON, Canada; Melanie J. Calvert, PhD, University of Birmingham, Birmingham, United Kingdom; Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia AIMS: Patient-reported outcomes (PROs) in clinical trials provide unique and important information from the patient perspective. However the potential for PROs to inform clinical guidelines and practice may be limited by high levels of missing data. While statisticians may attempt to account for missing data at the analysis stage, this is far from ideal. The broader trial team can assist in reducing the instance and impact of missing PRO data through improved study design and methodology. In the 1990s, this issue received significant attention in the literature, yet problems with missing PRO data have since persisted. Advances in PRO data collection methodology have further complicated missing PRO data issues. Thus, we undertook a systematic review to identify and collate strategies to minimize the problem of missing PRO data in clinical trials. METHODS: We searched Medline and CINAHL databases (inception-Mar2014) for English articles providing strategies to reduce the instance or impact of missing PRO data, or recommendations for transparent reporting of missing PRO data. Two reviewers independently screened titles and abstracts against pre-defined inclusion and exclusion criteria. Discrepancies were discussed with the research team. One investigator extracted data and a second investigator checked all extractions. RESULTS: Our database search retrieved 5996 unique articles. 163 articles were obtained in full and 22 articles are included in this review to date. A hand-search of reference lists of included articles is ongoing. Design and methodological strategies for reducing the risk of missing PRO data include: incorporating PROs completely into the protocol, involving trial staff in study design, forming a clear PRO assessment schedule and end-rules, minimizing patient burden, collecting auxiliary data to inform analysis decisions, collecting reasons for missing data using standardized forms, appointing a PRO coordinator, PRO-specific training for trial staff, gaining site commitment, allowing alternate modes of PRO administration, and sending reminders to patients and site staff. Transparent reporting strategies include reporting of baseline PRO scores, rates and reasons for missing data and methods for handling missing PRO data. CONCLUSIONS: Many experienced trialists have provided recommendations for reducing rates and risks of missing PRO data. This review collates this guidance for PRO researchers’ ease of reference.

(1025) Quality of life (Qol) in patients with early laryngeal cancer (LC) treated with transoral laser microsurgery (TLM) or radiotherapy: a systematic review Michelle Verelst, Antwerp University, wilrijk, anterwerpen, Belgium; Elien Dewaele, Antwerp University Hospital, Edeem, Belgium; Pol Specenier, Antwerp University Hospital, Edegem, Belgium AIMS: To summarize the published evidence on Qol in LC patients treated with RT or TLM. METHODS: NGC, NICE, SIGN, SUMSearch 2 and Pubmed were searched for the key words ‘‘laryngeal neoplasm’’, ‘‘Radiotherapy’’, ‘‘laser surgery’’, ‘‘microsurgery’’ and ‘‘Quality of life’’ (MeSH terms), excluding : ‘‘thyroid’’ and ‘‘esophagus’’ published between 2004 and 2014 in English. Titles and abstracts were screened for relevance applying the following inclusion criteria: objective Qol measurement or functional outcome (voice or swallowing) and [2 cohorts treated with TLM or RT. RESULTS:

Fig. 1 . 203 publications were retrieved, with 6 duplicates. 189 were excluded: 22 did not meet the inclusion criteria, 3 had questionable data, 164 were excluded based on title. Eight studies fulfilled the criteria and were inspected on full-text. Five were added through cross-referencing. 13 were finally included: 9 evaluated early LC, 1 included all LC stages. All had some form of voice outcome and/or quality of life questionnaire. 3 articles were reviews or guidelines. CEBM level of evidence for original studies: 2b: 6, 3a:1, 3b: 1, 4: 2. Qol was studied in 3. No major overall Qol differences were reported (Goor: COOP/Wonca, Laoufi: EORTC QLQ-HN35) albeit with some variations on sub-scales. Luo found a statistically significant difference in the SFWB and head and neck cancer-specific aspects of FACT-H&N. 9 studies focused on voice outcomes. 8 were restricted to early LC. Most studies favored RT: Osborn (trend in favor of RT for the physical function domain of V-RQO, no difference in overall V-RQOL score), Laoufi, Dinapoli and Kerr (better VHI), Mlynarek (better VHI and Visipitch parameters), Kujath (better VHI and PSSHN). Loughran found no difference apart from the emotional subscale of the VoiSS. Peeters reported a higher percentage of patients with an abnormal VHI following RT. Oridate included all TL stages and found no significant difference in V-RQOL or VHI-10 scores between (Chemo)RT and TLM; total laryngectomy patients scored significantly worse. CONCLUSIONS: No major overall Qol differences between RT and TLM were reported (Fig. 1). RT scored better on voice outcome scales in most studies. Abbreviations NGC: National Guideline Clearinghouse NICE: National institute for Health and Care Excellence SIGN: Scottisch Intercollegiate Guidelines Network VHI: Voice Handicap Index VPQ: Vocal Performance Questionnaire VoiSS: Voice Symptom Score FACT-HN: Functional Assessment of Cancer Therapy - Head and Neck PSS-HN: Performance Status Scale for Head and Neck Cancer Patients

123

82 V-RQOL: Voice-Related Quality of Life CEBM: Centre for evidence-based medicine Visipitch: Visipitch II computer program (Kay Elemetrics Corp., Lincoln Park, NJ). WONCA: World Organization of Family Doctors COOP: Primary Care Cooperative SFWB: scial functioning and well-being

(1031) Challenges in the conversion of quality of life data to utility values Faraz M. Ali, Clinical Research Fellow, Cardiff University, Cardiff, United Kingdom; Richard Kay, Visiting Professor in Medical Statistics, School of Pharmacy, Cardiff University, Cardiff, United Kingdom; Andrew Y. Finlay, Professor of Dermatology, Cardiff University, Cardiff, United Kingdom, Sam Salek, PhD RPh FFPM MRPSGB MCMS FESCP, School of Life & Medical Sciences, University of Hertfordshire, Hatfield and Director - Institute for Medicines Development, Cardiff, United Kingdom AIMS: Cost-utility analysis data are deduced from utility values which are generated by generic measures such as the EQ5D. These measures are able to produce the index values needed to generate QALYs. The aim of this study was to identify a suitable mapping approach to allow the accurate prediction of utility values from Health-Related Quality of Life (HRQoL) data. METHODS: A literature search was carried out using PubMed to identify studies that describe quality of life to utility value conversion techniques. RESULTS: The search identified 29 articles describing attempts at utility value conversions. The most frequently mentioned method was ‘transfer to utility’ or ‘regression analysis’. Another method, ‘‘direct revaluation’’, involves defining health states in the start measure which are then assigned weights using direct elicitation methods such as standard gamble (SG), time-trade off (TTO) and visual analogue scale (VAS). Two less common methods included response mapping and effect size translation. A range of statistical techniques to facilitate conversion were identified, including ordinary least squares (OLS) regression, Tobit regression, censored least absolute deviation (CLAD) estimates, ordered probit regression, restricted maximum likelihood (REML) and non-linear models. Linear regression using OLS estimation was the most commonly used modeling method; despite giving the highest accuracy, is only likely to produce a line of best fit when the relationship between the dependent and independent variables is linear. Mapping is more likely to be successful where there is a conceptual overlap between the two measures. CONCLUSIONS: Several challenges must be considered before a suitable modeling technique may be implemented: predictive ability, comparing the relationship between disease-specific and generic health measures, sample sizes and modelling assumptions. ‘Ordinal logistic regression’ is a potential approach as it recognises that the responses of dimensions may be ordered, as in the EQ5D. A large data set is currently being explored which would allow the internal and external validation of predicted EQ-5D responses based on the ordinal logistic regression model and utility values from a disease-specific HRQoL tool.

E-Health (1033) COPD symptom data-usability and validation of electronic, tablet-based implementations of the SGRQ-C and CAT questionnaires Paul O’Donohoe, CRF Health, London, United Kingdom; Karl McEvoy, CRF Health, London, United Kingdom; Jessica Thilaganathan, CRF Health, London, United Kingdom; Rebecca L.

123

Qual Life Res (2015) 24:1–191 Two, BA (Hons), PharmaQuest Ltd, Banbury, United Kingdom; Abby Cox, PharmaQuest Ltd., Banbury, United Kingdom; Hayley L. Simpson, BA, PharmaQuest Ltd, Banbury, United Kingdom, Fiona E. Miller, BA (Joint Hons), PharmaQuest Ltd, Banbury, United Kingdom AIMS: There are believed to be 64 million people in the world with Chronic Obstructive Pulmonary Disease (COPD). As there is no known cure for the disease, capturing accurate data about patient’s symptoms in clinical trials is vital. Patient reported outcome (PRO) measures like the St. George’s Hospital Respiratory Questionnaire for COPD Patients (SGRQ-C) and the COPD Assessment Test (CAT) can provide valuable insight into the patient experience. With the move towards capturing clinical outcome assessment (COA) data on electronic platforms there is a need to ensure these electronic versions are faithful migrations of the original, paperbased questionnaires. The aim of this study was to explore the usability of electronic versions of the SGRQ-C and CAT on a tablet computer in a representative sample of COPD patients to ensure it was intuitive and burden-free to use. The conceptual equivalence of the paper and electronic versions of the questionnaires were also investigated. METHODS: Electronic implementations of the SGRQ-C and CAT were developed for a tablet computer. Usability testing and cognitive interviewing were conducted on the two questionnaires in patients with COPD; 9 patients with the CAT and 5 with the SGRQ-C. Participants were asked to complete the tablet versions of the questionnaires and were then probed on their user experience as well as how they would interpret and respond to the electronic version compared to the original concepts of the questionnaire. RESULTS: During the cognitive debriefing, the respondents provided feedback that indicated they interpreted the electronic version in the same way as the original paper version. The usability testing of the questionnaires on a tablet computer indicated some minor issues with the layout of the questions on the SGRQ and CAT, which were addressed in the final build. CONCLUSIONS: Usability testing and cognitive interviewing found the electronic, tablet-based versions of the SGRQ-C and CAT to be user-friendly and intuitive. The conceptual equivalence of the paper and electronic versions were also demonstrated. Well migrated, electronic versions of PROs are vital to allow for the capture of the patient experience of a disease and treatment, without adding undue burden in a clinical trial.

(1035) The Daily Experience Sampling Questionnaire (DESQ): a new approach to measuring subjective well-being Christine Blome, University Medical Center Hamburg-Eppendorf, Hamburg, Germany AIMS: Subjective well-being can be measured with the Experience Sampling method: Patients rate their well-being on a mobile device several times a day. This momentary assessment prevents recall bias. A less costly, but time-consuming alternative is the Day Reconstruction Method (DRM): Respondents divide the complete preceding day into episodes and retrospectively rate each episode for well-being. We aimed to develop a measure of subjective well-being that combines advantages of Experience Sampling and DRM by using a oncedaily assessment via questionnaires (making the measure affordable) and assessing only a random sample of moments (making it short). METHODS: In the Daily Experience Sampling Questionnaire (DESQ), participants retrospectively rate their well-being at six randomly selected single moments of the preceding day (e.g., 10:15 a.m.). The DESQ was developed and pre-tested in four steps, including (a) development of response options for the well-being rating scale within the DESQ, based on a survey among 26 healthy persons; (b) interviews on the well-being rating scale with 10 healthy

Qual Life Res (2015) 24:1–191 persons (both convenience samples); (c) pre-test and revision of the DESQ in interviews with 11 patients with psoriasis; (d) completion of the DESQ by six patients for seven consecutive days, followed by telephone interviews on questionnaire feasibility. RESULTS: All participants of the 7-day-pre-test stated that they had filled in the DESQ at each day of the week and that they did not have any difficulties completing or understanding it. None of them found the daily completion burdensome or time-consuming. Only one participant suggested a minor improvement. CONCLUSIONS: The DESQ may be a time- and cost-reducing alternative to measuring patient wellbeing with Experience Sampling or DRM. We plan to determine criterion validity in future studies, using Experience Sampling as a gold standard.

(1037) Integrating quality of life assessments into acute care for older adults with chronic life-limiting illnesses Kara Schick-Makaroff, PhD, University of Alberta, Edmonton, AB, Canada; Glenda King, RN, MN, Trinity Western University, Vancouver, BC, Canada; Esther Laforest, RN, PhD student, McGill University, Montreal, QC, Canada; S. Robin Cohen, PhD, McGill University and Lady Davis Institute, Montreal, QC, Canada; Charlene Neufeld, Fraser Health, Vancouver, BC, Canada; Judy Lett, Fraser Health, Vancouver, BC, Canada; James Voth, Intogrey, Vancouver, BC, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada AIMS: Older adults who have advancing chronic and life-limiting illnesses present significant challenges for healthcare delivery in hospital-based settings. Patients’ and family caregivers’ concerns can be effectively addressed and made more visible to healthcare professionals through routine assessment of their perceived healthcare needs and their quality of life (QOL). The aim of this study was to gain input from healthcare providers on the design and use of an electronic quality of life assessment and practice support system (QPSS–built on the Turio [TM] platform developed by Intogrey Research and Development Inc.) for older adult patients and their family caregivers on acute care units. METHODS: A user-centered design was used in 4 focus groups with 6–9 acute care clinicians including palliative care physicians, registered nurses, administrators, a social worker, and a spiritual health practitioner. In follow-up to the focus groups, an additional 4 individual interviews were conducted with clinicians who were on a unit where patient-reported outcome measures were mandated to be used in their daily practice. Qualitative data analysis was undertaken using interpretive description, an established methodology used in applied clinical fields. RESULTS: Three main themes were identified. First, ‘‘ideal characteristics of QOL assessment instruments’’ included allowing both patients and families to express their experiences without undue burden, and facilitating holistic assessment of symptom management and QOL. Second, ‘‘desired features of the feedback system’’ encompassed ease of use, dissemination of information in real-time at the point of care, prevention of charting duplication, and visual representation of data over time. Third, ‘‘integration with interdisciplinary care planning’’ included referring patients to other members of the acute care team, and tracking interdisciplinary interventions. CONCLUSIONS: Clinicians’ perspectives helped determine which QOL assessments were most applicable to their practice setting, how these could be integrated into their practice, how results could be meaningfully reported, and which feedback mechanisms would be helpful in real-time. These findings have informed and facilitated the subsequent usability testing and implementation of the QPSS in routine care for older adult patients, family caregivers, and clinicians.

83 (1039) Linguistic validation of Clinical Outcomes Assessment (COA) questionnaires administered via an interactive voice response system (IVRS) Barbara Brandt, Corporate Translations Inc., East Hartford, CT, United States, Matthew Talbert, Corporate Translations Inc, Chicago, IL, United States; Mary Gawlicki, Corporate Translations, East Hartford, CT, United States; Shawn McKown, Corporate Translations Inc., East Hartford, CT, United States AIMS: The objective of this study was to explore best practices for conducting cognitive debriefing on subjects with translated Clinical Outcomes Assessments (COAs) intended for administration using interactive voice response systems (IVRS). METHODS: Subjects were debriefed on two (2) unique instruments, one intended for migraine sufferers in Spanish (US), and the other for patients with osteoarthritis of the knee in French (Canada) and Spanish (US). Interviews were conducted with five (5) subjects per language. Interviews were completed in-person, with an interviewer playing each recorded voice prompt for the subject, then instructing the subject to paraphrase the item in his/her own words. If a subject had comprehension difficulty, he/she was asked to suggest revisions to the translation to enhance comprehension. The interviewer also captured comments made by the subjects regarding other qualities of the recording, such as speech pace, enunciation and pronunciation. RESULTS: Cognitive debriefing with migraine sufferers yielded five (5) changes to the text in order to improve item comprehension and cultural appropriateness. Re-recording of prompts with slower speech and clearer enunciation was recommended for seven (7) items. Cognitive debriefing with knee osteoarthritis patients yielded three (3) changes to the text in order to improve item comprehension and cultural appropriateness. Re-recording of prompts was recommended for eight (8) items to correct pronunciation of words and numbers, improve enunciation, and slow the pace of speech. CONCLUSIONS: When conducting cognitive debriefing interviews of translated IVRS questionnaires on subjects, it is recommended to conduct debriefing in the same way that the questionnaire would be administered in actual clinical trials, with respondents listening to the audio prompts. Simulation of the actual respondent experience is shown to identify improvements to the voice recording such as decreasing the rate of speech, adding pauses, improving enunciation, and correcting pronunciation of terms.

(1041) Open Access telePRO: a website for patient-initiated tele Patient-reported Outcome (telePRO) follow-up in outpatient clinics Liv Marit Valen Schougaard, RN, MHSc, Regional Hospital West Jutland, Herning, Denmark; Caroline T. Mejdahl, RN, MNSc, Aarhus University Hospital, Aarhus, Denmark; Klaus Hvam Petersen, B. Eng., Regional Hospital West Jutland, Herning, Denmark; Niels Henrik Hjollund, MD,PhD, AmbuFlex, Herning, Denmark AIMS: In telePRO, outpatients report information on health status and symptoms from home before or instead of visiting the outpatient clinic. Typically, telePRO outpatient follow-up activity is initiated by clinicians, who prompt the patient to answer, collect the data and assess the results. The Open Access term indicates that contacts to the outpatient clinic are initiated solely by the patient and contact may be in the form of a PRO assessment at a time decided by the patient. The aim of this study was to develop a prototype website to collect telePRO in patient-initiated outpatient follow-up and link the data with the Danish National Health Website ‘Sundhed.dk’. METHODS: A research team, including experts in telePRO, patient involvement research, software technology, clinical epilepsy, and outpatients with epilepsy, provided inputs to design the prototype

123

84

Qual Life Res (2015) 24:1–191 of students transitioning to post-secondary education should be addressed. A content analysis was applied to analyze the collected data. The learning platform was required to strength students’ mental health. With joining ‘‘Life Coaching’’ concept into the LCA as the learning platform, it covered the interactive learning package in which we aimed at enhancing post-secondary students’ mental health, then helping them to going through the transition period healthier. It mainly consisted of several learning modules, i.e. (1) Identification of one’s mental health needs and their impacts on health; (2) Explanation of causes of health impacts; (3) Appropriate strategies to cope with one’s mental health needs; (4) Interpretation of one’s postcoping progress. RESULTS: Students’ positive feedback was explored. Their autonomy and freedom were respected in the use of LCA that could achieve the ownership of self-tailored learning. There was no pressure to learn more about stress management. Importantly, the continuing self-checking mechanism could help students knowing the warning signs of their stress and anxiety. CONCLUSIONS: In sum, the innovative e-Health/IT interactive and multimedia approach in the development of LCA for enhancing post-secondary students’ mental health was successful.

Fig. 1 The Open Access telePRO website. Note Lables were translated from Danish

website. The research team developed the initial website specifications, constructed the website and elicited feedback from epilepsy outpatients (n = 6), using cognitive interviewing techniques to study the manner in which the patients understood and responded to the website. RESULTS: The website was customized for patient use. The result of the website construction is shown in Fig. 1. Patients emphasized the importance of a user-friendly interface with clear and short information. Patients were interested in tracking change over time and in using the website because of potential to facilitate communication with their clinicians at a time decided by their own. They found it conceivable that access to their previously questionnaires could give them a better understanding of their chronic disease. Finally, they pointed out the need of a telephone call in the case of immediate need of contact. Patients had few problems assessing and using the site. The website is linked with the Danish National Health Website ‘Sundhed.dk’ and patients can access their PRO overview via secure login. CONCLUSIONS: Preliminary reports from patients suggest that the website could be an efficient method to collect telePRO in a patient-initiated telePRO outpatient follow-up activity. Further evaluation on the effect of quality of care and patient experiences will be documented in two parallel PhD projects with quantitative and qualitative methods.

(1043) Development and evaluation of interactive ‘‘life coaching’’ app for enhancing post-secondary students’ mental health in Hong Kong Ka Fai Wong, Assistant Professor, The Open University of Hong Kong, Hong Kong, China AIMS: In Hong Kong, post-secondary students regard their academic life as distressful and demanding. Students’ stress is not restricted to their studies but it could come from various sources such as health, financial, family and romantic relationship. The high level of stress might cause negative effect on students’ mental health, like depression. The study aims at developing and evaluating the interactive ‘‘Life Coaching’’ App (LCA) for enhancing post-secondary students’ mental health in Hong Kong. METHODS: By the documentary analysis and the interviews with students in Hong Kong, health needs

123

(1047) Successful integration of electronic patient reported outcomes to improve the quality of life of patients with prostate cancer in busy genitourinary outpatient clinics Alyssa Macedo, MScOT, University Health Network - Princess Margaret Cancer Centre, Toronto, ON, Canada; Doris Howell, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Michael Brundage, MD, MSc, Kingston General Hospital, Queen’s University, Kingston, ON, Canada; Sarah Stevens, Cancer Care ON, Toronto, Canada; Michael Lima, University Health Network - Princess Margaret Cancer Centre, Toronto, Canada; Anika Petrella, University Health Network Princess Margaret Cancer Centre, Toronto, Canada; Michael Nesbitt, University Health Network - Princess Margaret Cancer Centre, Toronto, Canada; Fred Cohen, University Health Network - Princess Margaret Cancer Centre, Toronto, Canada; Rupi Purewal, University Health Network - Princess Margaret Cancer Centre, Toronto, Canada; Iryna Tymoshyk, University Health Network - Princess Margaret Cancer Centre, Toronto, Canada; Andrew Bayley, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Robert Bristow, University Health Network Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Charles Catton, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Peter Chung, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Saibishkumar Elantholi Parameswaran, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Antonio Finelli, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Neil Fleshner, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Girish Kulkarni, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Padraig Warde, MD, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Tran Truong, University Health Network - Princess Margaret Cancer Centre, Toronto, Canada; Terri Stuart-McEwan, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada; Andrew Matthew, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada AIMS: Routine collection of patient reported outcomes (PROs) is essential to (1) guide quality assurance, (2) enhance program

Qual Life Res (2015) 24:1–191 development, and (3) inform novel research, with the overall goal to improve patient quality of life. This abstract outlines the electronic integration of the Expanded Prostate Cancer Index Composite for Clinical Practice (EPIC-CP), a validated prostate specific quality of life measure, in busy Genitourinary (GU) clinics. METHODS: Within the GU Site, which has the highest volume of unique patient visits (approx. 1600 per month) of the 15 disease sites, at Princess Margaret Cancer Centre (PM), EPIC-CP was added to the Distress Assessment and Response Tool (DART); an electronic platform for screening for distress. EPIC-CP is a 16-item prostate specific tool with 5 clinical domains: urinary incontinence, urinary irritation/obstruction, bowel, sexual, and vitality/hormonal symptoms. Several engagement strategies were utilized inclusive of a site based PROs kick-off event, development of a PROs collaborative, creation of educational toolkits and on-site support. Evaluations included: (1) screening rates, (2) EPIC-CP outcome data, and (3) a GU patient experience survey. RESULTS: Due to the limited access to electronic interfaces, the GU site has historically completed all PROs on paper. After 9 months of planning, DART-EPIC was launched on 20 iPads. After 6 months, the average screening rate was 90 % (8493/9392); 75 % electronic and 25 % paper. Of patients screened the mean age was 67 years and the average completion time was 5–6 min. The most affected clinical domains were: sexual symptoms (55 %), urinary irritation/obstruction (15 %) and vitality/hormonal symptoms (13 %). The cumulative average global EPIC-CP quality of life score was 12 (range 0–60). Based on a recent GU patient experience survey, 98 % (49/50) reported that EPIC-DART helps improve communication with their health care team. Additionally, 96 and 91 % of patients reported that their physical symptoms and emotional concerns were addressed, respectively. CONCLUSIONS: A change management approach to implementation of electronic PROs in busy, outpatient settings is necessary. DART-EPIC facilitates communication between patients and the health care team. The success of electronic PROs collection promotes confidence in a sustainable approach to continually improve patient quality of life.

(1051) Capturing high-quality ecological momentary assessments by integrating a smartphone application with a trial management system Victor B. Boesen, BSc, Copenhagen University Hospital Rigshospitalet and Gentofte Hospital, Copenhagen, Denmark; Per Cramon, MD, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark; Stine B. Nissen, M.S. in Psychology, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark; Steen J. Bonnema, MD, PhD, DMSc, Odense University Hospital, Odense, Denmark; Jakob B. Bjorner, MD, PhD, QualityMetric; University of Copenhagen, Copenhagen, Denmark; Thea Christoffersen, BSc, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark; Ulla Feldt-Rasmussen, MD, DMSc, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark; Mogens Groenvold, MD, PhD, DMSc, University of Copenhagen and Bispebjerg University Hospital, Copenhagen, Denmark; Laszlo Hegedus, MD, DMSc, Odense University Hospital, Odense, Denmark; Aase K. Rasmussen, MD, DMSc, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark; Torquil Watt, MD, PhD, Copenhagen University Hospital Rigshospitalet and Gentofte Hospital, Copenhagen, Denmark AIMS: To develop and implement a system to capture electronic ecological momentary assessments (EMA) and classical retrospective patient-reported outcomes, with focus on data security and on optimizing response timing and response rates. METHODS: The development took place within a multi-disciplinary setting, involving a software engineering company, university scientists and hospital

85

Fig. 1 . personnel. The software specifications followed current EMA standards (three semi-randomised prompts daily; repeated notification alarms; a 1-h fixed entry-period and livability functions enabling undisturbed time-interval, postponement of replies, and adjustment of diurnal rhythm) and was iteratively tested and revised among experts and patients. Data management and EMA capture was integrated with the PROgmatic trial management system for real-time data monitoring enabling study personnel notifications when relevant. RESULTS: Expert reviews and usability tests indicated a high usability of the developed application, but also revealed potential problems that might be addressed in the future. A secure and safe data management system fully integrated behind the hospital firewall provided timely distribution of questionnaires and automated email and text message reminders to study participants. In an on-going study, EMA and questionnaire response rates are 1488/1768 (84 %) and 120/123 (98 %), respectively. CONCLUSIONS: The system facilitates collection of high-quality EMA and questionnaire data (Fig. 1). The automated functions greatly reduce time consumption for study personnel. The system provides a template for collection of EMA data in clinical studies and practice.

(1053) Psychosocial, symptom, and quality of life predictors of participant retention in an internet-based symptom management study for women with recurrent ovarian cancer: A Gynecologic Oncology Group (GOG) Trial Heidi S. Donovan, PhD, RN, University of Pittsburgh School of Nursing, Pittsburgh, PA, United States; Susan M. Sereika, PhD, University of Pittsburgh School of Nursing, Pittsburgh, PA, United States; Sandra Ward, PhD, RN, University of Wisconsin-Madison, Madison, PA, United States; Catherine Bender, PhD, RN, University of Pittsburgh School of Nursing, Pittsburgh, PA, United States; Robert P. Edwards, MD, University of Pittsburgh School of Medicine, Pittsburgh, PA, United States; Judy Knapp, PhD, LCSW, University of Pittsburgh School of Nursing, Pittsburgh, PA, United States; Susan Nolte, PhD, RN, Abington Memorial Hospital, Abington, PA, United States; Michael Spring, PhD, University of Pittsburgh, Pittsburgh, PA, United States; Howard Stein, University of Pittsburgh School of Nursing, Pittsburgh, PA, United States; Joan Walker, MD, Oklahoma University Health Science Center, Oklahoma City, OK, United States; Kristin Zorn, MD, University of Arkansas for Medical Sciences, Little Rock, PA, United States; Lari Wenzel, PhD, University of California-Irvine, Irvine, CA, United States AIMS: The WRITE Symptoms (Written Representational Intervention to Ease Symptoms) Study is a 3-arm randomized controlled trial of an Internet-based intervention to improve symptom management for women with recurrent ovarian cancer conducted through the GOG. The aim of this analysis is to evaluate whether participants’ baseline psychosocial, symptom and quality of life (QOL) characteristics influenced study participation. METHODS: Women with recurrent or persistent ovarian, fallopian, or primary peritoneal cancer

123

86 who were experiencing 3 or more bothersome symptoms were eligible. Women (n = 497) were accrued from 53 GOG-affiliated sites. Following completion of valid, reliable measures of socio-demographics, optimism (LOT-R), depressive symptoms (CES-D), cancerand treatment-related symptoms (SRQ), and QOL (FACT-O), women were randomized to: enhanced care-as-usual, self-directed WRITE Symptoms, or nurse-delivered WRITE Symptoms. All participants completed follow-up measures monthly for 1 year. All cause study attrition variables (including death, disease progression, and drop-out) were created at 4 (end of primary aims) and 12 months. Baseline scores on key variables were then compared between those who were retained and those who were not. RESULTS: At 4 months, 380 women (76.5 %) were still on study. Those who were retained on study had lower baseline depressive symptoms (7.57 vs 10.31; p \ .001), higher optimism (23.21 vs 22.15; p = .009), more years of education (14.88 years vs. 14.25; p = .03), higher QOL (111.00 vs 100.10; p \ .001), and lower mean symptom severity (5.33 vs 6.04; p = .003) compared to those who were not retained on study. At one year, 287 women (58.1 %) had completed the full study. Only baseline depressive symptoms, QOL and symptom severity were associated with (p \ .01) participant retention for the full year. Age, marital status, income, children in the home, and employment status were not associated with participant retention (p [ .05). CONCLUSIONS: Lower baseline disease and treatment-related symptoms, lower depressive symptoms, and higher QOL were associated with early and long-term retention in this study. Higher optimism and years of education were associated with early, but not long-term retention. Further research is needed to identify ways to help cancer survivors with high physical and psychological symptom burden participate in Internet-based supportive care interventions.

(1055) A patient-centric, electronic solution for capturing concomitant medication usage Juha Matero, CRF Health, London, United Kingdom; Olli Kotiranta, CRF Health, London, United Kingdom; Nora Ibrahimova, CRF Health, London, United Kingdom; Paul O’Donohoe, CRF Health, London, United Kingdom, Jessica Thilaganathan, CRF Health, London, United Kingdom AIMS: In order to best understand the impact of a novel treatment it is important to ensure any additional medication a patient is taking alongside the study drug is also recorded. Historically this data has been captured retrospectively at site visits. The aim of this study was to develop an easy to use electronic solution that would provide more timely and accurate data on concomitant medication usage while reducing the burden on patients. METHODS: An electronic medication diary which allowed for the recording of medication name, route of administration and dosage was developed for a handheld platform. The solution included a pre-populated, tailored list of the patient’s most common drugs which is created and updated by the patient or site staff. This underwent iterative, one-on-one testing in 10 members of the general public aged between 60 and 82 who were unfamiliar with touch screen devices. The solution was updated and refined after each round of testing. RESULTS: The solution proved to be userfriendly and intuitive. The iterative approach proved an efficient method to gain usability maturity, allowing for the pragmatic application of changes to the solution and immediate re-evaluation of them. Over 3 rounds of testing a number of updates were made based on usability findings. The patient flow of the diary was improved, with the final solution requiring only 7 separate screens to report a medication intake. The terminology of the solution was fine-tuned to be more natural to the general public and especially to the elderly population. The training module was also improved to display animated rather than static guidance images to demonstrate how the

123

Qual Life Res (2015) 24:1–191 interactions for the solution worked. CONCLUSIONS: Concomitant medication usage has traditionally been collected with significant time delay. We developed an electronic medication diary for handheld devices and refined it through iterative testing in elderly patients to create a user friendly solution that made it easy to provide accurate, near real-time data on concomitant medication usage.

(1057) Patient reported outcomes (PRO) of acceptability with the use of a diabetes mobile application Astrid Torbjørnsen, MSc, Oslo and Akershus University College of Applied Sciences, Oslo, Norway; Lis Ribu, PhD, Oslo and Akershus University College of Applied Sciences, Oslo, Norway; Marit Rønnevig, RN MNSc, Oslo and Akershus University College of Applied Sciences, Oslo, Norway; Astrid Grøttland, BCS, Norwegian Centre for Integrated Care and Telemedicine, Tromsø, Norway; Sølvi Helseth, Oslo and Akershus University College of Applied Sciences, Oslo, Norway AIMS: To assess how patients report their acceptability and use of a diabetes mobile application for self-management, and further to assess their interactions with others (relatives, health care personnel) related to the use of the app. METHODS: The study sample are persons with type 2 diabetes living at-home, and who voluntarily participated in a study investigation mobile health and the use of a diabetes app. The eligibility criteria were adults C18 years old with type 2 diabetes, HbA1c 7.1 % or above, and ability to use the equipment and to fill in questionnaires in Norwegian language. This study is a part study of a three-armed randomized controlled trial with one control group and two intervention groups where both intervention groups received a smartphone with a digital diabetes diary and a blood glucose meter connected with Bluetooth. The app were provided outside usual care. The patients in the present study were recruited from the two intervention-groups (N = 50 + 51). We are using a qualitative method with in-depth interviews conducted successively after the participants had finished the one-year follow-up. The recruitment started in March 2012 and went on until saturation of data was achieved. Data were recorded and transcripted and qualitative content analysis was used according to their use of the app. The effects of PRO measures in in chronic care management are assessed with a conceptual framework (Santana, 2013). RESULTS: A total of 26 interviews were conducted with ca 1 1/2 h duration. Analysis is ongoing, and preliminary findings indicate that many of the persons’ were satisfied with the app, although they reported stress with the use of the app caused by rather outdated technology during the study. They also reported the need for more support according to difficulties with behavior change according recommendations, and according to a large disease burden. More detailed analysis will be presented at the ISOQOL conference. CONCLUSIONS: Our findings suggest that the participants were relative satisfied with the app, but less satisfied with the technology. Findings may indicate that the participants were in need of a higher intensity intervention with more support.

(1059) Implementation evaluation of a tablet-based waiting room survey of patient reported outcomes in an integrated youth health clinic Chris G. Richardson, PhD, University of British Columbia/St Paul’s Hospital, Vancouver, BC, Canada; Skye P. Barbic, Post-Doctoral Fellow, Faculty of Medicine, Vancouver, BC, Canada; Steve Mathias, MD, FRCPC, Providence Health Care, Vancouver, BC, Canada AIMS: There is increasing interest in measuring the patient reported outcomes (PROs) of youth in primary care settings. Two important uses of these PRO’s are to provide clinicians with a broad overview of

Qual Life Res (2015) 24:1–191

87

psychosocial functioning and to identify specific areas of concern that clinicians should discuss in more detail. Traditionally this data would be collected as part of a structured clinical interview; however, recent advances in technology have led to the development web-based waiting room surveys that provide flagged summary reports to clinicians in real time to inform their discussions with patients. We recently developed a waiting room survey derived from the HEADSS (Home, Education, Activities, Drug use and abuse, Sexual behavior, Suicidality and depression) interview guide that includes the GAINSS (Global Appraisal of Individual Needs Short Screener)—a tool to identify patients as potentially having one or more behavioral health problems. This aim of this implementation evaluation is to provide preliminary data on the impact of this system on patients, clinicians and organizational functioning. METHODS: Starting in the summer of 2015, approximately 1000–2000 youth attending the Granville Youth Health Centre in Vancouver Canada will complete the tabletbased waiting room survey prior to meeting with a clinician. Clinicians will be provided with a summary report that flags areas of concern and strength to inform their clinical interview with patients. Clinicians and patients perceptions of the benefits and weaknesses of the system will be assessed via qualitative and quantitative methods. RESULTS: Quantitative and qualitative findings related to the benefits and weaknesses of the system will be presented from the perspective of patients (e.g., satisfaction, engagement, identification of conditions requiring intervention), clinicians (perceived utility of brief reports to facilitate clinical interview, improved identification of conditions requiring medical intervention) and organizational functioning. CONCLUSIONS: We plan to discuss the implementation of this platform and describe how the collection of PRO’s via a waiting room survey can improve the primary health care delivered to youth in terms of earlier diagnosis of problematic conditions, increased engagement with the primary care practioner and overall increase in client satisfaction.

usability issues included using the stylus to select responses and glare on the screen. Differences in response between modes were not due to the differences in format. No changes were made between rounds. In round 2, issues with glare were found, but most patients preferred the tablet. Differences in response were found again. One participant attributed differences to the larger scaling of the VAS on the tablet which led to consistently different responses compared to paper; the participant considered these responses equivalent. CONCLUSIONS: The study showed excellent qualitative equivalence between paper and electronic versions of 8 PRO instruments with only minor usability issues reported. This study also supported usability of the tablet with an older OA population which expressed a strong preference for the electronic version.

(1061) Qualitative equivalence between paper and electronic tablet versions of 8 PRO instruments for osteoarthritis

AIMS: Many patients live with multiple chronic conditions requiring ongoing care and coordination across specialties. Many electronic tools attempt to facilitate ongoing care and coordination, but lack the integration of the patient as an equally-capable and committed care team member. In many cases, the patient’s voice is unheard and underappreciated in their unique journey—one that is often stressful, emotionally-charged, and life-altering. In response to these shortcomings, we are creating a digital care plan that integrates patientcentered methods within a user-centred design (UCD) development strategy to create a truly patient-centred solution. METHODS: We are following a proven UCD set of research and development methods, in which target users (in this case, the patients) are meaningfully included in the entire design process. As part of the first research phase, we conducted focus groups with patients who self-reported to have one or more chronic health conditions. In recognition of the complexity and emotionally-charged aspects of patient health-care experiences, we collaborated with a patient and community engagement research (PaCER) team to gain a clear understanding of what chronic disease patients really need. PaCER used patient-centred qualitative research methods from within the patient engagement research framework. The inclusion of these patient-centered research methods within a UCD platform is a natural fit to reach the goals of this project. RESULTS: Several qualitative analysis approaches led to the realization that study participants expressed perspectives that aligned perfectly with a salutogenic analysis model, which emphasizes the patient’s autonomy, strengths, and capabilities. Using this salutogenic lens, we were able to create 19 design requirements, written in the first person, that retain the personal nature of the patient’s health care experiences. They include: ‘‘My life has changed because of this condition. Help me find a new normal’’ and ‘‘Involve

Sonya Eremenco, MA, Evidera, Inc., Bethesda, MD, United States; Sonja Stringer, MPH, Evidera, Inc., Bethesda, MD, United States; Katelyn Cutts, Evidera, Inc., Bethesda, MD, United States; Michael Iacobelli, Evidera, Inc., Bethesda, MD, United States AIMS: This study evaluated the qualitative equivalence between paper and electronic tablet versions of the following 8 patient-reported outcome (PRO) instruments used in osteoarthritis (OA): SF36v2, WOMAC VAS, Pain VAS, Global Disease VAS, Brief Pain Inventory Interference, Sleep Assessment, Rating of Daytime Functioning, and Cognitive Function in Patients with Chronic Pain. Although equivalence for the WOMAC and SF-36v2 using different technology was previously published, additional equivalence evaluation of these and other instruments in a tablet with stylus was needed to document suitability of this mode of data collection for upcoming clinical trials. METHODS: A cross-sectional qualitative study was conducted involving cognitive and usability interviews with patients diagnosed with OA of the knee. The 8 instruments were administered on a TrialSlateTM tablet and paper versions, with each participant completing a subset of instruments on both modes. Participants were randomized to one of three instrument subgroups and also to order of mode completion. Interviews were conducted in two rounds to allow for evaluation of issues between rounds. RESULTS: Mean age of the sample (N = 25) was 62 years, (range 43–75), 60 % over 60 years old; 60 % were female; 92 % were white; 48 % were retired; 44 % had completed secondary school or some college, while 56 % had completed college or a post-graduate degree. In round 1, participants found the device easy to use and preferred the tablet to paper. Minor

(1063) Integrating patient-centered research methods within usercentered design: a patient-centered digital care plan Julie Babione, MSc, W21C Research and Innovation Centre/ UCalgary, Calgary, AB, Canada; Marlyn Gill, Patient and Community Engagement Researcher (PaCER), University of Calgary, Calgary, AB, Canada; Maria J. Santana, Research Ass. Professor, University of Calgary, Calgary, Canada; Katherine Skora, MS, MPH, CGC, Alberta Health Services, Calgary, Canada; Chloe De Grood, W21C Research and Innovation Centre/UCalgary, Calgary, AB, Canada; Colin Penman, Patient and Community Engagement Researcher (PaCER), University of Calgary, Calgary, AB, Canada; Dilshaan Panjwani, BHSc, W21C Research and Innovation Centre/ UCalgary, Calgary, AB, Canada; Jill De Grood, MSc, W21C Research and Innovation Centre/UCalgary, Calgary, AB, Canada; Jaime Kaufman, PhD, W21C Research and Innovation Centre/ UCalgary, Calgary, AB, Canada; William Ghali, MD, FRCPC, MPH, W21C Research and Innovation Centre/UCalgary, Calgary, AB, Canada; Peter Sargious, MD, FRCPC, MPH, W21C Research and Innovation Centre/UCalgary, Calgary, Canada

123

88 me in the decision-making process.’’ CONCLUSIONS: The integration of patient engagement research methods within a UCD development strategy has revealed the underlying needs of the patients and translated them into actionable development requirements. This will create a successful care planning solution that truly supports chronic disease patients to achieve wellness, goal-reaching empowerment, and in control of their health.

(1065) The usability and comparability of an app-based questionnaire across a range of screen sizes Paul O’Donohoe, CRF Health, London, United Kingdom; Juha Matero, CRF Health, London, United Kingdom; Petra Meinander, CRF Health, London, United Kingdom; Antti Heikkila, CRF Health, London, United Kingdom; Rauha Tulkki-Wilke, CRF Health, London, United Kingdom AIMS: Electronic data capture, in the form of electronic Patient Reported Outcomes (ePRO), has rapidly established itself as best practice across a wide range of clinical trial settings. Traditionally, ePRO has involved provisioning patients a handheld device with the study questionnaires already loaded on it, which provides consistency of administration across all participants. ‘‘Bring Your Own Device’’ (BYOD) is the concept of allowing patients to complete study questionnaires using their own devices, either via an app or a webbased system. A significant outstanding question for the successful implementation of BYOD in a clinical trial is the impact of administering the same questionnaire on range of different devices. This study looked to explore participant’s self-reported interpretation and response to questions administered on a range of smartphones, as well as the overall usability of the system. METHODS: An app-based version of a vaccine symptoms diary was developed for Android and iOS devices. 20 participants (11 female and 9 male) from the general population, covering a wide age range and self-reported levels of comfort with technology, were observed and interviewed while installing and using an app-based vaccine symptoms diary on a range of mobile devices with differing screen sizes (small, medium and large), including, where possible, the participants own device. RESULTS: All participants were able to successfully interact with the app, although a learning curve was observed. The majority of participants reported that the different screen sizes would not impact how they interpreted and responded to questions. Only 3 participants suggested they might respond differently, and this largely revolved around issues of familiarity with the test devices. The usability of the system was generally viewed as positive, although challenges around installing the app emerged. CONCLUSIONS: BYOD offers an exciting new opportunity for capturing PRO data in clinical trials, potentially reducing hardware costs and improving the patient experience. This study explored one of the outstanding challenges to BYOD around the impact of administering the same questionnaire on a range of different devices. The findings suggest that concerns around equivalence in BYOD studies may be overstated, although challenges to a successful BYOD approach remain.

(1067) Overcoming oncology study challenges with patient-centric technology Jessica Thilaganathan, CRF Health, London, United Kingdom; Karl McEvoy, CRF Health, London, United Kingdom, Paul O’Donohoe, CRF Health, London, United Kingdom; Nora Ibrahimova, CRF Health, London, United Kingdom; Katie Garner, CRF Health, London, United Kingdom AIMS: Although oncology trials account for a large proportion of global trials, only 5 % of cancer patients actually participate in trials.

123

Qual Life Res (2015) 24:1–191 Issues with recruitment and the complex nature of studies can make it challenging for patients to be as involved as they might want and can leave sponsors struggling to run high-quality studies. Our aim was to develop an electronic solution to reduce burden on sites and patients, helping to simplify data collection and, at the same time, improve data quality. METHODS: With complicated protocols, poor patient retention and difficulty capturing quality endpoint and patient reported outcome (PRO) data, sponsors are seeking ways of reducing the burden on site staff and patients while still collecting the data needed to demonstrate the effect of new treatments. We collected feedback from pharmaceutical sponsors, best practices from previous studies and reviewed literature before designing and developing an electronic solution to help track key endpoints and manage site visits. RESULTS: Initial sponsor feedback showed the need to have an intuitive concomitant medication diary to capture patient’s analgesic use and track pain levels. To help measure these important endpoints we developed an electronic diary that could help patients easily report their medication. A site portal was also developed to allow investigators to track what medications patients were reporting. Feedback from studies also indicated that sites struggled with organizing visits and staying protocol compliant. An electronic visit scheduling tool was developed to help sites keep track of what order and at which visits questionnaires should be completed. This tool also allowed notifications to be sent via the patient’s diary or through an SMS to the patient’s phone to remind them of an upcoming visit. Developing these tools has resulted in high compliance with an average compliance of 86 % achieved in site-based global studies involving over 1600 patients. CONCLUSIONS: With an increasing amount of PRO data being collected in oncology trials, both at home and at site visits, a user friendly eDiary and intuitive tablet solution can significantly reduce patient and site burden, leading to high compliance, improved data quality and study efficiency.

Cancer (1069) Social support is associated with mental health and healthrelated quality of life in breast cancer survivors: results from the German population-based, multiregional CAESARStudy Annika Waldmann, University Hospital Schleswig–Holstein, Campus Luebeck, Luebeck, Germany; Maren Bredehoeft, University Hospital Schleswig–Holstein, Campus Luebeck, Luebeck, Germany; Lena Koch-Gallenkamp, German Cancer Research Center (DKFZ), Heidelberg, Germany; Heike Bertram, Cancer Registry of North Rhine-Westphalia (Muenster region), Muenster, Germany; Andrea Eberle, Bremen Cancer Registry, Leibniz Institute for Prevention Research and Epidemiology – BIPS, Bremen, Germany,; Bernd Holleczek, Saarland Cancer Registry, Saarbruecken, Germany; Sieglinde Schmid-Hoepfner, Hamburg Cancer Registry, Hamburg, Germany; Sylke R. Zeissig, Cancer Registry Rhineland-Palatinate, Mainz, Germany; Hermann Brenner, German Cancer Research Center (DKFZ), Heidelberg, Germany; Volker Arndt, German Cancer Research Center (DKFZ), Heidelberg, Germany AIMS: Aims: Negative consequences of dramatic events and traumata can be buffered by social support according to the ‘‘buffering hypothesis’’ [1]. Following the ‘‘relational regulation theory’’, talks and common activities are main factors for this effect [2]. We aimed to determine the prevalence and the potential effect of social support on mental health and quality of life in breast cancer survivors. METHODS: Participants of the CAESAR-study (Cancer Survivorship-a multiregional population-based study) were identified via six epidemiological caner registries in Germany. A postal survey was conducted between 2009 and 2011 among 6036 survivors of breast cancer (diagnosis at least 5 years ago). The questionnaire contained

Qual Life Res (2015) 24:1–191 the Lubben Social Network Scale-6 (social isolation defined as \12 points) [3, 4], EORTC QLQ-C30 [5] and the Geriatric Depression Scale (cut off for suspected depression [5 points; manifest depression [11) [6]. Response rate was 44 % (n = 2649). RESULTS: Participants had a mean age of 65 years. Their cancer was diagnosed 8 years before (5–7 years: 44 %; 8–10 years: 44 %, [10 years: 12 %). 25 % lived alone, 60 % in multi-person households. 72 % had a spouse. 84 % had children. 18.6 % of the women reported social isolation (domain family: 14.7 %; domain friends: 23.9 %) – proportions varied with age. Youngest women were least and oldest women were most often affected by social isolation (30–39 years: 0 %; 40–49 years: 13.2 %, 80 years or older: 31.1 %). Socially isolated women in all 10-year age-categories had a lower emotional function (difference in most categories [10) and global quality of life/health status than women who reported social support. Further, socially isolated women had more often a manifest or suspected depression (34.9 and 10.7 % vs. 18.9 und 3.3 % among those with social support). CONCLUSIONS: Results from other studies indicate positive associations between social support and mental health and health related-quality of life in cancer survivors [7–11]. We were able to confirm this association—independent of age—for a large populationbased sample of German breast cancer survivors. Physicians and health care workers should be aware of this association and inform survivors about local sources of social support such as self-help groups. (1.) Cobb S. Psychosom Med; 36:200-14. (2.) Lakey B, Orehek E. Psychol Rev; 188:482-495. (3.) Lubben JE. Fam Community Health; 11:42–52. (4.) Lubben J et al. The Gerontologist; 46:503–13. (5.) Aaronson NK et al. J Natl Cancer Inst; 85:365–76. (6.) Yesavage JA, Sheikh JI. Clin Gerontol; 5:165–73. (7.) Kroenke CH, al. Breast Cancer Res Treat; 139(2):515–27. (8.) Michael YL et al. J Psychosom Res; 52(5):285–93. (9.) Leung J et al. Psychooncology. 2014; 23(9):1014–20. (10.) Soares A et al. Support Cancer Care; 21(8):2153–9. (11.) Lim J, Zebrack B. Support Care Cancer; 14(2):185–92.

(1075) Development of a patient-centric patient reported outcome instrument to assess the quality of life of patients living with mycosis fungoides/se´zary syndrome-type cutaneous T cell lymphoma (MF/SS-CTCL) Julia Braverman, PhD, PatientsLikeMe, Cambridge, MA, United States; Auriell N. Towner, PatientsLikeMe, Cambridge, MA, United States; Priya L. Raja, PatientsLikeMe, Cambridge, MA, United States; Magdalena Harrington, PhD, Shire Pharmaceuticals, Cambridge, MA, United States; Kristina F. Simacek, MA, PatientsLikeMe, Cambridge, MA, United States AIMS: In collaboration with Actelion Pharmaceuticals Ltd., PatientsLikeMe (PLM) developed the Cutaneous T-Cell Lymphoma-Quality of Life (CTCL-QOL) instrument- a patient-centric patient reported outcome (PRO) measure to understand the key areas of impact of mycosis fungoides/Se´zary syndrome-type cutaneous T-cell lymphoma (MF/SS-CTCL) on patients’ quality of life. METHODS: PLM utilized Open Research Exchange (ORE), a platform that enables rapid and effective patient engagement in the PRO development process to develop this instrument. The initial conceptual framework was based on systematic literature review and interviews with 3 key opinion leaders. Based on the conceptual framework, a 10-item, openended survey examining patients’ experience of living with MF/SS was generated. This survey was distributed to 21 MF/SS patients through ORE. In addition, in-depth follow-up interviews with 10 engaged patients were conducted. The results were analyzed through thematic content analysis, with saturation reached after 15 participants. Thirty-one preliminary items were generated based on the results of this analysis and administered to 50 MF/SS patients for their

89 feedback. The instrument was revised based on this feedback. The next 14-item version was then administered to 105 MF/SS patients to establish its construct validity, internal consistency and test–retest reliability. Forty eight of patients completed the retest. Following the analysis, one item was removed due to low item-total correlation. RESULTS: The final 13 item instrument demonstrated high internal consistency (Cronbach’s alpha = .94), test–retest reliability (.92) and high construct validity, assessed as correlation with pre-existent Skindex-29 measurement (r = .88) and known-group validity (t = 2.27; p \ .03) across different MF stages. CONCLUSIONS: Thus, the CTCL-QOL is a reliable and valid instrument for assessment of patients’ experience living with MF/SS-CTCL.

(1077) Use of the Skin Cancer Quality of Life Impact Tool (SCQOLIT)—A feasibility study in non-melanoma skin cancer Rubeta N. Matin, Consultant Dermatologist, Oxford University Hospitals NHS Trust, Headington, Oxford, United Kingdom, Elizabeth Gibbons, University of Oxford, Oxford, United Kingdom AIMS: Patient-reported outcome measures (PROMs) provide validated evidence of health and quality of life (QoL) from the patient perspective. Several national PROMs programmes have been implemented in the National Health Service in the UK—specifically for common elective surgical procedures. Local implementation is varied across settings and populations. The incidence of non-melanoma skin cancer (NMSC) is rapidly increasing, posing considerable burden on UK healthcare resources, yet there is limited evidence of use of PROMs in NMSC and little information about patients’ perceived health and QoL. This study will explore feasibility of implementing a skin cancer-specific PROM—Skin Cancer Quality of Life Impact Tool (SCQOLIT) for NMSC. METHODS: Three hundred patients with a pathological diagnosis of NMSC undergoing all treatment modalities will be recruited to complete SCQOLIT questionnaires at baseline, at 3, 6 and 9 months. Participation and response rates, missing data and individual change scores will be analysed. Staff and patients will be interviewed to explore acceptability and feasibility of collecting PROMs data. RESULTS: Interim results of the project to date will be presented. Feasibility will be assessed by evaluating number of eligible patients, number of consenting patients, reasons for not consenting and participant number. Individual longitudinal change in scores, response rates and psychometric properties of the SCQOLIT will be reported. CONCLUSIONS: Acceptability and feasibility of the SCQOLIT tool has never been rigorously assessed in Dermatology clinics. A validated NMSC-specific PROM would help standardize multi-centred trials, allow robust evaluation of quality of care and more appropriately direct healthcare resources to improve QoL in patients with NMSC.

(1079) Using a personalized measure (Patient Generated Index (PGI)) to identify what matters to people with cancer Ala S. Aburub, MSc, McGill University, Montreal, QC, Canada; Bruno Gagnon, M.D., M.Sc, Universite Laval, Laval, QC, Canada; Ana Maria Rodrı´guez, PHD, McGill University, Montreal, QC, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada AIMS: The specific objective of this study is to identify, for people with advanced cancer, similarities and differences in ratings of global QOL between personalized and standard measures METHODS: A total of 192 patients completed five QOL measures at study entry: PGI, generic measures (SF-6D, EQ-5D), and cancer specific measures of QOL (McGill Quality of Life Questionnaire and Edmonton Symptoms Assessment Scale). Comparisons among total scores were compared using Generalized Estimating Equations (GEE).

123

90

Qual Life Res (2015) 24:1–191

RESULTS: Patients voiced 114 areas of QOL concerns by the PGI with the top three being fatigue, sleep and pain (39.2, 22.6 and 21.6 %, respectively). PGI total QOL score was 25 to 30 percentage points lower than those documented by the other measures, particularly when QOL was poor. Correlations between PGI and other measures were low. CONCLUSIONS: PGI allowed patients to express a wide range of QOL concerns, many that were not assessed by the other QOL measures. If only one QOL measure is to be included, either in a clinical setting or for research, the PGI would satisfy many of the criteria for ‘‘best choice’’. PGI could be considered a cancer-specific QOL measure. This study provides evidence that the PGI would be a good measure for patients and clinicians to use together to identify areas of concern that require attention and monitor changing needs.

(1081) Negative effects on quality of life during Dutch cervical screening program are small Ida Korfage, PhD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands; Inge M. de Kok, PhD, Erasmus University Medical Center, Rotterdam, Netherlands; Wilbert B. van den Hout, PhD, Leiden University Medical Center, Leiden, Netherlands; Theo J. Helmerhorst, MD, PhD, Erasmus University Medical Center, Rotterdam, Netherlands; J. Dik F. Habbema, PhD, Erasmus University Medical Center, Rotterdam, Netherlands; Marie-Louise Essink-Bot, MD PhD, University of Amsterdam, Academic Medical Center, Amsterdam, Netherlands; Marjolein van Ballegooijen, MD PhD, Erasmus University Medical Center, Rotterdam, Netherlands AIMS: Quality of life (QoL) effects, ranging from small but frequent Qol losses due to cervical cancer screening to large but rare Qol gains due to prevention of cancer, are key to cost-effectiveness analyses of cervical cancer screening. We aimed to complete the matrix and fill the gap in available empirical data. METHODS: Utility scores (range 0 = death to 1 = perfect health) were empirically obtained for eight groups of women: invited for screening (n = 1023); having mildly dyskaryotic pap test results (n = 270); referred for colposcopy (n = 155); treated for precursors (n = 81); diagnosed with invasive cervical cancer (n = 77); disease-free after cervical cancer therapy (n = 285); having advanced cancer (n = 384), and a reference group (n = 835). Two utility scores (SF-6D and EQ-5D) were compared. RESULTS: If no invasive cancer was diagnosed, differences in utility scores between or within groups were undetectable. The reference group reported a mean utility score of 0.83 (95 % confidence interval (CI) 0.82–0.84). QoL losses (SF 6D) were large in women with invasive cancer (FIGO 2+, 0.63, 95 % CI 0.57–0.70). EQ-5D utility scores of patients with advanced cancer were worst (B0.36), implying large gains if these stages are prevented by screening. EQ-5D and SF 6D utility scores were almost always significantly correlated. CONCLUSIONS: This is the first study to empirically obtain utility scores for quality of life associated with cervical cancer screening and treatment in unselected populations in all relevant health states, enabling optimally valid estimates of cost-effectiveness of cervical cancer screening programs. If cervical cancer screening entails negative effects on quality of life during phases of screening, triage, and gynaecological evaluation, these will on average be small.

(1083) Prognostic value of health-related quality of life for death risk stratification in patients with unresectable glioblastoma Brice Paquette, MD, Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸on; Department of Digestive Surgery and Liver Transplantation, University Hospital of

123

Fig. 1 . Besanc¸on France, Besanc¸on, France; Dewi Vernerey, Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸on France, Besanc¸on, France; Ame´lie Anota, PhD, Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸on, Besanc¸on, France; Bruno Chauffert, Prof., Department of Medical Oncology, University Hospital, EA 4666, Amiens, France; Tienhan Sandrine Dabakuyo, PhD, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France; Loic Feuvret, MD, Department of Radiotherapy, PitieSalpetriere University Hospital, Paris, France; Luc Taillandier, MD, Department of Neurology, University Hospital, Nancy, France, Didier Frappaz, MD, Department of Oncology, Leon Berard Centre for Fight against Cancer, Lyon, France; Herve´ Taillia, Department of Neurology, HIA Val de Grace, Paris, France; Roland Schott, Department of Oncology, Paul Strauss Centre for Fight against Cancer, Strasbourg, France; Michel Fabbro, Department of Oncology, Val d’Aurelle Center for Fight against Cancer Montpellier France, Montpellier, France; Isabelle Tennevet, MD, Department of Oncology, Henri Becquerel Center for Fight against Cancer, Rouen, France; Francois Ghiringhelli, Prof., Department of Oncology, GF Leclerc Center for Fight against Cancer, Dijon France, Dijon, France; Chantal Campello, Department of Neurology, University Hospital, Nimes France, Nimes, France; Ce´cile Dalban, Biostatistics and Quality of life unit, Centre Georges Franc¸ois Leclerc, Dijon, France; Jerome Skrzypski, Biostatistics and Quality of life unit, Centre Georges Franc¸ois Leclerc, Dijon, France; Olivier Chinot, Prof., Department of Neuro-Oncology, University Hospital La Timone, Marseille, France; Franck Bonnetain, Prof., Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸onQuality of Life in Oncology National Platform, Besanc¸on, France AIMS: Glioblastoma is the most frequent primary brain tumour in adults with an extremely poor prognosis and is often unresectable (UGB). Health-Related Quality of Life (HRQoL) is a major subject of concern for these patients, who are often symptomatic at diagnosis. The aim of this study was to assess the independent prognostic value of HRQoL in UGB patients for death risk stratification. METHODS: The study was based on 134 patients with UGB enrolled in the TEMAVIR phase II trial. TEMAVIR evaluated bevacizumab and irinotecan as neo-adjuvant and adjuvant treatment combined with temozolomide-based chemoradiation for UGB. Individual HRQoL was evaluated at baseline using the EORTC QLQ-C30 and BN20

Qual Life Res (2015) 24:1–191

91

brain cancer-specific module. Clinical and HRQoL parameters were evaluated in univariate and multivariate Cox analysis as prognostic factors for overall survival (OS). Performance assessment and internal validation of the final model were evaluated with Harrel’s C-index, a calibration plot and a bootstrap sample procedure. RESULTS: Two independent predictors of OS were identified by the final multivariate analysis: future uncertainty and sensory deficit. This model exhibited good calibration and acceptable discrimination (C-statistic = 0.63). The internal validity of the model was verified with robust uncertainties around the hazard ratio (Fig. 1). The prognostic score identified 3 groups of patients with distinctly different risk profiles with median OS estimated at 16.2, 9.2 and 4.5 months. CONCLUSIONS: We demonstrated the additional prognostic value of HRQoL in UGB patients for death risk stratification and provided a simple prognostic score (PROSUG) that may help to guide clinical management and stratification in future clinical trials. Fig. 1 Symptom Distress ratings during aE versus E (1085) Quality of Life (QOL) Results of the UK TACT2 Trial: More Intensive Chemotherapy for Early Breast Cancer (EBC) Has a Measurable Impact on Patient-Reported Symptoms and Functioning (CRUK/05/019) Galina Velikova, MD PhD, Leeds Institute of Cancer Studies and Pathology University of Leeds, Leeds, United Kingdom; James Morden, ICR-CTSU, Division of Clinical Studies, The Institute of Cancer Research, London, United Kingdom; Peter Barrett-Lee, Academic Breast Unit, Velindre NHS Trust, Cardiff, United Kingdom; David Bloomfield, Sussex Cancer Centre, The Royal Sussex County Hospital, Brighton, United Kingdom; Murrey Brunt, Department of Oncology, University Hospital of North Staffordshire, Stoke-on-Trent, United Kingdom; Peter Canney, Department of Oncology, Beatson Oncology Centre, Glasgow, United Kingdom; Robert Coleman, Academic Unit of Clinical Oncology, Weston Park Hospital, Sheffiled, United Kingdom; Susan Russel, Cancer Clinical Trials Unit Scotland (CaCTUS), Edinburgh, United Kingdom; Mark Verrill, Department of Medical Oncology, Northern Centre for Cancer Care, Freeman Hospital, Newcastle upon Tyne, United Kingdom; Andrew Wardley, Department of Medical Oncology, The Christie Hospital NHS Trust, Manchester, United Kingdom; Judith Bliss, ICR-CTSU, Division of Clinical Studies, The Institute of Cancer Research, London, United Kingdom; David Cameron, Edinburgh Cancer Research Centre, University of Edinburgh and NHS Lothian, Edinburgh, United Kingdom AIMS: TACT2, a multicentre randomised phase III trial in patients with EBC with E-CMF as control has a 2 9 2 factorial design, testing: (1) accelerated epirubicin (aE) improves outcomes compared to standard epirubicin (E); and (2) oral capecitabine (X) gives similar efficacy but preferential side-effect profile and better QOL to CMF. Primary outcome results showed no benefit for accelerated epirubicin and non-inferiority of X over CMF in time to recurrence. We report the global comparisons of QoL, symptoms, functioning and toxicity for the different chemotherapy regimens. METHODS: Between 2005 and 2008, 4391 patients were randomised to receive either E (100 mg/ m2 9 4) 3 weekly or aE (100 mg/m2 9 4 plus pegfilgrastim 6mg) 2 weekly; followed by CMFx4 4 weekly or Xx4 (2500 mg/m2/day 14 days) 3 weekly. 1179 pts (from 1493 approached, 79 %) completed at least one QoL questionnaire (EORTC QLQ-C30/BR23, HADS, fatigue and TACT2-specific toxicity questions) at baseline, end of aE/E, end of X/CMF, 12, 24 months post-randomisation. Changes from baseline to each timepoint were compared between groups using ANCOVA, adjusting for baseline scores. GEE models were used to analyse data longitudinally across all time points. Results reported here are those found both statistically and clinically significant (p \ 0.01 from ANCOVA and comparison of proportions

with deterioration of C10 points on EORTC QLQ-C30/BR23). RESULTS: During treatment (after 4 cycles aE/E) patients reported worse symptoms with dose-dense chemotherapy aE than E (nausea/ vomiting, appetite loss, constipation, systemic side-effects) and deterioration of functioning (global QoL, role functioning) but the differences did not persist to 12 or 24 months. Patients reported worse symptoms with CMF than X at the end of 8 cycles (fatigue, dyspnoea, insomnia, constipation and systemic side-effects) and deterioration of functioning (global QoL, physical functioning). The differences often persisted up to 24 months. Agreement between clinician-reported and patient-rated toxicities was moderate with highest agreement for vomiting (71 %); for all side-effects the patients’ ratings indicated worse toxicity (Fig. 1). CONCLUSIONS: The hypothesis that CMF is more toxic and with worse impact on QOL than X was confirmed, with persistent long-term impact. More intensive chemotherapy (aE than E) led to worse symptoms only during treatment. Patient-reported data provided valuable evidence beyond clinical efficacy and clinician-reported toxicity.

(1087) Clustering based on unsupervised binary trees to define levels of symptom severity in cancer Pierre Michel, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Zeinab Hamidou, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Laurent Boyer, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Badih Ghattas, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Pascal Auquier, EA 3279 SelfPerceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France AIMS: Studies have suggested that the interpretation of symptom severity, functional and quality of life (QoL) data is not so intuitive. Implementation strategies of this type of measurements in clinical routine imply that consensual norms and guidelines regarding data interpretation are available in clinical practice. The aim of this study was to define levels of symptom severity for a cancer-specific questionnaire (EORTC QLQ-C30) using a method of interpretable clustering that uses unsupervised binary trees, and test their validity regarding clinical and functional outcomes. METHODS: The patients were classified using a top-down hierarchical method: Clustering using Unsupervised Binary Trees (CUBT). We considered a three-

123

92

Qual Life Res (2015) 24:1–191 Table 1 .

Fig. 1 .

Fig. 2 .

Fig. 3 .

group structure: ‘‘high’’, ‘‘moderate’’, and ‘‘low’’ level of symptom severity. The clustering tree was based on three stages using the 9 symptom scale scores of the EORTC QLQ-C30: a maximal tree was first developed by applying a recursive partitioning algorithm, the tree was then pruned using a criterion of minimal dissimilarity, and finally the most similar clusters were joined together. Inter-cluster comparisons were performed to validate the sample partition on clinical (duration of illness, type of cancer, presence/absence of metastasis) and functional (SF-36, EORTC QLQ-C30 Global Health Score and functional scale scores) data. RESULTS: Two hundred and 35 patients with different types of cancer were included. The threecluster structure has classified 143 patients with ‘‘low’’, 46 with ‘‘moderate’’ and 46 with ‘‘high’’ level of symptom severity (Figs. 1, 2, 3). This partition was mainly explained by cut-off values on Fatigue and Appetite Loss scores. The three clusters consistently differentiated patients on the clinical (duration of illness, type of cancer, presence/absence of metastasis) and functional outcomes (SF-36, EORTC QLQ-C30 Global Health Score and functional scale scores). CONCLUSIONS: Our study suggests that CUBT is relevant to define levels of symptom severity in cancer. This method may be very useful because the tree structure makes use of the original symptom data and not external variables, as in supervised classification, where the membership of an individual to a cluster is a priori defined (Table 1). This finding may have important implications for helping clinicians to interpret symptom scale scores in clinical practice.

123

(1089) The experiences of health-related quality of life in patients with non-specific symptoms that undergo a diagnostic work-up for cancer Ellen Moseholm Larsen, MScH, PhD Student, Copenhagen University Hospital, Nordsjælland, Hilleroed, Denmark; Bjarne O. Lindhardt, MD, DMSc, Copenhagen University Hospital, Hvidovre, Denmark; Susan Rydahl-Hansen, Associate professor, PhD MScN, Bispebjerg and Frederiksberg University Hospital and Aarhus University, Copenhagen, Denmark AIMS: The introduction of a diagnostic fast track programme is a new initiative in Denmark. The purpose is to confirm or refute serious illness within 22 days in patients with non-specific or vague symptoms of cancer. The diagnostic phase of cancer can affect healthrelated quality of life (HRQoL), however little is known about how patients with non-specific symptoms experience the time around diagnosis. The aim of this study was therefore to investigate how patients with non-specific cancer symptoms experience health-related quality of life during the pre-diagnostic phase. METHODS: This study is part of a large multicentre mixed methods study. Participants had participated in a convergent survey study investigating HRQoL in

Qual Life Res (2015) 24:1–191 the diagnostic phase. HRQoL was measured by The European Organisation for Research and Treatment of Cancer Quality of Life questionnaire (EORTC-QLQ-C30). 21 participants, that had completed a diagnostic fast-track programme at one of four hospitals in the Capital Region of Denmark were interviewed 2–4 weeks after completed diagnostic work-up. The interviews were semi-structured and were supported by an interview guide based on the same themes as in the EORCT-QLQ-C30 scale. Data analysis was based on content analysis, were the process of generating codes, sub-categories and categories was conducted. RESULTS: The analysis generated the following categories; emotional, physical, cognitive, role and social functioning, symptoms, the experience of QoL, the meaning of contextual factors and relations with health professionals. The categories describe different aspects of HRQoL during diagnostic investigations, and how participants struggle to be in control and maintain a healthy and social life despite dealing with symptoms of serious illness and diagnostic investigations. Undergoing diagnostic investigations for suspected cancer is frightening, and has an impact on everyday life. Social support, time to diagnosis and health professionals taking responsibility was important to how the participants experienced the diagnostic process. CONCLUSIONS: Undergoing diagnostic investigations for suspected cancer has an effect on several aspects of HRQoL, much in accordance with the EORTC-QLQ-C30. However, the interview data provided a more comprehensive understanding of how patients experience HRQoL during diagnostic investigations, which can be used in the interpretation of the EORTC-QLQ-C30 scores.

(1091) Identification of Patient-Relevant Non-Small Cell Lung Cancer (NSCLC) Symptoms through Semi-Structured Qualitative Interviews Kelly P. McCarrier, PhD, Health Research Associates, Inc, Seattle, WA, United States; Michael Scanlon, Health Research Associates, Inc., Seattle, WA, United States; Mona L. Martin, RN MPA, Health Research Associates, Inc., Seattle, WA, United States; Stephen J. Coons, PhD, Critical Path Institute, Tucson, AZ, United States; Alicyn K. Campbell, Genentech, Inc., South San Francisco, CA, United States; Kendra DeBusk, Genentech, Inc., South San Francisco, CA, United States; Dagmar Kaschinski, Boehringer Ingelheim GmbH, Ingelheim, Germany; Jean Marie Arduino, Merck & Co., Inc., Kenilworth, NJ, United States; Astra M. Liepa, Eli Lilly and Company, Indianapolis, IN, United States; Juliane Lungershausen, Boehringer Ingelheim GmbH, Ingelheim, Germany, On behalf of the Patient-Reported Outcome (PRO) Consortium’s Non-Small Cell Lung Cancer Working Group AIMS: To use qualitative interviews to identify patient-relevant NSCLC symptom concepts for assessment in new PRO measures. METHODS: Qualitative interviews were conducted at six U.S. clinical sites for concept elicitation in a sample of adults (18+) with NSCLC (stage I–IV; performance status 0–2). Semi-structured individual interviews used open-ended questions and day-reconstruction exercises to elicit spontaneous reports of NSCLC-related symptom concepts. Subsequent probing was used to assess concepts not spontaneously reported, and symptoms arising via probing were noted for analysis. Patients then rated the peak severity and bothersomeness associated with each expressed symptom using a 0 (‘‘not severe/bothersome at all’’) to 10 (‘‘extremely severe/bothersome’’) numeric rating scale. Interviews were audio-recorded, transcribed, and coded using Atlas.ti for content analysis. Rating results were summarized using descriptive statistics. RESULTS: A total of 51 patients participated in the interviews. Patients had a mean age of 65 years (range 46–86), were 51 % female, and 75 % white (non-Hispanic). At the time of interview, 12 % had Stage I, 37 % had Stage III, and 51 % had Stage IV NSCLC.

93 Twenty (39 %) patients were treatment-naive. In total, 1897 patient expressions of symptoms were coded from transcripts, and saturation of concepts (with no unique concepts appearing subsequently) was achieved after the 27th interview. The most frequently-expressed concepts were: Cough (by 80 % of patients [with 85 % of those reporting the symptom spontaneously]); Tiredness (77 % [55 %]); Shortness of Breath (69 % [82 %]); Poor Appetite (51 % [82 %]); Weight Loss (43 % [69 %]); Difficulty Breathing (41 %[82 %]; Fatigue (39 % [64]; Hoarseness (39 % [17 %]); General (Non-Chest) Pain (35 % [73 %]); and Chest Pain (33 % [82 %]). Among these expressed symptoms, General Pain (mean = 8.0 [SD = 2.3]), Poor Appetite (7.5 [2.4]), and Chest Pain (7.1 [2.7]) were rated the most severe. Poor Appetite (6.9 [1.9]), Shortness of Breath (6.9 [2.5]) and Tiredness (6.8 [2.2]) were rated the most bothersome. CONCLUSIONS: Relevant patient-reported NSCLC symptom concepts were identified through qualitative interviews. Support for relevance of concepts is demonstrated through both spontaneous and probed expression, patient-rated severity, and level of bother. Strength of these findings is supported by achievement of concept saturation. These identified symptom concepts should be considered in the development of new NSCLC-specific PRO instruments.

(1095) Cancer registry-linked data with patient-reported outcomes and patient experiences with care: an overview of findings from the centers for medicare & medicaid and national cancer institute SEER-MHOS and SEER-CAHPS data resources Erin E. Kent, PhD MS, National Cancer Institute, Rockville, MD, United States; Gigi Yuan, MS, Information Management Services, Inc., Calverton, MD, United States; Marie S. Topor, BS, Information Management Services, Inc., Calverton, MD, United States; Neetu Chawla, PhD MPH, Kaiser Permanente Northern California, Oakland, CA, United States; Sarah Gaillot, PhD, Centers for Medicare & Medicaid Services, Baltimore, MD, United States; Kimberly DeMichele, PhD, Centers for Medicare & Medicaid Services, Baltimore, MD, United States AIMS: New opportunities to conduct surveillance research of patientreported outcomes (PROs) research in cancer survivors are emerging, in part due to the now existing linkage of the National Cancer Institute’s Surveillance Epidemiology and End Results (SEER) program with two Centers for Medicare & Medicaid quality improvement initiative survey studies: the Medicare Health Outcomes Survey (MHOS) and the Consumer Assessment of Healthcare Providers and Systems (CAHPS) surveys. SEER-MHOS provides PRO data on a nationwide sample of individuals 65 and older and disabled individuals enrolled in Medicare Advantage Organizations (MAO) managed care health plans. The MHOS is an ongoing study that has been recruiting multiple cohorts since 1998. Individuals ages 65 and over who are enrolled in participating MAOs are randomly sampled, administered the survey by mail or telephone, and then re-surveyed two-years later. Fourteen cohorts, representing over 126,000 patients linked to the SEER cancer registry and over 1.9 million MAO enrollees without such history. The SEER-MHOS public data resource has produced to date 40 data use agreements and 17 publications. SEER-CAHPS links cancer registry data with cross-sectional survey data of Medicare beneficiaries (both Fee-for-Service and Medicare Advantage) that contain information on multiple aspects of patient experiences with care, including access to needed and timely care, doctor communication, and health plan customer service. The current linkage contains survey data from 1998 to 2010 and includes 150,750 respondents with a history of cancer and 571,318 without such history. METHODS: This presentation will provide an overview of the methodology, sample characteristics, and findings to date of

123

94 both datasets. RESULTS: Findings to date will be reviewed, including the impact of diagnosis and treatment on health-related quality of life in older cancer survivors, including physical health deficits among survivors of many cancers and variation in HRQOL by cancer treatments received, racial/ethnic variations in HRQOL, and the impact of cancer on activities of daily living. Findings from SEERCAHPS related to variations in care ratings between participants with and without cancer across the survivorship continuum and at end-oflife will also be described. CONCLUSIONS: The presentation will close with future opportunities, challenges, and research priority areas for conducting research with SEER-MHOS and SEER-CAHPS.

(1097) Incorporating the patient’s voice into instrument development: how do patients describe the impact of Non-Small Cell Lung Cancer (NSCLC) on their breathing? Michael Scanlon, Health Research Associates, Inc., Seattle, WA, United States; Kelly P. McCarrier, PhD, Health Research Associates, Inc, Seattle, WA, United States; Stephen J. Coons, PhD, Critical Path Institute, Tucson, AZ, United States; Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States; Alicyn K. Campbell, Genentech, Inc., South San Francisco, CA, United States; Kendra DeBusk, Genentech, Inc., South San Francisco, CA, United States; Juliane Lungershausen, Boehringer Ingelheim GmbH, Ingelheim, Germany; Dagmar Kaschinski, Boehringer Ingelheim GmbH, Ingelheim, Germany; Jean Marie Arduino, Merck & Co., Inc., Kenilworth, NJ, United States; Astra M. Liepa, Eli Lilly and Company, Indianapolis, IN, United States, On behalf of the PatientReported Outcome (PRO) Consortium’s Non-Small Cell Lung Cancer Working Group AIMS: Breathing-related symptoms are commonly reported with NSCLC, but the language patients use to describe breathing problems is not well understood. This study aims to identify the different expressions that patients with NSCLC use to describe their breathingrelated symptoms in order to facilitate the incorporation of the patient’s voice and perspective into new PRO measures. METHODS: Qualitative interviews were conducted at six U.S. clinical sites for concept elicitation in a sample of adults (18+) with NSCLC (stage I– IV; performance status 0–2; with and without comorbid chronic obstructive pulmonary disease [COPD]). Individual interviews were conducted by researchers using a semi-structured interview guide designed to elicit responses about NSCLC symptoms and related impacts. Interviews were audio-recorded, transcribed, and coded within Atlas.ti software for content analysis. Subjects rated symptoms for severity and bothersomeness using a numeric rating scale from 0 (‘‘not severe/bothersome at all’’) to 10 (‘‘extremely severe/bothersome’’). RESULTS: Fifty-one interviews were conducted. Participants had a mean age of 65 years; 51 % were female, 75 % were white (non-Hispanic), and 35 % had comorbid COPD. A total of 281 breathing-related sign and symptom expressions were identified in the transcripts and grouped into three distinct sub-concepts, Shortness of Breath (example patient language: I had shortness of breath), Difficulty Breathing (I had a hard time breathing), and Wheezing (I can hear the wheezing). Saturation of novel respiratory symptom/sign expressions was achieved within the first 27 interviews. The subconcept with the greatest number of coded expressions was Shortness of Breath (152 of 281 [54 %]), followed by Difficulty Breathing (94 [34 %]) and Wheezing (35 [13 %]). The mean severity and bothersome ratings, respectively, were: Shortness of Breath—6.7 (SD = 2.0) and 6.9 (2.5); Wheezing—6.2 (2.5) and 6.4 (3.2); and Difficulty Breathing—6.1 (2.5) and 6.0 (2.5). CONCLUSIONS: While a range of specific terms was used by patients to describe breathing-related signs and symptoms, ‘‘shortness of breath’’ was the most frequently-reported sub-concept and had the highest observed

123

Qual Life Res (2015) 24:1–191 mean severity and bothersomeness ratings among elicited sub-concepts. These findings provide support for ‘‘shortness of breath’’ as relevant patient language for assessing breathing-related problems within a PRO instrument for NSCLC.

(1099) Evaluate the impact of GVHD on quality of life, health, sexuality and fatigue of long term survivors after hematopoietic stem cell transplant (HSCT) Maribel P. Doro, Psychologist, ISOQOL, Curitiba, Parana´, Brazil; Julita M. Pelaez, psychologist, psychoanalysis, Curitiba, Parana´, Brazil; Vaneuza A. Funke, physician, Bone Marrow Transplant, Curitiba, Parana´, Brazil; Iris M. Okumura, student in clinical psychology at the Hospital, psychology, Curitiba, Parana, Brazil; Ricardo P. Pasquini, Professor and Phisycian, Member of the International Society of Hematology; Member of the International Society of Hematology; American Society for Blood and Marrow Transplantation; American Society of Clinical Oncology; International Society for Experimental Hematology; Socie, Curitiba, Parana´, Brazil AIMS: We designed this study in order to evaluate the impact of GVHD on QoL health, sexuality and fatigue in long-term survivors. METHODS: This is a prospective sectional study of 214 survivors, which were divided in two groups for comparison: Group 1 (G1 = 89) survivors with GVHD e Group 2 (G2 = 125) without GVHD. G1 was further divided into three groups according to GVHD classification: Group a (Ga)—acute GVHD; Group c (Gc)—chronic GVHD; Group ac (Gac) acute and chronic GVHD. Level of satisfaction about QoL, sexuality, health and fatigue were evaluated. All patients had at least 18 years old. Scales used were: WHOQOL, Fatigue FACT-F, Karnofsky and Socioeconomic demographic survey. Level of significance in all scales was 95 % (p B 0.05) for all risk factors for QoL. RESULTS: Among survivors from groups G1 and G2, there were more males (61.8 and 67 %), married (63, 54.8 %), and low socio economic status (67.4 and 61.3) patients. Both groups had similar percentage of survivors with more than 8 years of scholarship (67 %). Median age of survivor from G1 was 39 ± 10 years and from G2, 32 ± 9.6 years. Most of G1 survivors (67.4 %) were 25.9 ± 10.8 years old at the time of transplant versus 19.5 ± 10 years old for G2. Malignant diseases were predominant at G1 (60.7 %) when compared to G2 (29.8 %). At the time of the study, 73 % of survivors from G1 and 82 % from G2 had 10 to 15 years after HSCT. Pre and post HSCT Karnofsky scores were similar for both groups: G1 (89.9 % e 98.9 %), G2 (83.9 % e 99.2 %). Survivors from both groups reported satisfaction with QoL, health, and sexuality. 65.2 % from G1 and 75.4 % from G2 reported absence of fatigue. CONCLUSIONS: These results show that level of satisfaction of long term survivors with their QoL, health, sexuality and fatigue were favorable in spite of having been diagnosed with acute or chronic GVHD. Transplantation provides an opportunity for changes in subjectivity and in facing life’s adversities. It is therefore crucial that the QOL assessment can be part of the clinical protocol.

(1101) Leukemia treatment strategies: what are the minimally important differences in patient quality of life Emily McPherson, MA, BC Cancer Agency, Vancouver, British Columbia, Canada; Sonya Cressman, PhD MBA, BC Cancer Agency, Vancouver, British Columbia, Canada; Paulos Teckle, PhD, BC Cancer Agency, Vancouver, British Columbia, Canada; Dean A Regier, PhD, Canadian Centre for Applied Research in Cancer Control, Vancouver, British Columbia, Canada; Stuart J Peacock, DPhil, Canadian Centre for Applied Research in Cancer Control,

Qual Life Res (2015) 24:1–191 Vancouver, British Columbia, Canada, Alexa Evans, BC Cancer Agency, Vancouver, BC, Canada AIMS: Evaluating health-related quality of life (HRQL) is crucial when considering alternative chemotherapy treatments that long-term impacts on patient functioning and emotional status, but health utility data for haematological malignancy is limited for a number of reasons including the patients ability to complete questionnaires during times of extreme illness, such as the conditioning treatment before hematopoietic stem cell transplantation (HCT). We aim to determine the minimally clinically important difference (MCID) in HRQL for patients being treated for acute myeloid leukemia and myelodysplastic syndrome. METHODS: We administered two HRQL surveys (EQ-5D and FACT-leukemia) to acute myeloid leukemia and myelodysplastic syndrome-afflicted patients residing in five Canadian cities. Data were collected from 2012 to 2015 at Baseline, Month 3, 6, 12, 18 and 24. Canadian tariffs were used to calculate utility scores from the EQ-5D responses. We will use a combined anchor-based and distribution-based approach to calculate the MCID for the entire cohort and for three individual treatment strategies—consolidation chemotherapy, HCT or supportive care. RESULTS: By the time of the conference we will have analyzed longitudinal data for 30 patients treated with the different therapeutic strategies in order to produce estimated MCID ranges for EQ-5D. In reporting those ranges we will improve our knowledge basis about quality of life during the initial phases of leukemia treatment. CONCLUSIONS: As therapeutic delivery methods improve, e.g., offering chemotherapy in the outpatient setting, or using oral versus IV administration, improvements in patient HRQL are predicted to increase. The MCID values we will present can show how to provide more valuable therapeutics by going beyond statistical significance to determine the level of change in HRQL that is meaningful to patients.

Poster Session 1002: Thursday Poster Session 2 Kids and Young Adults (1008) What about the patient? Comprehensibility and user acceptance of the Kids-CAT among chronically ill children and adolescents Kathrin I. Fischer, M.Sc.med., Charite - Universitaetsmedizin Berlin, Berlin, Germany; Dana Barthel, Dipl.-Psych., University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Christiane Otto, PhD, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Ute Thyen, MD, University Medical Center Schleswig– Holstein, Lu¨beck, Germany; Marcus Klein, MD, University Medical Center Schleswig–Holstein, Kiel, Germany; Otto Walter, PhD, MD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Silke Schmidt, Prof., Ernst-Moritz-Arndt University Greifswald, Greifswald, Germany; Holger Muehlan, PhD, Ernst-Moritz-Arndt University, Greifswald, Germany; Matthias Rose, MD PhD, Charite Universitaetsmedizin Berlin, Berlin, Germany; Ulrike RavensSieberer, PhD, MPH, University Medical Center HamburgEppendorf, Hamburg, Germany; Sandra Nolte, PhD, Charite Universitaetsmedizin Berlin, Berlin, Germany AIMS: The Kids-CAT is a computer-adaptive test to assess healthrelated quality of life (HRQoL) in children and adolescents. The focus of this investigation was placed on mode of administration as well as comprehensibility of the Kids-CAT. The purpose is to outline the user perspective of the newly developed instrument to facilitate compliance and motivation of young patients to report their HRQoL in clinical routine. METHODS: 270 children and adolescents diagnosed with asthma, diabetes or rheumatoid arthritis completed the Kids-

95 CAT at two pediatric clinics in Germany. Group characteristics and questions on comprehensibility and user acceptance were assessed via a short electronic questionnaire, which was administered after completion of the Kids-CAT. Descriptive analyses were performed stratified by age and sex. RESULTS: 92.9 % of children and adolescents completed the Kids-CAT at the clinic using a laptop (68.4 %) or a PC (29.8 %). Due to the clinical setting, only one in five respondents were alone while filling out the questionnaire. However, the vast majority (88.7 %) reported not feeling interrupted while completing the Kids-CAT despite the real life condition of our study. Mean duration to complete the Kids-CAT was 7:46 min. 89.5 % of the children and adolescents valued the Kids-CAT as easy or very easy to complete and 80.1 % reported that they did not need any help while filling out the questionnaire. In general, we observed that adolescents (12–17 years) valued the Kids-CAT as easier to complete and needed less help compared to the group of children (7–11 years); they also needed less time to complete the questionnaire. CONCLUSIONS: The findings of this study support the use of the KidsCAT in clinical practice. Generally, children and adolescents from 7 to 17 years were able to complete the questionnaire quickly and evaluated it as comprehensible and easy to fill out. However, younger children may need some help when providing HRQoL data; therefore, clinical staff should be available in person (clinical assessment) or via phone (home assessment) in case questions arise during the test. Our study emphasizes the importance of incorporating the user perspective in HRQoL research.

(1010) Development of parent proxy and self-report healthrelated quality of life instruments for deaf children with cochlear implants Michael Hoffman, M.S., University of Miami, Coral Gables, FL, United States; Alexandra L. Quittner, PhD, University of Miami, Coral Gables, FL, United States; Ivette Cejas, PhD, University of Miami Miller School of Medicine, Miami, FL, United States AIMS: Severe to profound hearing loss is associated with measureable deficits in health-related quality of life (HRQoL), reflecting the broad effects of deafness and its concomitant effects on oral language, cognitive, social, and behavioral development. A cochlear implant (CI) is a surgically implanted, electronic device that provides sound to severely to profoundly deaf children; these devices have been shown to improve oral language in these children. To date, there are no CIspecific HRQoL measures for young children or their parents. This study will present preliminary findings from the development of the first HRQoL instruments for school-age deaf children, ages 6–12 years, and their parents. METHODS: Measure development will follow the procedures outlined in the FDA Guidance on Patient-Reported Outcomes. Phase I included a systematic review of the literature to hypothesize an initial conceptual framework and discussion guide to elicit information via focus groups with key stakeholders (e.g., otolaryngologists, speech pathologists), and open-ended, qualitative interviews with children and their parents. During Phase II, stakeholders (n = 80) as well as children with CIs and their parents (n = 24 dyads) will be recruited from CI Clinics at the University of Miami and Children’s Hospital of Philadelphia to complete the interviews. These data will be uploaded into Atlas.ti for qualitative coding. Coding of common themes will involve: (1) group consensus to identify and define initial codes, (2) independent coding of each transcript by two coders, followed by calculation of interrater agreement, and (3) final review of themes completed by the Investigators to ensure comprehensiveness, clarity and frequency of endorsement. Saturation of content will be calculated by assessing the frequency with which each item is mentioned by children and parents, and the point at which no new content is documented. RESULTS:

123

96 This study is currently awaiting IRB approval with focus group and open-ended interviews scheduled to begin in June. Results from Phases I and II will be presented. CONCLUSIONS: This is the first study to develop HRQoL measures for school-age children with CIs and their parents. CI-specific measures will enable us to track longterm outcomes, identify key targets for intervention, and analyze costeffectiveness.

Qual Life Res (2015) 24:1–191 Table 1 CEPI severity scale English translation (Spanish version validated) Extension Score Affects 0 % of 0 if total Body Surface area (BSA)

Intensity

Funcionality

No evident lesions

Not affected Can perform all daily activities Does not feel ashamed of disease

(1012) Validity and reliability of a composite measure of disease severity and quality of life in children with atopic dermatitis for use in primary care Andrea Na´jera, General Practitioner, Centro de la Piel, Ecuador., Quito, Ecuador; Julia E. Cabezas, Dermatology Resident, Centro de la Piel, Quito, Ecuador; Andrea C. Cueva, Attending Physician/ Researcher, Centro de la Piel, Quito, Ecuador; Marcos Serrano, Director-Department of Biostatistics, Hospital Carlos Andrade Marin, Centro de la Piel-Quito, Quito, Ecuador; Santiago A. Palacios, Dermatologist, Centro de la Piel. Universidad Cato´lica. Universidad Central, Quito, Ecuador AIMS: To validate a new scale that assesses both atopic dermatitis (AD) severity and quality of life (QoL) impairment. METHODS: The CEPI scale (CEPI-S) is a 9-item, 9-point likert scale instrument that measures disease severity including area of involvement and intensity, and rates QoL by assessing impact on the patient’s physical and psychosocial behavior. Two trained physicians prospectively recruited 196 children with AD following a standardized clinical history, physical examination and Scoring AD (SCORAD). The physicians also completed the CEPI-S and guided patients or carers to complete two other questionnaires, the Children’s Dermatology Life Quality Index (CDLQI) and the Center for Epidemiological Studies Depression Scale for Children (CES-DC). Data were analyzed using SPSS (version 18.0); the intrinsic characteristics of CEPI scale (reliability, precision and validity) were assessed with Cronbach’s alpha and Spearman’s Rank Correlation. The extrinsic characteristics (sensitivity, specificity, predictive values and likelihood ratios) were also analyzed with a 95 % CI. The local ethics committee approved the study and written informed consent was obtained from participants. RESULTS: A total of 196 patients (mean age = 8.52, SD = 3.83; female = 99) participated in the study and completed the CEPI-S. The mean CEPI-S score was 3.41 ± 1.7, SCORAD 25.90 ± 15.08, CDLQI 5.92 ± 4.86 and CES-DC was 7.56 ± 8.78. Cronbach’s alpha for the CEPI-S was 0.73; this did not improve upon item elimination. Extension correlated with intensity (0.46) and functionality (0.41); intensity correlated with functionality (0.58). In terms of precision, it showed a standard error of measurement of 0.85. Spearman’s rank Correlation Coefficient between CEPI-S and SCORAD was 0.86, and between CEPI-S and CDLQI (0.69) indicating convergence, and between CEPI-S and CES-DC was 0.35, indicating divergence. Analysis of extrinsic characteristics demonstrated that CEPI-S discriminated only between two groups of AD patients: mild and moderate/severe (p \ 0.0001). CONCLUSIONS: The CEPI-S demonstrated adequate correlation with SCORAD and CDLQI confirming that this valid scale can evaluate severity of AD as well as QoL impairment. It appears that the CEPI-S cannot discern between moderate and severe impairment and QoL affection, which requires further investigation. Nevertheless, it is a simple, valid and reliable tool that can be used for assessment of AD and QoL in primary care. ANNEX 1. CEPI SEVERITY SCALE The scale was created in 2002 in Quito, Ecuador. It evaluates three aspects of the disease: extension, intensity, and functionality. The scale is based in four premises: No affection (score of 0), mild (score of 1), moderate (score of 2) or severe affection (score of 3) in regards

123

Value Affects 1 to of 1 \30 % of if BSA

Dryness, erythema, desquamation

Causes mild discomfort Feels mildly ashamed Mild pruritus

Score Affects 30–60 % Dryness, erythema, 2 if of BSA desquamation, excoriation

Cannot perform some activities such as Play a certain sport Use a certain type of clothing Focus on work or studies Disturbs sleep

Score Affects over 3 if 60 % of BSA

Dryness, erythema, desquamation, exudate, and liquenification

Invalidates, patient Feels very ashamed, sad, or depressed Cannot sleep Cannot perform daily activities Does not want or cannot leave the house

to extension, intensity and functionality of the condition (Table 1). The scale has been critiqued as it discerns between mild and moderate disease, but not between moderate and severe; thus a score between 0 and 2 considers mild AD and a score of 3–9 considers moderate or severe AD.

(1014) Improving transition of adolescents with chronic health conditions: the impact of an educational program on transition competence and Quality of Life (QoL) according to age Henriette Golke, Dipl. Psych., Ernst-Moritz-Arndt-University Greifswald, Germany, Greifswald, Germany; Carsten HermannGaritz, Ernst-Moritz-Arndt-University Greifswald, Greifswald, Germany; Franziska Bomba, University Medical Center Schleswig– Holstein, Lu¨beck, Germany; Ute Thyen, MD, University Medical

Qual Life Res (2015) 24:1–191 Center Schleswig–Holstein, Lu¨beck, Germany; Silke Schmidt, Prof., Ernst-Moritz-Arndt University Greifswald, Greifswald, Germany AIMS: The transition from child-centered to adult-oriented healthcare systems poses a significant challenge for adolescents with chronic conditions. There is increasing evidence that inadequate transition arrangements can have adverse effects (e.g. higher morbidity). The aim of this study is to investigate the effect of an educational program to improve transition competence and the impact of QoL on different groups of age. METHODS: Transition competence (TCS) and QoL (EuroHis-8) were measured in a German sample of 242 adolescents and young adults from 13-20 years (M = 16, 58 years) with chronic conditions (diabetes (n = 153), inflammatory bowel disease [IBD] (n = 89) at baseline (t0) and six months thereafter (t2: intervention group [IG], n = 125; control group [CG, no training], n = 117). Repeated measurement analysis of variance with t0-scores and t2-scores in the IG and CG were calculated with the total sample regarding age. The participants were divided into age groups (\16, 16, 17,[18). RESULTS: IG and CG differed significantly with regard to change of TCS global score (F(1,232) = 164,85, p [ .00, g2 = .42). Looking at the age groups and the TCS score there is no effect of age (F(3,232) = .53, p \ .66, g2 = .01). There is a global effect on QoL (F(1,216) = 4.80, p [ .03, g2 = .02) between t0-scores and t2-scores. Regarding age there is a significant effect in the CG (F(3,96) = 2,78, p [ .05, g2 = .08), whereas in the IG there is not such an effect (F(3,120) = .57, p \ .64, g2 = .01). The testing of the effect of age in the CG shows that only the group of age \16 has a marginal significant effect (t(14) = 2.06, p \ .06). CONCLUSIONS: The educational program improved the transition competence for chronically ill subjects, with age of the participants having no impact on transition competence. Therefore the educational program is qualified for all investigated groups of age. The educational program did not lead to a significant better QoL score. The subjects of the CG exhibited a decline, especially the group under 16 years, in QoL after the 6-month period. The educational program can bolster these processes and stabilize QoL in the IG.

(1016) Secular trends in well-being in young Swedes: the ‘grow up Gothenburg’ studies Ebba Brann, PhD student, Inst. of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Monica Leu, PhD, Inst. of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Agneta Sjo¨berg, PhD, Department of Food and Nutrition, and Sport Science, University of Gothenburg, Gothenburg, Sweden; Aimon Niklasson, MD, PhD, Associate Professor, Inst. Clinical Sciences, University of Gothenburg, Gothenburg, Sweden; Kerstin Albertsson-Wikland, MD, PhD, Professor, Inst. of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Lauren Lissner, PhD, Professor, Inst. of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; John Chaplin, PhD, AFBPsS, CPsychol., Inst. Clinical Sciences, Sahlgrenska Academy at the University Gothenburg, Gothenburg, Sweden AIMS: The aim was to compare well-being in two school-based cohorts of Swedish adolescents, born 1974 and 1990, with attention to gender and weight status. METHODS: Two different birth cohorts of 18-year-olds completed a self-reported questionnaire including the Gothenburg Wellbeing scale for Adolescents (GWBa). The GWBa consists of 39 visual analogue scales and six dimensions: mood (8 items), self-esteem (7 items), physical condition (6 items), psychosocial functioning (5 items), stability (4 items), coping (3 items), and a total score. Height and weight were recorded in the schools by trained study teams and previously reported secular comparisons of overweight and obesity showed a significant increase in males but not in females. RESULTS: Height, weight

97 and well-being were recorded in the 1974 birth cohort (n: 4364, 50 % girls, mean age: 18.1 years) and in the 1990 birth cohort (n: 5151, 49 % girls, mean age: 18.7 years). In both cohorts, males reported significantly higher well-being compared to females in all dimensions. In the sexstratified analysis, the well-being for males was significantly lower on all dimensions in the later-born cohort. For females in the later-born cohort, scores were also lower in four dimensions, but there was no difference between the cohorts in coping, while self-esteem was significantly higher in the later-born cohort (p \ 0.01). Comparing the cohorts, the total wellbeing score was lower for the 1990 cohort, 60.7, compared to 63.1 (p \ 0.001). The greatest difference was seen for stability (feeling calm, relaxed, not-stressed and unconcerned), 51.3 (1990) compared to 60.5 (1974), (p \ 0.001). After adjusting for weight status and gender the significant differences between the cohorts remained (p \ 0.001). CONCLUSIONS: In males, a negative secular trend in well-being was observed in all dimensions and in four dimensions among females. The only secular improvement was in self-esteem among females. The total well-being score was lower in the later-born cohort, with the greatest decrease in the stability dimension. Even though prevalence of overweight and obesity increased in males and was stable in girls, the observed differences in well-being were still present after adjusting for weight status.

(1018) Parent–child reporting discrepancies in childhood epilepsy Nora Fayed, PhD, McMaster University, Hamilton, ON, Canada; Aileen Davis, PhD, University Health Network, Toronto, Canada; David Streiner, PhD, McMaster University, Hamilton, ON, Canada; Peter Rosenbaum, MD, McMaster University, Hamilton, ON, Canada; Charles E. Cunningham, PhD, McMaster University, Hamilton, Canada; Gabriel Ronen, MD, MSc, McMaster University, Hamilton, ON, Canada AIMS: Children with chronic conditions are increasingly being invited to provide information about outcomes that can only be obtained through self-report such as quality of life (QOL) or its predictors [e.g., mental health (MH)]. These reports are often complementary to, but occasionally different from, those of parents. To understand child and parent report, we sought to determine: (1) whether one might expect systematic differences for certain or all QOL and MH constructs; (2) if reporting differences were indicative of stable child-parent dyad subgroups; and (3) what traits characterized membership in one type of dyad group (i.e., ‘child positiveparent negative’, ‘parent positive-child negative’ or ‘child-parent agreement’). METHODS: We obtained two reports, a mean of 8.2 months apart, from 480 Canadian dyads involving children with epilepsy aged 8–14 and a parent, using an epilepsy-specific measure (CHEQOL) a generic measure of QOL (KIDSCREEN 52); and validated standardized measures of child MH components of attention, conduct, cooperativeness and depression (from Canadian Population Health Scales). Each QOL subscale or MH component total parent score was deducted from the respective child score and displayed as a box plot at each time point. Membership in discrepancy groups was based on score differences greater than 1 SD away from the mean. Stability of membership in a dyad group was determined across time and across constructs with pairings of agreement using Cohen’s kappa. Reporting patterns associated with dyad groups was assessed using a latent class analysis approach. RESULTS: Overall, children were significantly more likely to report positive overall life satisfaction, positive friendships, lower depression scores, and higher parental supportiveness than their parents. Remaining QOL and MH domains reflected overall agreement. Stability of membership in a discrepancy dyad ‘child positive-parent negative’, ‘parent positivechild negative’ or ‘parent–child agreement’) ranged from 0.61 to 0.71

123

98

Qual Life Res (2015) 24:1–191

across constructs and 0.63–0.72 across time points. Characteristics associated with membership in the ‘parent positive-child negative’ and ‘child positive-parent negative’ dyads were identified. CONCLUSIONS: Certain constructs within QOL and MH reporting are associated with general differences noted among children and parents. ‘Child positive-parent negative’ and ‘parent positive-child negative’ dyads exist and should be accounted for in patient reported outcomes research.

(1020) Health-Related Quality of life in Brazilian patients with osteogenesis imperfecta: a cross-sectional study using PedsQLTM Ana Paula Vanz, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Juliana Van de Sande Lee, Joana Gusma˜o Hospital, Floriano´polis, Brazil; Bruna Pinheiro, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Marina Zambrano, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Evelise Brizola, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Neusa Sicca Rocha, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Ida Vanessa D. Schwartz, Federal University of Rio Grande do Sul, Clinical Hospital of Porto Alegre, Porto Alegre, Brazil; Teˆmis Maria Fe´lix, Federal University of Rio Grande do Sul, Clinical Hospital of Porto Alegre, Porto Alegre, Brazil AIMS: Osteogenesis imperfecta (OI) is a genetic disorder of collagen biosynthesis, characterized by low bone mineral density leading to recurrent fractures. In general, clinical symptoms lead to functional limitations in most patients. Purpose: To evaluate the Health-Related Quality of Life (HRQOL) of children and adolescents with OI in two South Brazilian Reference Centers. METHODS: A cross-sectional study was conducted in Clinical Hospital of Porto Alegre, located in the state of Rio Grande do Sul, and in Joana Gusma˜o Hospital, located in the state of Santa Catarina, using the PedsQLTM instrument (Pediatric Quality of Life Inventory). The PedsQLTM is an instrument to evaluate generic HRQOL that has been validated in Brazil. This instrument results in four domains: physical functioning, emotional functioning, social functioning and school functioning. RESULTS: The sample consisted of 52 children and adolescents with OI (5–17 years), of these, 26 were OI type I, 13 type IV, 12 type III, and 1 type V. In general, the domain with the smallest score was physical functioning (62.07 ± 2.9). For data analysis, due to the highly variable clinical expression seen in OI, the sample was divided in moderate/severe forms (types III, IV and V) and mild form (type I). There was a significant difference in physical

functioning (p B 0.01) between mild OI (69.23 ± 3.3) and moderate/severe OI (54.56 ± 4.4). When stratified by sex, emotional functioning showed a noteworthy difference (p = 0.08), and the lowest score was found in males (62.75 ± 03.01). Regarding the treatments received, no significant differences between the study domains were found. A positive correlation was found between mobility and physical functioning (p B 0.01), and a negative correlation was found between pain and physical functioning (p = 0.025). CONCLUSIONS: Overall, this study found differences in physical functioning associated to disease severity and mobility. This data reinforces the importance of the clinical management of these patients aiming their functional improvement leading to a better quality of life.

(1022) Quality of life and functional performance of children born preterm at preschool age: longitudinal study Martina E. Vieira, University of Sa˜o Paulo, University of Goia´s State, Goiaˆnia, Goia´s, Brazil; Cibelle K. Formiga, University of Goia´s State, Goiaˆnia, Goia´s, Brazil; Maria Beatriz M. Linhares, University of Sa˜o Paulo, Ribeira˜o Preto, Sa˜o Paulo, Brazil AIMS: This prospective-longitudinal study aimed to examine prediction models of quality of life (QoL) and functional performance in children born preterm at preschool age by considering clinical, socioeconomic, and neurobehavioral variables as predictor factors at the neonatal phase, development in the first year of age, and health, socioeconomic, developmental, and functional indicators at preschool age. METHODS: The sample included 24 low-risk children born with gestational age of 33 weeks, in mean and low birth weight (mean = 1692 g). The neurobehavioral performance was evaluated by Neurobehavioral Assessment of Preterm Infant-NAPI prior to term age. The psychomotor development during the first year of age was evaluated by three different tools (Denver Development Screening Test II-DDST-II, Test of Infant Motor Performance-TIMP, and Alberta Infant Motor Scale-AIMS). At pre-school age, the functional performance was assessed by Pediatric Evaluation of Disability Inventory (PEDI) and the QoL was evaluated by Child Health Questionnaire-50 Parent Form (CHQ-PF50). The children’s health history and the socioeconomic status were also examined. Multiple linear regression analysis was performed regarding the QoL and the functional performance as final outcomes. RESULTS: The children displayed deficits in 7/15 of QoL domains. The prediction models showed that the worst children’s QoL was associated with the following risk factors: sexually transmitted disease in the mother, male gender of the

Table 1 Prediction models of the CHQ-PF50 summary scores QoL outcome

R2

Predictors (phase)

b

p

CI (95 %) Lower

Physical summary score

NAPI—alertness and orientationa (neonatal)

0.73

0.37

0.004

0.04

0.20

Respiratory disease (preschool)

-0.62

\0.001

-20.03

-9.11

Severe accidentsb (preschool)

-0.23

0.05

-6.94

0.13

b

Psychosocial summary score

Upper

NAPI—percent asleepa (neonatal)

0.52

DDST-II—personal-socialc (preschool) Educational level of the fatherd (preschool)

0.36

0.023

0.01

0.14

-0.46

0.005

-12.89

-2.63

0.40

0.013

1.21

9.04

R2 = multiple correlation coefficient squared adjusted based on the sample size; b = standardized partial regression coefficient; p = t-test; CI = confidence interval; NAPI = Neurobehavioral Assessment of Preterm Infant; DDST-II = Denver Development Screening Test II a

NAPI: higher score represents better neurobehavioral performance in all clusters, excepted in the percent asleep

b

Severe accidents and respiratory disease: No/Absence = 0 (zero), and Yes/Presence = 1

c

DDST-II: Normal = 0 (zero), and Risk = 1

d

Educational level of the father: Below high school = 0 (zero) and Above high school = 1

123

Qual Life Res (2015) 24:1–191

99

Table 2 Prediction models of the PEDI domains Functional outcome

R2

Predictors (phase)

b

p

CI (95 %) Lower

Self-care

Mother’s employmenta (neonatal)

0.38

CHQ-PF50—Parental impact-Emotionalb (preschool) DDST-II—Personal-socialc (preschool) Mobility

Gestational age

0.29

DDST-II—Personal-socialc (preschool) Social function

CHQ-PF50—Parental impact-Timeb (preschool)

0.21

Upper

0.32

0.080

-0.53

0.31

0.076

-0.01

8.62 0.18

-0.37

0.049

-12.37

-0.03

0.41

0.031

0.76

14.09

-11.47

0.010

-19.90

-3.04

0.48

0.014

0.11

0.84

2

R = multiple correlation coefficient squared adjusted based on the sample size; b = standardized partial regression coefficient; p = t-test; CI = confidence interval; PEDI = Pediatric Evaluation of Disability Inventory; CHQ-PF50 = Child Health Questionnaire-50 Parent Form; DDST-II = Denver Development Screening Test II a

Mother’s employment: unemployed = 0 (zero), and employed = 1

b

CHQ-PF50: higher scores indicates better QoL

c

DDST-II: Normal = 0 (zero), and Risk = 1

offspring, lengthy stay in hospital, corticotherapy, poor alertness and orientation in NAPI, motor development at risk (4–6 months), respiratory disease, use of medications, history of severe accidents, personalsocial behavior at risk (preschool age), and poor social function at preschool age, and low socioeconomic status (Tables 1, 2). In addition, the prediction models indicated that the worst children’s functional performance was associated with the following risk factors: gestational age \32 weeks, personal-social behavior at risk (preschool age), high emotional impact on family and also on parents’ time in QoL assessment at preschool age, and unemployed mother. CONCLUSIONS: The QoL and the functional performance of a sample of children born preterm at preschool age were associated with clinical, socioeconomic, and developmental factors throughout the trajectory from birth to preschool age. Preventive follow-up programs were recommended to detect these risk factors for developmental problems and also to effectively manage protecting the children from the potential negative impact on their QoL and functionality.

(1024) Identifying outcomes for a new paediatric CFS/ME PROM: a meta-ethnography of qualitative studies Roxanne M. Parslow, BSc (HONS), University of Bristol, Bristol, United Kingdom; Sarah Harris, MSc, University of Bath, Bath, United Kingdom; Jessica Broughton, MSc, University of Bath, Bath, United Kingdom; Alda Alattas, University of Bristol, Bristol, United Kingdom; Esther Crawley, BA(Hons), BM BCh, MRCP, FRCPCH, PhD, University of Bristol, Bristol, United Kingdom; Kirstie Haywood, Senior Research Fellow, Warwick Medical School, Warwick University, Coventry, United Kingdom; Alison Shaw, BA, MSc, PhD, University of Bristol, Bristol, United Kingdom AIMS: Paediatric Chronic Fatigue Syndrome/Myalgic Encephalomyelitis (CFS/ME) is common and disabling. However, little is known about patients’ experiences and priorities in terms of illness related outcomes. Consequently, children’s views have not been incorporated into healthcare evaluations. Our aim was to improve our understanding of aspects of health that are most important to children with CFS/ME to inform the conceptualisation of a new Patient Reported Outcome Measure (PROM). METHODS: A systematic review and metaethnography of studies exploring the experiences and/or perspectives of children (aged \18 years of age) with CFS/ME was carried out. Four key databases were searched: MEDLINE, EMBASE, PsychINFO and CINAHL. Hand-searching of key journals, grey literature, reference lists of included studies and contacting first authors was also

undertaken. Titles and abstracts (1432) and full texts (78) were double screened according to inclusion/exclusion criteria. Study quality was assessed using the Critical Appraisal Skills Programme (CASP) checklist. RESULTS: 1432 studies were identified of which 10 studies involving 82 children (8–18 years) were included: seven studies were based in the UK, two in Norway and one in South Africa (in Afrikaans). Five key themes emerged: (1) Uncertainty and disbelief. Unfamiliarity of the illness results in an uncertain trajectory and disbelief from others. (2) Opposing views in the search for legitimacy. Children use opposing views (physical and psychological) to explain CFS/ME and prove legitimacy. (3) Loss: Physical, social and personal. Children live with a new restrictive body, due to isolation become detached from normal social experiences. CFS/ME takes away who children ‘used to be’ resulting in identity confusion. (4) Coping. Reducing uncertainty through diagnosis, advice on illness management and validating the illness within children’s social networks helped children cope with the illness. (5) Recovery. Physical, functional and personal growth changes occur in children feeling better or recovered. CONCLUSIONS: There is evidence that CFS/ME has a significant impact on the health-related quality of life of children. Conceptualisation is a key initial stage in PROM development; the identified themes will inform a developing conceptual framework for a child-specific PROM in CFS/ME which seeks to accurately capture what really matters to children.

(1026) The impact of having a child with respiratory insufficiency on the family Robert J. Graham, Boston Children’s Hospital, Boston, MA, United States; Angie Mae Rodday, Tufts Medical Center, Boston, MA, United States; Yudy Muneton-Castano, Boston Children’s Hospital, Boston, MA, United States; Ruth Ann Weidner, Tufts Medical Center, Boston, MA, United States; Susan K. Parsons, Tufts Medical Center, Boston, MA, United States AIMS: Among the population of children with special healthcare needs, those with chronic respiratory insufficiency (CRI) represent an especially vulnerable group. Families of these children are required to assume an altered parental role, serving as parent, care coordinator and advocate, active caretaker, and any other required function. We assessed whether family characteristics and clinical factors were associated with impact on family when managing severe CRI at home. METHODS: Parents of children enrolled in a comprehensive program for CRI completed the Impact on Family Scale (IFS), the Child Health Ratings Inventories (CHRIs)-General parent proxy measure, and demographic information.

123

100 While higher IFS scores indicate greater family impact, higher CHRIsGeneral scores indicate better quality of life. Clinical information was obtained from medical chart review. Multivariable linear regression analysis was used to determine factors associated with IFS scores. RESULTS: 158 parents consented (81 %) and 118 completed measures at study entry. Median child age was 8.5 years and 50 % were female; parent respondents had a mean age of 40 and 81 % were female. The mean IFS score was 40 (SD = 10; possible range = 15–60). Multivariable modeling showed that insurance type, parent emotional functioning, parentrated child health, and physician-rated clinical severity were significantly associated with IFS score. Compared to those with both public and private insurance, those with private only had lower IFS scores (b = -4.7, p = 0.02). High scores of parent emotional functioning (b = -1.6 for ‘ SD, p \ 0.001) and parent-rated child health (b = -1.3 for ‘ SD, p = 0.003) were both associated with lower IFS scores. In contrast, worse physician-rated clinical severity was associated with higher IFS scores (b = 0.7, p = 0.04). Family characteristics, including income, education, job status, race/ethnicity, parent age, parent gender, and number of siblings were not associated with IFS scores. CONCLUSIONS: Families of children with CRI are greatly impacted by their child’s health and associated needs. Factors related to the disease, as reflected in physician- and parent-rated measures, insurance status, and parent emotional functioning were associated with IFS scores. Interestingly, many other family characteristics were not associated with IFS scores. Future research should identify interventions or resources to acknowledge and attenuate the impact of having a child with CRI on the family.

(1028) Immigrant mothers balancing care for chronically ill or disabled children and labour market participation Lisbeth G. Kvarme, PhD, Associate Professor, Oslo, Norway; Elena Albertini-Fruh, RN, Associate Professor, Oslo, Norway; Hilde Liden, PhD, Research Professor, Oslo, Norway AIMS: The aim of this study is to contribute to a better understanding of the incentives and barriers to balanced family life and employment among immigrant mothers caring for chronically ill or disabled children. Furthermore, we discuss these implications for mothers‘ quality of life. METHODS: The study is based on an exploratory qualitative design, and is a part of a larger research project including different methods (register data, interviews, and documents). The qualitative study is based on 27 individual semi-structured interviews and three focus groups with immigrant parents from Poland, Pakistan, and Vietnam. RESULTS: Several aspects of family life will affect the mothers’ options to work such as: childcare needs, the organization of family life, family structure, and the mothers’ health. The mothers‘ language skills, education, and previous work experience are vital to employment opportunities. Accessibility to public services, assistance and financial support is essential for labour marked participants. Most mothers are satisfied with the support they get from kindergartens and schools. However, taking the child to specialist treatment is timeconsuming, and often the public healthcare system is inadequate. The immigrants report that they lack information about their rights and the healthcare system. Often the support is inadequate or not adapted to the families’ needs. Many feel isolated, experiencing that social networks are reduced due to childcare needs. The mothers fail to have full-time or even part-time jobs because of these factors. To cover the families’ financial needs the fathers become the main breadwinners, working full-time and even overtime while the mothers take responsibility for childcare. One-third of our sample consists of single parent families, where the mother is either involved in education, works part-time, or plans to do so. CONCLUSIONS: While most mothers want to work part-time while continuing to care for their children, the majority experience that this is not an option. Mothers expect that employment will contribute to improve both mothers’

123

Qual Life Res (2015) 24:1–191 quality of life and health, their financial situation and their self-esteem, as well as integration into the society.

(1036) Quality of life in children with ADHD Montserrat Hernandez-Martinez, Pediatrician, Primary Care Center Collblanc, CSI, Hospitalet de Llobregat (Spain), L’Hospitalet de Llobregat, Barcelona, Spain; Nuria Pastor-Hernandez, Psychologist, Faculty of Psychology. University of Barcelona, Barcelona, Barcelona, Spain; Xavier Pastor-Duran, Pediatrician, Faculty of Medicine, Hospital Clinic, University of Barcelona, Barcelona, Spain; Cristina Boix-Lluc, Psychologist, Neurology Dep. Hospital Sant Joan de Deu, Esplugues de Llobregat, Barcelona, Spain AIMS: The Attention Deficit and Hyperactivity Disorder (ADHD) has a direct impact in the patients with such condition and all those who surround them with repercussion in the cognitive, emotional and social scope of the affected children, their family, schoolfellows and friends. The concept of ‘‘Quality of Life’’ (QoL) refers to the perceived patients’ health status as well as the degree of wellness or satisfaction with their condition. BiblioPro offers objective tools to measure this subjective feeling by the patients by themselves of by third persons (family, friend, etc.)The objective of this study is to evaluate the impact of ADHD on the QoL of children having this disorder in comparison with an equivalent control group without ADHD. METHODS: This is a cross prospective case–control study. An information sheet about the study was explained by the pediatricians to them and their parents. 58 children or both sexes with ages between 8 and 16 years-old were entered at the study after the signature of the informed consent. Inclusion criteria for the ADHD group were 28 children recently diagnosed without beginning any treatment. In the 30 children of the control group ADHD. The KINDL questionnaire was used to know the self-perceived QoL by all the children. Kid template was used for 8–12 years-old. Kiddo was the template for 13–16 years-old. This questionnaire evaluates 6 dimensions: 1—Physical wellness; 2—Emotional wellness; 3—Self

Table 1 Results of KINDL analysis in children with and without ADHD

Qual Life Res (2015) 24:1–191 esteem; 4—Friends; 5—Family; 6—School. Statistical data analysis has been done with IBM SPSS v. 22 using specific algorithms provided by BiblioPro to compute and normalize the dimensional scores. RESULTS: Each of the six dimensions as well as the global score analyzed with the Student’s t test showed highly significant differences (p \ 0.02) between the two groups. In all the scores, the best (highest value) fell onto the control group (without ADHD). Detailed results, normalized to 100 points, can be seen on the Table 1. CONCLUSIONS: Self-perceived QoL is worse in children with ADHD An early diagnosis, intervention and follow-up can be done at Primary Care.

(1038) Ability to predict healthcare utilization using healthrelated quality of life scores among children with respiratory insufficiency Angie Mae Rodday, Tufts Medical Center, Boston, MA, United States; Robert J. Graham, Boston Children’s Hospital, Boston, MA, United States; Ruth Ann Weidner, Tufts Medical Center, Boston, MA, United States; Norma Terrin, Tufts Medical Center, Boston, MA, United States; Laurel K. Leslie, Tufts Medical Center, Boston, MA, United States; Susan K. Parsons, Tufts Medical Center, Boston, MA, United States AIMS: Children with chronic respiratory insufficiency (CRI) associated with a variety of medical conditions, including muscular dystrophy, spinal muscular atrophy, skeletal dysplasia, and spinal cord injury, may require intermittent or continuous mechanical ventilation. These children are at risk for developing severe acute illness that may result in unscheduled and costly hospitalizations. Our aim was to determine whether parent-proxy reported health-related quality of life (HRQL) predicts future healthcare utilization. METHODS: Pediatric participants were aged 30 days–22 years with CRI and were enrolled in a coordinated care program at Boston Children’s Hospital. Parents of these children completed a global HRQL scale and general health item from the Child Health Rating Inventories (CHRIs) on behalf of their child at study entry and 6 months later. HRQL was categorized as poor/fair, good, and good/excellent. The two outcomes were total healthcare contact days and inpatient days in the 6-month windows following collection of the HRQL measures; healthcare utilization came from BCH hospital billing data. Negative binomial regression was used to estimate the effect of HRQL on future utilization controlling for child age, years enrolled in the coordinated care program, and physician-rated clinical severity. RESULTS: Median child age was 7 years and 46 % were female. Approximately 75 % of children had any healthcare utilization in the 6-month windows following the HRQL assessment; more than one-quarter had any inpatient days. Adjusted models indicated that children with poor/fair global HRQL had 3.1 times (p \ 0.001) more total healthcare contact days than those with very good/excellent global HRQL. Children with poor/fair global HRQL had 8.8 times (p \ 0.001) more inpatient days than those with very good/excellent global HRQL. Similarly, children with poor/fair general health had 2.7 times (p \ 0.001) more total healthcare contact days and 8.3 times (p \ 0.001) more inpatient days than children with very good/excellent global HRQL. CONCLUSIONS: Parent-proxy ratings of their child’s HRQL were associated with future total healthcare contact days and inpatient days. Future studies should investigate whether HRQL may help to identify children and families who would be appropriate targets for earlier intervention, improving outcomes, patient and family experience, and shifting care from more to less costly services.

101 Diabetes and Kidney (1040) How does it feel to donate a kidney? Kathe Meyer, Oslo University Hospital, University of Oslo, Oslo, Norway; Anders Hartmann, MD, Oslo University Hospital, University of Oslo, Oslo, Norway; Carolyn Schwartz, ScD, Tufts University Medical School; DeltaQuest Foundation, Inc., Oslo and Akershus University College,, Concord, MA, United States; Marit H. Andersen, PhD, Oslo University Hospital, University of Oslo, Oslo, Norway AIMS: The number of kidney transplantations performed with live donors are increasing. Live donors are healthy persons, and even though donors’ good physical and mental health generally persists, other health aspects might be impaired. Post-donation outcomes have been associated with psychosocial aspects. However, little is known about how live donors describe their experiences 10 years after donation. We aimed to explore donor experiences in a longitudinal perspective. METHODS: We conducted a cross-sectional survey of 217 Norwegian kidney donors 10 years after donation. The questionnaire about self-reported health outcomes included four openended questions; (1) How has your life been affected by the donation? (2) Why would you choose not to donate in hindsight? (3) What has been important for your experience of health and quality of life as it relates to being a kidney donor? (4) What would your recommendations be for potential donors? We thoroughly read through the donor responses to obtain an overall impression. The next steps were to categorize the responses, and then synthesize the essence. RESULTS: (1) For most donors life was unaffected and the donation was described as a positive experience. Positive aspects cited were the recipients’ improved health and the impact on the recipient’s family. Negative aspects were changed relations within families because of jealousy or little support. Siblings denoted their obligation to donate. (2) Losing touch with the recipient after donation and negative recipient outcome were reasons for regretting donation. (3) Extensive information, thorough work-up and follow-up were reassuring. Poor information and lack of interest after donation were perceived as negative experiences. Medical problems like elevated blood pressure were considered a trifle when compared to the recipient’s improved health. (4) Most donors stressed that it was important not to be forced into donation. Both health personnel and previous donors were considered important in the decision phase. CONCLUSIONS: Donation was a positive experience because of the positive impact on other people’s lives. However, our results demonstrate the complexity of live donation. Negative experiences were poor information, lack of interest after the donation, and changes in relationships. The decision had to be taken by the donors themselves after careful deliberation.

(1042) Diabetes—literacy, knowledge, empowerment and quality of life Pedro L. Ferreira, University of Coimbra, Coimbra, Portugal; Carminda S. Morais, CEISUC/ESS-IPVC, Viana do Castelo, Portugal; Rui E. Pimenta, CEISUC/ESTSP-IPP, Vila Nova de Gaia, Portugal; Jose´ M. Boavida, Ministry of Health, Lisboa, Portugal AIMS: Diabetes is a silent disease with slow evolution to chronicity and has been experiencing an increasing prevalence worldwide. This study aimed at evaluating diabetes literacy and knowledge people with type 2 diabetes have about their disease, ability of self-control (empowerment) and quality of life (QoL). METHODS: The population was constituted by people with type 2 diabetes followed in ambulatory consultations, from a total of 954 elements. A questionnaire of socio-demographic and clinical characterization, and the Portuguese versions of Diabetes Empowerment Scale-Short Form (DES-SF), Diabetes Knowledge Test (DKT), and EuroQol EQ-5D

123

102 were administered to a random sample of 260 individuals with the following inclusion criteria: diagnosis of type 2 diabetes more than 1 year before, accompanied on multidisciplinary consultation for at least 3 months, and aged 18 years or more. RESULTS: The majority of respondents (53.8 %) was female, between 40 and 92 years of age, lived with someone (92.3 %), had as education the first grade (84.6 %) and was retired (55.2 %). The results showed that people feel able of self-management, with a mean ± SD score of 3.7 ± 0.7. DKT registered 63.4 ± 12.3 % of correct answers among diabetes noninsulin-treated individuals and 65.7 ± 12.4 % among the insulintreated (p \ 0.001). QoL presented a mean value of 0.65 ± 0.3. We evidenced a significant positive correlation between the ability to control, the knowledge and the QoL. CONCLUSIONS: This study showed that there is a relatively low level of knowledge, a higher perception of self-control and QoL in men, in the younger age groups and more educated ones. We also found that some risk factors are under-perceived vis-a`-vis the measurements taken. The conclusions obtained may help promote QoL and literacy enhancement, as well as the empowerment of individuals with type 2 diabetes. In this sense, it is important that the intervention be adjusted to individuals but also to small groups, bearing in mind a variety of equity-promoting strategies that differences in literacy, per se, already entail. There is thus a need for an intervention process of recognized complexity and of collective responsibility that must therefore be built with people, with the professionals, with community leaders and with the political power.

(1044) Preference for pharmaceutical formulation and treatment process attributes Katie D. Stewart, MA, Evidera, Bethesda, MD, United States; Louis S. Matza, PhD, Evidera, Bethesda, MD, United States; Joseph A. Johnston, MD, MSc, Eli Lilly and Company, Indianapolis, IN, United States; Sarah Curtis, Eli Lilly and Company, Indianapolis, IN, United States; Heather Gelhorn, PhD, Evidera, Bethesda, MD, United States; Henry Havel, Eli Lilly and Company, Indianapolis, IN, United States; Stephanie Sweetana, Eli Lilly and Company, Indianapolis, IN, United States AIMS: Pharmaceutical formulation and treatment process attributes, such as dose frequency and route of administration, can have a meaningful impact on quality of life, treatment adherence, and ultimately disease outcomes. The aim of this literature review was to identify and summarize studies quantifying the impact of pharmaceutical formulations and treatment process on treatment preference in patients with diabetes, oncology, osteoporosis, and autoimmune disorders. METHODS: The literature search focused on identifying published studies reporting preferences for pharmaceutical formulation and treatment process. To meet inclusion criteria, studies were required to use formal quantitative preference assessment methods, such as health state utility valuation, conjoint analysis including discrete choice experiments, willingness to pay methods, or contingent valuation. Searches were conducted using Medline, EMBASE, Cochrane Library, Health Economic Evaluation Database, and National Health Service Economic Evaluation Database (January 1993–October 2013). RESULTS: A total of 40 studies met inclusion criteria: seventeen diabetes, nine oncology, five osteoporosis, and nine autoimmune. Across disease states, treatments associated with shorter treatment duration, less frequent administration, greater flexibility, and less invasive routes of administration were preferred over more burdensome or complex treatments. Relative importance of treatment process was highly dependent on the nature of the disease condition (e.g., acute vs. chronic) and patient characteristics (e.g., injection experience), and willingness to pay often varied by geography. In general, efficacy and safety were assigned greater relative importance than formulation and treatment process attributes;

123

Qual Life Res (2015) 24:1–191 however, in a few studies, particularly in diabetes and autoimmune, certain formulation and process attributes were found to be equally or more important. For example, in one study patients with moderate or severe psoriasis assigned a higher relative importance to treatment location than probability of benefit or adverse events. CONCLUSIONS: Summarizing findings of multiple studies is challenging given the heterogeneity in study methodology and design. Nevertheless, in addition to efficacy and safety, pharmaceutical formulation and treatment process have a quantifiable and meaningful impact on preference and willingness to pay for treatment. Understanding patients’ preferences for treatment attributes, including formulation and treatment process, can inform drug formulation development decisions and lead to better patient satisfaction and improved outcomes.

(1048) Evaluating the reliability and validity of the ANMS gastroparesis cardinal symptom index-daily diary in patients with diabetic and idiopathic gastroparesis Dennis Revicki, PhD, Evidera, Bethesda, MD, United States; Henry Parkman, MD, Temple University School of Medicine, Philadelphia, PA, United States AIMS: Patient-reported scales that capture gastroparesis-related symptom severity are necessary for evaluating treatments for gastroparesis. The ANMS Gastroparesis Cardinal Symptom Index-Daily Diary (GCSI-DD) was developed to assess daily symptoms in patients with gastroparesis. Most clinical trials comparing treatments for gastroparesis have been performed in diabetic gastroparesis. We evaluated the psychometric characteristics of the ANMS GCSI-DD in patients with idiopathic (IG) and diabetic gastroparesis (DG). METHODS: Three datasets of patients with diabetic or idiopathic gastroparesis were analyzed in this study: Revicki et al. (2012; n = 69), Phase 2 study (n = 79), and Parkman et al. (2013; n = 18). Gastroparesis diagnosis was based on compatible symptoms and delayed gastric emptying based on scintigraphy. Patients completed the GCSI-DD each day for 14–28 days, assessing 5 symptoms: severity of nausea, early satiety, postprandial fullness, and upper abdominal pain, and number of vomiting episodes. We conducted confirmatory factor analysis (CFA), and evaluated internal consistency, test–retest reliability, convergent validity, known groups validity, and responsiveness. RESULTS: 166 patients (66 IG, 99 DG, mean age 47 ± 15 years, 72 % women) completed the GCSI-DD. The CFA demonstrated a single factor model (CFI = 0.95, factor loadings 0.56–0.86). Cronbach’s alpha was 0.81, and test–retest reliability was 0.89–0.96. ANMS GCSI-DD scores were significantly correlated with overall gastroparesis severity, upper abdominal pain (r = 0.70), nausea/vomiting (r = 0.63), and fullness/early satiety scores (r = 0.65), and clinician-rated severity (r = 0.46). Mean ANMS GCSI-DD scores varied in relation to patient-rated and clinician-rated overall symptom severity (all p \ 0.0005). Responsiveness was seen related to clinician and patient-rated severity improvement groups (p \ 0.001), with responders reporting a 0.50 point improvement. CONCLUSIONS: The results of this psychometric analysis provides evidence to support the unidimensionality, reliability, validity, and responsiveness of the ANMS GCSI-DD to assess gastroparesis symptoms in either idiopathic or diabetic gastroparesis.

(1050) Mediating role of psychosocial support between mental health status and quality of life of type 2 diabetic patients Ranjan Pattnaik, Doctoral Research Fellow, Indian Institute of Technology Kharagpur, Kharagpur, West Bengal, India; Rabindra K.

Qual Life Res (2015) 24:1–191 Pradhan, Associate Professor, Indian Institute of Technology Kharagpur, Kharagpur, West Bengal, India AIMS: Diabetes in India is perceived as a social burden and a state of personal suffering among diabetic patients. This in turn results in a disturbed state of psychosocial well-being, where psychosocial support mechanism fails in maintaining their mental health and Quality of Life (QoL). Keeping this in view, the present examines the role of the mental health status in the QoL of Type-2 diabetic patients in India. The study further examines the mediating role of psychosocial support between mental health status and QoL. METHODS: A crosssectional study was conducted on a random sample of 170 Type-2 diabetic patients who have no history of psychological illness and attending a specialized diabetes care hospital in Eastern India. Mental Health status was measured using General Health Questionnaire (GHQ-28), psychosocial support was measured by using multidimensional scale of perceived social support assessment scale and Quality of Life was assessed by using Quality of Life Instrument for Indian Diabetes patients (QOLID). Data were analyzed by the help of SPSS (version 20.0). RESULTS: Results reveal that mental health status significantly influences QoL. Psychosocial support significantly mediated the relationship between mental health status and QoL. CONCLUSIONS: The theoretical and practical implications of the study are discussed in detail in the context of diabetes management.

(1052) Relationship between physical activity and health-related quality of life in Alberta’s type 2 diabetes population Danielle Thiel, University of Alberta, Edmonton, AB, Canada; Fatima Al Sayah, PhD, University of Alberta, Edmonton, AB, Canada; Steven Johnson, PhD, Athabasca University, Edmonton, AB, Canada; Jeffrey A. Johnson, PhD, University of Alberta, Edmonton, AB, Canada AIMS: To examine the association between meeting guidelines for physical activity (PA) and health-related quality of life (HRQL) in individuals with type 2 diabetes. METHODS: Baseline data from the Alberta’s Caring for Diabetes cohort was used in this analysis. Data was collected using self-report questionnaires distributed to type 2 diabetes patients over the age of 18 living in Alberta, Canada. PA was measured using a modified version of the Godin-Shephard LeisureTime Physical Activity Questionnaire. Participants were classified as ‘‘meeting guidelines’’ if C150 min of moderate-vigorous PA per week was reported, and ‘‘insufficiently active’’ otherwise. HRQL was measured using the SF-12 version 2 physical and mental component summaries (PCS/MCS) and the EQ-5D-5L. A multivariable linear regression model adjusted for age, income, number of comorbidities, diabetes duration, smoking status and depressive symptoms was used for this analysis, which was stratified by sex. RESULTS: Data were available for 2027 respondents; the average age was 64.4 (SD 10.7) years, 54 % were male, and the average diabetes duration was 12 (SD 10) years. The majority of respondents were classified as inactive (males: 76 %, females: 82 %). Mean (SD) HRQL scores were 0.80 (0.15), 44.4 (10.8) and 48.1 (9.9) for EQ-5D index score, PCS and MCS, respectively. Meeting guidelines for PA was associated with a 3.51 point higher mean adjusted PCS score for males (p \ 0.001) and a 3.98 point difference in mean PCS score for females who meet guidelines compared with those who don’t (p \ 0.001). In females, meeting PA guidelines was associated with a 2.69 point higher mean MCS score (p \ 0.001); there was no significant association in males. Meeting PA guidelines was associated with a 0.034 point higher EQ-5D-5L index score for males (p = 0.002), but the difference (0.026) was not statistically significant for females (p = 0.086). CONCLUSIONS: Participants in this study reported poor rates of adherence to PA guidelines and slightly diminished HRQL, consistent with previous research in type 2

103 diabetes patients. While recognizing the limitations of cross-sectional designs and potential reverse causality, we observed significant small to moderate positive associations of PA with physical health in both men and women, and similar associations with mental health in women only.

(1054) Health literacy and health-related quality of life in adults with type 2 diabetes: a longitudinal study Fatima Al Sayah, PhD, University of Alberta, Edmonton, AB, Canada; Weiyu Qiu, MS, University of Alberta, Edmonton, Canada; Jeffrey A. Johnson, PhD, University of Alberta, Edmonton, AB, Canada AIMS: To examine the association of health literacy (HL) with changes in health-related quality of life (HRQL) among patients with type 2 diabetes (T2D). METHODS: Data from a cohort study of T2D patients in Alberta, Canada, were used. HL was assessed using 3 questions (scored 1–5), with cut-off C9.0 for the summative score indicating inadequate-HL. HRQL was measured using the EQ-5D-5L and SF-12-version2 (physical and mental composite summary scores (PCS, MCS)). The associations of baseline HL with the changes in EQ-5D index score, PCS, MCS over one year, and the direction of changes (no change; declined; improved) were examined using adjusted linear regression and multinomial logistic regression respectively. Multiple imputation (MI) was employed along with sensitivity analyses. RESULTS: Average age of participants (N = 1948) was 64.5 (SD = 10.7) years, 45 % were female, and 12.5 % had inadequate-HL. After using MI for missing data in both years, participants had mean decrements of 0.01 in EQ-5D, 1.0 in PCS, and 1.2 in MCS over 1 year. Participants with adequate or inadequate-HL had similar mean changes in EQ-5D and PCS; however, those with inadequate-HL had significantly greater decrements in MCS. Compared to those with adequate-HL, a higher proportion of inadequate-HL patients had a clinically important decline in both PCS (24 vs 32 %) and MCS (26 vs 37 %), and a smaller proportion had improvements in MCS (19 vs 15 %). In adjusted analysis (Table 1), HL was not associated with changes in EQ-5D over one year (beta = 0.01, p = 0.146); however, patients with adequate-HL had 2.1 points greater increase in PCS (p \ 0.001), and 3.1 points greater increase in MCS (p \ 0.001) compared to those with inadequate-HL. Patients with adequate-HL were less likely to have a decline in EQ5D (RRR = 0.66; 95 % CI 0.44, 0.98), PCS (0.51; 0.34, 0.76) and

Table 1 Beta (b) coefficients from multivariable linear regression and relative risk ratios (RRR) from multinomial logistic regression of the association of HL (reference: inadequate) with the changes in HRQL and the direction of changes over one year (base-case analysis and sensitivity analyses)

123

104 MCS (0.49; 0.33, 0.72) compared to those with inadequate-HL. Patients with adequate-HL were more likely to have an improvement in MCS compared to those with inadequate-HL (1.78; 1.04, 3.04); such associations were not observed for PCS or EQ-5D. These results were robust for all outcomes in sensitivity analyses. CONCLUSIONS: Inadequate-HL was independently associated with worsening HRQL in adults with T2D, particularly in the mental health domain.

(1056) Comparison of health-related quality of life between hemodialysis and peritoneal dialysis patients in Jeddah, Saudi Arabia Abdulhameed A. Alharbi, MD, MPH, JBCM, Bader Health District, Almadinah Almunawarah, Saudi Arabia; Rajaa M. Alraddadi, MD, ABCM, RICR, Public Health, Jeddah, Saudi Arabia; Alwaleed A. Alharbi, MBBS, King Abdulaziz University, Jeddah, Saudi Arabia; Mohammad S. Alghamdi, MD, KFUF(famco), Center of Higher Education, Family and Community Medicine, Jeddah, Saudi Arabia; Yazeed A. Alharbi, King Abdulaziz University, Jeddah, Saudi Arabia; Adel M. Ibrahim, MD, MDPH, Public Health, Jeddah, Saudi Arabia AIMS: There are two major dialysis modalities, hemodialysis (HD) and peritoneal dialysis (PD) which, differ both technically and functionally. Each dialysis modality has its advantages and disadvantages and has a different degree of impact on patients’ health. Health-related quality of life (HRQoL) has become an important outcome indicator in research on patients with chronic illnesses. This study aims to compare the HRQoL and related factors between HD and PD patients. METHODS: This was a cross-sectional comparative study conducted in five dialysis centers. A 342 stable dialysis patients (HD, n = 267; PD, n = 75), completed a questionnaire containing demographic questions, dialysis duration, and the 36-item Short Form Health Survey (SF-36). The data were analyzed using t-tests, ANOVAs, and multiple linear regression analysis. RESULTS: Of the 342 patients, 53.8 % were male and 46.2 % were female. Their mean age was 46.1 ± 16.5 years. Compared to the general population, all study participants showed lower scores on all subscales of quality of life; their lowest scores were in role limitations both physical (32.46 %) and emotional (39.28 %). However, over all HRQoL was significantly better among PD than among HD patients (p = 0.026). HRQoL was affected by factors such as gender, marital status, education level and duration of dialysis. Among HD patients, HRQoL was predicted by gender (beta = -0.155, p = 0.018) and age in years (beta = -0.145, p = 0.027), explaining 4.7 % of variance; while among PD patients, it was predicted by age in years (beta = -0.305, p = 0.014), explaining 9 % of variance. CONCLUSIONS: Role limitation both physical and emotional makes HD inferior to PD. Current study supports opinions that encourage greater reliance on PD.

Musculoskeletal (1060) Differential associations of pain treatment satisfaction and pain-specific psychosocial factors with pain adjustment outcomes among Chinese patients with chronic pain Wing S. Wong, Dr, The Hong Kong Institute of Education, Hong Kong, China; Richard Fielding, The University of Hong Kong, Hong Kong, China; Steven Wong, Queen Elizabeth Hospital, Kowloon, Hong Kong, Hong Kong, China; Y F. Chow, Queen Elizabeth Hospital, Kowloon, Hong Kong, China; P P. Chen, MBBS, Alice Ho Miu Ling Nethersole Hospital, Tai Po, Hong Kong, China AIMS: Objectives. This study examined the differential associations of patient satisfaction, and pain-specific psychosocial factors with

123

Qual Life Res (2015) 24:1–191 three pain adjustment outcomes in a sample of Chinese patients with chronic pain. METHODS: Methods: A total of 178 patients with chronic pain were assessed at baseline, 3-, and 6-month following baseline using standardized questionnaire which comprised of the following measures: the Chronic Pain Grade (CPG), the Participant Compliance Reporting Scale (PCRS), the Pain Treatment Satisfaction Scale (PTSS), the depression subscale of the Hospital Anxiety and Depression Scale (HADS-D), the Pain Catastrophizing Scale (PCS), and the Tampa Scale for Kinesiophobia (TSK) respectively. Linear mixed effects models (LME) were fitted to assess associations between variables and to identify predictors of adjustment outcomes. RESULTS: Results: After adjusting for sociodemographic and pain factors, results LME analyses showed that pain-related fear (std b = 0.10, p \ 0.05), activity engagement (std b = -0.46, p \ 0.001), and satisfaction with medical care (std b = 0.13, p \ 0.001) were significantly associated with depression (Model 1). Higher activity engagement and higher dissatisfaction with current pain medication were associated with higher pain disability (Model 2). Five significant predictors emerged for pain intensity (Model 3) after controlling for sociodemographic and pain factors: pain-related fear (std b = 0.13, p \ 0.01), activity engagement (std b = -0.15, p \ 0.01), satisfaction with impact of current pain medication (std b = 0.15, p \ 0.001), satisfaction with side effects of medication (std b = 0.10, p \ 0.05), and satisfaction with efficacy (std b = 0.12, p \ 0.05). CONCLUSIONS: Conclusions: Our data suggest there may be interaction between psychosocial factors and pain treatment satisfaction in impacting adjustment outcomes among Chinese patients with chronic pain.

(1062) Sensitivity to change of the Oxford Hip Score and the Forgotten Joint Score in patients after total hip arthroplasty Johannes M. Giesinger, Department of Psychiatry and Psychotherapy, Medical University of Innsbruck, Innsbruck, Austria; Jonas Egeter, Medical University of Innsbruck, Innsbruck, Austria; Karlmeinrad Giesinger, Kantonsspital St. Gallen, St.Gallen, Switzerland; Hamish Simpson, Royal Infirmary of Edinburgh, Edinburgh, United Kingdom; Colin Howie, Royal Infirmary of Edinburgh, Edinburgh, United Kingdom; David F. Hamilton, Royal Infirmary of Edinburgh, Edinburgh, United Kingdom AIMS: To assess sensitivity to change over time and ceiling effects of the Oxford Hip Score (OHS) and the Forgotten Joint Score-12 (FJS12) in patients 6 and 12 months after primary total hip arthroplasty (THA). METHODS: Patients undergoing total hip arthroplasty in a 6 month period (January–June 2013) at a large orthopaedic teaching hospital were assessed 6- and 12-months post-surgery using the OHS and the FJS-12. Sensitivity to change was analysed by calculating effect sizes (Cohen’s d) for measured change. Ceiling effects were described with percentages of patients reaching the highest possible score on the questionnaires. RESULTS: Data from 234 patients were available for analysis (59.8 % female; mean age 67.2, SD 11.5). Mean OHS improved from 40.1 (SD 8.0) at 6 months to 40.9 (SD 7.3) at 12 months. Mean FJS-12 improved from 56.3 (SD 30.4) at 6 months to 61.2 (SD 29.3) at 12 months. Effect size of change was 0.10 for the OHS and 0.16 for the FJS-12. This means that detecting statistically significant change (in a t-test for dependent samples; alpha = 0.05; power = 0.80) between 6 and 12 months requires a sample size of n = 788 for the OHS and n = 309 for the FJS-12. At 12 months ceiling effect was 20.7 % for the OHS and 10.9 % for the FJS-12. 48.6 % (OHS) vs. 22.2 % (FJS-12) of patients scored within the top 10 % of the score ranges. CONCLUSIONS: Our findings suggest that the FJS-12 is more sensitive to change between 6 and 12 months after

Qual Life Res (2015) 24:1–191 total hip arthroplasty than the OHS. The measured ceiling effect for the OHS was twice that of the FJS-12. The difference in effect size of change results in substantial differences in required sample size if aiming to detect change between these two time points. This has important implications for powering clinical trails with PRO tools as the primary outcome.

(1064) Therapy adherence and quality of life of patients with rheumatoid arthritis Michael Koller, University Hospital Regensburg, Regensburg, Germany; Jens U. Ru¨ffer, German Fatigue Society, Cologne, Germany; Karolina Mu¨ller, University Hospital Regensburg, Regensburg, Germany; Florian Zeman, University Hospital Regensburg, Regensburg, Germany; Jens G. Kuipers, Rotes Kreuz Krankenhaus Bremen, Bremen, Germany AIMS: Adherence to the treatment of rheumatoid arthritis (RA) is an essential component for therapeutic success. The factors contributing to therapy adherence and its consequences are still under debate. The aim of this analysis was to investigate the relationship between therapy adherence and patients’ quality of life. METHODS: The survey included a representative, nationwide sample of German physicians specialized in RA and patients with RA. The physician questionnaire assessed disease activity (DAS28) and patients’ adherence to therapy. The patient questionnaire assessed quality of life (SF-12), health education literacy (i.e., the understanding and use of medical information, HELP), activities of daily living (MDHAQ), and fatigue (EORTC QLQ-FA13). The survey was anonymous, but corresponding patient and physician questionnaires could be linked through a common identification number. RESULTS: A total of 671 pairs of patient and physician questionnaires were analyzed. The mean age of the patients, of whom 73 % were women, was 60 years (SD = 12.00). At the time of assessment, 67.2 % of the patients showed little disease activity, 28.8 % medium disease activity, and 4.0 % high disease activity. Physicians rated patients’ adherence to therapy as very high in 50.2 % of the cases. This was taken as the cutoff for further analyses and patients with high versus low adherence were compared. Patients with high adherence reported higher physical (M = 42.08 vs. M = 41.10, p \ .02) and psychological (M = 49.66 vs. M = 46.21, p \ .001) quality of life, lower physical fatigue (M = 2.27 vs. M = 2.44, p \ .01), less problems to understand and apply medical information (M = 1.84 vs. M = 2.10, p \ .001), and less problems with activities of daily living (M = 0.69 vs. M = 0.79, p \ .05). CONCLUSIONS: This study showed that patient adherence to therapy (as rated by the treating physician) was associated with a higher level of quality of life. Despite being small, the effects were consistent across a number of different psychosocial domains. These findings and further analyses may help to detect patients who are at risk not to comply with treatment and eventually to optimize patient education and counseling.

(1066) Neuro-QOL scores and number of items required to complete a CAT varies by work status in persons with MS: implications for clinical implementations of the Neuro-QOL measures Deborah Miller, PhD, Cleveland Clinic, Cleveland, OH, United States; Lyla Mourany, Cleveland Clinic, Cleveland, Oman; Karon F. Cook, PhD, Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; David Victorson, Northwestern University, Evanston, IL, United States; Nicolas Thompson, Cleveland Clinic, Cleveland, OH, United States; Fan Youran, Cleveland Clinic, Cleveland, OH, United States;

105 Irene Katzan, Cleveland Clinic, Cleveland, OH, United States; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: Determine if persons with MS grouped by employment status varied in their T-scores on 4 Neuro-QOL domains (Upper Extremity Function (UEF), Lower Extremity Function (LEF), Fatigue Severity (FS), and Sleep Disturbance (SD) and to assess number of items required to reach a final score for each group. METHODS: Four Neuro-QOL item banks, UEF) (20 items), (LEF) (19 items), (FS) (19 items) and (SP) (8 items) were delivered by computer adaptive testing (CAT) to consecutive MS patients at a large US MS Clinic. Respondents were classified by work status categories, Working (W), Unemployed (U), Retired (R) Disabled/Unable for medical reasons to Work (D). T-scores and number of items to complete CATs were compared. CAT stop logic included a minimum of 4 items, a maximum of 12 items except for SP (8 items) and response precision of Standard Error = 0.3. We hypothesized that the greatest differences in T-scores and number of responses would be between W and D, with U approximate to W and R to D. RESULTS: Data were collected between 4/2014 and 12/2014, with 2012 respondents. Participants were predominately female (72 %), mean age = 44.62, with relapsing remitting MS (71 %). As hypothesized W T-scores were statistically significant different than the D T-scores scales and the U and R ranked as expected but not always significantly different. The number of completed items varied by working group with W requiring 9.74 items to complete UE and D required on 5.78 items. Results were similar for LE with the W requiring 7.32 items while D required on 4.43 items. There was a small but statistically significant difference for the FS with the W 4.48 items while DG required on 4.21 items. For the SI score, each of the working groups required all 8. CONCLUSIONS: T-score comparisons between working groups for each scale was as expected. The patterns revealed in the number of items needed to complete the UE, LE and FS scales indicates the items are focused on tasks less difficult for the W result in reduced respondent burden for those in the D. Response patterns for the Sleep scale suggest that adding additional items may be warranted.

(1070) Development of quality indicators for hip and knee arthroplasty rehabilitation Marie D. Westby, PT, PhD, University of Alberta, Edmonton, AB, Canada; Linda C. Li, PT, PhD, University of British Columbia, Vancouver, BC, Canada; Deborah Marshall, Associate Professor, University of Calgary, Calgary, AB, Canada; Jane Squire Howden, BScN, Edmonton Musculoskeletal Centre, Edmonton, AB, Canada; DWC (Bill) Johnston, MD, FRCS(C), University of Alberta, Edmonton, AB, Canada; Allyson Jones, PhD, University of Alberta, Edmonton, AB, Canada AIMS: This project aimed to develop and evaluate core sets of quality indicators (QIs) reflecting the minimum acceptable standard of rehabilitation care before and after elective total joint arthroplasty (TJA) for osteoarthritis. METHODS: We used a modified RAND-UCLA Delphi approach and online technology to assemble an 18-member Canadawide panel of clinicians, researchers and patients and establish consensus on 82 proposed QIs (40 for hip, 42 for knee). Panelists completed 2 rounds of rating using Fluid Survey interspersed by a moderated online discussion forum over a 3-month period. Prior to rating, panelists received the synthesized literature supporting each QI and were instructed to review this before accessing the survey. Each QI was rated for its importance and validity on a 9-point Likert scale. Individual and group ratings from Round 1 were summarized and provided to each panelist in order to inform the subsequent online discussion. During Round 2, panelists were asked to participate in the forum anonymously,

123

106 share their views on the proposed QIs and debate each other’s comments. This was immediately followed by the final round of ratings. Those QIs with median ratings of 7 or higher for both importance and validity, and with no disagreement were included in the final sets. RESULTS: Fifteen panelists representing 7 provinces and varied practice settings completed the Delphi process. Of the 82 proposed QIs, 67 (82 %) were rated as both important and valid (31 for hip, 36 for knee). For the hip, 14 pre-op, 6 acute and 8 post-acute QIs were accepted. For the knee, 16 pre-op, 10 acute and 8 post-acute indicators were accepted. Two of 3 across-continuum QIs were rated appropriate for both joints. Engagement was high with 83 % of panelists participating in the discussion forum. CONCLUSIONS: Online technology facilitated broad geographical representation, panelist anonymity and engagement, and decreased panelist burden and administration costs compared to the traditional RAND/UCLA method. However, time to develop and administer the online survey, moderate the discussion forum, monitor panelist engagement and send timely reminders remained significant. The QIs will be further tested for reliability and feasibility before being widely disseminated in clinical settings and used to assess care gaps.

(1072) Path to participation in people with multiple sclerosis Shahnaz Shahrbanian, PhD, McGill University/Buali Sina University, Hamedan, Montreal, QC, Canada; Pierre Duquette, University of Montreal, Montreal, QC, Canada; Sara Ahmed, BSc MSc PhD, McGill University, Montreal, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada AIMS: Multiple sclerosis (MS) is the most common disabling neurological disease among young adults in North America. Participation and social activities are key domains affecting quality of life (QOL). Symptoms such as fatigue, pain, and impairments of physical and mental capacity, can make participation in life’s roles a challenge for people with MS. However, how factors that predict participation of people with MS, interact with each other to influence participation is still unknown. The objective of this study, therefore, was to contribute evidence as to how the different MS impairments, psychological variables, activity limitations, and individual characteristics interact to affect participation. METHODS: This was a cross sectional study. A centre-stratified random sample of 188 patients registered at the 3 MS clinics in Montreal was drawn. Participants completed a battery of self-report and performance-based measures that assessed participation and domains affecting participation. In order to understand the relationships between study variables a theoretical conceptual framework based on the Wilson & Cleary Model that posits specific relations between different levels of the health outcomes was tested. Structural Equation Model was considered the most appropriate option of analysis aimed at identification of the direct and indirect predictors of participation within the hypothesized theoretical model. RESULTS: Overall, the results of fit indicators (Chi Square = 113.8, df = 94, p value = 0.08; RMSEA = .90, p value = 0.03; CFI = .98; TLI = .98; SRMR = 0.05) showed that the model fit the data adequately. The results suggested fatigue (b = 0.63, p = 0.00), physical function (b = 0.37, p = 0.00), and psychological variables (b = 0.15, p = 0.04) as significant predictors of participation. Pain (b = 0.4, p = 0.00) and age (b = 0.12, p = 0.00) were significant indirect predictors through fatigue and physical function, respectively. Together these effects explained 88 % of the variance of participation. CONCLUSIONS: This study provided support for fatigue, physical function, pain, and psychological variables as most important contributors for participation in persons with

123

Qual Life Res (2015) 24:1–191 MS. A clear understanding of participation repercussions and so application of appropriate intervention for their removing or reducing would maximize health related quality of life of people with MS.

(1074) Exploring the relationship between demographic and disease related variables and health-related quality of life in patients with axial spondyloarthritis Gudrun Rohde, Faculty of Health and Sport Sciences/Department of Clinical Research, University of Agder and Sorlandet Hospital, Kristiansand, Norway; Kari Hansen Berg, RN, PhD-student, University of Agder, Grimstad, Norway; Anne Proven, Martina Hansens Hospital, Bærum, Norway; Glenn Haugeberg, MD, PhD, Sorlandet Hospital and Norwegian University of Science and Technology, Kristiansand, Norway AIMS: Axial Spondyloarthritis (ax-SpA) is a chronic disorder with onset of back pain in early adulthood and do frequently also cause fatigue, stiffness and loss in physical function. The psychological burden of having ax-SpA may therefore impact health quality of life (HRQOL). The aim of the study was to explore the relationship between demographic and disease related variables and HRQOL in patients with axSpA including patients with both ankylosing spondylitis (AS) and nonradiographic ax-SpA. METHODS: Demographic, disease and HRQOL data were collected. Disease measures included Bath Ankylosing Spondylitis Disease Activity Index (BASDAI), BAS functional index (BASFI), BAS General (BAS-G), Maastricht Ankylosing Spondylitis Enthesitis Score (MASES), health assessment questionnaire (HAQ) and co-morbidity. HRQOL was assessed using; SF-36 (range 0-100) and the utility measures SF-6D and 15D (range 0-1). Univariate and multiple linear regression analyses (GLM) were applied. RESULTS: We examined 380 patients, 127 women (mean age = 46 years, SD = 12) and 253 men (mean age = 45 years, SD = 12) with Ax-SpA. Seventy-five percent of the patients were married/co-habiting and 73 % were working. In adjusted analyses low SF-36 Physical Component Summary was associated with exercise\1 h.pr.week (B = -2.72 (95 % CI -4.8, -0.58), p = 0.013), high HAQ (B = -4.61 (95 % CI -7.61 to –1.93), p = 0.001), high BASFI (B = -1.05 (95 % CI -1.90 to -0.25), p = 0.010), and high BAS-G (B = -0.91 (95 % CI -1.44 to -0.37), p = 0.001). Low SF-36 Mental Component Summary was associated with drinking less alcohol (B = 4.72 (95 % CI 0.77–8.42), p = 0.019), and high BAS-G (B = -1.75 (95 % CI -2.53 to -0.97), p \ 0.001). The same pattern persisted using the SF-6D measure. Finally, low 15D was associated with exercise \1 h.pr.week (B = -0.034 (95 % CI -0.054 to -0.014), p = 0.003), co-morbidity (B = -0.014 (95 % CI -0.023 to -0.005), p = 0.004), high HAQ (B = -0.050 (95 % CI -0.075 to -0.025), p \ 0.001), and high BAS-G (B = -0.010 (95 % CI -0.015 to -0.005), p \ 0.001). CONCLUSIONS: In patients with ax-SpA exercising less than 1 h per week, high HAQ (impaired physical function) and high BAS-G (impaired general disease wellbeing) were independently and consistently associated with decreased HRQOL across several HRQOL measures.

(1078) Reproducibility: reliability and agreement of short version of Western Ontario Rotator Cuff Index (Short-WORC) in patients with rotator cuff disorders Neha Dewan, PhD student, McMaster University, Hamilton, ON, Canada; Joy MacDermid, Professor, McMaster University, Hamilton, ON, Canada; Norma MacIntyre, Associate professor, McMaster

Qual Life Res (2015) 24:1–191 Table 1

107 Table 5 Reproducibility statistics: agreement parameters of Short-WORC and WORC (N = 43)

Patient baseline characteristics (N = 153)

Variable

N/percentage

Age in years (mean ± SD)

(56.78±9.97) 153/100 %

Sex

92/60 %

Measure

Males Females

61/40 %

Hand dominance

19/12 %

Right

MDC90individual

MDC90group

Short-WORC

8.8 (7.3, 11.1)

20.4 (18.3, -22.4)

5.1

WORC

7.7 (6.4, 9.8)

17.8 (16.5, -19.1)

4.5

Short-WORC, Short version of Western Ontario Rotator Cuff Index WORC, Western Ontario Rotator Cuff Index CI, Confidence Interval SD, Standard Deviation ICC, Intra class Correlation Coefficient MSE, Mean Square Error SEM, Standard Error Measurement d, Mean Difference MDC, Minimal Detectable Change n, sample size

Left 134/88 %

Affected Shoulder

SEMagreement (95 % CI)

70/46 %

Left Right

83/54 %

SD standard deviation

Table 2

Floor and ceiling effects for test-retest scores of Short-WORC and WORC (N = 43) Test ShortWORC

Retest WORC

ShortWORC

WORC

Minimal scores

6/43 = 14 %

5/43 = 12 %

6/43 = 14 %

3/43 = 7 %

Maximal scores

4/43 = 9 %

4/43 = 9 %

3/43 = 7 %

3/43 = 7 %

SHORTWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index

Table 3

Reproducibility Statistics: cross sectional reliability of Short-WORC and WORC

Measure

Internal Consistency (Cronbach’s alpha) Baseline (n = 153) Cronbach’s alpha (95 %CI)

Month 3 post-operatively (n = 147) Inter-item correlation range

Cronbach’s alpha (95 %CI)

Inter-item correlation range

Short-WORC

0.84 (0.79, 0.87)

0.28–0.64

0.89 (0.85, 0.91)

0.43–0.71

WORC

0.90 (0.88, 0.93)

0.04–0.66

0.95 (0.94, 0.96)

0.20–0.78

Short-WORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, CI Confidence Interval

University, Hamilton, ON, Canada; Ruby Grewal, Associate Professor, Western University, Roth | McFarlane Clinical Research Laboratory, Hand and Upper Limb Centre, London, ON, Canada

AIMS: Recently, a shorter version of Western Ontario Rotator Cuff Index (Short-WORC) was proposed as a subset of 7 items from the original 21-item WORC. However, the reproducibility of the ShortWORC has not been established. Thus, the purpose of our study was to determine reproducibility (reliability and agreement) of ShortWORC among patients with rotator cuff pathology. METHODS: Patients (n = 153) diagnosed with RCD completed the WORC at baseline and at 3 months post-operatively (n = 146). The ShortWORC was extracted from the full version of WORC. From this retrospective cohort, 43 patients were retested within 5 weeks, if they remained stable. Cronbach’s alpha (a) and intra class correlation coefficients (ICC2,1) were used to assess internal consistency and test–retest reliability respectively. Standard error measurement (SEM), minimal detectable change (MDC90) and Bland–Altman (BA) plots were used to assess agreement. RESULTS: No floor and ceiling effects were reported for either the Short-WORC or WORC. Cronbach’s a were 0.84 and 0.90 at baseline and 0.89 and 0.95 at 3 month of follow up for ShortWORC and WORC respectively (Tables 1, 2, 3, 4, 5). The ICC2,1 were 0.89 and 0.91 for the Short-WORC and WORC respectively. The agreement parameters for the Short-WORC were: SEMagreement = 8.8, MDC90individual = 20.3, MDC90group = 5.1. We found substantial agreement between the two versions of WORC on BA plots within minimal (mean difference (d) \1) systematic differences between them (Figs. 1, 2, 3). The LOA between two versions of WORC were similar across sessions and fell within range of -11.7 to 13.25 points at test and -14.68 to 14.68 points at retest. CONCLUSIONS: Short-WORC and WORC demonstrates strong reproducibility and can be used for group and individual comparison of HRQoL among patients with RCD. Wider limits of agreement may be expected when using the ShortWORC for individual patient assessment. Reproducibility data is

Table 4 Reproducibility statistics: longitudinal reliability of Short-WORC and WORC (N = 43) Variable

Test retest reliability Test mean (SD)

Retest mean (SD)

d (SD)

95 % CI

(95 % LOA)

Linear relationship (Pearson r)

ICC2,1 (95 %CI)

Non-transformed data

Transformed data

Non-transformed data

Transformed data

Short-WORC

38.1 (26.1)

40.1 (27.6)

-2.1 (12.4)

-5.9, 1.7

22.3, -26.5

0.89** (0.82,0.93)

0.89** (0.82,0.93)

.89** (.81,.94)

.89 ** (.80,.94)

WORC

38.8 (25.3)

41.1 (25.8)

-1.3 (10.9)

-4.6, 2.1

20.1, -22.7

0.91** (0.85,0.95)

0.92 ** (0.87,0.95)

.91** (.84,.95)

.92** (.85,.95)

Short-WORC, Short version of Western Ontario Rotator Cuff Index WORC, Western Ontario Rotator Cuff Index d, Mean difference (test retest) SD, Standard Deviation CI, Confidence Interval LOA, Limits of Agreement ICC, Intraclass Correlation Coefficient ** All correlation coefficient were statistically significant at p \ 0.001 (1-tailed)

123

108

Fig. 1 Bland–Altman Limits of Agreement (LOA) Plot between test and retest of Short-WORC and WORC (n = 44; with outlier): The difference between test retest scores plotted against the mean of test and retest scores for Short-WORC and WORC. On each plot, the central line represents the mean of intra individual differences (d) and two horizontal lines represent 95 % LOA. The 95 % LOA shows that 95 % of the intra individual differences are expected to lie between d ± 1.96 SD of mean difference

123

Qual Life Res (2015) 24:1–191

Fig. 2 Bland–Altman Limits of Agreement (LOA) Plot between test and retest of Short-WORC and WORC (n = 43; without outlier): The difference between test retest scores plotted against the mean of test and retest scores for Short-WORC and WORC. On each plot, the central line represents the mean of intra individual differences (d) and two horizontal lines represent 95 % LOA. The 95 % LOA shows that 95 % of the intra individual differences are expected to lie between d ± 1.96 SD of mean difference

Qual Life Res (2015) 24:1–191

109 Logan Trenaman, MSc, University of British Columbia, Vancouver, BC, Canada; Stirling Bryan, PhD, University of British Columbia, Vancouver, BC, Canada; Jeffrey Johnson, PhD, School of Public Health, University of Alberta, Edmonton, AB, Canada AIMS: With routine measurement of patient-reported outcome measures (PROMs) becoming commonplace, we sought to understand how this data could be used to improve decision-making at the individual patient level. METHODS: We conducted 2 focus groups in 15 individuals over the age of 55. We first provided participants with an existing decision aid for patients with osteoarthritis considering knee arthroplasty. It provided information on the average risks and benefits of surgery, including the probability of problems with surgery, post op satisfaction, and mortality. Previous research suggests patients often have incomplete information about the benefits and risks of knee replacement surgery and their alternative options, and when they become more informed a proportion choose to delay or avoid surgery. We next provided participants with interactive results from the UK PROMs initiative which visually described changes in EQ-5D dimensions for similar patients (based on sex, age group, and pre-surgical EQ-5D scores) who had underwent knee arthroplasty. For example, these data enabled us to describe of the 3007 males between 60 and 70 who had moderate pain prior to surgery, 1135 (38 %) improved with no pain 6 months post-surgery, 1768 (59 %) still had moderate pain, while 104 (3 %) worsened to have severe pain. RESULTS: Without prompting, both groups expressed concerns that the statistics were not relevant to their personal context (e.g. biased by elderly patients). Groups also sought information on the influence of surgery on daily living including mobility and self-care. When individualized PROMs data were provided, participants understood it, preferred it to the conventional statistics, and suggested it might change their decision. Later discussion involved topics on: trust in the data—where it was from and how recent, a desire for longer-term data beyond 6 months, and PROMs for similar patients who underwent alternative options to surgery. CONCLUSIONS: This formative and developmental research helps understand the potential use for PROMs data in informing patients about important medical decisions. While PROMs collection to date has predominantly had quality/safety and policy goals, we suggest the design of PROM data collection should consider its role in informing patients, having longer-term follow up when relevant, and including patients who opt for alternative options.

Fig. 3 Bland–Altman Limits of Agreement (LOA) Plot between Short-WORC and WORC (n = 43; without outlier): The difference between scores on the SHORTWORC and WORC plotted against the mean of sum scores of Short-WORC and WORC for the test and retest sessions. On each plot, the central line represents the intra individual mean differences (d) between Short-WORC and WORC and two horizontal lines represent 95 % LOA. The 95 % LOA shows that 95 % of the intra individual differences are expected to lie between d ± 1.96 SD of mean difference essential, but should be supplemented by validation in samples representing the spectrum of RCD. Key words: Rotator cuff tear, reliability, Western Ontario Rotator Cuff Index, quality of life, shoulder pain, minimal detectable change

(1080) Using routine Patient Reported Outcome Measures to enhance patient decision making: a proof of concept study Nick Bansback, PhD, University of British Columbia and Centre for Health Evaluation and Outcomes Sciences, Vancouver, BC, Canada;

(1084) Linking of the American academy of orthopaedic surgeons distal radius fracture clinical practice guidelines to the international classification of functioning disability and health (ICF) and ICF core sets for hand conditions Saravanan Esakki, PT, McMaster University, Hamilton, ON, Canada; Saipriya Vajravelu, PT, McMaster University, Hamilton, ON, Canada; Joy MacDermid, Professor, McMaster University, Hamilton, ON, Canada AIMS: To analyse and describe the scope and focus of the American Academy of Orthopaedic Surgeons (AAOS) distal radius fracture (DRF) clinical practice guidelines (CPG) using the International Classification of Functioning, Disability, and Health (ICF) and International Classification of Disease (ICD-10) as a basis for content analysis, and to compare the content of the CPG to the ICF Hand Core Sets as reference standard. METHODS: AAOS DRF CPG is readily available to clinicians, patients and policy-makers. ICF provides a framework for describing the impact of health conditions and ICD-10 is a classification system to classify health conditions as specific disease or disorders. ICF Hand (Brief and Comprehensive) Core Sets represent an accepted reference standard for the key health impacts for hand conditions. Established linking rules were used by 2 independent raters to analyse the 29

123

110 recommendations of the AAOS DRF CPG. Summary linkage statistics were used to describe the results to ICF and the Hand Core Sets RESULTS: Among the 29 recommendations of the AAOS DRF CPG, five meaningful concepts were linked to the ICF focusing on impairments. Of these 5 codes appeared on the Comprehensive ICF Core Set and only 3 codes appeared in the Brief ICF Core Set. And 7 conditions were covered in ICD-10 codes. CONCLUSIONS: The AAOS DRF CPG focuses on surgical interventions and has minimal linkage to the constructs of the ICF and ICD-10. It does not address activity or participation (disability), and is not well linked to key concepts relevant to hand conditions. While appropriate for surgeons, the AAOS DRF CPG does not take a rehabilitative perspective. A CPG specific to rehabilitation of DRF is needed.

Mental Health (1086) The relationship between body mass index, psychological distress and vitality: a population-based study exploring mediators and gender differences ˚ s, Norway, Neha Agnihotri, Norwegian University of Life Science, A Jocelyne Clench-Aas, Senior Scientist, Norwegian Institute of Public Health, Oslo, Norway; Grete G. Patil, Associate Professor, Norwegian ˚ s, Norway; Ragnhild B. Nes, Research University of Life Sciences, A Scientist, Norwegian Institute of Public Health, Oslo, Norway AIMS: Although mental health problems among those with a nontypical body weight are being increasingly studied, the role of positive mental health has been less investigated. This study aimed to i) examine associations between six standardized categories of body mass index (BMI) and both negative (psychological distress) and positive (vitality) indicators of mental health, and to ii) explore to which extent lifestyle, sense of mastery and social support mediated these relationships. METHODS: The study was based on self-report cross-sectional data from The Level of Living Surveys, conducted in Norway in 2008 and 2012. The study sample included 7909 men and women (18–79 years), and was weighted to account for differences in response rates. Analysis included logistic regression (using the complex sample module) and mediation analyses using the PROCESS-macro for SPSS. Psychological distress was assessed by the Mental Health Inventory-5 (MHI-5) and vitality by the vitality subscale of the SF-36 using standard cut-points (MHI-5: \56; Vitality: \62). RESULTS: Only underweight (BMI \18.5 kg/m2) women reported high levels of psychological distress as compared to normalweighted women (OR 2.59, P \ 0.001). Low vitality was associated with all categories of BMI in men and women alike, apart from among overweight (BMI [25.0–29.9 kg/m2) women. Physical activity significantly mediated the association between BMI and psychological distress for men, whereas social support mediated the relationship for women. For vitality, physical activity mediated the association for both genders, and vegetable consumption had a small mediating effect for men. Sense of mastery was not a significant mediator between BMI and either psychological distress or vitality. The overall findings concerning mediation were modest. CONCLUSIONS: The findings indicate that BMI is differentially associated with positive and negative indicators and is mediated by partly different factors in men and women.

Qual Life Res (2015) 24:1–191 Institute; McGill University, Montre´al, QC, Canada; Paul Emmelkamp, Netherlands Institute for Advanced Study; The Center for Social and Humanities Research, Jeddah, Saudi Arabia, Wassenaar, Netherlands; Toshi A. Furukawa, Kyoto University Graduate School of Medicine/School of Public Health, Kyoto, Japan; Erik Hedman, Karolinska Institutet, Solna, Sweden; Maria Kangas, Centre for Emotional Health, Department of Psychology, Macquarie University, Sydney, Australia; Alain Lesage, Institut universitaire en sante´ mentale de Montre´al & Department of Psychiatry, University of Montreal, Montre´al, Canada; Doris M. Mwesigire, Makerere College of Health Sciences, Kampala, Uganda; Sandra Nolte, PhD, Charite Universitaetsmedizin Berlin, Berlin, Germany; Vikram Patel, London School of Hygiene & Tropical Medicine; Public Health Foundation of India, Sangath, Goa, India; Paul Pilkonis, PhD, University of Pittsburgh, Pittsburgh, PA, United States; Harold A. Pincus, Department of Psychiatry, Columbia University, New York, NY, United States; Roberta A. Reis, Universidade Federal do Rio Grande do Sul - Federal University of Rio Grande do Sul, Grupo CNPq - USP - GPEMSA - Sa˜o Paulo University/Health, Sao Paulo, Brazil; Graciela Rojas, Clinical Hospital, Department of Psychiatry, University of Chile, Santiago de Chile, Chile; Cathy Sherbourne, The RAND Corporation, Santa Monica, CA, United States; Caleb Stowell, International Consortium for Health Outcomes Measurement (ICHOM), Boston, MA, United States; Matthias Rose, MD PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany AIMS: Treatment of mental disorders such as depression and anxiety disorders are one of the most important health challenges today. If we are to meet this challenge we need agreed measures of severity of such disorders that are assessed before and after treatment. Recently, the International Consortium for Health Outcomes Measurement (ICHOM) was founded with the aim to propose consensus-based measurement tools and documentation for 50 different diseases. The aim of this project was to provide a core battery for the assessment of health outcomes for the treatment of patients with depression and/or anxiety disorders. METHODS: Patient advocates, outcome researchers and clinical experts from all continents were invited to a consensus-building exercise to develop an evidence-based assessment of health outcomes. The group followed a structured consensus-driven modified Delphi technique. RESULTS: The group recommends a core battery of patient reported outcome (PRO) measures including demographic factors, functional and clinical status, and diseasespecific symptom burden. Follow-up data on symptom reduction and treatment side effects are recommended to be collected over a period of 2 years for continued monitoring and comparability of treatment success. The annual outcome set contains broader assessment compared to the treatment monitoring set which is relatively short and allows for quick application in daily clinical routine. The total number of items ranges between 21and 47 (baseline set) items for depressive and general anxiety disorder. Due to the lack of availability of short instruments, assessment of specific anxiety disorders contains up to 82 items. Core recommended tools include the PHQ-9, GAD-7, and WHODAS 2.0 (Fig. 1). Symptom measures for specific anxiety disorders are also recommended. The group recommends scoring the main symptom tools (i.e. PHQ-9, GAD-7) on standardized metrics developed by the Patient-Reported Outcomes Measurement Information System (PROMIS) to increase comparability with other outcome measures (Table 1). CONCLUSIONS: An international

(1088) A proposal for standardization of health outcome measures for depression and anxiety Alexander Obbarius, MD, Charite´ - Universita¨tsmedizin Berlin, Berlin, Germany; David Clark, University of Oxford, Oxford, United Kingdom; Anne G. Crocker, Douglas Mental Health University

123

Fig. 1 Follow-up timeline for the depression and anxiety standard set

Qual Life Res (2015) 24:1–191 Table 1 Measures in the depression and anxiety standard set

111 greater QOL in the Social domain at the end of treatment (Pearson = 0.21, p\ 0.05). This correlation, however, did not persist after adjusting for demographic and clinical variables (Beta = 0.08, p = 0.35). CONCLUSIONS: Positive Religious Coping was predictor of a less significant improvement in the Antecipatory Anxiety symptom in this sample of PD patients. Future studies may clarify whether this relationship can be explained due to these patients have a greater severity of disease. Also some religious practices may have mobilized more anxiety. Finally, Religious Coping was not predictor of QOL in this sample.

(1092) Quality of life in patients with locked-in syndrome: evolution over a 6-year period

standard of health outcome assessment has the potential to improve clinical decision-making, enhance health care for the benefit of patients, and facilitate scientific knowledge.

(1090) Can religious coping predict clinical outcome and quality of life in patients with panic disorder? An open clinical trial Roge´rio R. Zimpel, MSc, PhD Student, Federal University of Rio Grande do Sul - Brazil, Porto Alegre, Rio Grande do Sul, Brazil; Raquel G. Panzini, PhD, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Denise R. Bandeira, PhD, Federal University of Rio Grande do Sul - Brazil, Porto Alegre, Brazil; Elizeth Heldt, PhD, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Gisele G. Manfro, PhD, Federal University of Rio Grande do Sul, Porto Alegre, Brazil; Marcelo P. Fleck, PhD, UFRGS, Porto Alegre, Brazil; Neusa S. Rocha, PhD, Federal University of Rio Grande do Sul, Porto Alegre, Brazil AIMS: Religiosity has a double-edged capacity to enhance or damage health and quality of life (QOL). Scales which assess the opposing aspects can help to understand how these two extremes occur. This study aims to evaluate if religious coping can predict clinical outcomes and better QOL in an open trial with patients diagnosed with Panic Disorder (PD). METHODS: A sample of 101 patients with PD was accompanied in treatment for a period of 12–16 weeks. The sample was divided into two groups: Individual Drug Treatment and Group Cognitive-Behavioral Treatment. Scales were applied at baseline and at the end of the treatments to evaluate Spirituality (WHOQOL-SRPB), Religious Coping (RCOPE Scale) as well as clinical evaluation (Panic Inventory, BDI, HAM-A and CGI) and QOL (WHOQOL-Bref). RESULTS: There was a general improvement in patients, regardless of treatment. When analyzing Religious Coping as a predictor and the PD as an outcome, higher scores of Positive Religious Coping at baseline presented inverse correlation (Spearman’s rho = -0.20, p \ 0.05) with an improvement in the Anticipatory Anxiety symptom at the end of the treatment. This correlation remained and even increased its value after adjustment in a Multiple Linear Regression (Beta = -0.31, p = 0.01). In this adjustment other variables which were also measured at baseline were included: demographic (marital status, education, use of religion as coping) and clinical variables (anxiety/panic symptoms, type of treatment and QOL). When QOL was analyzed as outcome, there was a correlation between Negative Religious Coping at baseline with

Marie-Christine Rousseau, Fe´de´ration des Hoˆpitaux de Polyhandicap et Multihandicap Hoˆpital San Salvadour, Assistance Publique Hoˆpitaux de Paris, Paris, France, Karine Baumstarck, EA 3279 SelfPerceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Marine Alessandrini, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Thierry Billette de Villemeur, Fe´de´ration des Hoˆpitaux de Polyhandicap et Multihandicap Hoˆpital San Salvadour, Assistance Publique Hoˆpitaux de Paris, Paris, France; Pascal Auquier, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France AIMS: Although mortality is high at the onset of the locked-in syndrome (LIS), life expectancy can be significantly improved using appropriate care when the patients become medically stable. Improved knowledge of quality of life (QoL), evolution of QoL over time, and middle- and long-term QoL predictors in LIS patients would have implications for managing their care and assist clinicians in choosing the most appropriate interventions. We performed a survey of a population of LIS patients to describe the course of LIS patients’ QoL over a 6-year period, and determine potential predictive factors of the QoL change over time. METHODS: This is a longitudinal study performed on a 6-year period among individuals with a LIS diagnosis according to Plum and Posner’s description. The patients were contacted by mail in 2007 and 2013. The following data were recorded: (i) Sociodemographic data; (ii) Clinical data related to the LIS (type, duration, etiology), physical/handicap status (gastrostomy, tracheotomy, urinary probe, pain, wheelchair, communication…), psychological status; (iii) Self-report QoL using the Anamnestic Comparative Self-Assessment (ACSA); (iv) Integration in the life using French Reintegration to Normal Living Index (RNLI). RESULTS: Among the 67 patients included in 2007, 39 (58 %) patients returned their questionnaire in 2013. The QoL of the patients was relatively satisfactory compared to populations in other severe conditions. The QoL did not statistically differ between the 2 periods (p [ 0.05). Twenty-one (70 %) individuals reported a stable (n = 10) or improved (n = 11) QoL between 2007 and 2013. The physical/handicap statuses of 2007 and 2013 did not be related to the 6-year QoL, except the communication way: the individuals reporting the use of the yes–no code compared to the non-users reported a statistical lower QoL in 2013. Patients with suicidal ideas reported a significant lower QoL level than others. CONCLUSIONS: In opposition to a general and widespread opinion, LIS persons report a relatively satisfactory QoL level that stays stable over time, suggesting that life with LIS is worth living. Preservation of autonomy and communication may probably help them to live as normal life as possible.

123

112

Qual Life Res (2015) 24:1–191

(1094) Patient-reported health related quality of life is strongly correlated to an objective index of comorbidity: results from a study of elderly U.S. patients Jesse Cambon, Humana, Louisville, KY, United States; Tristan Cordier, Humana, Atlanta, GA, United States; Lane Slabaugh, PharmD, MBA, Humana, Louisville, KY, United States; Gilbert Haugh, Humana, Louisville, KY, United States; Bill Glasheen, Humana, Louisville, KY, United States; Vipin Gopal, PhD, MBA, Humana, Louisville, KY, United States; Andrew Renda, MD, MPH, Humana, Louisville, KY, United States AIMS: The Charlson Comorbidity Index (CCI) is a widely used composite index of an individual’s comorbid conditions that helps predict hospitalization and mortality. Healthy Days is a validated four question quality of life survey developed by the U.S. Centers for Disease Control and Prevention that measures the number of days within the last 30 that an individual’s physical or mental health was not good. This investigation examines the correlation between CCI (both total score and component parts) with physically and mentally unhealthy days (PUHD/MUHD) in an elderly U.S. population with Medicare Advantage (MA) insurance. METHODS: A cross-sectional survey using a stratified random sample of individuals from a large national insurer was performed November 24–December 24, 2014. MA respondents aged [65 years with continuous medical plan coverage in the 365 days prior to the survey were included in the study. CCI was calculated using claims data with dates of service in the coverage period. Survey analysis methods were used to estimate variance. Linear regressions were used to calculate adjusted means and to assess the adjusted magnitude of an individual comorbidity’s association with MUHD and PUHD. Age, sex, rural urban commuting areas (RUCA), and Medicaid eligibility were used to adjust for confounding. Inflated experiment-wise error was controlled using Benjamini-Hochberg adjustment. RESULTS: A study sample of 58,975 persons were eligible for analysis (50.5 % female; mean age 74.9 years). Mean PUHD in the population was 7.2 days (95 % CI 7.0, 7.3) and mean MUHD was 4.0 days (95 % CI 3.9, 4.1). Adjusted analyses showed each point increase in CCI was linearly associated with an increase in PUHD (+0.76 days; 95 % CI 0.71, 0.81) and MUHD (+0.38 days; 95 % CI 0.33, 0.42). Conditions with the strongest adjusted association were: chronic pulmonary disease (+2.4 PUHD, +4.2 MUHD), severe liver disease (+3.6 PUHD, +2.0 MUHD), and peripheral vascular disease (+2.5 PUHD, +1.7 MUHD); all adjusted P \ 0.05. CONCLUSIONS: There is evidence of correlation between physically and mentally unhealthy days and CCI, both in composite and reduced form. Increasing comorbidity has a significant impact on both PUHD and MUHD.

(1098) Development and psychometric testing of the Japanese Version of the 13-item Sense of Coherence Scale-Revised (SOC13-RJ) among Japanese community-dwelling elderly Junko Sakano, Health and Welfare Science, Okayama Prefectural University, Soja, Okayama, Japan; Yuki Yajima, Niimi College, Niimi, Okayama, Japan; Yoshihiko Yamazaki, Nihon Fukushi Univerisity, Nagoya, Japan; Asuka Seki, INCAjinomoto Co., Inc., Kawasaki, Japan; Kaori Ono, INCAjinomoto Co., Inc., Kawasaki, Japan; Naoki Hayashi, INCAjinomoto Co., Inc., Kawasaki, Japan AIMS: The Japanese version of 13-item sense of coherence scale (SOC13-J) is complicated and difficult to use in questionnaire survey especially for elderly. We revised the SOC13-J to make it more easily understood by modifying item wording and by changing the response format to a five-point Likert scale ranging from ‘‘not at all’’ to ‘‘always’’. The aim of this study was to investigate the reliability and validity of the Japanese version of the SOC13-J-revised (SOC13-RJ)

123

Fig. 1 Parameter estimated of the Model 5 (strong factorial invariance model).  Identification constraints, *All standardized factor loadings were significant (p \ 0.05)

Table 1 Item means and reliablity coefficients of the SOC13-RJ (n = 429)

Table 2 Summary of multigrout factor analysis of the SOC-13RJ across gender

among Japanese community-dwelling elderly. METHODS: A selfadministered questionnaire including demographic information, the SOC13-RJ developed by the researchers, and the short-form health survey (SF-8) to assess multidimensional health-related quality of life (QOL), was mailed to 735 persons (aged 60 or older), recruited from senior clubs in A prefecture. The questionnaire was returned by 504 individuals (68.6 %) and 429 responses (58.3 %) were valid. The mean age of subjects was 75.6 years (60–90 years) and 53.8 % were female. Reliability of the SOC13-RJ was assessed by internal consistency using Cronbach’s alpha and corrected item-total correlations. Structural validity was determined by confirmatory and multigroup factor analyses. Construct validity was assessed by multiple regression analysis, to test whether SOC13-RJ predict theoretically expected outcome (health-related QOL). RESULTS: Chronbach alpha of the SOC13-RJ was 0.82 (Fig. 1). Item-total correlations were between 0.32 and 0.60. The hypothesized model comprised of three first-order factors (comprehensibility, manageability, and meaningfulness) and

Qual Life Res (2015) 24:1–191 single second-order factor (SOC) had acceptable fit (male: GFI = .919, CFI = .925, RMSEA = .060; female: GFI = .921, CFI = .897, RMSEA = .067), with all parameters significant. Multigroup confirmatory factor analysis showed strong factorial invariance across gender (Tables 1, 2). The SOC13-RJ scores had significant and theoretically consistent effects on two summary scores of physical (standardized beta = .230, p \ .001) and mental component (standardized beta = .382, p \ .001), after controlling for age and gender. CONCLUSIONS: These findings indicate the SOC13-RJ had good reliability and validity. The SOC13-RJ will be a useful measure in investigations for associations between SOC and various health outcomes among Japanese community-dwelling elderly.

(1100) How do recreational drugs influence quality of life? Extending the WHOQOL-BREF to assess substance users Elena Swift, MRes, BC Mental Health and Substance Use Services, Vancouver, BC, Canada; Suzanne M. Skevington, PhD, CPsychol, FBPsS, University of Manchester, Manchester, United Kingdom AIMS: To examine empirical support for extending the WHOQOLBREF to assess quality of life in recreational substance users, applying international WHOQOL findings and methods. METHODS: Cross-sectional online survey; internet recruitment from drug use communities. Completion by 121 adults (mean age 26; 31 % women; 42 % British). Five candidate quality of life (QoL) items piloted for an international facet on ‘dependence on non-medicinal substances’ were modified for substance users, and administered with the WHOQOL-BREF, dependency, and mood measures. RESULTS: Dependency was strongly, negatively associated with all four QoL domains, but weakly with substance use. The ‘best’ item, ‘SubsQoL’, was identified through item reduction procedures. Excellent internal consistency reliability (a = .913) was obtained when SubsQoL was added, compared to the WHOQOL-BREF alone (a = .909). In PCA, the SubsQoL loaded with physical items on medication dependency and pain; also cognitions and physical safety (12 %). SubsQoL scores discriminated between levels of substance use, and dependency. After controlling for dependency and health status, substance use accounted for additional variance in SubsQoL scores. Multiple regression confirmed that the predictive power of the WHOQOL-BREF SubsQoL exceeds the WHOQOL-BREF. CONCLUSIONS: Substance use impacts on QoL cannot be explained by dependency alone, especially for occasional users. The new SubsQoL item demonstrated excellent psychometric properties when tested on adults with diverse substance use patterns. This extension adds specificity and accuracy to the WHOQOL-BREF for substance using populations, with promise for research and clinical use.

(1102) Environmental aspects of functioning in people with intellectual disabilities Juliana Bredemeier, Dean of Psychology Program, UFRGS, FADERGS, Porto Alegre, RS, Brazil; Marilyn Agranonik, Professor, UNISINOS, Porto Alegre, Brazil; Markus Bredemeier, MD, Hospital Conceic¸a˜o, Porto Alegre, Brazil; Tatiana S. Perez, FADERGS, Porto Alegre, Brazil; Marcelo P. Fleck, PhD, UFRGS, Porto Alegre, Brazil AIMS: To investigate the impact of quality of care, attitudes towards disabilities, depression, satisfaction with life, parental care, social support and other sociodemographic variables in the perception of disability in people with intellectual disabilities (ID) using the biopsychosocial framework from International Classification of Functioning, Disability and Health (ICF). METHODS: A total of 156 people with ID respectively were enrolled in a prospective crosssectional study. Disability was measured using the WHO Disability

113 Assessment Schedule (WHODAS-II). A multivariable linear regression model (MLRM) using a stepwise method selected the variables that were independently and additively associated with the WHODAS-II score. A p value B 0.05 was considered statistically significant. RESULTS: The variables Social Domain (WHOQOL-Bref), Disability Visibility, Perspectives Factor (ADS-ID), and Physical Domain (WHOQOL-Bref) were significantly associated to the WHODAS-II score in the MLRM for people with ID (adjusted R2 = 0.39, p \ 0.001). CONCLUSIONS: Conclusions: The MLRM built in this study partially explained the determinants of functioning for ID. Further research is needed to clarify the variables that determine functioning and help the development of intervention strategies more appropriate to the real needs of this population.

New Investigator Poster Award Finalist Abstracts (2001) Does the EQ-5D capture the effect of physical and mental health on subjective well-being among older people? Eithne Sexton, Trinity College Dublin, Dublin, Ireland; Caitriona Cahir, Economic and Social Research Institute, Dublin, Ireland; Tom Fahey, Royal College of Surgeons in Ireland, Dublin, Ireland; Kathleen Bennett, Trinity College Dublin, Dublin, Ireland AIMS: Evidence increasingly suggests that the EQ-5D measure of health-related quality of life does not adequately capture mental health. This study examined the extent to which the EQ-5D captures the effect of anxiety and depressive symptoms on life satisfaction in older adults, compared with the effect of physical health (chronic disease burden and disability). METHODS: Retrospective cohort study of 870 patients aged C70 years from 15 general practices (GP) in Ireland. Patients’ GP medical records and medication dispensing history (pharmacy claims database) were reviewed for chronic diseases. Patients completed a questionnaire measuring quality of life outcomes. Path analysis was used to evaluate the direct and indirect effects of: (i) chronic disease burden (Rx Risk score); (ii) disability (basic and instrumental activities of daily living); (iii) anxiety symptoms and; (iv) depressive symptoms (Hospital Anxiety and Depression Scale) on life satisfaction (Life Satisfaction Index Z), via a utility score based on the EQ-5D. Utility-weights derived using a time-trade off study in the UK population were applied to raw EQ-5D scores to derive a utility score for each participant. Covariates included age, socioeconomic status and social support. RESULTS: The path model fitted the data well (chi-sq = 7.5, df (7), p = 0.37). EQ-5D fully mediated the effect of chronic disease burden and disability on life satisfaction (total indirect effect of disability = -0.04, p = 0.005). However, the effect of anxiety and depression was only partially mediated by utility score, with both retaining a significant direct effect on life satisfaction (total direct effect of depression = -0.38, p \ 0.001). In addition, the direct effect of utility score on life satisfaction was small in size (0.09, p = 0.004), compared with the effects of anxiety and depression. CONCLUSIONS: The EQ-5D does not adequately capture the effects of anxiety and depression on overall well-being among older adults, suggesting that it may under-estimate the effectiveness of interventions which improve mental health.

(2003) Second-order confirmatory factor analysis of the work role functioning (WRF-26) questionnaire applied to firefighters Kenneth Tang, McMaster University, Hamilton, ON, Canada; Joy MacDermid, Professor, McMaster University, Hamilton, ON, Canada; Kathryn Sinden, McGill University, Montreal, QC, Canada; Rob D’Amico, Hamilton Professional Firefighters Association, Hamilton, ON, Canada; Karen Roche, Hamilton Fire Department,

123

114 Hamilton, ON, Canada; Colin Grieve, Hamilton Professional Firefighters Association, Hamilton, ON, Canada AIMS: Valid approaches to assess health-related on-the-job problems among firefighters are needed. The aim of this study is to examine the factorial validity of the Work Role Functioning questionnaire (WRF26) among firefighters. METHODS: Active firefighters from the City of Hamilton Fire Services (n = 293) were recruited to participate in a 1-year prospective cohort study as part of the Firefighter Injury Reduction Enterprise-Wellness Enabled Life and Livelihood (FIREWELL) program to examine the musculoskeletal and cardiovascular health and overall well-being of firefighters. At baseline, participants completed a survey that includes the WRF-26. The WRF-26 aims to quantify and differentiate health-related on-the-job difficulties over five work domains (i.e. subscales): work scheduling (4 items), output demands (7 items), physical demands (8 items), mental demands (4 items), and social demands (3 items). Five Likert response options are available for each item (difficulty all of the time to none of the time). A second-order confirmatory factor analysis (CFA) was performed to evaluate whether empirical data offer support for WRF-26s intended domain structure. This approach hypothesized that WRF-26 item responses would be explained by five first-order factors corresponding to the proposed work domains, and also that the interrelationship between these factors would be explained by an underlying higherorder construct (i.e. overall work role functioning). CFA was performed with a robust weighted least squares estimator with mean and variance adjustment (WLSMV) estimator using Mplus version 6.0. Traditional model fit indices (CFI [[0.95 expected], TLI [[0.95 expected], RMSEA [\0.08 expected], WRMR [\1 expected]), saliency of factor loadings (b [ 0.3 and p \ 0.05 expected), and correlations among WRF-26 subscales were concurrently assessed to inform psychometric performance. RESULTS: CFA revealed excellent model fit (CFI = 0.999, TLI = 0.999, RMSEA = 0.043, WRMR = 0.794) after minor initial adjustments to the WRF-26 measurement model to facilitate model convergence. All first- and second-order factor loadings were highly salient (all b [ 0.8) and statistically significant (p \ 0.001). Spearman correlations between the five subscales ranged between 0.71 and 0.83. CONCLUSIONS: Evidence of factorial validity of the WRF-26 was demonstrated, affirming that its domain (subscale) structure is sound when applied to firefighters. Subscale-level scoring is therefore appropriate, which can provide a unique opportunity to differentiate problems over distinct work domains.

(2005) Using mobile communication technology to improve social participation in frail elderly residing in long-term care: an exploratory group concept mapping study Rozanne Wilson, PhD, University of Washington, Seattle, WA, United States; Dagmar Amtmann, PhD, University of Washington, Seattle, WA, United States; Jeff Small, PhD, University of British Columbia, Vancouver, BC, Canada; Alex Mihailidis, PhD, P.Eng., University of Toronto, Toronto, ON, Canada AIMS: Communication is critical to social participation and quality of life (QOL). For frail elderly persons living in long-term care (LTC) with chronic conditions that affect communication (e.g., dementia), there is a high risk of social isolation and unmet care needs. Mobile communication technology (MCT) (e.g., smartphone, tablet) has the potential to serve as an innovative tool to support the delivery of care and staff-resident communication. Focus group (FG) method is useful for investigating stakeholder perceptions; however, in the context of LTC, conducting FGs can be challenging due to time and scheduling constraints. This study aims to explore the use of Group Concept Mapping (GCM) method as a novel alternative to FGs to examine care staff’s (e.g., care aides, nurses)

123

Qual Life Res (2015) 24:1–191 perceptions about the potential use of MCT to improve participation (e.g., communication, self-care, relationships) of frail elderly persons in the LTC setting. METHODS: Two LTC facilities are participating in this study. In each facility, 20 community-of-care stakeholders (e.g., care aides, nurses, managers) will participate in GCM. GCM is a participatory mixed-methods approach that is implemented in an online environment. GCM involves three stages: brainstorming, sorting, and rating. During the brainstorming stage, participants generate items by responding online to an open-ended question: ‘‘What roles do you see MCT playing in the delivery of care to support participation of frail elderly residents and how do you see increased participation improving the quality of life of frail elderly residents?’’ Once brainstorming has concluded, participants go online again to sort the items into categories and rate the importance of each item using a 5-point Likert scale. RESULTS: Sorted and rated items (e.g., ‘‘MCT improves communication’’) will be analysed using Concept Systems Global Max Software to produce conceptual frameworks. Multi-dimensional scaling (i.e., 2-dimensional map of the relationships among all statements) and hierarchical cluster analysis will be used to identify key emergent themes around MCT and participation. CONCLUSIONS: Findings from this study will describe care staff’s perceptions about the use of MCT as a tool to improve the QOL of frail elderly residents in LTC and evaluate the potential usefulness of GCM as a research tool in LTC settings.

Poster Session 2001: Friday Poster Session 1 Reviews (2007) Patient-reported outcomes in randomised controlled trials of head and neck cancers: a systematic review of reporting quality from the PROMOTION registry Rebecca Mercieca-Bebber, University of Sydney, Sydney, NSW, Australia; Alessandro Perreca, Data Center and Health Outcomes Research Unit, Italian Group for Adult Hematologic Diseases (GIMEMA), Rome, Italy; Andrew Macann, Department of Radiation Oncology, Auckland Regional Cancer and Blood Service, Auckland City Hospital, Auckland, New Zealand; Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia; Katie Whale, University of Bristol, Bristol, United Kingdom; Salvatore Soldati, Data Center and Health Outcomes Research Unit, Italian Group for Adult Hematologic Diseases (GIMEMA), Rome, Italy; Marc Jacobs, Academic Medical Center, University of Amsterdam, Department of Medical Psychology, Amsterdam, The Netherlands, Amsterdam, Netherlands; Fabio Efficace, Data Center and Health Outcomes Research Unit, Italian Group for Adult Hematologic Diseases (GIMEMA), Rome, Italy AIMS: To determine the extent to which reporting of patient-reported outcomes (PROs) in randomized clinical trials (RCTs) of head and neck cancer (HNC) interventions adhered to the International Society for Quality of Life Research (ISOQOL) PRO reporting standards, and the potential of these reports to facilitate translation to clinical practice. METHODS: Two reviewers systematically and independently searched Medline(OVID)/PubMed, Cochrane CTR, PsycINFO and PsycARTICLES databases for English-language, Phase II-III HNC RCTs of biomedical interventions with PRO endpoints published Jan 2004–May 2014. Two investigators independently: (1) coextracted clinical, PRO and demographic RCT data; (2) assessed adherence to the ISOQOL PRO reporting standards. One ‘point’ was allocated to each ISOQOL checklist item addressed in RCT publication/s, giving a maximum score of 29 and 18 each for RCTs with a primary or secondary PRO endpoint respectively. Reviewers’

Qual Life Res (2015) 24:1–191 screening discrepancies were settled with a senior investigator. RESULTS: We screened 4491 abstracts and obtained 161 full-text articles. Fifty-four RCTs (82 publications) were eligible, enrolling overall 12,162 patients. Of these RCTs, 18 included a primary PRO and 36 included a secondary PRO endpoint. Adherence to the ISOQOL PRO reporting standards was higher for RCTs with primary PRO endpoints: M 12.6 (SD 3.7) than for secondary PRO endpoints: M 5.5 (SD 3.8). A weak correlation between publication year and % adherence to ISOQOL reporting checklist was not significant, r(53) = .25, p = .07. Some items on the ISOQOL PRO reporting checklist were addressed frequently; 45 RCTs (83 %) identified a PRO endpoint in the abstract, 41 RCTs (75.9 %) described the intended PRO data collection schedule. Of the 18 RCTs with a primary PRO endpoint, 16 (88.9 %) included a summary of past PRO research, 12 (66.7 %) described baseline PRO scores and 15 (83.3 %) reported results of all PRO domains assessed. Other items were reported less frequently; only 8/54 (14.8 %) RCTs stated a PRO hypothesis; 11 (20.4 %) justified their chosen PRO measure; rates or reasons for missing PRO data were reported for 16 RCTs each(29.6 %) and 30 (55.56 %) discussed PRO findings in the context of other trial outcomes. CONCLUSIONS: Many HNC RCTs, particularly those with secondary PROs, did not report sufficient details about PRO endpoints to facilitate transparent interpretation and clinical translation.

(2009) Measuring outcomes in young patients with conditions associated with a facial difference: are PRO instruments asking the right questions? Natasha Wickert, McMaster University, Hamilton, ON, Canada; Karen Wong, MD MSc FRCSC, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; Mark Mansour, McMaster University, Hamilton, ON, Canada; Timothy Goodacre, MBBS BSc FRCS, University of Oxford, Oxford, United Kingdom; Christopher Forrest, MD MSc FRCSC, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; Andrea Pusic, MD MPH FACS, Memorial Sloan-Kettering Cancer Center, New York City, NY, United States; Anne Klassen, DPhil, McMaster University, Hamilton, ON, Canada AIMS: The aim of this systematic review was to identify patientreported outcome instruments (PROs) used in research with children and youth, with conditions associated with facial differences in order to identify the health concepts measured. METHODS: MEDLINE, EMBASE, CINAHL, and PsycINFO were searched from 2004 to 2014 to identify PRO instruments used in acne vulgaris, vascular malformations, congenital melanocytic nevi, burns, protruding ears, microtia, facial asymmetries, and facial paralysis patients. We performed a content analysis whereby the items were coded based on major and minor themes. A conceptual framework was developed, and items were categorized as having positive or negative content and/or phrasing. RESULTS: 6921 articles were screened, and PRO instruments were used in 125 publications. A total of seven generic and 11 condition-specific instruments were identified, corresponding to four major domains, 11 sub-domains, and 85 unique health concepts. Condition-specific instruments were found for: acne (4); oral health (2); dermatology (1); facial asymmetries (2); microtia (1); and burns (1). No measures were found for facial paralysis, birthmarks, or protruding ears patients. A content analysis of the condition-specific instruments generated 287 items with most addressing psychological health (n = 130). Instruments measure appearance-related distress, but not appearance itself (i.e. what patients think about how they look). 75 % of the identified items were considered to have negative content or phrasing (e.g. ‘‘Because of the way my face looks I wish I had never been born’’). The YQOL-FD and Acne-QoL measures

115 closely followed guidelines for PRO instrument development and validation. Generic PRO measures were found to lack content validity for patients whose conditions involved facial function and/or appearance. CONCLUSIONS: Treatments can change facial appearance and function, sometimes quite dramatically. This review draws attention to a problem with content validity in existing PRO instruments. Our team is now developing a new PRO instrument called FACE-Q KIDS to address this concern.

(2011) Digital methodologies in the development of clinical outcome assessments Chloe Tolley, Adelphi Values Ltd, Cheshire, United Kingdom; Chris Marshall, Adelphi Values Ltd, Cheshire, United Kingdom; Adam Gater, Adelphi Values Ltd, Cheshire, Select State, United Kingdom; Gavin Dickie, Adelphi Values Ltd, Cheshire, United Kingdom; Sarah Kilgariff, Adelphi Values Ltd, Cheshire, United Kingdom AIMS: Clinical Outcome Assessments (COAs) are designed to measure how patients feel or function and should reflect the ‘real-life’ or ‘lived’ experience of the population of interest. Historically, such evidence has been generated using traditional qualitative research techniques such as interviews or focus groups, however, the evolution of digital technologies represents an exciting opportunity to gain ‘here-and-now’ insights into the daily lives of patients, caregivers and clinicians. METHODS: Examples of digital techniques which can be utilized in the development of a COA will be identified through a comprehensive literature review and based on experience of using such methods in primary and secondary research. RESULTS: Digital methods of greatest relevance to the COA development process include: Mobile apps to elicit data from patients in the form of tasks set by the researcher can increase patient engagement and removes the artificial interview setting. This real-time data collection improves ecological validity by not having to rely on patients recalling experiences over a long period of time. Qualitative data obtained through internet based research (e.g. Forums, online communities, blogs and vlogs) can be used to supplement or replace data elicited via traditional qualitative methods using passive observation of existing data or moderated primary research. Bespoke on-line panels can provide quick responses to research questions or feedback on minor revisions to instruments following traditional cognitive debriefing. Web/Video communication software (e.g. Skype) can overcome difficulties faced in conducting face to face interviews if patients are hard to reach and reduce patient burden if unable to attend an interview due to their condition or circumstances. Finally, instant messaging (e.g. google talk) can be used where patients are embarrassed to talk face to face or have communication difficulties. CONCLUSIONS: Digital technologies can provide flexibility and convenience for patients, facilitate patient engagement in the research process and help to elicit qualitative data with greater depth and ecological validity. While such methods are an excellent addition to the COA developer’s ‘toolkit’, careful consideration must be given to the goals of the research, the population of interest and the strengths and weaknesses of each approach to ensure an appropriate selection of any digitally-enhanced research methods.

(2015) The use of patient reported outcome measurements (PROMs) as performance indicators in lung cancer treatment Majken M. Broenserud, MD, PhD student, University of Southern Denmark and Odense University Hospital, Odense City, Denmark; Erik Jakobsen, MD, MPM, Chief Surgeon, Clinical associate professor, Odense University Hospital, Odense C, Denmark; Liv H. Dørflinger, Master of Public Health Science (MScPH), Health and

123

116 Quality Consultant, The Danish Cancer Society, Copenhagen East, Denmark AIMS: The project ‘‘Patient reported quality of life in lung cancer patients—The use of patient reported outcome measurements (PROMs) as performance indicators’’ aims to investigate the use of PROMs in evaluation of clinical quality, including an evaluation of data quality (validity and completeness). Furthermore, we wish to illustrate the relations between lung cancer manifestations, the treatment given, the patients’ socioeconomic situation and PROMs during the clinical course. Hereby we will be able to evaluate if PROMs can be applied as indicators in a clinical database. METHODS: The cancer generic questionnaire ‘‘Quality of Life Questionnaire Cancer 30’’ (EORTC_QLQ-QC30) and the lung cancer specific questionnaire ‘‘Quality of Life Questionnaire Lung Cancer 13’’ (EORTC_QLQ_ LC13) are applied as tools for collecting PROMs. Paper questionnaires are sent out to the patients 4 times during the clinical course: At diagnosis before treatment, and then 3, 6 and 12 months after. All patients registered with a lung cancer diagnosis in Danish Lung Cancer Registry (DLCR) in the period from January 1, 2014 until December 31, 2015 are included in the study (approximately 4500 patients a year). RESULTS: As part of the study, a systematic review is made during the summer of 2015, to illustrate what previously have been published about the use of PROMs as performance indicators in lung cancer treatment. The results from the review and preliminary results from the questionnaire study will be presented at the ISOQOL 22nd Annual Conference 2015. CONCLUSIONS: The integration of PROMs will contribute to secure, that Danish lung cancer patients’ evaluation of treatment, course of treatment and its effects on their health and health related quality of life, will be variables in the clinical quality database in terms of benchmarking, quality monitoring, support of the patients’ clinical courses before-and especially after-treatment. Furthermore, collection of PROMs will create more knowledge about sequelae after treatment for lung cancer in Denmark. Combined with other information in DLCR, this knowledge can be put in relation to the patient’s disease phase, comorbidity and treatment, by which identification of risk groups, where special attention is needed, can be possible.

(2017) The quality of systematic reviews of health-related outcome measurement instruments Caroline B. Terwee, VU University Medical Center, Amsterdam, Netherlands; Cecilia A. Prinsen, PhD, Amsterdam, Netherlands; Maria G. Ricci Garotti, MSc, Bologna, Italy; Arnela Suman, MSc, Amsterdam, Netherlands; Henrica C. de Vet, VU University Medical Center, Amsterdam, Netherlands; Lidwine B. Mokkink, PhD, Amsterdam, Netherlands AIMS: Systematic reviews of outcome measurement instruments are important tools for the selection of instruments for research and clinical practice. Our aim was to assess the quality of systematic reviews of health-related outcome measurement instruments, and to determine whether the quality has been improved since our previous study in 2007. METHODS: A systematic literature search was performed in MEDLINE and EMBASE on June 19, 2014 to identify all systematic reviews of health-related outcome measurement instruments published between July 1, 2013 and June 19, 2014. The quality of the reviews was rated using a standardized checklist. RESULTS: A total of 102 reviews were included. In many reviews the search strategy was considered not comprehensive; in only 59 % of the reviews a search was performed in EMBASE and in about half of the reviews we had doubts about the comprehensiveness of the search terms for measurement instruments. In 41 % of the reviews, compared to 30 % in our previous study, the methodological quality of the included studies was assessed. In 58 %, compared to 55 %, the quality of the included outcome measurement instruments (i.e. their

123

Qual Life Res (2015) 24:1–191 measurement properties) was assessed. In 43 %, compared to 7 %, a data synthesis was performed in which the results from multiple studies on the same instrument were combined according to predefined criteria. CONCLUSIONS: Despite a clear improvement in the quality of systematic reviews of outcome measurement instruments in comparison with our previous study in 2007, there is still room for improvement in all above mentioned domains.

(2019) Content mapping of patient-reported outcomes (PROs) important to patients with non-muscle invasive bladder cancer (NMIBC) against available PRO measures (PROMs) Claudia Rutherford, PhD, University of Sydney, Sydney, NSW, Australia; Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia; David Smith, PhD, NSW Cancer Council, Sydney, NSW, Australia; Daniel Costa, Dr, The University of Sydney, Sydney, NSW, Australia; Manish Patel, PhD, University of Sydney, Sydney, Australia AIMS: NMIBC is a chronic condition (5-year survival[80 %) requiring intensive follow-up, repeated endoscopic examinations, tumour resections and intravesical treatments. In this clinical context, quality of life (QOL) is a critical concern for patients and their managing clinicians, with enormous potential for PROs to be integral to treatment assessment and recommendations for NMIBC. We aimed to: (1) develop a conceptual framework of PROs important to patients receiving treatment for NMIBC; (2) identify PROMs used in NMIBC research; and (3) assess PROM’s conceptual coverage against the framework to determine suitability and appropriateness for use in NMIBC research. METHODS: We conducted a systematic review searching six electronic databases (inception-Dec 2014) and semi-structured qualitative interviews with patients with NMIBC (n = 16), treating physicians (n = 8) and specialist nurses (n = 9). From these sources we compiled an exhaustive list of clinically relevant issues, producing an empirically-based conceptual framework of PROs important for assessing the impact of contemporary treatments for NMIBC. RESULTS: We found 170 papers investigating various treatment options and schedules for NMIBC, identifying several local and systematic side effects. However, only 13/170 assessed PROs and 5/13 used a standardized instrument. Other limitations include: lack of comparison group; poor adjustment for preillness functioning; and failure to distinguish between risk profiles and treatments in the analysis. Consequently, key data on the impact of contemporary treatments for NMIBC on patients’ QOL is lacking. Assessing the content of PROMs used in NMIBC research against our framework revealed they fail to cover important content (e.g. rash, joint pain, stiffness) and include irrelevant (e.g. bloating, flatulence) and nonNMIBC-specific domains (e.g. worried about future health). Our clinician interviews revealed that PROMs do not distinguish local (common and expected) from systemic (rare but warranting immediate action) side effects and that separate subscores would increase clinical utility of a PROM. CONCLUSIONS: Few studies assessed PROs when evaluating treatment for NMIBC. Where assessed, PROMs that fail to capture some outcomes important and relevant to contemporary treatments for NMIBC were used, suggesting existing PROMs may not be suitable PRO endpoints for treatment effectiveness trials in NMIBC. Our conceptual framework provides the basis for selecting PRO endpoints in future clinical trials.

(2021) Enrolling and keeping participants in self-management interventions: a meta-analysis Alaa M. Arafah, PhD Candidate, McGill University and King Saud University, Montreal, QC, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada

Qual Life Res (2015) 24:1–191 AIMS: Self-management is one of the few interventions that directly target health-related QOL but nothing is effective if not done. For people with Multiple Sclerosis (MS), self-management becomes an important element in their care because of distressing symptoms and limitations in daily activities. Even though the importance of selfmanagement interventions for people with MS is well documented, rate of participation and completion varies across studies. The objective of this study is provide an estimate of expected enrollment and attrition rates based on published studies of existing self-management interventions for people with MS, and to identify contributing factors and impact on outcomes. METHODS: A structured literature search was conducted using multiple electronic-databases. Experimental and quasi-experimental studies that met specified criteria were included. A random effect meta-regression analysis was conducted to estimate the overall enrollment and attrition proportions, effect of person and study-related factors, and impact on study outcomes. RESULTS: A total of 43 studies, comprising 646 persons were identified. The estimated enrollment rate was 49.3 % (95 % CI 48.6–50.1) and the estimated attrition rates in the intervention and control groups were 17.4 % (95 % CI 16.8–17.9) and 13.9 % (95 % CI 13.3–14.5), respectively. The main reported reason for refusing to participate was lack of interest (71.6 %), while the reported reasons for dropping-out were mainly due to medical issues (29.5 %), disliking the intervention (18.2 %) and time constraint (13.3 %). The features of the intervention that promoted remaining in the study were individual contact, web- or telephone-based, with a total contact time of [12 h. The sub-groups most likely to remain in the intervention were women, under the age of 50, within 7–11 years of diagnosis, and with RRMS. Studies that had a 10 % higher attrition rate had an effect size that was larger by 0.14 (95 % CI 0.12–0.19). CONCLUSIONS: Greater understanding of the factors associated with acceptance and attrition rates would help in improving the design of future clinical trials and increasing their validity and generalizability. Understanding the reasons for not participating in trials and not completing interventions would also provide valuable information that will help in planning and developing a more appealing self-management intervention that patients can easily accept and incorporate into their everyday lives.

(2023) How do health care providers respond to PROMs data to improve patient care? A realist synthesis Joanne Greenhalgh, University of Leeds, Leeds, United Kingdom; Sonia Dalkin, University of Leeds, Leeds, United Kingdom, Elizabeth Gibbons, University of Oxford, Oxford, United Kingdom; Nick Black, PhD, London School of Hygiene and Tropical Medicine, London, United Kingdom; Jose M Valderas, MD, PhD, MD, University of Exeter, Exeter, United Kingdom; Judy Wright, University of Leeds, Leeds, United Kingdom; David Meads, University of Leeds, Leeds, United Kingdom; Liz Lingard, North East Quality Observatory System, Newcastle upon Tyne, United Kingdom; Laurence Wood, PPI representative, Leeds, United Kingdom; Jane Jackson, Director, JJ consulting, London, United Kingdom AIMS: To understand when, how and why providers respond to publicly released data from patient reported outcome measures (PROMs) in order to improve patient care. METHODS: A realist synthesis of quantitative and qualitative literature examining the feedback of aggregated data from PROMs to hospitals. We analysed policy documents, editorials, opinion articles, letters and websites to identify the underlying ideas and assumptions (‘programme theories’) of how providers are expected to respond to PROMs data. We then searched the empirical literature, electronic databases, Google Scholar and used citation tracking to identify studies to test these theories. This identified 1617 papers; 55 studies were included in our synthesis. We extracted data for each study and compared, contrasted and explained similarities

117 and differences between studies to refine our programme theories. RESULTS: Providers are motivated to respond to PROMs data because they seek to protect their professional reputation and be as good or better than their peers. Providers identified as poor performers are more likely respond than average or high performers. Providers take steps to improve patient care only when they trust the PROMs data. Trust is enhanced if the perceived reason for the data collection aligns with clinical concerns, data sources are perceived to be accurate, case mix adjustment is adequate, the data supports providers’ self-assessment of their performance and data is fed back in a timely and interpretable way. PROMs data alone does not enable providers to improve the quality of care for patients. Providers also need additional locally collected data and resources to investigate the cause of poor care and develop solutions to address this. CONCLUSIONS: Healthcare providers appear motivated to use PROMs data but would benefit from guidance on how to interpret PROMs data, how to integrate PROMs data with other locally and nationally collected performance data and what steps they could take to investigate the cause of their outlier status. Focusing on reducing negative outliers may not realise the full potential of PROMs data to improve the quality of care for all patients.

(2025) A systematic review of quality of life measurement in studies investigating treatments of psoriasis Faraz M. Ali, Clinical Research Fellow, Cardiff University, Cardiff, United Kingdom, Andrea C. Cueva, Attending Physician/Researcher, Centro de la Piel, Quito, Ecuador; Ausama A. Atwan, MD, Cardiff University School of Medicine, Cardiff, United Kingdom; Jui Vyas, Clinical Senior Lecturer, Cardiff University, Cardiff, United Kingdom; Vincent Piguet, Professor of Dermatology, Cardiff University School of Medicine, Cardiff, United Kingdom; Sam Salek, PhD RPh FFPM MRPSGB MCMS FESCP, School of Life & Medical Sciences, University of Hertfordshire, Hatfield and Director - Institute for Medicines Development, Cardiff, United Kingdom; Andrew Y. Finlay, Professor of Dermatology, Cardiff University, Cardiff, United Kingdom AIMS: Systematic reviews (SR) are unique in helping to form the basis for developing practice guidelines and identify gaps in knowledge. The aim of this study was to generate evidence of the use of quality of life (QoL) measures to assess treatments for psoriasis. METHODS: The methodology for this SR followed PRISMA guidelines. All available articles describing randomized placebo and active-controlled trials (RCTs) of therapies for psoriasis that included QoL measurements published up to November 2014 were identified. Six databases were examined with 388 search terms, and grey literature was identified. Abstracts of articles were reviewed independently by two assessors (FA, AC) and included if they met the inclusion criteria. A third adjudicator (AA) resolved any differences of opinion. Two authors (FA, AC) independently reviewed risk of bias with the JADAD scale and extracted data using the Cochrane Handbook for SR form. RESULTS: Of 3597 screened article abstracts, 329 articles were selected for detailed review, 102 trials met eligibility criteria for inclusion. The Dermatology Life Quality Index (DLQI) was the most commonly used QoL tool (number of studies = 84, 82.4 %), followed by the Short Form 36 (SF-36) (32, 31.4 %), EuroQol (EQ-5D) (15, 14.7 %), Psoriasis Disability Index (PDI) (14, 13.7 %) and Skindex (5,5 %); 46 trials used more than one QoL tool for assessment. The most widely investigated interventions that included QoL assessment were: topical calcipotriol (13 trials), systemic methotrexate (6), etanercept (14) and phototherapy (9). The areas of origin of these studies were: Europe (47 trials) including Germany (9), UK (9); and USA/Canada (40) trials including USA (21), USA/Canada (12) and Canada (7). The number of psoriasis treatment studies that included QoL measurement has

123

118 gradually increased throughout the years: 1998–2004 = 13; 2005–2009 = 33; and 2010–2014 = 56. CONCLUSIONS: Where QoL is measured in the assessment of interventions in psoriasis, the DLQI is the most commonly used method. There is an increasing recognition of the value of QoL assessment in monitoring the efficacy of interventions in psoriasis, as demonstrated by the increasing reporting of QoL measurement.

Translation (2027) Clinical interpretation of the Chinese version of functional digestive disorders quality of life questionnaire Zheng-kun HOU, MD, The First Affiliated Hospital of Guangzhou University of Chinese Medicine, Guang Zhou, China; Feng-bin LIU, MD, The First Affiliated Hospital of Guangzhou University of Chinese Medicine, Guang Zhou, China, Zhuo-qun CHEN, Guangzhou University of Chinese Medicine, Guangzhou, China AIMS: The functional digestive disorders quality of life questionnaire (FDDQL) is a useful and well recognized health assessment instrument for functional dyspepsia (FD) patients. It had been translated into Chinese language. The purpose of this study was to describe its score interpretation for clinical practice. METHODS: Data of FDDQL Chinese version and social geography characters from the FD patients aged C18 years admitted to the hospital between November 2009 and April 2013 were enrolled prospectively. Five approaches were used to clinically interpret outcome data. First, the baseline and responsiveness of the questionnaire were estimated among Completed group, Missing group and Health group. Second, the standard error of the estimate was used to construct the 90 % confidence interval for each score estimate. Third, percentile ranks were applied to each domain and total scores. Four, statistically reliable change, contains Paired t-statistic, Effect Size, standardized response mean, standard error of measure (SEM) and SEM95, minimal detectable change (MDC) and MDC90, MDC95, MDC99 were used to define individual subject-level change. Five, clinically important change was analyzed with receiver operating characteristic curve. At last, a comprehensive analysis of above mentioned methods were performed to define the final clinical interpretation. RESULTS: The data of 202 FD patients and 150 health persons were enrolled for analysis, and 25 missing data were used for baseline analysis. Compared with the intake FD patients, the discharged and health persons have significant better health status in all domains (P \ 0.001, expect Discomfort in discharged people, P = 0.142), totally contrast to missing data. On average, precision of a single score was estimated by health score ±1 both for intake and discharge. Based on score distribution, 25th, 50th and 75th percentile ranks corresponded to intake scores of 49, 58, and 66, and discharge scores of 59, 65, and 72, respectively. An MDC95 value of 11 or more represented statistically reliable change. Receiver operating characteristic analyses supported that changes in scores of 6 or more represented minimal clinically important improvement. The total evidence quality for the final recommendation level was high. CONCLUSIONS: The score interpretation of the Chinese version of FDDQL could assist clinicians during therapy practice.

(2029) The DISABKIDSÒ-37 questionnaire for Brazilian children and adolescents with chronic conditions showed cross-cultural Brazilian regions validity Serlyjane P. Nunes, Universidade Federal Do Maranha˜o, Sa˜o Luı´s, Maranha˜o, Brazil; Jenny Kellma F. Silva, Universidade Federal Do Maranha˜o, Sa˜o Luı´s, Maranha˜o, Brazil; Fernanda M. dos Anjos, Universidade Federal Do Maranha˜o, Sa˜o Luı´s, Maranha˜o, Brazil;

123

Qual Life Res (2015) 24:1–191 Roberta A. Reis, Universidade Federal do Rio Grande do Sul, Porto Alegre, Rio Grande Do Sul, Brazil; Cla´udia Fegadolli, Universidade Federal De Sa˜o Paulo, Sa˜o Paulo, Brazil; Monika Bullinger, Vice Chief of Department, University of Hamburg, Hamburg, Germany, Claudia B. dos Santos, Universidade De Sa˜o Paulo, Ribeira˜o Preto, Sa˜o Paulo, Brazil AIMS: The aim was to test the performance of a cross-culturally developed disease-generic quality of life (QoL) measure for children with chronic diseases in a multiregional study. METHODS: The study included 306 participants who were children and adolescents with asthma or diabetes mellitus type 1, aged between 8 and 18 years old, from the Southeast and Northeast regions of Brazil, and their parents or caregivers. It was conducted in 2014 and was approved by the Ethics Committee of the University of Sa˜o Paulo at Ribeira˜o Preto College of Nursing, Brazil. RESULTS: The Analysis of Differential Item Functioning showed the absence of this effect for the vast majority of items between different subgroups for the self-version, meaning that boys and girls of all ages, residing in the Southeast and Northeast, living in conditions diabetes or asthma, are able to answer them exhibiting the same level of understanding. Results with similar characteristics were evident with respect to understanding the items of proxy version of the Questionnaire. CONCLUSIONS: The psychometric performance of both items and scales of the DISABKIDS condition oriented measure was good across cultures, despite a range of cross-regional Brazilians differences. The results presented are sufficient to give it further to the completion of the validation process DCGM-37 for Brazil, with the inclusion of children and adolescents and their parents or caregivers of the North, Midwest and South regions. Finally, DISABKIDS 37 module for Brazilian children and adolescents with chronic conditions is a useful tool to analyze Quality of Life.

(2031) Evaluating the impact of defined difficulty ratings on consistency of translatability assessment results for PRO measures Sarah Basse, Health Research Associates, Inc., Seattle, WA, United States; Alexa Chhay, Health Research Associates, Inc., Seattle, WA, United States; Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States AIMS: Translatability Assessment (TA) is employed with increasing frequency as part of the PRO development process to help refine the wording of new questionnaire items in a way that facilitates clear and equivalent concept reflection (translatability) into other languages. There are multiple approaches to conducting TA, with no current consensus on standards for reliable, replicable results. We aim to understand how use of a standardized, defined rating scale for assessing translation difficulty of questionnaire items impacts the consistency and overall quality of TA results. METHODS: Documentation from two separate TA projects was reviewed and compared. In project one, language consultants used a 5-point scale to rate the predicted degree of difficulty for translating each questionnaire item into their language. In project two, consultants were oriented with a detailed definition for each level of the rating scale and were provided examples that demonstrated various types of translation difficulties that fit those ratings. RESULTS: Project one results demonstrated lack of uniformity in ratings for the same type of translation difficulty from one rater to another, and approximately 44 % of the questionnaire items were given a rating that was not consistent against standardized definitions. Evaluation of the documentation in project two showed greater consistency within translation difficulty levels assigned, leaving fewer ratings (25 %) that failed to match the standardized definitions and required negotiations between the project manager and consultant to resolve. Furthermore,

Qual Life Res (2015) 24:1–191 having pre-set standardized definitions supported the project manager with guidance during negotiations with consultants to move those initial ratings into alignment with the standardized definitions provided for translation difficulty. CONCLUSIONS: A reference document that gives clear definitions, examples, and guidance for the assignment of translation difficulty ratings was shown to be an effective training tool for preparing TA reviewers, and also a reliable source of information to be used by project managers in checking the accuracy of assigned ratings. Use of defined rating levels for assessing the translation difficulty of PRO items can aid in providing more meaningful descriptions of translation challenges that may affect data collection, interpretation and comparability.

(2033) Challenges related to the linguistic validation of culturally sensitive questions Fiona E. Miller, BA (Joint Hons), PharmaQuest Ltd, Banbury, United Kingdom; Hayley L. Simpson, BA, PharmaQuest Ltd, Banbury, United Kingdom; Rebecca L. Two, BA (Hons), PharmaQuest Ltd, Banbury, United Kingdom AIMS: Guidelines for the methods and good practice in the translation and linguistic validation of patient reported outcome (PRO) measures are known and readily accepted. The aim of the translation and linguistic validation process is to ensure conceptual equivalence and cultural relevance in the target country. However, difficulties can arise in the process where the measure includes questions of a sensitive nature. For example, quality of life questions concerning the respondent’s sex life, intimate body parts or demographics may be taboo subjects in some countries. However, people from other cultures may be much more open about the topics and willing to talk about them. We aim to discuss some issues of cultural sensitivity experienced during the linguistic validation process and ways in which these can be overcome. METHODS: We analysed a number of previous projects and collected examples of culturally sensitive questions that posed difficulties during the translation and linguistic validation process, specifically during the cognitive debriefing stage. RESULTS: Whilst in some cases, respondents made clear that they were unwilling to discuss sensitive topics any further, some made the point that they would refuse to discuss them in an interview setting during pilot testing, but that they would be happy to complete the questionnaire in a real life situation, as long as the questionnaire was self-administered and not interviewer-administered. In many cases, women can be unwilling to discuss female issues with a male interviewer and vice versa. CONCLUSIONS: Although it might be expected that some respondents may be unwilling to provide their true answers to sensitive questions, it seems that in many cases it is the discussion about the topic during linguistic validation that poses a bigger problem. In-country investigators can be encouraged to discuss problematic questions in an indirect way when cultural sensitivity becomes an issue, and it is advisable for women to be interviewed by a female interviewer and vice versa for questionnaires including sensitive, gender-specific items. In summary, a number of alternative approaches can be employed by the in-country investigator and by the translation vendor in order to overcome cultural difficulties that arise during cognitive debriefing.

(2035) Linguistic validation interviews in a pediatric cerebral palsy population: approaches for meaningful feedback Valeska Kantzer, Health Research Associates, Inc., Mountlake Terrace, WA, United States; Sarah Basse, Health Research Associates, Inc., Seattle, WA, United States; Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States;

119 Thorin L. Geister, PhD, Merz Pharmaceuticals, Frankfurt am Main, Germany AIMS: The Questionnaire on Pain caused by Spasticity (QPS) is a modular patient- and observer-reported outcome measure (PRO/ ObsRO) designed for children with spasticity-related pain due to cerebral palsy (CP). One module is interviewer-administered for children who are too young to read, or older children with CP-related motor or cognitive deficits that prevent them from using the selfadministered module. Linguistic validation interviews require cognitive participation in order to test the equivalency of a new translation. In this population, unique methods were required to make the interview process feasible with compromised children and still obtain the required documentation to support a PRO translation. METHODS: Standard linguistic validation interview methods assessing concept equivalence (requiring the patient to verbally identify the concept being asked about) were revised to focus on interviewer observation of the child’s ability to provide a meaningful response to the PRO item exactly as it appeared in the preliminary translation. When responses were not clear or conflicted with previous answers, the interviewer used simple re-phrased versions of the translated questions to try to determine whether the issue was related to the child’s cognitive issues, or to the wording of the translation. RESULTS: A total of 56 children participated in testing the translated wording of the 11-item interviewer administered QPS module, across 29 different language versions. Of the possible 616 items that were assessed, 493 were confirmed as eliciting meaningful responses from the children interviewed. By asking simple, rephrased versions of the questions, interviewers determined 27 cases of no meaningful answer due to CP-related deficits and 96 cases to be related to the complexity of the translation. In those cases, the rephrased wording that successfully elicited a meaningful answer from the child was offered as a revision to improve the translation of the questionnaire item for the pediatric CP population. CONCLUSIONS: While interviews with children might pose a challenge, alternative approaches to achieving linguistic validation can still produce helpful feedback to be considered for refining the wording of items in the target language, and ensuring the linguistic viability of translated PROs for pediatric populations with condition-specific deficits.

(2037) Psychometric evaluation of the Suicide Ideation Scale in mentally ill Korean patients dwelling in communities Jin Sil Han, MPH, Gimpo Mental Health Center, Gimpo, South Korea; Eun-Hyun Lee, PhD, Ajou University, Suwon, South Korea; Tong Woo Suh, PhD, Gimpo Mental Health Center, Gimpo, South Korea; Chang Hyung Hong, PhD, Ajou University, Suwon, South Korea AIMS: The Suicide Ideation Scale (SIS) is a self-reported instrument, measuring the intensity of suicidal ideation form covert suicidal thoughts to more overt or intense ideation and actual suicide attempts. This study was to evaluate the psychometric properties of the Korean version of the Suicide Ideation Scale (SIS-K) in patients with chronic mental disorders dwelling in community. METHODS: The SIS-K was translated to Korean using a translation and back-translation technique. A sample of 396 Korean patients with mental disorders was recruited from 6 regional mental health centers in Korea. Factorial construct validity was tested using a cross-validation approach involving both exploratory factor analysis (EFA) and confirmatory factor analysis (CFA). The sample was randomly split into two subsamples. Subsample 1 was used for EFA to identify the underlying structure of the items, while subsample 2 was used for CFA to test whether the underlying structure was supported by actual data. Itemconvergent and -discriminant validity, known-groups validity, and internal consistency reliability were tested using multitrait/multi-item

123

120 correlation, t-test, and Cronbach’s alpha, respectively. RESULTS: EFA extracted 9 items clustered into 2 subscales (suicidal desire, and resolved plans/preparation), which accounted for 68.59 % of the variance. One item (‘‘I feel there is no solution to my problems other than taking my own life’’) was deleted. The construct of the twosubscales was supported by CFA (Chi squared value = 39.14, p = 0.04, CIMIN/DF = 1.57, RMR = 0.03, GFI = 0.96, NFI = 0.96, CFI = 0.98, RMSEA = 0.05). The scaling success rates of item-convergent validity and item-discriminant validity were all 100 % for the two subscales. The SIS-K scores were significantly higher in patients who had suicidal attempt experience, implying satisfactory known-groups validity. The Cronbach’s alpha of the total SIS-K was 0.91, and those of the subscales were ranged from 0.85 to 0.89. CONCLUSIONS: The study provided the excellent psychometric properties of the ISI-K. It is thus appropriate for use with respect to reliability and validity in practice and clinical trials for Korean patients with mental disorders.

(2039) Gender differences in cognitive debriefing of clinical outcomes assessments (Coa) and the impact of the interviewer Matthew Talbert, Corporate Translations Inc, Chicago, IL, United States; Barbara Brandt, Corporate Translations Inc., East Hartford, CT, United States; Carolyn Schulz, Corporate Translations, East Hartford, CT, United States; Shawn McKown, Corporate Translations Inc., East Hartford, CT, United States; Mary Gawlicki, Corporate Translations, East Hartford, CT, United States AIMS: A previous study analyzed the difference between male and female contributions to cognitive debriefing, and whether the gender of subjects has an impact on the final translation of Clinical Outcomes Assessments (COA). While the previous study deemed that subject gender made no impact on final translations resulting from cognitive debriefing, the gender of the interviewer was not factored into that analysis. This follow-up study seeks to test the hypothesis that the interviewer will have an impact on the final translation. METHODS: The total number of comments made during cognitive debriefing was tabulated, with a focus on comments which resulted in a revision to the final translation. Four groups were analyzed using a Chi Square test: Male subjects interviewed by a male interviewer, male subjects interviewed by a female interviewer, female subjects interviewed by a female interviewer and female subjects interviewed by a male interviewer. The alpha for this study was 0.05 with a degree of freedom of 1. The Chi Square test was run twice, once for total comments made and again for comments that resulted in a revision to the final translation. RESULTS: For total comments made, the Chi Square result was 34.25. Given the alpha of 0.05, this shows that the genders of the subjects and the interviewer are significant with regard to the number of comments made overall, some of which may be stylistic revisions or suggestions that are not equivalent to the source. It was observed that more comments appeared to be made when a male interviews female subjects. For comments resulting in a change to the translation, the Chi Square result was 0.165, showing that the genders of the subject and interviewer will not impact the number of comments made that result in translation revisions. CONCLUSIONS: The gender of the interviewer and subjects may impact total comments made during cognitive debriefing. However, the results indicate that the interviewers’ and subjects’ genders may not impact the number of comments that result in translation revisions.

123

Qual Life Res (2015) 24:1–191 (2041) Calibration and validation of the Dutch-Flemish PROMIS pain behavior and pain interference item banks in patients with rheumatoid arthritis Martine H. Crins, Amsterdam Rehabilitation Research Center | Reade, Amsterdam, Netherlands; Caroline B. Terwee, VU University Medical Center, Amsterdam, Netherlands; Rene Westhovens, Department of Development and Regeneration, Skeletal Biology and Engineering Research Center, KU Leuven,, Leuven, Belgium; Dirkjan van Schaardenburg, Jan van Breemen Research Institute | Reade, Amsterdam, Netherlands; Niels Smits, Department of Clinical Psychology and Department of Methodology, the EMGO Institute for Health and Care Research, VU University Medical Center, Amsterdam, Netherlands; Johan Joly, Department of Development and Regeneration, Skeletal Biology and Engineering Research Center, KU Leuven, Leuven, Belgium; Patrick Verschueren, Department of Development and Regeneration, Skeletal Biology and Engineering Research Center, KU Leuven, Leuven, Belgium; Kristien van der Elst, Department of Development and Regeneration, Skeletal Biology and Engineering Research Center, KU Leuven, Leuven, Belgium; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Karon F. Cook, PhD, Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Joost Dekker, Department of Rehabilitation Medicine and Department of Psychiatry, VU University Medical Center, Amsterdam, Netherlands; Maarten Boers, VU University Medical Center, Amsterdam, Netherlands; Leo D. Roorda, Amsterdam Rehabilitation Research Center | Reade, Amsterdam, Netherlands AIMS: The aims of current study were to calibrate the Dutch-Flemish translation of the PROMIS Pain Behavior (DF-PROMIS-PB) and Pain Interference (DF-PROMIS-PI) item banks in Dutch and Flemish patients with rheumatoid arthritis (RA), and to evaluate the reliability, construct validity and Dutch vs. Flemish cross-cultural validity. METHODS: 1182 Dutch and 650 Flemish RA patients completed a paper-and-pencil or web-based survey, including the full DF-PROMIS-PB (39 items) and DF-PROMIS-PI (40 items). One-factor confirmatory factor analysis assessed unidimensionality. Item response theory (IRT) models evaluated the item characteristics of the two item banks, to facilitate future development of computer adaptive tests (CAT). A graded item response model (GRM) was fitted. Construct validity was studied by comparing the results with scores on two legacy instruments (the Health Assessment Questionnaire (HAQ) and the Short Form 36 Health Survey (SF-36)). Ordinal regression models evaluated Differential Item Functioning (DIF) for e.g. language (Dutch vs. Flemish) to analyse cross-cultural validity. RESULTS: The results support unidimensionality of the DF-PROMIS-PB and DFPROMIS-PI (CFI = 0.976; 0.997 respectively and TLI = 0.975; 0.997 respectively). The first factor accounted for 50 % (DF-PROMIS-PB) and 80 % (DF-PROMIS-PI), respectively, of the total questionnaire variance. Fourteen of 741 (1.9 %) DF-PROMIS-PB item pairs and 12 of 780 (1.5 %) DF-PROMIS-PI item pairs were marked as possibly locally dependent. The data of the two item banks fit the GRM. The threshold-parameters ranged from -1.68 to 3.78 (DF-PROMIS-PB) and from -0.46 to 5.39 (DF-PROMIS-PI) respectively; in both item banks some response categories were rarely used. The item banks showed good cross-cultural validity: only 1 DFPROMIS-PB and no DF-PROMIS-PI items showed DIF between Dutch and Flemish language. Furthermore, the item banks showed good reliability (Cronbach’s alpha = 0.985 (DF-PROMIS-PB); 0.989

Qual Life Res (2015) 24:1–191 (DF-PROMIS-PI). Analyses of construct validity are in progress and will be presented at the conference. CONCLUSIONS: The DFPROMIS-PB and DF-PROMIS-PI can be used to develop two CATs for measuring pain behavior and pain interference respectively, in Dutch and Flemish RA patients.

(2043) Psychometric validation of the Spanish ManchesterOxford Foot Questionnaire (MOXFQ) in patients with foot or ankle surgery Sabine Goldhahn, MD, AO Foundation, Duebendorf, Switzerland; Juan B. Gerstner Garces, MD, Centro Me´dico Imbanaco Cali, Cali, Colombia; Ian Winson, MB ChB FRCS, Avon Orthopaedic Centre, Southmead Hospital, Westbury on Trym, Bristol, United Kingdom; Andrew K. Sands, MD, New York-Presbyterian Hospital – Lower Manhattan Hospital, New York, NY, United States AIMS: There is a lack of foot and ankle specific patient-reported outcome measures in Spanish. Therefore, the Manchester-Oxford Foot Questionnaire (MOXFQ) has been cross-culturally adapted and validated in Spanish (Colombia) for use in patients undergoing foot and ankle surgery. METHODS: The MOXFQ was translated and cross-culturally adapted to Colombian Spanish based on established guidelines. This validation study included 120 patients (73.3 % female) with a mean age of 50.4 ± 17.0 years who completed the MOXFQ and the SF-36 Health Survey twice before surgery, and during follow-up examinations at 6 and 12 months after surgery. In parallel, surgeons completed the American Orthopaedic Foot and Ankle Society (AOFAS) Clinical Rating System. Internal consistency, test–retest reliability, floor and ceiling effects, construct validity and responsiveness were assessed for the three MOXFQ dimensions Pain, Walking/standing (WS), Social interaction (SI), and the MOXFQ Index. A general transition item was asked about perceived changes related to foot and ankle between the status before surgery and at follow-ups. RESULTS: The Spanish MOXFQ demonstrated consistency with Cronbach’s alpha values between 0.65 (SI) and 0.90 (WS and MOXFQ Index), and is highly reliable (intraclass correlation coefficient [ICCs] [ 0.95 for all dimensions). A moderate to strong correlation was demonstrated between the WS dimension and the related domains of the SF-36 (|r| [ 0.6), the AOFAS Ankle-Hindfoot Scale (|r| [ 0.47) and Hallux-MTP-IP Scale (|r| [ 0.64). The Spanish MOXFQ showed excellent responsiveness between Baseline and postoperative FU visits, with large effect sizes (ES) (ES [ 2.1 for all dimensions and MOXFQ Index). ES and standardized response mean were higher for the MOXFQ and the AOFAS than for the SF-36. On a 90 % confidence level, the respective minimal detectable change (MDC90) was 22.64, 17.47, 32.59 and 14.18 for the MOXFQ Pain, WS and SI dimensions, and the MOXFQ Index. The mean scores of all MOXFQ dimensions and the MOXFQ Index changed in the same direction as the general transition item. CONCLUSIONS: The Spanish (Colombia) version of the MOXFQ showed good psychometric properties in patients with foot and ankle disorders and can help to bridge the gap between the large Spanish-speaking population worldwide and the current lack of foot and ankle specific PROMs in Spanish.

(2045) Challenges in translating the adult version of the transfusion-dependent QoL (TranQoL) Questionnaire into Arabic, French, Greek, Italian and Turkish Cedric Roulliat, Mapi, Lyon, France, Robert J. Klaassen, University of Ottawa, Children’s Hospital of Eastern ON, Ottawa, ON, Canada; Dalia Mahmoud, Celgene, Summit, NJ, United States; Nazilabee Toraubully, Mapi, Lyon, France

121 AIMS: The Transfusion-Dependent Quality of Life (TranQoL) questionnaire was developed to assess the QoL of children and adults with transfusion-dependent thalassemia. The adult version is self-reported and includes 37 items, rated on a 5-point scale (never to always), with a ‘‘past week’’ recall period. The objective of this study was to identify the main challenges encountered during the translation of the TranQoL adult version into Arabic, French, Greek, Italian and Turkish. METHODS: In each country, the linguistic validation was conducted using the standard forward/backward methodology with reviews by the author of the questionnaire, and cognitive interviews with 10 patients. The basis for discussion was the list of concepts elaborated with the author. For each country, the history grid of the translation process was analyzed to identify the most problematic items across countries. RESULTS: Twelve items raised a lot of discussion either during the forward step or during the interviews, and required additional input from the author. Most of the issues were semantic, e.g., finding the appropriate equivalent for words or expressions such as ‘‘emotional health,’’ ‘‘peers,’’ ‘‘social events,’’ ‘‘significant others,’’ ‘‘academic performance’’, or ‘‘iron removal therapy.’’ The syntactic structure of items 5 and 6 (i.e., passive form: ‘‘I was limited in my ability to do the kind of [moderate (item 5)/ vigorous (item 6)] work I take part in.’’) could not be kept in Arabic, Greek and Turkish. In addition, the concept of usual activity which was more or less implicit (i.e., I take part in) in the original had to be explicit in French and Greek. Items 29/30 about the disease financial burden were easy to understand, but generated discussion since French and some Italian respondents argued that they were not concerned since their expenses were covered at 100 % by Social Security. However, it was clear for them that, in this case, they had to answer ‘‘never.’’ CONCLUSIONS: The involvement of the author (development of the list of concepts and continuous reviews), as well as the cognitive interviews, proved to be critical to the translation process and enabled the development of conceptually and culturally relevant translations of the TranQoL.

(2047) Danish translation and linguistic validation of the PatientReported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) Christina Baeksted, PhD Student, MSc in Public Health, Rigshospitalet, Copenhagen University Hospital and Danish Cancer Society, Copenhagen, Denmark; Aase Nissen, MSc Pharm., MPH, Danish Cancer Society, Copenhagen, Denmark; Helle Pappot, MD DMSc, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark; Pernille E. Bidstrup, MA, PhD, Danish Cancer Society Research Center, Copenhagen, Denmark; Sandra A. Mitchell, PhD, CRNP, AOCN, National Cancer Institute, Rockville, MD, United States; Ethan Basch, MD, MSc, University of North Carolina, Chapel Hill, NC, United States; Susanne O. Dalton, MD, PhD, Danish Cancer Society Research Center, Copenhagen, Denmark; Christoffer Johansen, Professor, MD, PhD, Dr Med Sci, Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark AIMS: The Common Terminology Criteria for Adverse Events (CTCAE) is the basis for standardized clinician-based grading and reporting of adverse events in patients participating in cancer clinical trials. The US National Cancer Institute has developed a companion measure, the Patient-Reported Outcomes version of the CTCAE (PRO-CTCAE), to incorporate patient self-reporting of symptomatic toxicity into adverse event reporting. The aim of the current study was to translate and linguistically validate a Danish version of PROCTCAE. METHODS: The US English language PRO-CTCAE was translated into Danish using a forward and backward procedure with reconciliation. The linguistic validity of the PRO-CTCAE was

123

122 Table 1 Characteristics of the 56 patients who participated in the linguistic validation

examined in two successive rounds of semi-structured cognitive interviews with 56 patients equally distributed by gender and cancer site (prostate, head and neck, lung, breast, gynecological, gastrointestinal and hematological cancer); all were currently undergoing cancer treatment (Table 1). RESULTS: In the translation of PROCTCAE into Danish, no substantial differences were observed between the two translators in the forward or in the backward translation. In the first round of linguistic validation (n = 42), the phrasing of five symptom terms was adjusted, and the refined phrasing was tested in a second round of interviews (n = 14). For these five symptom terms, consensus around phrasing that was both culturally acceptable and semantically comprehensible was achieved in the second round of interviewing. Across the two rounds, statements from participants describing the meaning of the PRO-CTCAE terms support the conceptual equivalency to the English version. CONCLUSIONS: The availability of the NCI PRO-CTCAE in languages other than English will support international congruence in self-reporting of side effects of cancer treatment. A rigorous methodology was used to develop and test a Danish language version of PRO-CTCAE; results provide support for the use of PRO-CTCAE in cancer clinical trials that include Danish speakers.

(2049) Improving the reporting quality of instrument crosscultural adaption: the IRICA statement Zheng-kun Hou, MD, The First Affiliated Hospital of Guangzhou University of Chinese Medicine, Guang Zhou, China; Feng-bin Liu, MD, The First Affiliated Hospital of Guangzhou University of

123

Qual Life Res (2015) 24:1–191 Chinese Medicine, Guang Zhou, China; Yuan-kun Tan, PhD, Guangzhou University of Chinese Medicine, Guangzhou, China AIMS: Reporting quality is vital for the researchers and physicians to prepare the paper and assess the scientific properties for translated instruments. The article aims to propose the preferred reporting items for instrument cross-cultural adaption. METHODS: Abiding by the guidance recommended by the Enhancing the quality and transparency of health research (EQUATOR) Network, we identified the need for a guideline, performed a literature review, obtained funding for the guideline initiative and identified participants. RESULTS: A comprehensive literature search were performed in PubMed, IEEE Xplore, and A catalogue of reporting guidelines for health research in EQUATOR Network, we found few studies focused on the reporting items for instrument cross-cultural adaptation. With the same literature search strategy, two independent researchers and one coordinator extracted the study information with a standardized survey form. Excluding 1532 articles, the study group included 19 instrument translate research guidelines and 30 high quality reviews for preliminary item pool establishment, and formed a 70 items checklist. After three waves of focus groups discussion (July and August 2014, February 2015, Guangzhou), a checklist draft about the preferred reporting items for instrument cross-cultural adaption was put forward, which contains 25 items subheadings under 6 topics: title and structured abstract are subheading under ‘Title and Abstract’, rationale and objective are subheading under ‘Instruction’, Authorization, Participants Criteria, Forward Translations, Forward Synthesis, Backward Translations, Backward Synthesis, Experts Qualitative Review, Pilot Testing, International Harmonization, Field Testing, Statistical methods are subheading under ‘Methods’, Participants, Series Instruments, Main results, Other analyses are subheading under ‘Results’, Summary of evidence, Comparison of evidence, Limitations, Final Permission, Clinical attentions, Conclusions are subheading under ‘Discussion’, Appendix, Funding are subheading under ‘Other information’. Now, the online survey systems were established for worldwide Delphi exercise, and the expert information from the International Society for Quality of Life Research (ISOQOL) Translation & Cultural Adaptation Special Interest Group, The Cochrane Collaboration Patient-Reported Outcomes Methods Group, et al., were identified for further Delphi survey. CONCLUSIONS: The preferred reporting items assist researchers to report their results more accurately and evaluate the quality of research methods. The international Delphi survey is ongoing and a more strict preferred reporting checklist was expected

(2051) Chinese–English language equivalence of the short form 12-item health survey Daniel Y. Fong, PhD, The University of Hong Kong, Hong Kong SAR, Hong Kong SAR, China; Janet Y. Wong, PhD, RN, The University of Hong Kong, Hong Kong, Hong Kong SAR, China; Cannas Kwok, Lecturer, University of Western Sydney, Penrith NSW, Australia; Kwok-Fai Lam, The University of Hong Kong, Hong Kong, Not, China AIMS: The Short Form 12-item health survey (SF-12) was originally developed in English, which is also available in Hong Kong (HK) Chinese. While both language versions had their measurement properties well assessed in their respective populations, their equivalence in scores has not been examined. Therefore, we aimed to assess the language equivalence of the English and HK Chinese versions of the SF-12. METHODS: We conducted a cross-sectional study on individuals aged 18 years or above in a university campus via both an online platform and papers. Those who were bilingual in English and Chinese were randomly assigned to self-complete either

Qual Life Res (2015) 24:1–191 the standard English SF-12v2 or the HK Chinese SF-12v2, in additional to questions on demographics. Scores of the eight scales and the two US norm-based components from the two language versions of the standard SF12v2 were compared by linear regression, with and without adjustment for age and gender that were known to be associated with quality of life. The equivalence in a scale would be concluded if the 95 % confidence interval (CI) for the difference between the two language versions completely fell within the minimal clinically important difference of ±3 units. RESULTS: A total of 795 participants bilingual in English and Chinese consented their study participation. Of which, 756 completed the sf12v2 (273 in English and 483 in HK Chinese), with mean age of 24 years (Range 18–68) and 508 (67 %) were female. Only the mental health (MH) scale, physical component scale (PCS) and mental component scale (MCS) had their 95 % CIs (MH: -2.85 to 2.52; PCS: -0.99 to 1.01; MCS: -1.44 to 1.52) completely fell within the ±3 units. The other seven scales had their 95 % CIs covered ± 5 units or more. After adjusting for age and gender, the 95 % CIs closely resembled those without adjustment. CONCLUSIONS: Language equivalence of the SF-12v2 was established mainly in the two component scales, but generally not in the eight scales. The PCS and MCS of the SF-12v2 may be used as quality of life outcomes in cross-cultural studies involving both Chinese and English speaking individuals.

(2055) Professional proofreading: a new step in the EORTC translation procedure to further improve output quality Dagmara Kulis, EORTC, Brussels, Belgium; Cheryl Whittaker, EORTC, Brussels, Belgium; Edite Fiskovica, EORTC, Brussels, Belgium; Eva Greimel, University Hospital Graz, Graz, Austria, Andrew Bottomley, EORTC Headquarters, Brussels, Belgium; Michael Koller, University Hospital Regensburg, Regensburg, Germany AIMS: The European Organisation for Research and Treatment of Cancer (EORTC) Quality of Life Group (QLG) is a leading developer of patient-reported outcomes measures in oncology [1]. One of the strengths of their questionnaires is broad language coverage [2]. New translations are done according to the EORTC Translation Manual [3], which conforms to the ISPOR Principles of Good Practice for the Translation and Cultural Adaptation of PRO Measures [4], striving for the highest quality possible, based on evidence in the literature. In order to enhance the quality of academic translations, a new step was introduced to the procedure: proofreading by a professional translator of the reconciled translation before pilottesting. The aim of this abstract is to show the benefits of the introduction of this independent proofreading. METHODS: New language versions are developed within the module development process or upon request from external users. Depending on the nature of the request (academic or commercial), they are done by volunteers (academic) or translation services providers (commercial), and always coordinated by the EORTC Translation Unit (TU). The EORTC Translation Procedure [3,5] includes two forward translations, reconciliation, two back translations, review by the TU, pilot-testing, review by the TU, and preparation of the final version. Translations done by translation services providers include multiple proofreading and other quality assessment steps, while translations done by academics included only reviews of the back translations by the TU. The step in which the academic translation is proofread by a professional subcontracted translator was introduced to remedy this situation and improve the quality of output. To show the actual benefits, we analysed all the proofreading reports for translations following the new procedure. RESULTS: 23 academic translations have been proofread up to April 2015. A total of 777 segments (instructions, response scales, time frames, items, and other

123 elements of the questionnaire) were proofread. Proofreaders commented on 395 segments, and academic collaborators agreed with 364 of those. CONCLUSIONS: The introduction of the additional proofreading has already enabled the identifying and solving of various problems in new language versions developed by non-linguists. With the high rate of acceptability of proofreaders’ comments, we can conclude they are correct and necessary. BIBLIOGRAPHY [1]

[2]

[3]

[4]

[5]

Bottomley A. and Aaronson NK. International perspective on health related quality of life research in cancer clinical trials: The European Organisation for Research and Treatment of Cancer Experience. Journal of Clinical Oncology 2007;25(32): 5082-5086. Kulis´ D., Arnott M., Greimel ER., Bottomley A., Koller M. Trends in translation requests and arising issues regarding cultural adaptation. Expert Rev Pharmacoecon Outcomes Res. 2011;11(3):307-14. PMID: 21671700. Dewolf L., Koller M., Velikova G., Johnson C., Scott Neil, Bottomley A. EORTC Quality of Life Group: Translation Procedure. 3rd Ed. Brussels: Quality of Life Study Group Publications. EORTC Publication; 2009. ISBN 978-2-930064-38-3. Wild D. et al. Principles of Good Practice for the Translation and Cultural Adaptation Process for Patient-Reported Outcomes (PRO) Measures: Report of the ISPOR Task Force for Translation and Cultural Adaptation. Value in Health 2005; 8(2). Koller M., Kantzer V., Mear I., Zarzar K., Martin M., Greimel E., Bottomley A., Arnott M., Kulis´ D. The ISOQOL TCA-SIG. The process of reconciliation: evaluation of guidelines for translating quality-of-life questionnaires. Expert Rev Pharmacoecon Outcomes Res. 2012 Apr;12(2):189-197.

(2057) Why two are better than one: the added value of the second back translation in the EORTC translation procedure Dagmara Kulis, EORTC, Brussels, Belgium; Cheryl Whittaker, EORTC, Brussels, Belgium; Edite Fiskovica, EORTC, Brussels, Belgium; Eva Greimel, University Hospital Graz, Graz, Austria; Andrew Bottomley, EORTC Headquarters, Brussels, Belgium; Michael Koller, University Hospital Regensburg, Regensburg, Germany AIMS: The back translation (BT) is widely accepted as an essential step in the translation of patient-reported outcomes measures. However, the ISPOR Principles on Good Practice for Translation and Cultural Adaptation of PRO Measures include a BT step without specifying how many BTs are needed. While some guidelines state that one BT suffices, the European Organisation for Research and Treatment of Cancer (EORTC) Translation Manual requires two BTs. The aim of this abstract is to show the scientific need based on the added value of two BTs. METHODS: The 20 most recent BT reports (finalized between December 2014 and April 2015), all consisting of two forward translations, a reconciled translation and two BTs, were reviewed and the differences between the two BTs were analyzed. RESULTS: In all 20 reports differing BTs were present, which is a natural result of the BT process. In 18 reports we found cases in which one of the two BTs was significantly different, with the discrepancy being big enough to lead the reviewer to ask for confirmation of which BT was correct and closer to the source. In 5 reports we found 8 items in which the two BTs differed significantly, leading to questions from the reviewer and, upon confirmation that the reconciled translation

123

124

Qual Life Res (2015) 24:1–191

was wrong, to eventual rewording of the reconciled version (Table 1). CONCLUSIONS: In almost all reports, the differences between the two BTs led to questions from the reviewer regarding the correct wording of the reconciled translation. These questions, and the responses to them, provide an essential quality-control tool in the translation process. The two BTs approach has proven beneficial in achieving a high quality of translations and offers an additional view on the reconciled translation: its wording and structure. Therefore, despite the fact that a two BT process is costand resource-intensive, the update of the EORTC Translation

Manual will continue to require two separate BTs in the translation process.

Table 1 .

AIMS: To validate the performance of a Hindi version of Chronic obstructive pulmonary disease Assessment Test (CAT) for assessing health status of north Indian patients with chronic obstructive pulmonary disease (COPD). METHODS: 178 patients of stable COPD completed the Hindi version of CAT twice at 4 weeks interval. All eight items of the CAT were scored on a five-point Likert scale and responses summated to obtain total score. Patients also self-completed Hindi versions of abbreviated World Health Organization Quality of Life questionnaire (WHOQOL-Bref) and St. George’s Respiratory Questionnaire (SGRQ) at initial assessment. Baseline clinical details and spirometric data were also recorded. Construct validity was assessed by noting the correlation between individual item and total CAT scores, and other objective measures of disease severity (forced expiratory volume in first second [FEV1], dyspnea grade) and scores of WHOQOL-Bref and SGRQ questionnaires. Internal consistency was evaluated using Cronbach’s alpha (correlation of individual items with each other) on data from the baseline administration of CAT. Test–retest reproducibility was assessed, on a patient subgroup with stable disease between the two clinic visits, using random effects intraclass correlation coefficient (ICC) between two CAT administrations. Responsiveness was assessed through effect size among patients who had deteriorated between the two assessments. RESULTS: There were 167 (93.8 %) men and 159 (89.3 %) smokers in the study population. 127 (71.3 %) patients completed the second assessment, of whom 19 had deteriorated. Each of the eight CAT items correlated strongly with total CAT score (Pearson coefficients 0.59-0.73). Total CAT score correlated moderately with FEV1 and dyspnea grade, and strongly with all WHOQOL-Bref and SGRQ domain scores. Cronbach’s alpha coefficient had high value of 0.83. ICC for 108 patients having stable disease between two assessments was high (0.91), but effect size in the 19 patients who deteriorated was moderate (0.45). CONCLUSIONS: The Hindi version of CAT has good validity and reliability and can be used to quantify impact of COPD on the life of north Indian patients.

(2059) Psychometric validation of a Hindi version of chronic obstructive pulmonary disease assessment test (CAT) Ashutosh N. Aggarwal, Professor, Postgraduate Institute of Medical Education and Research, Chandigarh, India; Isak Lallawmkima, Junior Resident, Postgraduate Institute of Medical Education and Research, Chandigarh, India; Debasish Basu, Professor, Postgraduate Institute of Medical Education and Research, Chandigarh, India

Acknowledgements: This abstract presents independent research funded by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research Programme (Reference Number RP-PG-0611-20008). The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR or the Department of Health. We also acknowledge support from the MRC ConDuCT-II Hub.

Cancer (2061) The newly derived four-factor structure of the FACIT-SpEx: comparing cancer survivors to other cancer/community groups Hayley S. Whitford, PhD, Cancer Council Australia, Sydney, New South Wales, Australia; The University of Adelaide, Adelaide, South Australia, Australia; Ian Olver, MD, PhD, FRACP, Cancer Council Australia, Sydney, New South Wales, Australia; Sansom Institute for

123

Qual Life Res (2015) 24:1–191 Health Research, University of South Australia, Adelaide, South Australia, Australia AIMS: This study aimed to determine if the four newly-derived factors of the 23-item expanded Functional Assessment of Chronic Illness Therapy—Spiritual Well-Being (identified in a sample of cancer survivors), were reliable in other cancer/community groups, and whether such groups differed across spiritual well-being domains. METHODS: Of 2,290 online survey respondents administered the FACIT-Sp-Ex or the FACIT-Sp-Ex-NI (non-illness version with slight wording alteration to 2/23-items), 52.1 % cancer survivors were compared to 5.4 % chronically ill, 9.4 % informal cancer caregivers, 4.1 % cancer health professionals, and 29.0 % general population participants. RESULTS: Reliability (Cronbach’s a’s) of the FACITSp-Ex subscales were high and comparable to those of the FACIT-SpEx-NI; Peace (.83 vs .80), Meaning (both .87), Faith (.92 vs .91), Positive Interaction (.86 vs .83), and Total Spiritual Well-Being (.91 vs .90). Of basic demographics, only age was controlled in analyses comparing survivors and other groups, given significant univariate associations. Cancer survivors evidenced significantly higher means compared to the chronically ill on Meaning (p = .01, partial g2 = .01) and informal caregivers on Peace (p = .03, partial g2 = .004). There were no significant differences between survivors and general population participants. However, survivors showed significantly lower means than health professionals across Peace (p = .005, partial g2 = .01), Meaning (p = .008, partial g2 = .01), Faith (p = .000, partial g2 = .05), Positive Interaction (p = .04, partial g2 = .004), and Total Spiritual Well-Being (p = .000, partial g2 = .01). The controlled impact of age was significant in most instances, with the highest effects reported for Peace across all group comparisons (2.8–3.6 % of the variance). CONCLUSIONS: Despite a previously difficult-to-interpret factor analysis of the FACIT-Sp-ExNI in non-ill populations, compared to cancer survivors on the FACIT-Sp-Ex, the resulting subscales were equally reliable in patient/survivor and non-ill groups. It is not unexpected to find some higher spiritual well-being outcomes for survivors compared to chronically ill and informal caregiver groups, given some studies report poor or equally challenged psychological well-being to survivors, in these specific samples. More interestingly, survivors were found to have comparable spiritual well-being outcomes to the general population with only health professionals showing higher outcomes. Although this may indicate greater existential well-being in those working with survivors, the magnitude of effects should be noted as only small-to-negligible.

(2063) Defining responders on the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life questionnaire (30-item core module) (QLQ-C30) subscales Kim Cocks, KCStats Consultancy, Leeds, United Kingdom; Jacqueline Buchanan, Onyx Pharmaceuticals, Inc., an Amgen subsidiary, South San Francisco, CA, United States AIMS: A responder can be defined as an individual who reports a change in a HRQOL score that would provide evidence of a treatment benefit. This allows the interpretation of HRQOL data by comparing the proportion of patients between treatment groups who meet the responder definition (RD). One can also consider time to deterioration/improvement based on the RD. Responder analysis is concerned with change at the individual patient level so it is important to understand how this translates on different HRQOL scales. Most HRQOL scores are based on underlying discrete Likert scales and minimal important differences (MID) derived from group averages may not translate to a plausible change score on an individual basis. Using the QLQ-C30 we aim to provide guidance on appropriate RDs.

125 METHODS: The QLQ-C30 scoring manual was used to calculate all possible scores for each scale. The minimum possible change in score can inform the choice of a RD. Existing literature was searched to identify the currently used RDs. RESULTS: It is common to use change scores of 5, 10 and 20 points or 10 % to define responders across QLQ-C30 subscales based on MID. The minimum change an individual can achieve on the QLQ-C30 subscales ranges from 6.7 (physical functioning scale comprising 5 items) to 33.3 (single item scales) (see Table). These minimum changes are achieved when one item on the subscale changes by one category on the Likert response criteria assuming all other items on the subscale have remained the same. For all QLQ-C30 subscales a RD of 5 points therefore would consider any patient who had this minimal change as a responder. For subscales with single, two (4-point Likert), or three items, the 5- and 10-point RDs are identical since no patient can achieve a change below 10 points on these subscales. CONCLUSIONS: The definition of responders on HRQOL scales requires careful consideration. Published group MID definitions are not directly applicable to individual changes and should not be automatically applied to define responders. Consideration of the discrete nature of the HRQOL scores must be accounted for when defining responders.

(2065) Utilization of patient reported outcomes (PROs) tools in studies that include pancreatic cancer Agnes Hong, Pfizer Inc, New York, NY, United States; Giovanni Zanotti, PharmD MSc, Pfizer, New York, NY, United States; Shrividya Iyer, PhD, Pfizer Inc., New York, NY, United States AIMS: To summarize the patient-reported outcomes (PROs) measurement tools used and inclusion of quality of life (QoL) endpoints in studies with pancreatic cancer. METHODS: A literature search was conducted in PubMed and Google Scholar for clinical trials and observational studies that investigated pancreatic cancer in human subjects between 1990 and 2015. Full-text articles and abstracts were found using the following keywords: ‘‘pancreatic cancer’’, ‘‘patient reported outcomes’’, ‘‘quality of life’’ and ‘‘symptoms’’. Each article was evaluated for frequency, type of PROs and inclusion of QoL as endpoints in the study. Studies were excluded if the disease type was other than pancreatic cancer, evaluated non-drug related treatments or were questionnaire validation studies. RESULTS: The review resulted in the selection of 57 studies, 34 (60 %) of which included PROs tools to measure QoL. 12 studies included QoL as primary endpoints, 17 studies included QoL as secondary endpoints and 5 studies included QoL as general endpoints. For the 34 studies, PROs were used 63 times, and an average of 2 PROs tools were used per study with a range of 1–4. A total of 22 different types of PROs measurement tools were identified. 13 of the tools were symptom-specific and most commonly assessed pain that was often reported as abdominal and back pain. Other diseaserelated symptoms measured throughout the studies were fatigue, depression, mood and anxiety. QoL was assessed with 9 different validated questionnaires. Of the 63 times PROs tools were used in the studies, global QoL tools were used 26 (41 %) times, healthrelated QoL tools were used 16 (25 %) times, a generic QoL tool was used once (2 %) and symptom-specific evaluation tools were used 20 (32 %) times. CONCLUSIONS: PROs were included in 60 % of the clinical trials and observational studies reviewed. Over the last decade, there has been an increasing trend in publication of studies with pancreatic cancer patients that include PROs as endpoints. Furthermore, evaluation of PROs in pancreatic cancer patients will provide a more comprehensive understanding of disease symptoms, clinical benefits and QoL from a patient’s perspective.

123

126 (2067) Assessment of patient-reported adverse events after discharge from hospital following gastrointestinal cancer surgery: do EORTC disease-specific modules have sufficient coverage to monitor recovery and detect problems early? Elaine O’Connell Francischetto, Research Associate, University of Bristol, Bristol, United Kingdom; Kerry Avery, University of Bristol, Bristol, United Kingdom, Galina Velikova, MD PhD, Leeds Institute of Cancer Studies and Pathology University of Leeds, Leeds, United Kingdom; Jane Blazeby, University of Bristol, Bristol, United Kingdom AIMS: There is an increasing need for adverse event (AEs) monitoring after hospital discharge following surgery because of the widespread use of enhanced recovery programmes and short hospital stays. Currently, measures to assess patient-reported AEs are limited. It is possible that the European Organisation for Research and Treatment of cancer (EORTC) quality of life questionnaires (QLQ) (designed to cover symptoms and functions) are relevant in this setting. The aim of this research was to assess if AEs covered in EORTC QLQ upper gastrointestinal (GI) modules could be used to assess AEs post-discharge. METHODS: Questionnaire items and scales in the EORTC QLQ-C30 and the five validated upper GI modules were analysed in MS Excel. Items assessing chemotherapy and radiation treatment side effects and general psycho-social scales and items were excluded. Residual items and scales with overlapping content were scrutinised and, where published validation papers showed significant associations between module and core questionnaire scales and items, only the C30 items were retained. Semi-structured audio-recorded interviews were then conducted with a purposive sample of upper GI surgery patients to ensure the questionnaire covered relevant AEs. RESULTS: There were 134 items identified from the C30 and five upper GI modules, made up of 72 scales/single items. Methods initially excluded 17 items/scales relevant only to chemoradiation treatment or psycho-social issues. The remaining scales/items included 13 overlapping scales/items in the upper GI modules, so decisions were made on which scale to use. Associations between diseasespecific pain and fatigue scales were at least moderately correlated with scales in the C30, so only the C30 scales were retained. The shorter physical functioning scale in the EORTC QLQ C15-PAL was used instead of the five-item C30 scale. The final questionnaire comprised of 34 items, including 19 EORTC QLQ scales. Interviews with 18 patients showed sufficient coverage except for wound-related problems, which were not covered by EORTC questionnaires. CONCLUSIONS: Existing validated disease-specific EORTC modules can be utilised to form an upper GI questionnaire to measure AEs in surgical patients. Further work is underway to identify clinically important thresholds for these EORTC symptom scales that could be used to trigger an alert to medical staff.

(2069) Psychometric Assessment of the Functional Assessment of Cancer Therapy/Gynecologic Oncology Group-Neurotoxicity (FACT/GOG-Ntx) Questionnaire in Patients with Relapsed and Refractory Multiple Myeloma (RRMM) Receiving Carfilzomib Treatment Elizabeth Bacci, Evidera, Seattle, WA, United States; Kim Cocks, KCStats Consultancy, Leeds, United Kingdom; Dana Hurley, Onyx Pharmaceuticals, Inc., an Amgen subsidiary, South San Francisco, CA, United States, Sumeet Panjabi, Onyx Pharmaceuticals, Inc., an Amgen subsidiary, South San Francisco, CA, United States; Dennis Revicki, PhD, Evidera, Bethesda, MD, United States AIMS: Assess the psychometric properties of the FACT/GOG-Ntx questionnaire, a measure of health-related quality of life (HRQL) completed by patients with RRMM treated with single-agent

123

Qual Life Res (2015) 24:1–191 carfilzomib. METHODS: Post-hoc analyses of HRQL data from patients enrolled in a Phase 2 trial (PX-171-003-A1) (N = 266) who previously were treated with bortezomib and an immunomodulatory agent. FACT/GOG-Ntx was completed at screening and on Day 1 of each 28-day cycle (Cycles 3, 5, 7, 9, 11), and end of treatment (EOT). Psychometric properties of ‘well-being’ (WB) subscales, (Physical [PWB], Social [SWB], Emotional [EWB], Functional [FWB]), neurotoxicity symptoms (Ntx), and total scores (Total FACT-G, Trial Outcome Index [TOI]-Ntx, Total FACT/GOG-Ntx) were assessed (higher scores indicate better HRQL). Reliability was examined using Cronbach’s alpha at all cycles. Known-groups validity was evaluated by comparing mean scores at baseline by Eastern Cooperative Oncology Group (ECOG) performance status (PS 0, 1, or 2) and peripheral neuropathy (PN) grade (1 or 2 vs. none). Sensitivity to change was examined using analysis of variance to compare mean change between response groups (partial [PR], minimal response/ stable disease [MR/SD], or progressive disease [PD]) from baseline to Cycle 3. RESULTS: Median patient age was 63 years (37–87); 58 % were male. Cronbach’s alpha for all subscales and TOI-Ntx across all cycles suggested high reliability (a = 0.64–0.91) through Cycle 7, after which the limited number of remaining patients reduced the reliability for Total FACT-G (a \ 0.3). Patients with ECOG PS ‘0’ reported significantly higher total and subscale scores except on SWB, compared with PS ‘1’ or ‘2’ (p B 0.03). Patients with no PN reported higher mean Ntx subscale, TOI-Ntx, and Total FACT/GOGNtx scores versus those with PN Grade 1 or 2 (p \ 0.05). Total FACT-G, TOI-Ntx, Total FACT/GOG-Ntx, PWB, and FWB demonstrated sensitivity to change as mean improvement from baseline to Cycle 3 was significantly greater for the PR group versus the MR/SD (all p \ 0.05) and PD (all p \ 0.05) groups. Comparable findings were demonstrated for EWB, but not SWB. CONCLUSIONS: The FACT/GOG-Ntx has previously been shown to be valid in ovarian and gynaecological cancers. This study demonstrates reliability, known-groups validity, and sensitivity to change in patients with RRMM receiving carfilzomib.

(2071) Quality of life after patients progress in oncology clinical trials: the black box Kathryn Lasch, PhD, Pharmerit International, Newton, MA, United States; Amylou C. Dueck, PhD, Mayo Clinic, Scottsdale, AZ, United States; Ethan Basch, MD, MSc, University of North Carolina, Chapel Hill, NC, United States; Jie Zhang, Oncology Franchise Lead, Novartis Pharmaceuticals Corporation, East Hanover, NJ, United States; Vasudha Bal, Novartis Pharmaceuticals Corporation, East Hanover, NJ, United States; Erica G. Horodniceanu, Pharmerit International, Bethesda, MD, United States; Lavanya Sudharshan, Pharmerit International, Bethesda, MD, United States; Charles S. Cleeland, PhD, UT MD Anderson Cancer Center, Houston, TX, United States AIMS: While biomarkers and radiological evaluations are often used to evaluate progression in cancer patients, patient reported outcome (PRO) assessments are the only evidence of treatment benefit directly from patients. The value of PRO measures in cancer is increasingly recognized and used to support registration and reimbursement decisions as they help explain what clinical progression means from the patient perspective. However, collection of PRO data post-progression (i.e., when a patient’s disease has progressed based on clinical/radiologic markers), is rare. Such data are important to evaluate treatment benefit from many perspectives. This research aims to describe challenges and solutions to collecting/analyzing PRO data post progression. METHODS: A targeted literature review was conducted to investigate the current status of oncology trials collecting PRO data post progression. Three experts (clinical or

Qual Life Res (2015) 24:1–191 statistical) with extensive experience in the development, validation, collection, and analysis of PRO measures in oncology were interviewed twice to discuss current practices, value, challenges, and solutions of PRO data collection and analysis in the post-progression period. Literature review and individual expert interview results were shared and discussed via group WebEx with experts. The interviews and WebEx were recorded; data were analyzed qualitatively. RESULTS: Observational cohort studies have evaluated PRO measures most frequently in the post-progression period in breast and lung cancer. European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30) and Functional Assessment of Cancer Therapy (FACT) questionnaires were most often used. Data collection frequency ranged from 30 to 90 days postprogression. Challenges highlighted by experts included patient burden, data collection logistics costs, difficulty in tracking post-progression (patients lost to follow-up), treatment heterogeneity post trial, variability in cancer types and survival time. CONCLUSIONS: The value of collecting PRO data post-progression was acknowledged both in the literature and by experts. Experts suggested solutions to the identified data collection and analysis challenges. A key component to the success of such efforts will be standardizing outcomes used in evaluating patients in the post-progression period. Additional empiric work is needed to help understand feasibility and value of this information.

(2073) Using administrative data to identify eligible patients for the Consumer Assessment of Healthcare Providers and Systems (CAHPS) survey for cancer care prototype Kathleen J. Yost, Mayo Clinic, Rochester, MN, United States; Sue Visscher, Mayo Clinic, Rochester, MN, United States; Marlene Frost, Mayo Clinic, Rochester, MN, United States; Sarah Jenkins, Mayo Clinic, Rochester, MN, United States; Christian Evensen, American Institutes for Research, Chapel Hill, MN, United States; Melissa Robertson, Mayo Clinic, Rochester, MN, United States; James Naessens, Mayo Clinic, Rochester, MN, United States; Jason Egginton, Mayo Clinic, Rochester, MN, United States; Monica Van Such, Mayo Clinic, Rochester, MN, United States AIMS: The CAHPS Survey for Cancer Care Prototype (Cancer CAHPS, trademark pending) measures patients’ perceptions of their cancer care experience. Cancer care providers can compare their casemix adjusted performance scores to a benchmark of their peers to identify opportunities for improvement. Consumers can compare providers on performance to help decide where to receive care. Providers should use a standardized sampling approach when selecting patients to receive Cancer CAHPS. Otherwise, it is not possible to know whether observed differences in performance are due to real differences in cancer care quality or due to differences in sampling. Our objective is to describe the development and testing of a standardized sampling approach for Cancer CAHPS. This work was sponsored by the California Health Care Foundation. METHODS: The target population for Cancer CAHPS is very broad and includes adult patients who received surgery, radiation, or medical oncology treatment for cancer in the 6 months prior to being sampled. Previous attempts to standardize sampling of this population have proved challenging. We collaborated with stakeholders to identify sources of administrative data that would be readily available and comparable across cancer care providers. We developed a two-step approach: (1) create three separate sampling frames of eligible patients (surgery, radiation, or medical oncology), (2) draw random samples of patients from each frame to receive the survey. Duplicate cases who receive more than one modality of care are randomly selected to remain in only one sampling frame. The approach was tested using chart review. RESULTS: Our approach uses combinations

127 of ICD-9 diagnostic codes for malignant neoplasms, ICD-9 procedure codes, and Current Procedural Terminology (CPT) codes. Lists of codes were approved by clinical experts. Chart reviews confirmed that patients identified using the sampling approach met eligibility criteria for Cancer CAHPS. CONCLUSIONS: This standardized approach for identifying and selecting eligible patients for the Cancer CAHPS survey will facilitate the two purposes of the Cancer CAHPS survey. The feasibility of implementing the sampling approach in different cancer care settings is currently being assessed in a multi-site field test of Cancer CAHPS in community oncology practices in California, U.S.

(2075) Assessing Health Related Quality of Life (HQOL) in Mesothelioma: which measures are optimal? Jason M. Fowler, PhD Fellow, University of Sydney, Concord, NSW (New South Wales), Australia; Janette Vardy, Medical Oncologist, Centre for Medical Psychology and Evidence Based Decision (CeMPED), University of Sydney, Concord, New South Wales, Australia; Steven Kao, Staff Specialist, University of Sydney, Camperdown, NSW (New South Wales), Australia; Joe R. Coll, Senior Research Fellow, The University of Sydney, Camperdown, NSW, Australia; Anne Warby, Research Officer, University of Sydney, Sydney, NSW (New South Wales), Australia; Melanie A. Price, Senior Research Fellow, University of Sydney, Camperdown, New South Wales, Australia; Haryana Dhillon, Research Fellow, Centre for Medical Psychology & Evidence-based Decision-making, Camperdown, NSW (New South Wales), Australia AIMS: Little is known about HQOL of people with malignant mesothelioma (MM) outside clinical trials, and it is unclear which measures of HQOL are optimal. We aimed to compare measures of HQOL in mesothelioma patients for sensitivity to change over time, relative responsiveness and statistical efficiency. METHODS: Prespecified exploratory analysis of HQOL data from an ongoing longitudinal observational study of HQOL in MM at any stage in their disease trajectory (Table 1). Participants completed these measures every 6 weeks: EORTC QLQ-C30 and Lung Cancer Symptom Scale (LCSS). Preliminary results from baseline (T1) and week 6 (T2) are reported. Inter-instrument correlations between measures at baseline, inter-instrument correlations of change over time, standardised response means (SRMs) and relative efficiency (RE) (Liang et. al) were performed in SPSS 22. RESULTS: Thirty-five participants are included, mean age 66 (range 37–84) years, 29 (83 %) male and 17 (49 %) on active anti-cancer treatment. Inter-instrument correlations at baseline showed significant positive correlations between the LCSS global QoL question, LCSS single item symptom questions and corresponding scales on the QLQ-C30 (range .568–.854, p \ 0.001). Correlations of change over time showed moderate significant associations between the LCSS and QLQ-C30 (Table 2). Estimates of relative efficiency [RE] may indicate an advantage for the QLQ-C30 [RE = 0.60] over LCSS for global HQOL, an efficiency advantage for LCSS single item measures of fatigue [RE = 3.15] and appetite [RE = 1.17] and an advantage of the QLQ-C30 for measures of pain. [RE = 0.30] (Table 3). CONCLUSIONS: Our results suggest that, whilst there is some degree of overlap between the constructs measured by these two instruments, we are unable to determine which is optimal and both should continue to be used in our ongoing longitudinal study. Whilst, more data and longer follow up are required, our preliminary data suggest there may be differences in the size of the sample required by each to detect change in specific symptom measures. For example, for an intervention assessing fatigue, the LCSS may require approximately 1/3 the sample size of a study using the QLQ-C30.

123

128

Qual Life Res (2015) 24:1–191

Table 1 Inter-instrument correlations of function scores between QLQ-C30 and comparable single item scales on the LCSS at baseline (n = 35) Dimension

Pearson correlation (r)

Sig. (2 tailed)

Global HQOL

.746**

p \ .001

Fatigue

.568**

p \ .05

Pain Appetite

.854** .657**

p \ .001 p \ .001

Table 2 Inter-instrument correlations of change over timea between scores on QLQ-C30 and comparable single item scales on the LCSS at baseline (n = 35) Dimension

Pearson correlation (r)

Sig. (2 tailed)

Global HQOL

.627**

p \ .001

Fatigue

.456**

p .05

Pain

.604**

p \ .001

Appetite

.436**

p \ 0.05

a

Change over time represents the difference between scores between baseline (T1) and 6 week follow up (T2) intervals

Table 3

Inter-instrument standardised response means and relative efficiency estimates

(n = 35) Domain

Global

Instrument Mean SD Standardised change change response meana HQOL

LCSS

QLQ-C30

4.12

4.13

20.07

Difference in T responsiveness statistic changeb

Relative efficiency (REc)

0.18

0.02

1.216

0.20

1.573

prognosis of early stage tumors and because of treatment morbidity, outcomes other than ‘survival’ are increasingly important. This study describes and compares health-related quality of life (HRQOL) of prostate cancer patients who received either radical prostatectomy (nerve-sparing, nsRP, or non-nerve-sparing, nnsRP) or radiotherapy (external RT, brachytherapy, or both combined) for treatment of localized prostate cancer. METHODS: The prospective, multicenter cohort study (ProCaSP) included 529 patients. Questionnaires included the IIEF, QLQ-C30, and PORPUS-P. Data were collected before (baseline), 3, 6, 12, and 24 months after treatment. Differences between groups’ baseline characteristics were assessed; changes over time were analysed with generalised estimating equations (GEE). Missing values were treated with multiple imputation. Further, scores at baseline and end of follow-up were compared to German reference data. RESULTS: The typical time trend was a decrease of average HRQOL three months after treatment followed by (partial) recovery. RP patients experienced considerable impairment in sexual functioning. The covariate-adjusted GEE identified a significant—but not clinically relevant - treatment effect for diarrhoea (b = 7.0 for RT, p = 0.006) and PORPUS-P (b = 2.3 for nsRP, b = 2.2 for RT, p = 0.045) compared to the reference nnsRP. Most of the HRQOL scores were comparable to German norm values by Waldmann et al. 2013 [PLoS ONE 8(9): e74149]. CONCLUSIONS: We were able to reproduce findings from previous international research in a specific setting of a patient cohort in the German health care system. According to the principle of evidence-based medicine, this strengthens the messages regarding treatment in prostate cancer and its impacts on patients’ health-related quality of life. After adjustment for baseline HRQOL and other covariates, RT patients reported increased symptoms of diarrhoea, and nnsRP patients decreased prostate-specific HRQOL. RP patients experienced considerable impairment in sexual functioning. These differences should be taken into account by physicians when choosing the best therapy for a patient.

0.60

Appetite LCSS

Fatigue

Pain

6.89

31.50

0.24

QLQ-C30

8.00

39.61

0.23

LCSS

2.34

20.24

0.09

QLQ-C30

1.38

21.22

0.06

LCSS

1.92

22.09

0.06

-4.20

21.54

-0.13

QLQ-C30 a

15.48

-0.1

1.294

1.17

(2079) The effects of TCM on chemotherapy related fatigue and quality of life in lung cancer patients

3.15

Jie You, The Ninth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China; Tan XiangWen, Shanghai Tianshan Hospital of Traditional Chinese Medicine, Shanghai, China; Chen Xue Fen, Shanghai University of Traditional Chinese Medicine, Shanghai, China; Weiwei Chen, XinHua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China; Zheng Chen, Shanghai University of Traditional Chinese Medicine, Shanghai, China; Yuqing Zhang, McMaster University, ON, Canada; William Zhang, McMaster University, ON, Canada

1.195 -0.3

0.683 0.385

-0.19

0.513

0.30

-0.932

Mean change score/SD of change

b

Change from baseline (T1-T2)

c

RE = (tLCSS/tQLQ-C30)2

(2077) The Cohort Study ProCaSP—Quality of life outcomes of prostate cancer patients after radiotherapy or radical prostatectomy during a follow up period of 24 months Nora Eisemann, University of Luebeck, Luebeck, Germany; Sandra Nolte, PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Alexander Katalinic, University Luebeck, Luebeck, Germany; Volker Rohde, Justus Liebig University of Giessen & Medical Practice of Urology, Bad Schwartau, Germany, Annika Waldmann, University Hospital Schleswig–Holstein, Campus Luebeck, Luebeck, Germany AIMS: Prostate cancer is the most common malignancy in men in Germany with about 65,000 incident cases each year and a 5-year prevalence of about 280,000 men. The main therapeutic strategies are radical prostatectomy (RP), external radiation, and interstitial brachytherapy. Each of above therapies can achieve a five-year cancer-specific survival of more than 90 %. Because of the favorable

123

AIMS: Adopting FACT-G and F (Functional Assessment of Cancer Therapy-General and Fatigue) to evaluate the effects of TCM (Traditional Chinese Medicine, TCM) intervention on chemotherapy-induced fatigue and quality of life in primary lung cancer patients. METHODS: 63 primary lung cancer patients were randomly assigned into two groups according to the random number results of SPSS 17.0: TCM intervention group(treatment group) and the control group (therapies are as mentioned below). Treatment group used herbal decoction plus chemotherapy while control group only used chemotherapy. The observation period is 14 days. FACT-F(consists of FACT-G plus Additional Concern which focus on Fatigue, include 13 items) was used as evaluation tool at baseline and at the end of the treatment. The scores of items were recorded. Besides, erythrocytes, hemoglobin, leukocytes were also recorded at the same time. RESULTS: (1) Baselines. There was no statistical difference at scores of each sub-domain between the two groups before treatment at

Qual Life Res (2015) 24:1–191 baseline. (2) The physical well-being (PWB). The post treatment scores of both groups were lower than what they were before. Scores in treatment group was significantly higher than it was in the control group after treatment (P \ 0.0001). (2) Patients’ Social/Family WellBeing (SWB), Emotional Well-Being (EWB), Functional Well-Being (FWB) and Additional Concerns (AC, this part of items focus on Fatigue). Scores were statistical significance between the two groups in terms of before-after treatment. These changing scores in treatment group were statistical higher than it was in control groups before and after treatment. (3) The total scores. After treatment, the treatment group scores were statistically significantly higher than that of the control group. As for the differences before and after the treatment, the treatment group raised while the control group dropped, which shows statistical meanings. (4) There were no significant differences in hemoglobin and numbers of erythrocytes, leukocytes, neutrocytes between the two groups during the observation period. CONCLUSIONS: This study showed that TCM can relieve the degree of fatigue, decrease the side effects of chemotherapy, and improve Quality of Life of primary lung cancer patients.

(2081) The impact of comorbidity on quality of life in elderly patients with chronic myeloid leukemia Francesco Cottone, Data Center and Health Outcomes Research Unit, Italian Group for Adult Hematologic Diseases (GIMEMA), Rome, Italy; Johannes M. Giesinger, Department of Psychiatry and Psychotherapy, Medical University of Innsbruck, Innsbruck, Austria; Fabio Efficace, Data Center and Health Outcomes Research Unit, Italian Group for Adult Hematologic Diseases (GIMEMA), Rome, Italy AIMS: Our analysis aimed at investigating the impact of comorbid conditions on quality of life (QOL) in elderly patients with chronic myeloid leukaemia (CML). METHODS: We retrieved data from 174 CML patients with an age of at least 60 years from a larger data set of 448 CML patients treated with imatinib that was collected in a multicentre cross-sectional study at 26 centers in Italy. The data set comprised detailed sociodemographic and clinical variables, number and type of comorbidity, SF-36 scores and questions on specific symptoms. QOL and symptoms were compared between patients with no, one, or more than one comorbid condition using analysis of covariance. RESULTS: In our data set mean patient age 70.3 years and 55.2 % were male. Sokal risk classification was low in 37 % of the patients, intermediate in 51 % and high in 12 %. Thirty-six percent of the patients did not report any comorbidity at the time of QOL assessment, 32 % reported one comorbidity and 32 % reported two or more comorbidities. In the patient sample the presence of comorbidity had a statistically significant impact on all scales of the SF-36 (all p \ 0.013). The comparison of patients with no and with one comorbidity showed largest differences for General Health (13 points) and Physical Functioning (10 points). More pronounced differences were found for patients with no comorbidity and patients with more than one comorbidity with differences of 22–25 points for Physical Functioning, Role-Physical, Pain Index, General Health, and Role Emotional. With regard to specific symptoms, comorbidity had the strongest impact on fatigue, muscular cramps, musculo-skeletal pain, and edema. CONCLUSIONS: In elderly CML patients comorbid conditions have a strong impact on various domains of QOL and on a number of somatic symptoms. Especially, in patients with more than one comorbid conditions QOL was strongly impaired regarding most of the domains assessed with the SF-36. Our findings highlight the need to systematically collect comorbidity data in the diagnostic work-up of elderly CML patients.

129 (2083) Health-related quality of life and mammography and Pap test screening: disparity within race/ethnicity groups Sarah Hobbs, South College School of Pharmacy, Knoxville, TN, United States; Kimmie Ly, South College School of Pharmacy, Knoxville, TN, United States; William Gentry, South College School of Pharmacy, Knoxville, TN, United States; Connie Rust, South College School of Pharmacy, Knoxville, TN, United States; I-Chan Huang, PhD, St. Jude Children’s Research Hospital, Memphis, TN, United States, Pranav Gandhi, PhD, South College, Knoxville, TN, United States AIMS: Health-related quality of life (HRQoL) has shown to be an important predictor of future disability, morbidity and mortality. The purpose of the study was to examine whether the relationships of HRQoL and mammography and Pap test screening, respectively, differ within whites, blacks, Hispanics, and other race/multiracial. METHODS: This was a cross-sectional study that analyzed population-based data from the 2012 Behavioral Risk Factor Surveillance System (BRFSS) dataset. Per the recommendations of the US Preventive Services Task Force, women were adherent to mammography screening guidelines if they received a mammogram within the previous 2 years and adherent to Pap test screening guidelines if they received a Pap test within the previous 3 years. Separate logistic regression models were conducted to test the relationship between four HRQoL aspects (general health status (good or better health vs. fair or poor health), physical HRQoL, mental HRQoL, and activity limitation (\14 days compared to C14 days for other three HRQoL concepts) and mammography and Pap test screening within race/ ethnicity groups: whites, blacks, Hispanics, and other race/multiracial, adjusting for covariates. Covariates such as demographics, socioeconomic status, smoking status, physical activity, body mass index, and health care access were controlled for in individual regression models. Statistical analysis accounted for the complex sampling design of the BRFSS dataset and the a priori alpha error was set at p B 0.05. RESULTS: White women reporting better HRQoL in all four aspects were significantly more likely to receive mammography screening compared to their counterparts reporting poor HRQoL; HRQoL was not significantly associated with mammography screening within other race/ethnicity groups. White women reporting better general health status and physical HRQoL were significantly more likely to receive Pap test screening compared to their counterparts reporting poor HRQoL. HRQoL was not significantly associated with Pap test screening within other race/ethnicity groups, except blacks reporting good or better general health status were significantly more likely to receive Pap test screening compared to their counterparts reporting poor general health status. CONCLUSIONS: Disparities in mammography screening by HRQoL persisted among particular race/ethnicity groups. Efforts to increase mammography and Pap test screening should target women reporting poor HRQoL.

(2085) PROMIS cancer-specific reference values for the US Cancer Population Roxanne Jensen, PhD, Cancer Prevention and Control Program, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC, United States; Carol M. Moinpour, PhD, Public Health Sciences Division, Fred Hutchinson Cancer Research Center, Seattle, WA, United States; Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Tania Lobo, MS, Cancer Prevention and Control Program, Lombardi Comprehensive Cancer Center, Georgetown University, Washington, DC, United States; Arnold Potosky, Georgetown University, Washington, DC, United States

123

130

Qual Life Res (2015) 24:1–191

AIMS: Estimate Cancer-specific population PROMIS reference values by patient age, severity, and other clinically important characteristics representing all cancer patients in the US population. METHODS: We recruited patients diagnosed between 2010 and 2012 in the Measuring Your Health (MY-Health) study from four Surveillance, Epidemiology, and End Results (SEER) Program cancer registries (n = 5513). Patients were diagnosed 6–13 months prior with one of seven cancers (prostate, colorectal, non-small cell lung, Non-Hodgkin lymphoma, breast, uterine, or cervical). We oversampled by race-ethnicity and \50 years of age to ensure a diverse sample. MY-Health contained the following PROMIS measures: anxiety (11-items), depression (10-items), fatigue (14-items); and pain interference (11-items). PROMIS measures are standardized to the U.S. general population (50 = mean, 10 = SD). Higher scores reflect higher symptoms. To estimate reference values pertaining to all US cancer cases, we created post-stratification analytical weights for each cancer type using a raking weighting algorithm compensating for non-response and non-coverage within the MYHealth cohort (Tables 1, 2, 3, 4, 5, 6, 7). We applied the iterative raking procedure by race-ethnicity, age, and stage to generate weighted PROMIS scores that effectively adjust the proportion of

patients in the MY-Health cohort to match the overall distributions by cancer type among the total new (incident) cases reported by all 18 SEER registries. RESULTS: The mean score change after adjustment to the SEER population was less than 1 point for each symptom (anxiety: 0.86, depression: 0.75, fatigue: 0.88, pain: 0.70). The majority of values were adjusted lower. Stage specific reference values (all ages) reported larger adjustments than age-specific values across all symptoms (mean difference: 0.1–0.4 points). The highest SEER adjusted values are: lung (pain [55.3], fatigue [57.3], and depression [51.2]); and cervical cancer (anxiety: 55.6). CONCLUSIONS: Overall, our SEER adjusted reference values for PROMIS pain, fatigue, anxiety, and depression domains performed as anticipated. The adjusted values were slightly lower than the MY-Health sample, reflecting the oversampling of non-white race/ ethnicity and younger age, both groups who often report higher symptom severity. The SEER Program reflects approximately 26 % of all US incident cancer patients and is sociodemographically heterogeneous. These values enable clinically meaningful comparisons to better interpret PROMIS scores in research and clinical practice.

Table 1 Breast cancer unadjusted MY-health cohort mean scores and SEER weighted U.S. reference values MY-health Anxiety

Depression

Pain

Fatigue

N

Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted

1588

50.7

49.6

49.3

48.5

53.8

52.6

52.9

52.3

Stage I

755

49.5

48.4

48.4

47.6

52.4

50.7

51.4

50.6

Stage II

609

51.8

50.9

50.2

49.6

54.4

53.6

54.1

53.6

Stage III

186

51.8

50.3

50.5

49.1

56.6

56.1

54.5

54.6

Stage IV

38

50.5

51.8

48.5

48.8

57.7

57

55.6

56.1

21–49 years

580

52.7

52.3

50.7

50.6

54.1

53.6

54.2

54.1

50–64 years

556

51

50.1

49.8

48.8

54.2

53.1

53.2

52.9

65–84 years

452

47.8

47.4

47.1

46.9

52.9

51.6

50.9

50.8

53.3

Overall Breast Cancer By stage (all Ages)

By age group (all Stages)

By age and stage groups Age: 21–49 years Stage I

232

51.8

52.1

50.2

50.6

52.8

52.9

52.9

Stage II

255

53.5

53.6

51.2

51.5

54.6

54.2

55.1

54.8

Stage III

82

52.7

49.4

50.6

47.4

55.5

51.9

54.4

52.2

Stage IV

11

50.4

52.2

49.3

50.9

58.8

58.3

59.4

59.9

Age: 50–64 years Stage I

256

50

48.9

48.8

47.9

52.2

50.3

51.5

50.7

Stage II

214

51.6

50.4

50.4

49.4

54.9

54.3

54.3

54.4

Stage III

66

52.8

52.8

51.7

51.3

58.4

58.7

55.9

57.2

Stage IV

20

51.2

52.8

48.1

48.4

57.5

56.9

53.1

53.2

Stage I

267

47.1

46.6

46.5

46.2

52.2

50.3

49.9

49.5

Stage II

140

49

49.1

47.9

48.3

53.1

52.3

52.1

51.6

Stage III

38

48.1

47.7

48

47.6

55.8

56

52.4

53.2

Stage IV

7

48.6

48.9

48.2

47.6

56.6

56.2

56.9

59.6

Age: 65–84 years

** Cells less than 50 in RED

123

Qual Life Res (2015) 24:1–191

131

Table 2 Prostate cancer unadjusted MY-health cohort mean scores and SEER weighted U.S. reference values MY-health Anxiety

Depression

Pain

Fatigue

N

Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted

1141

46.3

Overall Prostate cancer

45.9

45.6

45.4

49.9

49

47.5

47.2

By stage (all ages) Stage I/ II

962

45.8

45.5

45.2

45

49.6

48.9

47

46.9

Stage III/IV

179

48.9

48.4

47.9

47.7

51.6

49.9

49.9

49.4

42–64 years 65–74 years

536 474

47.1 45.6

46.6 45.2

46.3 44.9

46.1 44.5

50.5 49.2

49.1 48.5

47.6 46.8

47.1 46.7

75–84 years

131

46

45.7

45.2

45.5

50.2

50

49

49.4

By age group (all stages)

By age and stage groups Age: 42–64 years Stage I/ II

435

46.6

46.1

45.8

45.7

50.2

49

47

46.6

Stage III/IV

101

49.2

48.8

48.5

48

51.4

49.3

50

49.3

Age: 65–74 years Stage I/ II

404

45.1

44.8

44.6

44.2

48.9

48.3

46.4

46.3

70

48.1

47.1

46.8

46.5

51.4

49.8

49.6

48.9

Stage III/IV Age: 75–84 years Stage I/ II

123

45.5

45.3

44.8

45.2

49.9

49.6

48.9

49.1

8

52.4

52.8

50.9

52.3

55.6

57.6

51.1

54.6

Stage III/IV ** Cells less than 50 in RED

Table 3 Colorectal cancer unadjusted MY-health cohort mean scores and SEER weighted U.S. reference values MY-health

Anxiety

Depression

Pain

Fatigue

N

Unadj.

SEER weighted

Unadj.

SEER weighted

Unadj.

SEER weighted

Unadj.

SEER weighted

890

50.2

48.8

49.2

48.1

54.1

52.9

53.2

52.6

Stage I/ II

442

49.1

47.4

48.2

46.9

52.8

51.4

51.2

50.3

Stage III

292

50.6

49.5

49.7

48.5

54.7

53.7

54.4

53.7

Stage IV

156

52.5

51.1

51.1

50.2

57

55.6

56.4

56.5

22–64 years

480

52.2

51

50.8

49.6

55.8

55.1

54.7

54.7

65–84 years

410

47.8

47

47.4

46.9

52.2

51.3

51.3

51

Stage I/ II

198

52.2

51.3

50.6

49.5

54.8

53.8

53

52.6

Stage III

178

52.1

50.8

50.9

49.7

55.6

54.8

55.4

54.9

Stage IV

104

52.5

50.9

50.8

49.7

57.8

57

56.7

57.2

Stage I/ II

244

46.5

45.3

46.2

45.6

51.1

50.1

46.5

49.1

Stage III

114

48.3

48.2

47.8

47.4

53.3

52.6

48.3

52.5

Stage IV

52

52.7

51.3

51.6

50.9

55.2

53.7

52.7

55.4

Overall Colorectal cancer By stage (all ages)

By age group (all stages)

By age and stage groups Age: 22–64 years

Age: 65–84 years

123

132

Qual Life Res (2015) 24:1–191

Table 4 Lung cancer unadjusted MY-health cohort mean scores and SEER weighted U.S. reference values MY-Health

Anxiety

Depression

Pain

Fatigue

N

Unadj.

SEER weighted

Unadj.

SEER weighted

Unadj.

SEER weighted

Unadj.

SEER weighted

697

51.7

51.4

51.2

51

55.7

55.3

56.6

57.3

Stage I/ II

309

51.3

50.8

50.4

50.1

54.8

54.1

54.9

54.5

Stage III/IV

388

52.1

51.7

51.8

51.4

56.4

55.8

58

58.4

25–69 years

432

53

52.6

52.4

52.3

57

57

57.3

57.6

70–84 years

265

49.7

50.1

49.2

49.6

53.6

53.4

55.5

56.9

Stage I/ II

176

53.1

53

51.9

51.9

56.2

55.9

56.5

56.6

Stage III/IV

256

52.9

52.5

52.7

52.4

57.5

57.3

57.8

57.9

Stage I/ II

133

48.9

48.9

48.5

48.5

53

52.6

52.7

52.7

Stage III/IV

132

50.5

50.7

49.9

50.1

54.1

53.8

58.4

58.9

Overall Lung cancer By stage (all ages)

By age group (all stages)

By age and stage groups Age: 25–69 years

Age: 70–84 years

Table 5 NHL cancer unadjusted MY-health cohort mean scores and SEER weighted U.S. reference values MY-Health Anxiety

Depression

Pain

Fatigue

N

Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted

445

50.5

Overall NHL cancer

49.9

49.6

49.2

52.4

51.9

52.3

52.4

By stage (all ages) Stage I/ II

228

50.7

49.7

49.4

48.6

51.8

51

52.1

52.1

Stage III/IV

217

50.4

50.1

49.8

49.7

53.1

52.6

52.5

52.6

21–59 years

212

52.2

51.8

50.4

50

53.1

52.2

52.3

52.1

60–84 years

233

49

48.8

48.8

48.7

51.9

51.7

52.3

52.5

117

52.8

52.3

50.7

50.3

52.4

51.1

52

51.6

95

51.5

51.3

50

49.8

53.9

53.2

52.7

52.5

111 122

48.5 49.4

48.1 49.4

47.9 49.6

47.6 49.6

51.2 52.5

50.9 52.3

52.2 52.4

52.4 52.6

By age group (all stages)

By age and stage groups Age: 21–59 years Stage I/ II Stage III/IV Age: 60–84 years Stage I/ II Stage III/IV

Table 6 Uterine cancer unadjusted MY-health cohort mean scores and SEER weighted U.S. reference values MY-health Anxiety

Depression

Pain

Fatigue

N

Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted

388

50

50.4

49

47.4

52.4

51.9

51.2

50.4

318

49.9

49.5

48.9

47.3

51.8

51

50.6

49.5

70

50.9

53.6

49.4

47.7

53.1

52.6

54.1

53.6

Overall Uterine cancer By stage (all ages) Stage I/ II Stage III/IV

123

Qual Life Res (2015) 24:1–191

133

Table 6 continued MY-health Anxiety Depression Pain Fatigue N Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted Unadj. SEER weighted By age group (all stages) 28–59 years

180

51.2

50.4

49.6

48

53.1

52.2

51.3

50.4

60–83 years

208

49

50.4

48.4

46.9

51.9

51.7

51.2

50.4

By age and stage groups 28–59 years Stage I/ II

147

51.1

49.8

49.8

48.5

52.4

51.1

50.6

49.8

33

51.8

52.4

48.5

46.4

53.9

53.2

54.1

52.4

Stage III/IV 60–83 years Stage I/ II

171

48.8

49.3

48

46.4

51.2

50.9

50.5

49.3

37

50.1

54.7

50.2

48.8

52.5

52.3

54.2

54.7

Stage III/IV

** Cells less than 50 in RED

Table 7

Cervical cancer unadjusted MY-health cohort mean scores and SEER weighted U.S. reference values MYHealth N

Depression

Pain

Unadj.

Anxiety SEER weighted

Unadj.

SEER weighted

Unadj.

SEER weighted

Fatigue Unadj.

SEER weighted

142

54.4

53.2

52.8

51.8

54.2

53.5

54.7

54.2

80

53.3

51.4

51.5

49.9

52.4

51

52.8

51.8

62

55.8

55

54.5

53.7

56.6

55.9

57.1

56.5

93

54.6

52.6

53.2

51.9

53.6

51.9

54.6

53.6

49

54

53.8

52

51.8

55.5

55.4

54.7

54.8

59

54.1

51.9

52.2

50.6

51.8

50.2

53

51.9

34

55.7

53.6

55

53.8

56.6

54.3

57.5

56.1

21

51.3

50.4

49.5

48.7

54.1

52.4

52.3

51.5

28

55.9

55.9

53.9

53.6

56.5

57.2

56.6

56.8

Overall Cervical cancer By stage (all ages) Stage I Stage II/III/ IV By age group (all stages) 24–49 years 50–81 years By age and stage groups Age: 24–49 years Stage I Stage II/III/IV Age: 50–81 years Stage I Stage II/III/IV

(2087) Psychosocial health among adolescent and young adult brain tumor survivors David Buchbinder, MD, MSHS, CHOC Chidren’s Hospital & University of California - Irvine, Irvine, CA, United States, Justin Wilford, PhD, University of California - Irvine, Irvine, CA, United States; Michelle Fortier, PhD, University of California - Irvine, Irvine, CA, United States; Kathryn Osann, PhD, University of California Irvine, Irvine, CA, United States; Violet Shen, MD, CHOC Children’s Hospital & University of California - Irvine, Irvine, CA, United States; Lilibeth Torno, MD, CHOC Children’s Hospital & University of California - Irvine, Irvine, CA, United States; Leonard Sender, MD, University of California - Irvine, Irvine, CA, United States; Susan K. Parsons, Tufts Medical Center, Boston, MA, United States; Lari Wenzel, PhD, University of California-Irvine, Irvine, CA, United States

AIMS: Brain tumor (BT) survivors suffer from chronic health conditions and neurocognitive impairment which results in adverse psychosocial health that may persist. Despite growing numbers of adolescent and young adult (AYA) BT survivors, little is known about their psychosocial health. Our aim was to describe the psychosocial health of AYA BT survivors and to identify factors associated with adverse psychosocial health. METHODS: A cross-sectional study of 31 parents representing 28 AYA BT survivors was utilized to assess parent-reported psychosocial health of survivors (mean age 15.1 years, standard deviation [SD] 3.3) who were on average 10.1 years (SD 4.8) post-diagnosis. Parent-reported psychosocial outcome measures included: Patient-Reported Outcomes Measurement Information System Depression and Peer Relationship measures, and the Pediatric Symptom Checklist (PSC)-17. Independent variables included: parent-reported cognitive functioning assessed using the PedsQL Cognitive Functioning Scale (CFS), as well as both parent-report (chronic health conditions) and medical record-report (sex, age at diagnosis, time since diagnosis, diagnosis, treatment exposures, recurrence) BT-related variables. Associations were assessed using Pearson product-moment correlation, Kruskal–Wallis tests, and regression. RESULTS: Parent-reported AYA BT survivor depression (mean 48.8; 95 % confidence interval [CI] 45.2–52.5) T-scores were within normal limits; however, peer relationship T-scores (mean 41.7; 95 % CI 37.0–46.4) were significantly lower than normative values (one sample t-test, mean 50, p = 0.001). Parent-reported PSC-17 scores (mean 7.87; 95 % CI, 6.3–9.4) were below the threshold of 15 documenting the absence of emotional/ behavioral problems. CFS scores were lower (mean 46.7; 95 % CI 36.4–56.9) than previous samples of childhood cancer survivors (mean 76.2, SD 22.3) and healthy children (mean 90.1, SD 14.7). Factors associated with survivor depression included history of recurrence (p = 0.04). Factors associated with survivor peer relationship health included the presence of speech/language problems (p = 0.02). Factors associated with survivor emotional/behavioral problems included higher CFS scores (p = 0.04). Age at diagnosis and radiation exposure was not associated with the measured outcomes. CONCLUSIONS: AYA BT survivors demonstrate impairment in psychosocial health, particularly in peer relationships. These groups can be predicted from cognitive functioning, presence of chronic health conditions, and history of recurrence. This has implications for identification of AYA BT survivors at high-risk of development of adverse psychosocial outcomes.

123

134 (2089) Factors determining the change of patient reported outcomes after 24 months follow-up in prostate cancer men, Victoria, Australia Rasa Ruseckaite, Monash University, Melbourne, Australia; Sue M. Evans, Monash University, Melbourne, Australia AIMS: To examine the factors associated with a change in urinary, sexual and bowel bother outcomes reported twelve and 24 months post-diagnosis in prostate cancer men in Victoria, Australia. METHODS: Twelve and 24 months post-diagnosis follow-up (FU) EPIC-26 scores for urinary, bowel and sexual bother were collected via telephone interviews from prostate cancer patients who did not opt out from the Victorian Prostate Cancer Registry (VIC-PCR). The bother scores were then compared to determine the proportion of men who reported better, same or worse at 24 months FU. Multinomial logistic regression was conducted to investigate demographic, clinical and treatment factors associated with the changes in outcome. RESULTS: From 3256, 52 % had the same urinary health outcomes, 27.3 % reported worse and 20.9 % felt better at 24 months FU. A positive change was mainly observed in those men (39.8 %) who were treated \12 months prior their FU. Bowel bother scores were examined only in those (n = 1300) men who undertook radiotherapy. In 55 % of them outcomes did not change. Similarly to the urinary bother scores, the highest proportion of men (32.6 %) reported better scores if treated \12 months prior the FU. Sexual bother was examined in 1784 patients who had a prostatectomy. Overall, 30.9 % of those men had worse outcomes, 48.2 % remain same and only 20.9 % reported positive scores. Multinomial regression results confirmed that a positive change in bother scores was determined by a type of treatment, a shorter time taken between treatment and FU, older patient’s age, private hospital type in a metropolitan area and low/intermediate cancer risk group. CONCLUSIONS: Findings from this study indicated that certain categories of prostate cancer men tend to have poorer quality of life outcomes, inclusive of sexual/urinary/ bowel bother post-surgery or radiotherapy. Those men who’s FU respond as having ‘‘a big sexual/urinary/bowel bother’’ will be offered to discuss their problems with the Care Coordinator to obtain a referral to specialized support services for further assessment/treatment. We hypothesize that men, referred to the Care Coordinator, will have different health outcomes at 24 months FU when compared to those men not being referred to the Coordinator.

(2093) Examining the associations between cancer survivorship status and health-related quality of life among African-American female adults Selina A. Smith, PhD, MDiv, Georgia Regents University, Augusta, GA, United States; Mechelle D. Claridy, MPH, Morehouse School of Medicine, Atlanta, GA, United States; Benjamin E. Ansa, MD, MSCR, Georgia Regents University, Augusta, GA, United States; Francesca Damus, MPH, Morehouse School of Medicine, Atlanta, GA, United States; Ernest Alema-Mensah, PhD, DMin, Morehouse School of Medicine, Atlanta, GA, United States; William W. Thompson, PhD, U.S. Centers for Disease Control and Prevention, Atlanta, GA, United States AIMS: The primary aim of this study was to examine health-related quality of life (HRQOL) among African American females (AAF), and to compare differences in HRQOL between AAF breast cancer survivors (AAF-BCS), AAF survivors of other cancers (AAF-SOC), and AAF with no history of cancer (AAF-NHC). METHODS: Using the 2010 National Health Interview Survey (NHIS), a cross-sectional nationally representative health survey, we selected AAF aged [35 years and examined differences in HRQOL by cancer status. We compared three groups; AAF-BCS (n = 62), AAF-SOC (n = 74) and

123

Qual Life Res (2015) 24:1–191 AAF-NHC (n = 1,566). The Patient Reported Outcomes Measurement Information System (PROMIS) Global Health Scale was used to assess differences in current physical and mental health. The two global health outcomes comprise four items each and the summary physical and mental health scores are estimated in t-score units. All analyses were weighted and adjusted for age, marital status, and education. RESULTS: For the summary physical measure, there were no statistically significant differences between AAF-BCS and AAFNHC while AAF-SOC reported poorer physical health relative to AAF-NHC [5.8 t-points 95 % CI 2.8–8.8]. For the summary mental health measure, there were no statistically significant differences between AAF-BCS and AAF-NHC while AAF-SOC reported poorer mental health relative to AAF-NHC [3.3 t-points 95 % CI 0.6–5.9]. At the item level, we found no statistically significant differences between AAF-BCS and AAF-NHC on any physical or mental health items. When comparing AAF-SOC to AAF-NHC, we found statistically significant differences for three physical health items where AAF-SOC reported poorer physical health relative to AAF-NHC (ability to carry out physical activities [OR = 3.4; 95 % CI 1.7–6.7], level of fatigue [OR = 2.0; 95 % CI 1.1–3.7], and level of pain [OR = 3.3; 95 % CI 1.3–3.9]). CONCLUSIONS: African American female breast cancer survivors reported similar levels of physical and mental health compared to individuals with no history of cancer, while survivors of other cancers reported poorer physical and mental health compared to those with no history of cancer. Better understanding the similarities and differences in HRQOL between African American female survivors of breast cancer and survivors of other cancers could help improve HRQOL among all African American females.

(2095) Impact of age-related socio-economic and clinical determinants of quality of life among long-term breast cancer survivors Pegwende Olivia Dialla, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France; Wai-on Chu, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France; Patrick Roignot, Pathology centre, Dijon, France; MarieChristine Bone-Lepinoy, Centre radiotherapie du Parc, Dijon, France; Marie-Laure Poillot, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France; Charles Coutant, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France; Patrick Arveux, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France, Tienhan Sandrine Dabakuyo, PhD, Centre Georges Franc¸ois Leclerc Comprehensive Cancer Centre, Dijon, France AIMS: The main purpose of this study was to identify age-related socioeconomic and clinical determinants of quality of life among breast cancer survivors 5 years after the diagnosis. The secondary objective was to describe quality of life in the studied population according to age. METHODS: A cross-sectional survey in five-year breast cancer survivors was conducted in women diagnosed with breast cancer in 2007 and 2008 in Coˆte d’Or. Quality of life was assessed with the SF-12, the EORTC-QLQ-C30 and the EORTCQLQ-BR23 questionnaires. Socio-economic deprivation was assessed by the EPICES questionnaire. Social support was assessed by the Sarason questionnaire and clinical features were collected through the Coˆte d’Or breast cancer registry. Age-related determinants of quality of life were identified using multivariate mixed model analysis for each SF-12 dimension. Bonferroni’s correction was used to adjust the a-risk to the eight multivariable models. The significance limit was then set at P \ 0.006 for multivariable models. To assess the impact of missing data on the determinants of quality of life outcomes, a sensitivity analysis was conducted with multiple imputations using

Qual Life Res (2015) 24:1–191 the fully conditional specification method and the predictive mean matching method. RESULTS: Overall 396 (62 %) women completed the questionnaires. Women aged\65 years had a better quality of life and a greater availability of social support than did women aged C65 years. Body mass index, relapse and EPICES deprivation scores were found to be determinants of quality of life in younger women (p \ 0.006) while those in older women were comorbidities and EPICES deprivation scores. CONCLUSIONS: Five years after breast cancer diagnosis, disease severity did not affect quality of life in breast cancer survivors

(2097) Value in breast and prostate cancers Pedro L. Ferreira, University of Coimbra, Coimbra, Portugal; Teresa Sequeira, CEISUC/UFP, Porto, Portugal; Augusta Silveira, CEISUC/ UFP, Porto, Portugal; Lara Ferreira, CEISUC/UALG, Faro, Portugal; Carlota Quintal, CEISUC/FEUC, Coimbra, Portugal; Suzete Gonc¸alves, CEISUC, Coimbra, Portugal; Lurdes Carvalho, IPOP, Porto, Portugal; Joana Teixeira, IPOP, Porto, Portugal; Ana Eberhardt, IPOP, Porto, Portugal AIMS: Breast cancer is the most prevalent cancer in women and the first cause of death from cancer among women. Despite its high incidence rates in Western countries, 89 % of women are still alive five years after their diagnosis. Also, prostate cancer is the second leading cause of cancer death in men. The incidence rate for prostate cancer has increased since 1980 with lower but increasing incidence in less developed countries. This study aimed to derive utilities for breast and prostate cancer patients after the implementation of a routine quality of life (QoL) assessment in the Portuguese Cancer Institute, in Porto to explore factors related to an impaired QoL. It also intended to demonstrate the importance of collection of routine data on these patients. METHODS: Consecutive patients admitted respectively at Breast and Prostate Units were asked to complete the health status measures EORTC QLQ-C30, the corresponding BR23 and PR25 modules, SF-12, and EQ-5D-3L. The Portuguese value set was used to derive the EQ-5D index and socio-demographic and clinical variables were also collected. RESULTS: 600 breast cancer patients had 52.5 ± 10.0 years old, 32.5 % had had the diagnosis for 1 year, 33.0 % between 2 and 4 years. Functional status dimensions measured by QLQ-C30 presented high values, and global QoL score 60.5 revealed high levels of overload. Body image (78.1) and the concern with hair loss (39.5) were the worst reported symptoms. EQ5D index was 0.62 ± 19.7 with 20 % of patients providing values greater than 0.8 and 0.9 % with scores worse than death. 300 prostate cancer patients had 71.2 ± 8.9 years of age, mostly with a diagnosis from 2 to 4 years and with very low literacy. QoL reached the score 58.3 and more burdensome symptoms were also insomnia, fatigue, financial difficulties and symptoms relating to sexual activity. EQ-5D index was 0.80 ± 0.29 and about 1 % of patients presented utility values considered worse than death. CONCLUSIONS: Elderly patients had always lower functional status values, greater symptoms overload and lower utility values. As a conclusion, we state that with a well-structured care organization and a good health information system, it is possible to measure the value a patient may gain in being treated.

Student Poster Award Finalist Abstracts (2002) Examining the factor structure of general and health selfefficacy scales with individuals who are homeless or vulnerably housed Sneha Shankar, PhD Student, University of British Columbia, Vancouver, Canada; Anita M. Hubley, PhD, University of British

135 Columbia, Vancouver, Canada; Bruno D. Zumbo, PhD, University of British Columbia, Vancouver, Canada AIMS: Self-efficacy (SE) refers to one’s perceived capability. SE beliefs impact the goals one sets, amount of effort exerted, outcomes one expects, how one responds to adversity, and one’s quality of life and mental health. Much research on homelessness focuses on the effects of making health resources more available and accessible. Often missing from this research is the role that SE may play in homeless individuals’ abilities to make use of health resources, stay motivated, as well as cope with and effect change. As a start, it is important to identify SE measures, such as the Generalized SelfEfficacy Scale (GSES) and Perceived Health Competence Scale (PHCS), that may be used with homeless individuals and examine their factor structure. Given the use of total scores, both scales are treated as unidimensional. This study aims to examine the factor structure of the GSES and PHCS with a sample of 330 men and women who are homeless or vulnerably housed from Vancouver’s Downtown Eastside. METHODS: Data were obtained from the Health and Housing in Transition (HHiT) study. Two confirmatory factor analyses were conducted separately on each scale, treating the 4- and 5-point response format data as ordinal or continuous. Model fit was determined using several fit indices. RESULTS: The GSES was found to be unidimensional only when the data were treated as ordinal. In the case of the PHCS, a one-factor model did not fit the data at all. Subsequently an exploratory factor analysis was run using principal axis factoring and promax rotation. Two factors, reflecting positively and negatively worded items and explaining 61.3 % of the variance, were extracted and correlated r = -.57. CONCLUSIONS: Use of SE measures with homeless individuals may provide important information about their ability to utilize available resources and facilitate change as well as the effectiveness of interventions. To determine this, one needs to have appropriate SE measures for this population. This study shows that the GSES has a unidimensional latent structure and use of a total score is appropriate. The PHCS showed a two-factor structure suggesting that separate subscale scores should be used for the positively and negatively worded items.

(2004) detection of response shift using a mixed approach combining: k-means and random forest methods Mohamed Boucekine, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Karine Baumstarck, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Pierre Michel, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Laurent Boyer, EA 3279 SelfPerceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Anderson Loundou, 1. AixMarseille University, EA 3279 – Public Health, Chronic Diseases and Quality of Life - Research Unit, 13005 Marseille, France; Badih Ghattas, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France; Pascal Auquier, EA 3279 Self-Perceived Health Assessment Research Unit, School of Medicine, Aix Marseille Universite´, Marseille, France AIMS: To investigate the Response Shift (RS) using a mixed approach combining k-means and Random Forest (RF) methods in a population with multiple sclerosis (MS). METHODS: This is a multicenter and longitudinal study, including MS patients. Quality of life (QoL) was assessed using both the MusiQoL and the 36-Item Short-Form (SF36) at baseline and every 6 months (t = M0, M6, M12, M18, M24). Disability (EDSS) was collected at each evaluation. At M0, the individuals were classified as low/high QoL level from the 9 QoL scores (MusiQoL) using the k-means. Similarly, the

123

136

Qual Life Res (2015) 24:1–191

individuals were classified as low/high disability (EDSS). Then, 2 classes of individuals were defined: 1. Concordant: individuals with low QoL/low disability, and high QoL/high disability; 2. Discordant: individuals with low QoL/high disability, and low QoL/high disability. We defined the variable Y as follows: Y = 0 if concordant and Y = 1 if discordant. We hypothesized that RS occurred for discordant individuals. The procedure was replicated for each time (t). Each individual i was characterized by a vector Vi = (YiM0, YiM6, YiM12, YiM18, YiM24). We used a GEE model to test the probability of an increasing of the event (Y = 1) over time. The model was: logit (P(Yit = 1)) = b0 + b1 9 t. We assumed RS occurrence if b1 is statistically significant, determining 2 groups (RS/no-RS, where RS individuals are concordant at M0 and discordant at Mj). To investigate if this RS is a reprioritisation, we applied the RF method in each group, based on the following model: MusiQoL_Index = f(Physical Score-SF36,Mental Score-SF36). RESULTS: At M0, 283 individuals were concordant and 152 were discordant. Based on the GEE model, RS was observed only between M0 and M6 (Figs. 1, 2, 3). Sixty concordant individuals at M0 became discordant at M6 (RS group). Using the RF, we observed in the RS group that the physical score did not contribute to the MusiQoL index score while the importance of the mental score increased over time. In the no-RS group, we observed that both physical and mental scores contributed to the MusiQoL index. CONCLUSIONS: The combination of k-means and RF methods can assist researchers to detect and identify RS occurrence over time.

Fig. 1 .

Fig. 2 .

123

Fig. 3 . (2006) Outcome reporting in UK-based maternity trials: a Systematic review of randomised controlled trials (The Pro-Maternity group)* Ayesha Mahmud, PhD student, University of Birmingham, Birmingham, United Kingdom; Tabassum Khan, Birmingham Women’s Hospital, Birmingham, United Kingdom; Sara Kenyon, University of Birmingham, Birmingham, United Kingdom; Kirstie Haywood, Senior Research Fellow, Warwick Medical School, Warwick University, Coventry, United Kingdom; Khaled Ismail, University of Birmingham and Birmingham Women’s Hospital, Birmingham, United Kingdom AIMS: Pregnancy is a physiological state which can significantly impact the health and well-being of the woman: a range of outcomes, including pathophysiological, life impact and resource use, are important for understanding the wide-ranging impact of pregnancy. However, guidance regarding the most appropriate outcomes to include in randomised clinical trials (RCTs) of maternity care does not exist. Assessment heterogeneity in clinical trials hinders data comparison and systematic reviews. This review aimed to describe the range of outcomes reported in Maternity RCTs. METHODS: A systematic literature review of RCTs related to pregnancy and childbirth between 2004 and 2014. Relevant articles were identified by applying a search strategy to Medline, Embase and Cinhal. Study eligibility was assessed by two reviewers and included: UK-based maternity RCTs reporting pregnancy and/or childbirth-focused primary outcomes; following CONSORT guidance. Trials reporting on medical disorders were excluded. RCTs were categorised as: antenatal; birth-related; postnatal. Further sub-categorisation focused on intervention: surgical/non-surgical; medicinal; behavioural. Data extraction was completed by two reviewers using a standardised form which captured primary and secondary outcomes, completion mode, timing and reproducibility. RESULTS: 11,836 articles were identified from the initial search; all titles were screened for eligibility. 68 RCTs met the inclusion criteria. This included antenatal (41 %), birth-related (47 %) and postnatal (12 %) trials. The largest proportion of trial interventions were medicinal (46 %), followed by behavioural (22 %), non-surgical (19 %) and surgical (13 %) interventions. Although a wide range of clinical and clinician-completed outcomes were identified there was inconsistency in the selection and reporting of outcomes among similar interventions. 48 trials used clinician reported outcomes as a primary outcome whereas, 20 trials used women reported outcomes as a primary outcome. CONCLUSIONS: This review highlights the significant heterogeneity and poor quality of reporting of health outcomes in maternity RCTs. The impact of pregnancy and child-birth as understood from the perspective of the mother was poorly assessed. These findings underpin the importance of seeking to better understand the experience of mothers and the outcomes that they value throughout pregnancy and childbirth, in the pursuit of a woman-derived pregnancy-related outcome measure. *Prof. Edwin van Teijlingen, Mr. Edward Morris, Prof. Pauline Slade,

Qual Life Res (2015) 24:1–191

137

Prof. Vanora Hundley, Annalise Weckesser, Gwyneth Eanor (Birth trauma association) & Miranda Dodwell (BirthChoice UK)

Poster Session 2002: Friday Poster Session 2

difficulties to complete the standard DSWAL-QoL or true lack of agreement between both parties. The latter can be due to underestimated communication problems when discussing subjective problems. This lack of agreement can compromise the healthcare provided to patients with problematic self-reporting. An adjusted DSWAL-QoL could increase the reliability of patients’ ratings.

Measure development (2008) Comparing patient and proxy agreement concerning impact of dysphagia on patients’ quality of life (QoL) as measured by the Dutch version of the Swallowing Quality-of-Life (DSWALQoL) questionnaire Ingeborg Simpelaere, MSc, University College Bruges, AZ Delta Roeselare-Menen, University of Antwerp, Zedelgem, West Flanders, Belgium; Jan Vanderwegen, ENT MD, Saint-Pierre University Hospital, Brussels, Belgium; Gwen Van Nuffelen, University Hospital Antwerp, Antwerp, Belgium; Marc De Bodt, University Hospital Antwerp, Antwerp, Belgium AIMS: In patients with difficult self-reporting due to insufficient cognitive/communication skills, proxies (healthcare providers or significant others) can provide information on perceived QoL. Reliability of this information should be established for each questionnaire. Our study evaluated agreement of proxies’ ratings of dysphagic patients using the standardized DSWAL-QoL. METHODS: 56 patients and their proxies were studied. Inclusion criteria: neurologic or mechanical dysphagia during at least 1 month and objectified by MASA/FEES. 25 patients suffered only dysphagia (Dys); the others had additional language impairment/cognitive problems (DysLC). Patients and proxies independently completed the DSWAL-QoL. Comparison of proxy and patient data was performed using Intraclass Correlation Coefficients (ICC) and Spearman rank correlation (rs) in SPSS20. RESULTS: Total scores showed significant agreement only for Dys (ICC = .72; rs = .73), not for DysLC (ICC = .27; rs = .21). All subscales (except eating duration and sleep) confirmed similar agreement exclusively in the Dys group, with moderate to good correlation (Table 1). CONCLUSIONS: Worrying discrepancies between DysLC-patients and their proxies indicate either DysLC Table 1 Intraclass correlation coefficient (ICC) and Spearman rank correlation coefficient (rs) for correlation proxy—patient for the 11 subscales of the DSWALQoL Subscale

Dys vs. proxy rs

ICC [95 % CI] 0.72 [0.38–0.87]

Total

0.73**

General burden

0.48*

0.70 [0.33–0.87]

Eating duration

0.20

0.37 [-0.38–0.72]

DysLC vs. proxy rs

ICC [95 % CI] 0.21

0.27 [-0.39–0.63]

0.28

0.35 [-0.18-0.67]

-0.30

-0.78 [-2.70-0.13]

Eating desire

0.63**

0.76 [0.46–0.89]

0.23

0.19 [-0.28–0.54]

Symptoms

0.61**

0.70 [0.33–0.87]

0.33

0.31 [-0.45–0.67]

Food selection

0.54**

0.71 [0.34–0.87]

0.05

0.11 [-0.65–0.54]

Communication

0.78**

0.87 [0.71–0.94]

0.17

0.35 [-0.36–0.69]

Fear of eating

0.48*

0.66 [0.25–0.85]

0.34

0.55 [0.06–0.78]

Mental health

0.56**

0.72 [0.38–0.88]

0.23

0.49 [-0.05–0.75]

Social functioning

0.54**

0.70 [0.33–0.87]

-0.08

-0.03 [-0.85–0.46]

Sleep

0.32

0.51 [-0.12–0.78]

0.46**

0.59 [0.17–0.80]

Fatigue

0.44*

0.56 [0.004–0.80]

0.35

0.50 [0.001–0.75]

(2010) A patient centred approach to the development of a patient reported outcome measure (PRoM) in idiopathic pulmonary fibrosis (IPF) Anne-Marie Russell, MSc RN RHV LPE PGCE, Imperial College London & Royal Brompton Hospital, London, United Kingdom; Anne-Marie Doyle, D.Clin.Psy BSc MA, Royal Brompton Hospital London, London, United Kingdom; Caroline Burdett, BA, ISoQol & Royal Brompton Hospital London, London, United Kingdom; Jennifer Gane, Patient Partner RN RHV, Royal Brompton Hospital London, London, United Kingdom; Zahra Aden, MSc BSc, NHLI Royal Brompton and Imperial College London, London, United Kingdom; Sharon Fleming, PhD RN, Royal Brompton & Harefield NHS foundation Trust London, London, United Kingdom; Paul Cullinan, MFPHM MD FRCP MRCP MB MSc, Imperial College London and Royal Brompton Hospital London, London, United Kingdom AIMS: Introduction: The FDA1 considers patient perspectives to be of great importance when evaluating the content validity of a PROM. We aimed to capture patient/carer perspectives as study participants in focus groups and Delphi surveys and also as co-investigators. We established a multidisciplinary patient/carer Research Support Group (RSG) to steer the development of a PROM for patients diagnosed with IPF. IPF is associated with significant morbidity and mortality and emotive narratives of the patient/carer journies METHODS: • RSG members: lead researcher (LR); research nurse; clinical psychologist; patients; carer and Patient and Public Involvement Lead • Patients self-selected to join the RSG representing the mild-moderate spectrum of IPF • Formal terms of reference and role descriptions were agreed for RSG members. • Patients reviewed and commented on the study protocol. All RSG members: • Reviewed focus groups transcripts actively contributing to thematic analysis • Using consensus methodology undertook a rigorous reduction of items generated from deconstructed questionnaires reported in the published literature of IPF research • Assessed the face validity of the electronic Delphi survey • Pilot tested each round of the Delphi survey • Contributed to the qualitative analysis of the comments captured in each round of the Delphi study RESULTS: Interim Results: 1. The LR applied a validation list to the questionnaires identified in the literature. 14 were retained and deconstructed yielding 1408 items. RSG members removed duplicate items thought to have the same meaning; using consensus methods 32 candidate items were identified for inclusion in the Delphi survey. 2. RSG members contributed a level of sympathy, insight, compassion and knowledge to the thematic analysis of the focus groups: a dynamic that enhanced the interpretation of transcripts. 318 statements were nominated for inclusion in the Delphi 3. Domains identified by a Nominal Group of ILD experts were reviewed by the RSG and nominated for inclusion/exclusion in the Delphi survey (n = 12) CONCLUSIONS: Discussion: We advocate the RSG model as feasible sustainable and enriching to the research experience, specifically for PRO development. Separate evaluation of the process will be undertaken to capture the patient/carer perspective of this experience. This study is ongoing.

* p \ .05; ** p \ .01

123

138 (2012) A community sample’s perception of cause-and-effect directionality among health domains Janel Hanmer, MD, PhD, University of Pittsburgh, Pittsburgh, PA, United States; David Feeny, PhD, Department of Economics, McMaster University, Hamilton, ON, Canada, Hamilton, ON, Canada; Baruch Fischhoff, PhD, Carnegie Mellon University, Pittsburgh, PA, United States; Ron D. Hays, PhD, UCLA, Los Angeles, CA, United States; Rachel Hess, MD, MS, University of Utah, Salt Lake City, UT, United States; Paul Pilkonis, PhD, University of Pittsburgh, Pittsburgh, PA, United States; Dennis Revicki, PhD, Evidera, Bethesda, MD, United States; Mark Roberts, MD, MPP, University of Pittsburgh, Pittsburgh, PA, United States; Joel Tsevat, MD, MPH, University of Cincinnati, Cincinnati, OH, United States; Lan Yu, PhD, University of Pittsburgh, Pittsburgh, PA, United States AIMS: Community members may consider health-related quality of life domains as dependent on each other but the direction of the dependence is unclear. We examined statements about pairs of health domains for the ‘‘root’’ (causal) domain and the corresponding affected domains. METHODS: Fifty adults from the community were asked to evaluate pairs of PROMIS domains. Each subject was presented with 20 randomly-selected pairs of domains and asked to consider 1 pair at a time. For a given pair, the subject was asked if it were possible to simultaneously have high functioning in domain 1 but low functioning in domain 2, and vice versa, to provide input on independence. Statements were independently evaluated by 2 coders for strong statements of causality. Strong statements were those that used words such as ‘‘will,’’ ‘‘would,’’ ‘‘need,’’ ‘‘can’t,’’ or ‘‘don’t’’ (e.g., ‘‘If you have pain, you can’t function cognitively.’’). Statements containing ‘‘could’’ or ‘‘might’’ were not included (e.g., ‘‘Physical function could cause anxiety.’’). Strong statements were coded for a root domain (cause) and affected domain. All differences in coding were discussed until consensus was reached. RESULTS: In 995 domain pair comparisons, there were 896 statements, of which 114 (13 %) were strong causal statements. Statements are summarized in the Table 1. CONCLUSIONS: Community members described cause and effect in pairs of associated health-related quality of life domains. When paired with other domains, Pain and Sleep were most likely to be perceived as impacting the other domain, whereas

Table 1 .

Qual Life Res (2015) 24:1–191 Cognition, Social Roles, and Sexual Function were most likely to be perceived as the affected domain. Fatigue, Anxiety, Depression, and Physical Function were considered as either the root or the affected domain.

(2014) Evaluating treatment effect based on personal outcomes of specific interest (POSI): a personalized patient-reported outcome strategy Sravan Jaggumantri, University of British Columbia, Vancouver, BC, Canada; Jean Paul Collet, MD, PhD, University of British Columbia, Vancouver, BC, Canada; William McKellin, PhD, University of British Columbia, Vancouver, BC, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; Sylvia Stockler, MD, PhD, BC Children’s Hospital, Vancouver, BC, Canada; Clara van Karnebeek, MD, PhD, BC Children’s Hospital, Vancouver, BC, Canada AIMS: Existing patient-reported outcome measures have limited abilities to assess the significance of therapeutic changes at an individual level. We conducted a feasibility study to demonstrate the utility of a personalized patient-reported outcome strategy that captures Personal Outcomes of Specific Importance (POSI), which evaluate treatment effectiveness in routine clinical care. A POSI can be a particular symptom or some aspect of daily function, which individual patients consider important in the context of their disease and personal values. METHODS: The POSI approach is interviewbased, using a patient’s narrative account, similar to those conventionally obtained by physicians. It is administrated in 4 stages: (1) POSI Identification through obtaining the patient’s account of the impact of the condition on daily life and a collaborative discussion between the patient and the physician to identify a treatment strategy (2) Finalize POSI that are therapeutically and personally realistic and desirable (3) Scaling POSI using Goal Attainment Scaling to enable measurable results (4) POSI follow up evaluation. The POSI strategy was tested in 5 patients in the Metabolic Division at BC Children’s Hospital with promising results. RESULTS: The identified POSI were useful to determine meaningful change from each patient’s perspective in their domain/activity/outcome of interest. It was noted that the POSI identified were for most instances did not correspond with any item/domain of a standard scale. Even if did correspond, it was only to a domain level and therefore lacked the sensitivity to capture the meaningful change. In all cases the standard scales were never normed in the target patient population challenging the validity of the scores (a situation common to rare diseases). POSI may be especially useful for diseases where there are no scales or established outcomes (Eg: Lesch Nyhan Syndrome). CONCLUSIONS: POSI is a unique generic strategy for direct patient reporting of personalized outcomes across multiple diseases and different drugs. It does not change the informal patient-physician encounter but rather enables a standardized process of personalized outcome evaluation and systematic data collection that will assist the physician to identify the most appropriate treatment for each patient. We are now conducting a study to establish the measurement properties of this strategy.

(2016) A framework for classifying root causes of missing data for patient reported outcomes in prospective studies Michael J. Palmer, PhD Candidate, Queen’s University, Kingston, ON, Canada; Rebecca Mercieca-Bebber, University of Sydney, Sydney, NSW, Australia; Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia; Melanie J. Calvert, PhD, University of Birmingham, Birmingham, United Kingdom; Harriet

123

Qual Life Res (2015) 24:1–191 Richardson, PhD, Queen’s University, Kingston, ON, Canada; Michael Brundage, MD, MSc, Kingston General Hospital, Queen’s University, Kingston, ON, Canada AIMS: Missing patient reported outcomes (PROs) data are a problem that can lead to biased results or to loss of power to detect betweentreatment differences. Awareness of root causes may help researchers reduce missing PRO data through study design and trial processes. The aim is to construct a conceptual framework of root causes of missing PRO data in the context of comparative studies. METHODS: The findings from a focus group with individuals involved in the collection of PRO data will be integrated with findings from a systematic literature review to construct a framework of potential root causes. Data extraction for the systematic review is ongoing. The focus group was conducted with experienced Clinical Research Associates (CRAs) to identify causes of missing PRO data in comparative studies. Participant CRAs actively collected PRO data, had knowledge of the clinic environment and had an understanding of patient-related processes. One research team member observed proceedings and made detailed notes. An experienced moderator led the discussion. This was audiotaped, transcribed and analyzed for themes. MEDLINE and CINAHL databases were each searched from inception to March 2014 for English articles examining predictors or causes of missing PRO data, including specific studies examining reasons for missing PRO data in actual trial datasets. Two reviewers independently screened titles and abstracts. Discrepancies were discussed with the research team prior to finalizing the list of eligible papers. One investigator extracted data and a second investigator checked all extractions. A hand-search of references was performed. RESULTS: Initial identified framework components include characteristics of: the PRO instruments, the clinical trials, the participating centres, trial participants, and patient caregivers as relevant to missing PRO data. This preliminary framework will be synthesized with complete extractions from the systematic review. 5996 unique articles were retrieved. 131 met inclusion criteria: 81 had congruence on independent screening, and 50 had research team consensus. CONCLUSIONS: A conceptual framework is being constructed to organize and describe root causes of missing PRO data. The conceptual framework will have application to research projects that focus on missing PRO data, and will be used to inform a mixed methods component of our research program.

(2018) Applying Eran Tal’s model-based account of the epistemology of measurement to the short form-36 Laura Cupples, University of South Carolina, Columbia, SC, United States AIMS: Following Eran Tal’s model based account of the epistemology of measurement, I ask what it might mean to give a model-based account of quality of life measures, and if it is possible for models to provide the same sort of epistemic support for claims about the validity, accuracy, and comparability of these measures as they do for measures in the physical sciences. METHODS: As a case study, I examine the Short Form-36 and ask what model of quality of life, if any, is inherent in the measure. I further ask what model describes the interaction of the instrument with the phenomenon in question. I examine the epistemic role these models play in light of Tal’s modelbased account of the epistemology of measurement. RESULTS: According to Tal’s account, models provide epistemic support for measures primary because they serve as local articulations of scientific theory. They mediate between abstract theory and concrete phenomena. However, many philosophers and quality of life

139 researchers have argued that quality of life measures lack theoretical grounding. There is no consensus account of what well-being entails, or how it varies with life circumstances or our adaptation to those circumstances. There is also limited theoretical work explaining how respondents understand and interact with these survey instruments, i.e., how these measures tap into the phenomenon in question. If models provide epistemic support for the SF-36, it is not because they are mediating between theory and phenomena as Tal argues they do in physical measures. There is no consensus theory of quality of life to serve as a target for mediation. Yet the SF-36 does seem in some ways to make use of a model of quality of life, or at least of health, even if that model is not grounded in consensus theory. CONCLUSIONS: I argue that a model of quality of life that is not grounded in theory cannot epistemically support what Eran Tal calls measurement outcomes, but that it may provide measurement indications. Through a process Hasok Chang calls epistemic iteration, we may be able to use these indications to develop a better theoretical understanding of quality of life.

(2022) Item banking in Sweden Bo Palaszewski, Centre of Registers Va¨stra Go¨taland, Gothenburg, Sweden; Evalill Nilsson, The Research & Development Department ¨ stergo¨tland, Sweden, Quality Register of Local Health Care, Region O Center QRC Stockholm, Stockholm, Sweden, Linko¨ping, Sweden; John Chaplin, PhD, AFBPsS, CPsychol., Inst. Clinical Sciences, Sahlgrenska Academy at the University Gothenburg, Gothenburg, Sweden AIMS: The aim is to evaluate the requirements for the integration of PRO data collection into future item banks in Swedish national quality registries (NQRs). NRQs are designed to be used in an integrated and interactive way for quality control, research, and clinical management. There are over 100 different NQRs in Sweden, each focused on a specific diagnosis, condition or intervention, and contain individual patient data. METHODS: The evaluation will consider the processes, goals, and outcomes of PROs in relation to the needs and structure of the NQRs in Sweden. Documentation reviews will be carried out. Consideration will be given to the values and cultural aspects of measurement; scalability, and performance. Processes and outcomes will be evaluated against the needs of the Swedish system. It is considered desirable that comparisons can be made across conditions and between national and international healthcare systems. RESULTS: In the NQR implementation of PROs there is a need for both generic and diseasespecific measures. Both electronic and paper-and pencil administrations will be needed. Therefore, a comprehensive approach to PRO measurement is required. Barriers to the inclusion of PROs in the NQRs include the availability of validated and reliable measures that are relevant and feasible for consistent and repeated use across Sweden. Item banks such as PROMIS are being considered. The PROMIS domain specific approach has advantages in that it allows comparisons within and across different diagnoses which facilitates the need for comparisons, meta-analyses, leading to the development of health policy. CONCLUSIONS: An item bank approach allows for flexible administration and different modes of administration. Further evaluation is required concerning how to integrate PRO item banks into the diverse Swedish NQR system. This will lead to improved clinical practice through the process of setting goals and objectives for treatment and monitoring response to treatment resulting in shared decision-making in clinical encounters. If the NQRs adopt PROMIS, this would mean that the major part of the Swedish healthcare system would be involved, forming a solid base for quality improvement and research. In the evaluation, PROMIS will be compared with other alternatives.

123

140 (2024) Reuniting functioning and wellbeing: the role of the ICF in the development of PROMs Michaela Dabs, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Janika Blo¨meke, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Rachel Sommer, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Julia Quitmann, researcher, University of Hamburg, Hamburg, Germany, Holger Muehlan, PhD, Ernst-Moritz-Arndt University, Greifswald, Germany; Monika Bullinger, Vice Chief of Department, University of Hamburg, Hamburg, Germany AIMS: The International Classification of Functioning, Disability and Health (ICF) is an universal WHO approach to assess disease consequences. Linking HrQoL measures to the ICF facilitates content comparisons, identification of relevant domains and selection of appropriate instruments. We introduce a novel approach by using the child and youth version ICF-CY to classify relevant patient concerns from focus group material in short statured children with Achondroplasia. METHODS: Fifteen focus group discussions (n = 65 participants) among children (8–12 years), adolescents (13–18 years), young adults (19–28 years) and their families were recorded, transcribed and content analyzed using the MAXQDA program. Statements were inspected and coded along the ICF-CY category system by two independent raters consenting to nine domains with up to ten subcategories, which were sorted to ICF-CY categories, listed according to their frequency and compared to an inductive content analysis. RESULTS: A total of 1953 statements were identified, which represented the content categories self-image, family, friends, school, strangers, health care, technical aids, recreation and physical problems. On the third of the four level ICF-CY coding system ‘‘e410 Individual attitudes of immediate family members’’ (Environmental Factors: Attitudes) received 145 codings, followed by ‘‘b180 Experience of self and time functions’’ (Body Functions: Mental Functions, 142 codings), ‘‘e445 Individual attitudes of strangers’’ (Environmental Factors: Attitudes, 136 codings) and ‘‘d240 Handling stress and other psychological demands’’ (Activities and Participation: General tasks and demands, 130 codings). Overall 35 ICF-CY categories in Activities and Participation, 24 categories in Environmental Factors, 14 in Body Structures and 11 in Body Function were represented on level three. CONCLUSIONS: A unique pattern of domains and dimensions appeared with ICF framework, suggesting relevance of statements that reflect patients’ experience of diversity. When comparing the information gleaned from the ICF-CY to an inductive qualitative approach, gaps in addressing wellbeing became apparent, while functioning issues were more comprehensively addressed with the ICF-CY. This supports uniting both approaches when attempting to reflect young patients’ concerns.

(2026) Qualitative development of a new clinician reported measure of multiple sclerosis status Glenn A. Phillips, PhD, Biogen, Cambridge, MA, United States; Louis S. Matza, PhD, Evidera, Bethesda, MD, United States; Katie D. Stewart, MA, Evidera, Bethesda, MD, United States; Jessica B. Jordan, PhD, Evidera, Bethesda, MD, United States AIMS: This qualitative study was designed to understand how patients with multiple sclerosis (MS) are typically assessed during office visits with neurologists. This information was gathered to support the possible development of outcome measures for patients with MS. METHODS: Concept elicitation interviews were conducted with patients diagnosed with MS as well as neurologists who treat MS. Patients and neurologists were recruited from MS centers and general neurology practices in the US. Patient interviews focused on MS symptoms and impact, followed by discussion of assessment procedures during office visits with neurologists. Neurologist interviews focused on identifying methods they

123

Qual Life Res (2015) 24:1–191 use to assess MS severity including symptoms, impact, and relapse. RESULTS: Interviews were conducted with 14 MS patients (42.9 % relapsing remitting, 42.9 % secondary progressive, 7.1 % primary progressive, 7.1 % chronic progressive) and nine neurologists between June and October 2014. Patients reported symptoms in a wide range of domains (e.g., walking, weakness, coordination/balance, sensory, fatigue, vision, bladder), as well as functional impairment (e.g., daily activities, mood/emotions, work performance, driving, self-care). When asked about office visits, patients reported that neurologists typically ask at least one general question about their MS and perform tests. Six patients (42.9 %) reported that their doctor also asks more targeted questions (e.g. specific symptoms and medication use). Ten participants (71.4 %) reported that their doctor does not ask about the impact of MS on their life. Neurologists reported assessing a wide range of MS signs, symptoms, and impacts. When evaluating MS status, neurologists rely primarily on a combination of patient history, neurological exam, performance-based measures (e.g., EDSS, timed walk tests, cognitive tests), and observation. Unstandardized questioning about patient history was the most common assessment method. CONCLUSIONS: Patients and neurologists said MS symptoms are typically assessed via general questions and tests including a neurological exam. Respondents indicated that standardized patient-reported and clinician-reported measures were not commonly used, and impact on functioning was not consistently assessed. A brief clinician-reported measure of MS symptoms and impact could help standardize and quantify these assessments.

(2028) A PRO surveillance feasibility pilot, the prevalence and correlates of symptoms-related PROs it collected and the quality improvement reports it produced Tenbroeck Smith, MA, American Cancer Society, Atlanta, GA, United States; Alyssa Troeschel, MPH, American Cancer Society, Atlanta, GA, United States; Kate Castro, RN, MS, AOCN, National Cancer Institute, Rockville, MD, United States; Neeraj Arora, PatientCentered Outcomes Research Institute, Washington, DC, United States; Kevin Stein, PhD, American Cancer Society, Atlanta, GA, United States; Ryan M. McCabe, PhD, American College of Surgeons, Chicago, IL, United States; Steven Clauser, PhD, MPA, Patient-Centered Outcomes Research Institute, Washington, DC, United States; Elizabeth Ward, PhD, American Cancer Society, Atlanta, GA, United States AIMS: Repeatedly captured facility or population level PROs—PRO surveillance—could play an important role in quality monitoring, benchmarking, policy making and research. The Patient Reported Outcomes Symptoms & Side Effects Study (PROSSES) had three aims. (1) To test the feasibility of using the Commission on Cancer’s Rapid Quality Reporting System (RQRS) to identify patients and manage study data flow for the collection of PROs. (2) To describe predictors of patient perceptions of symptom care. And, (3) to provide actionable cancer center quality improvement reports based upon patient reported data. METHODS: We sampled patients 4–12 months from diagnosis with locoregional breast/colon cancer from 17 hospital-based community cancer centers (CCs). Software automating case identification, sampling and reporting was developed and betatested. The questionnaire was adapted from validated NCI instruments and mailed using the Dillman method with online option. RESULTS: Hospital staff spent an average of 4 h per month working on PROSSES, mostly on recruitment activities. 2,487 eligible patients responded (RR = 61 %). 70 % of patients reported a health provider gave advice on coping with pain related to their cancer, 61 % for fatigue, 54 % for emotional distress (ED). 61 % reported being bothered by pain, 74 % by fatigue, 46 % by ED. Among those bothered, 42 % reported definitely getting the help they wanted for

Qual Life Res (2015) 24:1–191 pain, 40 % for fatigue, 46 % for ED. Multi-level multivariate logistic regressions revealed statistically significant correlates of the above variables including age, education, cancer type, months since surgery, and months since chemotherapy. CONCLUSIONS: The system for collecting PROs has potential for wider application as evidenced by low cancer center staff level of effort and high response rate. Onethird to half of patients did not recall getting needed advice or help for three common symptoms, suggesting room for improvement in symptom care. Each participating center received a custom quality improvement report containing study-wide and CC-specific estimates of patient perceptions of symptom care. Centers responded favorably to these reports. Efforts to improve care could take the form of professional/patient education or applying standards of care.

(2030) Measurement validation in patient-reported outcomes: challenges and opportunities Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; Eric K. H. Chan, PhD, Trinity Western University, Langley, BC, Canada; Bruno D. Zumbo, PhD, University of British Columbia, Vancouver, Canada; Clifton O. Bingham, MD, Johns Hopkins University, Baltimore, MD, United States; William Gardner, Children’s Hospital of Eastern ON, Ottawa, Canada; Jeffrey Jutai, University of Ottawa, Ottawa, Canada; Ayse Kuspinar, McGill University Health Centre - RI, Montreal, QC, Canada; Tolulope T. Sajobi, PhD, Departments of Community Health Sciences & Clinical Neurosciences, University of Calgary, Calgary, AB, Canada; Lisa M. Lix, PhD, University of Manitoba, Winnipeg, MB, Canada AIMS: Measurement validation is fundamental in the development, evaluation, and use of patient-reported outcomes (PRO) instruments. In this presentation, we provide an analysis of modern perspectives of measurement validity and address questions germane to the validation of PRO instruments. METHODS: We reviewed the literature on current interdisciplinary perspectives of measurement validation, including Messick’s unified view of construct validity, to address three questions regarding the use of PRO scores: (1) What evidence is needed to warrant comparisons between groups and individuals? (2) What evidence is needed to warrant comparisons over time? (3) What are the value implications, including personal and societal consequences? RESULTS: Measurement validity theories have expanded significantly, from the early days in which the practices of validation were dominated by criterion- and content-based procedures, to Cronbach and Meehl’s ideas of a nomological network, to Messick’s unitary view of construct validity. The Draper-Lindley-de Finetti (DLD) framework of measurement validation by Zumbo provides a theoretical foundation for the use of statistical methods (factor analysis, item response theory, latent variable mixture modeling) to evaluating the dimensionality of a PRO instrument and its measurement equivalence for different groups and individuals (question 1) by examining exchangeability of both measurement items and sampling units (groups or individuals). In addition to conventional methods for evaluation of test–retest reliability, theories and methods for the detection of response shift (including recalibration, reconceptualization, and reprioritization) provide a foundation for examining the comparability of PRO scores over time (question 2). Finally, based on the work of Messick and others, it is increasingly recognized that measurement validation involves careful consideration of underlying value implications, including personal and societal consequences (question 3). Because the use of PROs has both personal and societal consequences, we must evaluate the intended consequences as well as the unintended side effects. CONCLUSIONS: Measurement validation is about the process by which the validity of inferences or decisions based on PRO scores is established. It must be grounded in evidence regarding the comparability of scores across groups and

141 overtime, as well as an understanding of the values and consequences underlying the use of PRO scores.

(2032) Rasch analysis of patient-reported outcomes measure of pharmaceutical therapy for quality of life (PROMPT-QoL) Phantipa Sakthong, Associate Professor, Faculty of Pharmaceutical Sciences, Chulalongkorn University, Bangkok, Thailand, Chanadda Chinthammit, University of Arizona, College of Pharmacy, Tucson, AZ, United States AIMS: To assess the psychometrics of a new instrument for medication management: the Patient-Reported Outcomes Measure of Pharmaceutical Therapy for Quality of Life (PROMPT-QoL) using a Rasch analysis. METHODS: Outpatients who were 18 or over and had been taking medicines at least 3 months were purposively sampled from three university hospitals in Bangkok, Thailand, between July 2014 and March 2015. The 42-item PROMPT-QoL questionnaire included eight domains: Medicine and Disease Information, Satisfaction with Medication Effectiveness, Impacts of Medicine Sideeffects, Psychological Impacts of Medication Use, Convenience, Availability and Accessibility, Therapeutic Relationships with Healthcare Providers, and Overall QoL with 9, 3, 8, 9, 3, 4, 3, and 3 items, respectively. Each item was measured on a 5-point scale with response choices from ‘‘Not at all’’ to ‘‘Very Much.’’ The Rasch analysis employing WINSTEPS 3.81.0 was conducted to examine item and category fit statistics, person and item reliabilities, and response category ordering. RESULTS: A total of 1,156 patients responded to the PROMPT-QoL with a response rate of 84 %. All domains met the assumptions of unidimensionality and local independency. Out of 42 items, 41 items demonstrated acceptable item fit properties (Infit and outfit mean-squares range 0.5–1.5). Only one item ‘‘Trust the doctor in prescribing medicines’’ provided outfit mean-square of 1.64. All domains demonstrated very high item classification (Item separation [ 3.0; Item reliability [ 0.9), whereas five domains exhibited acceptable person reliability [ 0.8 and four domains showed good person classification (person separation index [ 2.00). Only two domains of Impacts of Medicine Side-effects and Availability and Accessibility yielded unacceptable person separation index of 0.93 and 1.09, respectively, and person reliability of 0.46 and 0.54, respectively. Possible reasons for the low person reliabilities and separation indexes were that only half of the sample had medicine side-effects and the three university hospitals had similar healthcare services. None of the 5 categories of all domains showed misfit since they provided outfit mean-squares \ 2.0. For all domains, category measures and thresholds increased monotonically with advancing category. CONCLUSIONS: The PROMPT-QoL questionnaire showed acceptable psychometrics using the Rasch analysis. Future research needs to be done in various kinds of healthcare settings.

(2034) The McGill Quality of Life Questionnaire-Expanded (MQOL-E) and the concept of overall quality of life Lara B. Russell, PhD, University of Victoria & Centre for Health Evaluation and Outcome Sciences, Vancouver, BC, Canada; Anne M. Gadermann, PhD, University of British Columbia, Vancouver, University, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; S. Robin Cohen, PhD, McGill University and Lady Davis Institute, Montreal, QC, Canada AIMS: The McGill Quality of Life Questionnaire (MQOL) was developed to measure quality of life (QOL) for people with a lifethreatening illness. The original MQOL has 16 items covering four domains (physical; psychological; existential; support), plus a Single-

123

142

Qual Life Res (2015) 24:1–191

Table 1 Estimated second-order factor loadings and correlations with SIS MQOL-E domains

Factor loading

Correlation with SIS

Physical

0.70

0.68

Psychological

0.68

0.45

Existential Relationships

0.83 0.58

0.52 0.30

Health care

0.45

0.32

Cognition

0.66

0.31

Feeling like a burden

0.40

0.31

Physical surroundings

0.50

0.31

Item Scale (MQOL-SIS) measuring global QOL. A revised version presented at the ISOQOL conference in 2014, MQOL-R, covers the same domains with some improvements. We recently created MQOLExpanded (MQOL-E) to add coverage of contributors to QOL found to be important in studies since MQOL’s creation. Six items covering new domains (cognition; physical surroundings; feeling like a burden; quality of health care) were added to MQOL-R for a total of 8 hypothesized domains and 21 items. The aim for this report was to examine the measurement structure of the MQOL-E and specifically determine the: (1) presence of a second-order latent factor representing overall QOL, (2) relative importance of the MQOL-E domains in relation to the second-order factor, and (3) correlations of the MQOL-E domains and second-order factor with MQOL-SIS. METHODS: Data came from four Canadian samples of people with life-threatening illnesses from various care settings (N = 869). Multiple imputation was used for missing data. Confirmatory factor analysis was applied to evaluate the first- and second-order factor models of the MQOL-E; standardized loadings for the second-order factor structure were estimated to evaluate relative importance. Correlations were calculated between the SIS and (a) the 8 MQOL-E latent factor domains, and (b) the second-order factor (overall QOL). RESULTS: There was good fit for the first-order (CFI = .963; RMSEA = .033) and second-order (CFI = .932; RMSEA = .042) factor models (Table 1). Standardized second-order factor loadings ranged from .83 (existential) to .40 (feeling like a burden). Correlations between the MQOL-SIS and the domains ranged from .68 (physical) to .30 (relationships) (Table 1), and .66 with the second order factor. CONCLUSIONS: The results support the use of an MQOL-E Total score (the mean of the 8 submeasure scores). Existential wellbeing, along with the physical and psychological domains, is an important contributor to QOL in people with life-threatening illness. MQOL-E is conceptually related to but not identical to global QOL.

(2036) Comparison of computerized adaptive testing and fixedlength short forms of the Prosthetic Limb Users Survey of Mobility (PLUS-MTM) Dagmar Amtmann, PhD, University of Washington, Seattle, WA, United States; Jiseon Kim, University of Washington, Seattle, WA, United States; Hyewon Chung, Chungnam National University, Daejeon, Korea, South; Ryoungsun Park, University of Texas, Austin, TX, United States; Rana Salem, University of Washington, Seattle, WA, United States; Brian J. Hafner, University of Washington, Seattle, WA, United States AIMS: The Prosthetic Limb Users Survey of Mobility (PLUSMTM) is an item bank developed to measure mobility of people

123

with lower limb amputation who use prosthetic limb(s) to ambulate. The purpose of this study was to compare the performance and efficiency of computerized adaptive testing (CAT) to two fixedlength short forms (SFs) (7- and 12-item). METHODS: PLUS-M item bank includes 44 items calibrated to item response theory (IRT) and the scores are on a T-metric (M = 50, SD = 10). The CAT and SFs were administered to adults with lower limb amputation. The minimum number of CAT items administered to each participant was 4 and maximum was 12. The CAT stopped when the standard error (SE) dropped below 3 or 12 items had been administered. We examined the number of items administered, standard errors (SEs) of the final scores, and the correlations between the CAT- and the SF-based scores. RESULTS: 199 lower limb prosthesis users responded to the CAT and SFs. On average, the CAT administered 5 items (range 4–12) with 59.8 % responding to 4 items, 20.6 % to 5 items, and 8.5 % to 6 items. 7 participants who chose the highest or lowest response to most items responded to the maximum number of items. The CAT was more efficient for scores between 40 and 60, with an average of 4 items (n = 155, SD = 0.7; range = 4–7) administered. Average PLUS-M scores were: CAT 50.8 (SD = 8.6); 7-item SF 50.5 (SD = 7.7); and 12-item SF 50.3 (SD = 7.9). Average SEs associated with these scores were 2.7 (SD = 0.3) for CAT; 2.9 (SD = 0.5) for the 7-item SF; and 2.2 (SD = 0.5) for the 12-item SF. For participants with SE [ 3.0, the averages scores were 64.8 (n = 5; an average of 12 items administered) for CAT; 59.0 (n = 60) for the 7-item SF; and 62.6 (n = 16) for the 12-item SF. Pearson correlations between the CAT score and the 7- and 12-item SFs scores were both 0.9. CONCLUSIONS: PLUS-M CAT required fewer items and estimated scores with greater precision than either fixed-length SF. The CAT could be made more efficient by implementing different stopping rules for those who select the highest or the lowest response category.

(2038) Refining items in the Long-Term Conditions Questionnaire (LTCQ): cognitive interviews, stakeholder feedback and translatability assessment Caroline Potter, University of Oxford, Oxford, United Kingdom; Cheryl Hunter, University of Leeds, Leeds, United Kingdom; Laura Kelly, University of Oxford, Oxford, United Kingdom; Elizabeth Gibbons, University of Oxford, Oxford, United Kingdom; Crispin Jenkinson, University of Oxford, Oxford, United Kingdom; Ray Fitzpatrick, University of Oxford, Oxford, United Kingdom; Michele Peters, PhD, University of Oxford, Oxford, United Kingdom AIMS: This research sought to clarify and reduce items within the 23-item draft Long-Term Conditions Questionnaire (LTCQ), a patient-reported outcome measure (PRO) designed to be applicable in mental and physical health and multi-morbidity. METHODS: The 23 items were generated through literature reviews, professional stakeholder interviews, and interviews with 42 patients who had at least one long-term condition (LTC). Thirteen of these patients (age range 36–88 years) participated in a cognitive interview. The interviews were audio-recorded and analysed for comprehensiveness, acceptability, interpretability, and feasibility of individual items and the LTCQ as a whole. The draft questionnaire was also sent to the professional stakeholders who had informed initial item generation; 13 stakeholders from across health and social care provided comments. A professional translatability assessment was undertaken to assess the LTCQ’s potential for use in diverse languages: Arabic, French, Polish, Punjabi, simplified Chinese, and Urdu in addition to the original English. RESULTS: Participants took 6 min on average to complete the questionnaire and expressed general support for the paper-based, self-administered format. All participants found the item content

Qual Life Res (2015) 24:1–191 relevant, but suggestions were made for deleting repetitive or ambiguous items and for simplifying the item content. Professional stakeholders also found the item content highly relevant and made suggestions for additional questions or item revisions. The translatability assessment raised no fundamental concerns but suggested minor revision to five items for greater clarity. Based on these three strands of work, six of the initial 23 items were deleted, six items were substantially revised to clarify meaning (including one original item being split into two new items), five items were refined through minor wording changes, and six items were retained unchanged. Items were also re-ordered as patients found it difficult to switch repeatedly between positively and negatively phrased items. This work resulted in a revised LTCQ of 18 items. CONCLUSIONS: The LTCQ is acceptable to people living with LTCs, and its potential value to professional stakeholders across health and social care has been reconfirmed. Next steps will include further cognitive interviews with patients ahead of testing via a large-scale survey.

(2040) Age-specific factor structures can improve the sensitivity of patient reported outcome measures: evidence from the Grow up Gothenburg study Sarah Hitz, MSc, Hamburg University of Applied Sciences, Hamburg, Germany; Ebba Brann, PhD student, Inst. of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Kerstin Albertsson-Wikland, MD, PhD, Professor, Inst. of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Zita Schillmo¨ller, PhD, Hamburg University of Applied Sciences, Hamburg, Germany, John Chaplin, PhD, AFBPsS, CPsychol., Inst. Clinical Sciences, Sahlgrenska Academy at the University Gothenburg, Sweden, Gothenburg, Sweden AIMS: To confirm and test the factor structure of the Gothenburg well-being scale for children (GWBc) in a school based pre-adult population (18–19 year olds). The GWBc was developed in 1994 as a measure of quality of life to be used in clinical studies of short stature. The 34 items were derived from a pool of 49 items following an exploratory factor analysis. It is a patient-reported questionnaire consisting of VAS (scales with bipolar adjectives as endpoints derived from a general school population (9–13 years). The GWBa has six domains: mood, self-esteem, physical condition, psychosocial functioning, stability and coping. It has been used consistently in studies of children’s quality of life in short stature. METHODS: 5002 students (51 % boys) in the final grade (school year 12) of high school in Gothenburg completed the questionnaire as part of a larger survey. A confirmatory factor analysis (CFA) (Amos 20.0) was carried out. RESULTS: Assumptions of normality were confirmed on all items. The CFA showed fit discrepancies (Chi square test p \ 0.001) with the original. The GFI and AGFI did not pass the threshold for acceptability (GFI = 0.80; AGFI = 0.77). The standardised root mean square residual (SRMR) was too high (SRMR = .075). RMESA had an almost acceptable fit value (RMSEA = 0.08). An alternative model for adolescents (GWBa) was developed excluding 7 items and including 8 items previously excluded from the original pool of 49 adjectives. Internal consistency was acceptable (Cronbach´s alpha [0.75). The domains constituted unitary concepts. Scale intercorrelations within the GWBc scale ranged from 0.17 to 0.53 indicating that the new model (GWBa) had generally higher correlated domains (0.277–0.861). CONCLUSIONS: The factor structure developed on a pre-teen population was not relevant to an older preadult population. This indicates the necessity of age specific instruments. The Gothenburg Well-being scale for Adolescence (GWBa) can be used in pre-adult populations to measure quality of life and well-being. It has applications in several areas of research including gender analysis and disability assessment.

143 (2042) Development of thresholds for clinical importance for four key domains of the EORTC QLQ-C30 Johannes M. Giesinger, Department of Psychiatry and Psychotherapy, Medical University of Innsbruck, Innsbruck, Austria; Wilma Kuijpers, The Netherlands Cancer Institute, Amsterdam, Netherlands; Teresa Young, Lynda Jackson Macmillan Centre, Mount Vernon Cancer Centre, Northwood, Middlesex, United Kingdom; Krzysztof Tomaszewski, Department of Anatomy, Jagiellonian University Medical College, Krakow, Poland; Elizabeth Friend, Basingstoke & North Hampshire Hospital, Basingstoke, United Kingdom; August Zabernigg, Kufstein County Hospital, Kufstein, Austria; Bernhard Holzner, Medical University of Innsbruck, Innsbru¨ck, Austria; Neil K. Aaronson, The Netherlands Cancer Institute, Amsterdam, Netherlands AIMS: The EORTC QLQ-C30 questionnaire is increasingly being used to monitor cancer patients’ quality of life in daily clinical practice. In such a setting the interpretation of questionnaire scores can be facilitated by using thresholds for clinical importance. Our study aimed at identifying thresholds for clinical importance for four EORTC QLQ-C30 scales that are often affected by cancer and its treatment: Physical Functioning (PF), Emotional Functioning (EF), Pain (PA) and Fatigue (FA). METHODS: For our study we recruited cancer patients at the Netherlands Cancer Institute (the Netherlands), Kufstein County Hospital (Austria), Mount Vernon Cancer Centre (UK), Basingstoke & North Hampshire Hospital (UK) and at Jagiellonian University Medical College (Poland). No restrictions were placed on diagnosis or type of treatment. Patients completed the QLQ-C30 and, for each of the four QLQ-C30 domains (PF, EF, PA and FA), three anchor items assessing burden, limitations in daily activities and need for help. We merged the three anchor items into a dichotomous external criterion to estimate thresholds for clinical importance. Statistical analysis relied on Receiver Operator Characteristic (ROC) curves. RESULTS: We recruited 548 patients with a mean age of 60.6 years. 54 % were female and the most frequent diagnoses were breast (26 %), colorectal (13 %) and lung cancer (12 %). The QLQ-C30 scales showed high diagnostic accuracy with regard to predicting burden, limitations and need for help related to PF, EF, PA and FA (all areas under the curve were above 0.86). We were able to estimate thresholds for clinical importance for these four QLQ-C30 scales. CONCLUSIONS: Thresholds for clinical importance will facilitate interpretation of scores at both an individual and group level. When used in daily clinical practice, these thresholds can help to identify patients with clinically relevant problems requiring further exploration and possibly intervention by health care professionals. The thresholds also allow converting QLQ-C30 scores into prevalence rates.

(2044) Improving patient centered care: implementation of PROMIS in a surgical spine practice Richard L. Skolasky, ScD, Johns Hopkins University, Baltimore, MD, United States; David Li, Johns Hopkins University, Baltimore, MD, United States; Brian J. Neuman, MD, Johns Hopkins University, Baltimore, MD, United States AIMS: Patients provide important information as experts in disease experience. Improved understanding of symptoms will result from combined patient-provider perspectives. Existing patient reported outcomes (PROs) were developed with variable levels of patient input. Significant changes in diagnosis and treatment of spine disease and patient expectations have occurred. It is critical to accurately assess patient experiences to integrate their assessment as measures of treatment outcome. The goals are: prioritize important aspects of disease experiences and identify corresponding PROMIS measures;

123

144 estimate correlation between existing PROs and PROMIS; and estimate clinically meaningful change and difference in PROMIS. METHODS: We enrolled 68 patients with degenerative lumbar spine disease [Mean (SD) age 56.5 (10.5) years; 51 % Women] treated with laminectomy with or without fusion. Pain numeric rating scale (NRS); Brief Pain Inventory, Pain Interference (BPI); Oswestry Disability Index (ODI); SF12 physical and mental health; Depression (PHQ8); and Anxiety (GAD7). PROMIS Pain interference; Physical function; Fatigue; Anxiety; Depression; Sleep disturbance; and Satisfaction with Social Role. Patient interviews were conducted to determine important aspects of disease experience. We assessed patients before and 6 weeks, 3, 6, and 12 months following surgery. Pearson correlation was estimated between PROs and PROMIS measures. Using published anchors, we estimated clinically meaningful change on PROMIS. RESULTS: Preoperative mean (SD): back pain NRS, 6.0 (2.6); leg pain NRS, 6.2 (2.9); pain interference, 12.1 (14.4); ODI, 43.9 (18.8); SF12 physical, 33.3 (10.4) and mental, 44.9 (11.2); depression (PHQ8), 7.9 (6.4); and anxiety (GAD7), 7.1 (6.8). Preoperative mean (SD) PROMIS measures: Pain interference, 64.1 (6.8); Physical function, 37.8 (8.5); Fatigue, 54.9 (10.1); Anxiety, 53.5 (10.9); Depression, 50.9 (10.5); Sleep disturbance, 53.9 (10.7); and Satisfaction with Social Role, 42.1 (7.8). At the preoperative visit, there were significant positive correlations between PROs and PROMIS measures (p \ .05 for all comparisons). Using ODI anchor, clinically meaningful improvement was associated with a 7-point change in PROMIS pain interference and physical function and a 9-point change in PROMIS depression. CONCLUSIONS: Patients were able to identify key aspect of their disease experience addressed by PROMIS. Patients reported satisfaction with electronic data capture. Moderate to strong correlation indicated convergent validity of PROMIS. PROMIS measures are sensitive to change following spine surgery.

(2046) Assessing meaning and purpose in life: development and validation of an item bank and short forms for the NIH PROMIS John M. Salsman, Wake Forest School of Medicine, Winston-Salem, NC, United States; Crystal L. Park, University of Connecticut, Storrs, CT, United States; Benjamin D. Schalet, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Elizabeth A. Hahn, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Michael F. Steger, Colorado State University, Fort Collins, CO, United States; Login George, University of Connecticut, Storrs, CT, United States; Mallory A. Snyder, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; David Cella, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: The Patient-Reported Outcomes Measurement Information System (PROMIS) is an NIH Roadmap initiative to improve and standardize PROs for patients with a range of conditions and demographic characteristics. In order to address a gap in the adult assessment of psychological well-being, we sought to develop and validate an item response theory (IRT)-based PRO tool of meaning and purpose in life (M&P) for inclusion in PROMIS. METHODS: Informed by a literature review and qualitative input from patients and content experts, a pool of 52 items was created to assess M&P concepts. The item pool was comprised of 18 items from the NIH Toolbox Meaning & Purpose Survey, 8 items from the PROMIS Pediatric Meaning & Purpose Short Form, and 26 newly identified and written items to ensure adequate content coverage. Items were administered to a general population sample (n = 1000) for calibration and validation purposes. We randomly split the sample into halves and conducted an EFA (n = 500) followed by several CFAs (n = 500). IRT analyses

123

Qual Life Res (2015) 24:1–191 included assessment of differential item functioning (DIF). RESULTS: In the EFA scree plot, there was one dominant factor with a second distinguishable but smaller factor comprised of negatively worded items. CFA had good fit for a 37-item bank of positively worded items (CFI = 0.962, TLI = 0.960, RMSEA = 0.085). All items were free of gender, age, education and race DIF. Item calibrations and content considerations informed the selection of 8-, 6-, and 4-item short forms. Internal consistency estimates ranged from alpha = .90 (4-item short form) to alpha = .98 (37-item bank). The 8-item M&P short form was correlated with the Meaning in Life Questionnaire-Presence of Meaning Subscale (r = .89), the Life Engagement Test (r = .79), the full PROMIS M&P item bank (r = .98), and the PROMIS M&P item bank less short form items (r = .96). CONCLUSIONS: The PROMIS M&P measure demonstrated sufficient unidimensionality and displayed good internal consistency, model fit, and convergent validity. The PROMIS M&P item bank and short forms were designed to aid clinicians and researchers to better evaluate and understand the potential role of positive psychological processes for individuals with chronic health conditions. Further psychometric testing in patients with chronic diseases, will help evaluate the generalizability of this new tool. (2048) The impact of item format on item information and resulting person parameters in patient-reported outcomes measuring physical function Gregor Liegl, Mag., Charite - Universitaetsmedizin Berlin, Berlin, Germany; Felix Fischer, ScD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Jakob B. Bjorner, MD, PhD, QualityMetric; University of Copenhagen, Copenhagen, Denmark; James F. Fries, MD, Stanford University School of Medicine, Stanford, CA, United States; Barbara Gandek, PhD, University of Massachusetts Medical School, Worcester, MA, United States; John Ware, PhD, University of Massachusetts Medical School, Worcester, MA, United States; Sandra Nolte, PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Matthias Rose, MD PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany AIMS: An essential step in the development of patient-reported outcomes (PROs) is the determination of the specific item format, including item content and wording. Using the example of measuring physical function (PF), respective item content can focus on either a person’s capability or on a person’s performance, while respective item wording may either be deficit- or resource-oriented. The aim of this study was to investigate if these aspects of item structure affect item information and resulting person parameters (thetas). METHODS: In the course of the PROMIS wave 1 data collection, participants rated their PF status based on five daily activities presented in four different item formats, resulting in a set of 20 items. Three formats aimed at capability (‘‘Are you able to…?’’, ‘‘How difficult is it for you to…?’’, ‘‘Does your health now limit you to…?’’); one asked for performance (‘‘Over the last 7 days, did you…?’’). Using Item-Response Theory (IRT), we applied a Graded Response Model (GRM) and a Generalized Partial Credit Model (GPCM) and compared the format-specific item information curves for each of the five activities. By fitting an IRT-based mixed-effect model, we examined the impact of the item format on the theta estimates. RESULTS: Using an incomplete blocked design, data were used from n = 15,721 participants who responded to a subset of the PROMIS PF item bank. Item information was similar for capability item formats with a maximum information between 4 and 9, depending on rated activity and the underlying IRT model. The performance item format led to considerably lower item information with a maximum information of \0.5. Capability formats did not differ in resulting theta estimates, while the performance format systematically underestimated person parameters (p \ .0001). The range of high item information along the

Qual Life Res (2015) 24:1–191

Fig. 1 .

145 Physical Function, Fatigue, Sleep-related Impairment, Headache, Depression, Anxiety, Anger, Alcohol Use, and (when it becomes available) Abuse/Misuse of Prescription Medication. A provider’s summary report enhances the clinical management of patients with pain. This report is a summary of an individual patient’s status based on that patient’s self-report. Patients’ self-reported goals are included on the report along with their self-rated progress towards these goals. Graphs and tables show opioid equivalents, pain maps, symptoms and outcomes, and progression toward goals over time in addition to concerning results from the most recent evaluation (e.g., screened positive for depression, alcohol misuse, PTSD). The report is a powerful tool to efficiently focus the limited clinical encounter time on specific treatment issues that the patient has identified as relevant. To date, 774 individuals have completed PROMIS measures at baseline; 420, 230, 129, and 52 have completed one, two, three, and four follow-up assessments, respectively. A retrospective study is underway to evaluate the validity and response burden of PROMIS measures. Results of these analyses will be presented. CONCLUSIONS: PASTOR has substantial potential for military medicine both as a mechanism for building a pain registry and as a clinical tool for helping patients living with pain.

(2052) Psychometric validation of the Questionnaire on Pain Caused by Spasticity (QPS) Fig. 2 . theta continuum was broader for the GRM, while the maximum information was higher for the GPCM. CONCLUSIONS: Different capability item formats used for the assessment of physical function lead to similarly high item information and to equal theta estimations. In contrast, a performance item format does not seem to be appropriate for measuring the PROMIS PF construct (Figs. 1, 2).

(2050) Leveraging PROMIS measures to build and pilot the DoD’s Pain Assessment Screening Tool and Registry (PASTOR) Karon F. Cook, PhD, Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Michael A. Kallen, PhD, MPH, Northwestern University, Feinberg School of Medicine, Missouri City, TX, United States; Chester (Trip) Buckenmaier, III, Uniformed Services University of the Health Sciences, Rockville, MD, United States; Richard C. Gershon, PhD, Northwestern University Feinberg School of Medicine, Chicago, IL, United States AIMS: To describe development and implementation of the US DoD’s Pain Assessment Screening Tool and Outcome Registry (PASTOR) and present results of retrospective analysis of the first year of data collection. METHODS: Identification of measures to include in PASTOR was accomplished with input from multiple stakeholders across clinical types and specialties. An initial version of PASTOR was integrated into a standalone AC-Lite application for piloting at Walter Reed Army Medical Center and at Madigan Army Medical Center. Based on additional feedback from clinicians and patients using the pilot, PASTOR measures are now being implemented within a military health system’s electronic health record system. RESULTS: Twelve PROMIS measures were selected for inclusion in PASTOR (Pain Intensity, Pain Interference, Pain Quality,

Donald M. Bushnell, MA, Health Research Associates, Inc., Seattle, WA, United States; Thorin L. Geister, PhD, Merz Pharmaceuticals, Frankfurt am Main, Germany; Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States; Zhenhuan Liu, MD, Guangzhou University of Traditional Chinese Medicine, Guangdong, Foshan, China; Yang Jie, MD, XiangYaBoAi Rehabilitation Hospital, Changsha City, Hunan, China; Yu-qiong Zhang, MD, MCH Hospital of Dongguan, DongGuan City, GuangDong, China AIMS: To report initial psychometric validation of the Questionnaire on Pain caused by Spasticity (QPS), a modular patient- and observerreported outcome on spasticity-related pain (SRP) in children with cerebral palsy (CP) and their parents. METHODS: The study was a prospective, observational study in 137 children with upper and/or lower limb spasticity and their parents at three sites in China. Two visits with a 3-week interval provided data to evaluate and establish administrative burden (ease of use), scoring of the QPS (factor analyses, Rasch analyses), reliability (Cronbach’s alpha, intraclass correlation coefficient), and validity (correlations with quality of life [PedsQL], motor impairment [Gross Motor Function Classification Score, Gross Motor Function Measure-66, Manual Ability Classification System], spasticity [Ashworth Scale, Modified Tardieu Scale]). RESULTS: Children (and parents) reporting more demanding activities also reported higher levels of associated pain (or observed pain behavior). Activity-related SRP items were combined for a total QPS score. Cronbach’s alpha was low for child self-reports (0.28 and 0.58), but was acceptable for interview-administered and parent reports on SRP behavior (0.70–0.91). Test–retest reproducibility was high for all modules (0.92–0.97). Depending on the module and child sub-population, the found hypothesized relationships to other outcomes were stronger between QPS and quality of life as measured by the PedsQL (0.33–0.69, p \ 0.01). Lower associations of the QPS were seen with motor impairment and spasticity due to the fact that SRP was ubiquitously distributed across content constructs. CONCLUSIONS: This initial study was successful in establishing the psychometric characteristics of the QPS. Despite complications with

123

146 varied levels of functional and cognitive compromise in the children with CP, clinic staff was able to administer the QPS versions easily, and both children and parents were able to complete the designated QPS in a proper way.

Qual Life Res (2015) 24:1–191 http://www.common-metrics.org facilitates the estimation of IRT latent trait estimates for researchers less familiar with advanced psychometric techniques. Additional IRT models, e.g. correcting for DIF in different age-groups or languages, can be easily added. Simple scoring on common metrics is an important step towards the establishment of instrument independent scales.

(2054) Continuity and change on the AQOL-MHS over 8 months Ligia Chavez, PhD, University of Puerto Rico – Medical Sciences Campus, San Juan, PR, United States; Pedro Garcia, Medical Sciences Campus Univ. of Puerto Rico, San Juan, PR, United States; Nyrma Ortiz, Medical Sciences Campus Univ. of Puerto Rico, San Juan, PR, United States; Patrick Shrout, New York University, New York, NY, United States AIMS: The Adolescent Quality of Life Mental Health Scale (AQOLMHS) is designed to measure quality of life in clinical samples of Latino adolescents aged 12–18 years. It was developed using a mixed method design and preliminary findings supports its reliability, validity and our conceptual model for the use of its three domains (Emotional Regulation—ER, Self-Concept—SC and Social Context—SoC). Our current aim tests the usefulness of the AQOL-MHS for tracking changes in HRQOL during the course of service use. METHODS: Three waves of data were collected from 59 participants who were recruited from public mental health clinics, and the waves were spaced 4 months apart. Participants were receiving services at baseline assessment and were tracked for follow up appointments regardless of treatment status. RESULTS: A Generalizability Theory (GT) analysis revealed that the AQOL-MHS domains can measure change reliably. The reliability estimates varied from .65 to .78. Although there was reliable change at the individual level, on the average the AQOL-MHS means improved only slightly over the three waves. CONCLUSIONS: QOL related to mental health problems in Latino adolescents can be measured over time. In future studies we will track associations of AQOL-MHS domain change with changes in symptom levels, impairment and social functioning.

(2060) Simple web-based IRT score estimation for common metrics: http://www.common-metrics.org Felix Fischer, ScD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Matthias Rose, MD PhD, Charite - Universitaetsmedizin Berlin, Berlin, Germany, Sandra Nolte, PhD, Charite Universitaetsmedizin Berlin, Berlin, Germany AIMS: item response theory allows calibration of Patient Reported Outcomes on the same scale. Such models have been for example recently used to calibrate the PROMIS measures for Anxiety and Depression on a common scale with other well established instruments such as the BDI, HADS, CESD, PHQ-9 and others. We developed a web-application (http://www.common-metrics.org) to allow IRT-score estimation for such models in a convenient way. METHODS: The webapplication is programmed using open source software R. It allows IRTscore estimation under the ML, MAP, WLE and EAP approach using item parameters from various published IRT models. Several different priors can be used. RESULTS: Researchers can easily upload raw data and receive latent trait score estimates, which can later be analyzed in standard statistical software packages. Compared to cross-walk tables, this approach tolerates missing data and takes all the information from the full response pattern into account. CONCLUSIONS:

123

(2062) The influence of chronic illness treatment and self-care on patient functioning and well-being: validating the impact scales of a new measure of treatment burden David T. Eton, Mayo Clinic, Rochester, MN, United States; Kathleen J. Yost, Mayo Clinic, Rochester, MN, United States; Jordan K. Rosedahl, Mayo Clinic, Rochester, MN, United States; Mark Linzer, Hennepin County Medical Center & University of Minnesota Medical School, Minneapolis, MN, United States; Deborah H. Boehm, Hennepin County Medical Center, Minneapolis, MN, United States; Azra Thakur, Hennepin County Medical Center, Minneapolis, MN, United States; Sara Poplau, Minneapolis Medical Research Foundation, Minneapolis, MN, United States; Roger T. Anderson, University of Virginia School of Medicine, Charlottesville, VA, United States AIMS: We recently developed a multi-dimensional measure of treatment burden. This study validates two scales from this measure: (1) role and social activity limitations and (2) physical/mental exhaustion from treatment and self-care. METHODS: We mailed surveys comprising the treatment burden measure and several standardized measures to 838 multi-morbid patients from two medical centers. Reliability and construct validity of the two scales were assessed, including Spearman’s correlations with standardized measures and mean score comparisons of known groups (ANOVAs). Multivariable linear regressions were used to determine how various aspects of burden are related to scores of the two scales. RESULTS: Surveys were completed by 331 patients (39.5 % response, age range 26–90 years, 56 % female, 73 % collegeeducated). Median number of diagnosed conditions was 3. Both the activity limitations and exhaustion scales showed excellent internal consistency reliability (Cronbach’s alphas = 0.95). Scores were strongly correlated with scores on standardized measures (all p values \.001). Specifically, more activity limitations and exhaustion from treatment and self-care were associated with more overall distress due to health problems (rhos = 0.64 and 0.67), less satisfaction with medications (rhos = -0.35 and -0.44), lower self-management selfefficacy (rhos = -0.40 and -0.50), lower general physical health (rhos = -0.55 and -0.59), and lower general mental health (rhos = -0.39 and -0.59). In known-groups analyses, patients reporting more (versus fewer) financial problems due to their health reported more activity limitations and exhaustion (ps \ .001); those with 18+ visits to a provider in the last 18 months (versus fewer visits) reported more exhaustion (p \ .005). Regressions identified several aspects of perceived burden that may predict activity limitations and exhaustion (adjusted R-squares = 0.43 and 0.54). Common to both models was a strong association between relationship tension (e.g., lack of understanding from family/friends) and worse outcomes (ps \ .001). CONCLUSIONS: The activity limitations and physical/mental exhaustion scales of this new measure of treatment burden are reliable and valid. These scales can help to characterize the overall impact of treatment and self-management burden on quality of life. Scaling and psychometric testing of the rest of the measure is ongoing.

Qual Life Res (2015) 24:1–191

147

Response Shift (2064) Testing response shift in the SF-36 physical functioning subscale using item response theory (IRT) models: results from the Canadian Multicentre Osteoporosis Study Lisa M. Lix, PhD, University of Manitoba, Winnipeg, MB, Canada; Xiuyun Wu, Dalhousie University, Halifax, NS, Canada; Wilma Hopman, Queen’s University, Kingston, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada; Tolulope T. Sajobi, PhD, Departments of Community Health Sciences & Clinical Neurosciences, University of Calgary, Calgary, AB, Canada; Juxin Liu, University of Saskatchewan, Saskatoon, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada AIMS: Differential item functioning (DIF) occurs when individuals with the same underlying level of health respond differently to the items that comprise a health measure. Changes in item functioning over time may indicate the presence of response shift (RS) in the data. The study purpose was to test for recalibration and reprioritization RS using item response theory (IRT) models for DIF in the SF-36 physical functioning (PF) subscale in a population-based cohort. METHODS: Study data were from the Canadian Multicentre Osteoporosis Study (CaMos), a prospective cohort study initiated in 1996 to provide national estimates of the prevalence and incidence of osteoporosis and fractures. Individuals were recruited regardless of disease history from nine sites. A two-parameter standard graded response IRT model was applied to the 10 items in the PF subscale at baseline and 5-year follow-up. Models that did and did not constrain the item thresholds and/or factor loadings to be equal across occasions was used test for RS. Goodness of fit was evaluated using likelihood ratio tests. Latent factor scores with and without adjustment for RS were computed. RESULTS: The CaMos cohort had 9423 respondents at baseline. A total of 7539 (80.0 %) had complete data at both occasions; this group comprised the analytic sample. The cohort was primarily female (70.9 %) and almost equally represented in younger (25–64 years: 56.2 %) and older age groups. A model that did not constrain parameters across measurement occasions was a significantly better fit (p \ .0001) than a constrained model, suggesting the presence of RS. Reprioritization was indicated in four items and

Table 2 Response shift models for the SF-36 physical functioning sub-scale items in the Canadian Multicentre Osteoporosis Study Model

RS effect tested

-2*(Difference in LL)

Difference in df

p value

Model 1

Unconstrained model (free item parameters)

147.71

28

<0.0001

Model 2

No RS model (constrained item parameters)

Reference

Reference

Models 3.1–3.10

Recalibration (uniform) RS

Model 3.1

RS for PF1- Vigorous activities

31.92

2

<0.0001

Model 3.2

RS for PF2-Moderate activities

0.24

2

0.885

Model 3.3

RS for PF3-Lifting or carrying groceries

8.25

2

0.016

Model 3.4

RS for PF4-Climbing several flights of stairs

4.46

2

0.107

Model 3.5

RS for PF5-Climbing one flight of stairs

4.99

2

0.082

Model 3.6

RS for PF6-Bending, kneeling or stooping

2.70

2

0.259

Model 3.7

RS for PF7-Walking more than a mile

11.22

2

0.004

Model 3.8

RS for PF8-Walking several blocks

5.93

2

0.052

Model 3.9

RS for PF9-Walking one block

1.39

2

0.498

Model 3.10

RS for PF10-Bathing or dressing self

13.68

2

0.001

Models 3.11–3.20

Reprioritization RS

Model 3.11

RS for PF1

4.44

1

0.035

Model 3.12

RS for PF2

3.29

1

0.070

Model 3.13

RS for PF3

0.46

1

0.496

Model 3.14

RS for PF4

17.45

1

<0.0001

Model 3.15

RS for PF5

0.23

1

0.629

Model 3.16

RS for PF6

7.61

1

0.006

Model 3.17

RS for PF7

8.56

1

0.003

Model 3.18

RS for PF8

7.99

1

0.005

Model 3.19

RS for PF9

0.82

1

0.364

Model 3.20

RS for PF10

3.37

1

0.066

Table 1

Distribution of item responses (%) for the SF-36 physical functioning items Canadian Multicentre Osteoporosis Study (N = 7539) Item

Baseline

Year 5

Response option

Response option

Limited a lot

Limited a Not little limited at all

Limited a lot

Limited a Not little limited at all

PF1: Vigorous activities

39.6

36.8

23.5

48.6

33.6

PF2: Moderate activities

9.1

23.9

67.0

14.1

26.3

59.6

PF3: Lifting or carrying groceries

6.2

21.0

72.8

9.9

23.1

67.0

11.6

29.4

59.0

18.1

30.5

51.4

PF5: Climbing one flight of stairs

3.8

14.0

82.2

7.1

16.6

76.4

PF6: Bending, kneeling or stooping

PF4: Climbing several flights of stairs

17.8

10.7

32.2

57.1

14.5

34.7

50.8

PF7: Walking more than 12.4 a mile

20.3

67.3

19.7

21.1

59.2

PF8: Walking several blocks

7.7

13.7

78.6

12.4

15.6

72.1

PF9: Walking one block

2.5

8.1

89.4

5.4

10.9

83.7

PF10: Bathing or dressing self

1.4

5.0

93.6

2.2

6.6

91.2

LL = log of the likelihood function; df = degrees of freedom; differences in LL and df were computed based on Model 2

recalibration in six items, with both forms of RS evident in items about walking more than one mile and vigorous physical activity. After adjusting for RS, a statistically significant decrease in latent factor means (standardized difference of 0.20) and increase in score variability was estimated over time. CONCLUSIONS: This study identified RS in the SF-36 PF subscale over a five-year period in population-based data, suggesting that RS occurs as a function of the aging process. Future research will test the association of population characteristics with the magnitude of bias in the item responses (Tables 1, 2).

(2066) Impact of response shift effect on the longitudinal analysis of health-related quality of life in oncology clinical trials: a simulation study Ame´lie Anota, PhD, Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸on, Besanc¸on, France; Antoine Barbieri, Biostatistic unit, Institut re´gional du Cancer de Montpellier (ICM) - Val d’Aurelle, Montpellier; Institut de

123

148 Mathe´matiques et de Mode´lisation de Montpellier, University of Montpellier 2, Montpellier, France; Marion Savina, INSERM, Clinical and Epidemiological Research Unit (CIC-EC 7) – CTD INCa, Institut Bergonie´, Bordeaux;INSERM CIC-EC7 Axe Cancer, Universite´ de Bordeaux, Bordeaux, France; Sophie GourgouBourgade, Biostatistic unit, Institut re´gional du Cancer de Montpellier (ICM) - Val d’Aurelle, Montpellier, France; Franck Bonnetain, Prof., Methodological and Quality of Life in Oncology Unit, EA 3181, University Hospital of Besanc¸on Quality of Life in Oncology National Platform, Besanc¸on, France; Caroline Bascoul-Mollevi, PhD, Biostatistic unit, Institut re´gional du Cancer de Montpellier (ICM) - Val d’Aurelle, Montpellier, France AIMS: An important challenge of the longitudinal analysis of Healthrelated quality of life (HRQoL) is the potential occurrence of a Response shift effect (RS) resulting in a recalibration, a reprioritization and a reconceptualization of HRQoL. A differential occurrence of the RS effect on each treatment arm should be investigated to assess if over- or under estimation of the treatment effect could occur due to RS. The objective is to investigate and compare the impact of the occurrence of the recalibration component of the RS effect on three statistical methods for the longitudinal analysis through a simulation study. METHODS: The methods proposed were the score and mixed model, the longitudinal partial credit model (LPCM) and timeto-event analysis approach based on the time to HRQoL score deterioration (TTD). Several reference scores were explored as the reference score in the TTD approach (baseline, best previous or immediately previous score) in order to take into account the occurrence of the RS effect. Simulations compared the methods regarding the bilateral type I error and statistical power of the test of an interaction effect between treatment arm and time. Longitudinal HRQoL data were performed using a LPCM considering that the latent trait follows a multivariate normal distribution. Five measurement times were simulated assuming that a RS effect occurred since time #3. Both uniform and non-uniform recalibration effect were simulated by varying the item difficulty parameter for one or both treatment arms, in the same direction as the time effect. RESULTS: When a uniform recalibration occurred in only one arm, type I error rate highly increased and the statistical power highly decreased for all statistical methods. When a non-uniform recalibration occurred in only one arm, type I error rate remained unchanged while the statistical power slightly decreased for all statistical methods except the TTD as compared to the best previous or the immediately previous score for which the statistical power increased. CONCLUSIONS: The occurrence of a differential RS effect seemed to bias the results for all statistical approach except for the TTD as compared to the best previous or the immediately previous score seemed in the presence of a non-uniform recalibration.

(2068) Response shift detection at the item level: a comparison of item response theory and structural equation models-based methods Myriam Blanchin, PhD, University of Nantes, Nantes, France; Alice Guilleux, University of Nantes, Nantes, France; Antoine Vanier, University of Nantes, Nantes, France; Jean-Benoit Hardouin, PhD, University of Nantes, Nantes, France; Veronique Sebille, Professor of Biostatistics, University of Nantes, Faculty of Pharmacy, France, NANTES, France AIMS: Issues regarding the identification of the most appropriate methodological approach for the identification of response shift (RS) occurrence and for its appropriate adjustment in the analyses of longitudinal Patient Reported Outcomes (PRO) data remains debated. Analyses are usually performed at the dimension level. Performing analyses at the item level could provide new insight to RS

123

Qual Life Res (2015) 24:1–191 investigation. Relative performances of item response theory (IRT) models and Structural Equation Modeling (SEM) both implemented at the item level for RS detection and adjustment are assessed and compared in a simulation study. METHODS: Simulation studies are performed to compare IRT and SEM for detection and adjustment of RS effects at the item level in several simulated settings according to: (a) absence or presence of RS, (b) absence or presence of true change, (c) number of items affected by RS, (d) different types of RS (uniform or non-uniform recalibration, reprioritization), (e) different sample sizes, (f) different number of items in the scale. 500 complete datasets are simulated using Generalized Partial Credit Models for all previous parameters combinations. The performance of IRT and SEM is assessed regarding the type I error (when no RS is simulated), the power (when RS is simulated), RS assessment (regarding the items and the types of RS that were detected). Model fit and convergence are also evaluated for both methods. RESULTS: Model fit and convergence is correct for IRT and SEM in most cases but it can be altered when many items are affected by RS. SEM seems to perform better for detection of uniform recalibration whereas IRT seems to perform better for detection of non-uniform recalibration. CONCLUSIONS: This extensive simulation study provides the possibility to formally compare IRT and SEM response shift analyses performed at the item level. The pros and cons of using one method or another as well as the potential of using both of them as complementary analyses for RS and true change assessment will be underlined.

Other Health Conditions (2070) Evidence-based health education in asthma: no short-term effects on health-related quality of life and satisfaction with programme but higher patient perceived trainer skills Eva M. Bitzer, Medical Doctor, Senior Public Health Researcher, University of Education Freiburg, Freiburg, Germany; Kathrin Baeuerle, Health Education Research, University of Education Freiburg, Freiburg, Germany; Janine Feicke, Health Research, University of Education Freiburg, Freiburg, Germany; Wolfgang Scherer, Senior Consultant, Rehabilitation Centre Utersum, Utersum, Germany; Ulrike Spoerhase, Biologist, University of Education, Freiburg, Germany AIMS: We hypothesize an evidence-based patient education programme targeted at adult asthma patients in medical rehabilitation based on the theoretical domains framework for use in behaviour change and criteria for good educational practice will enhance asthma control. We expect the programme to be effective via enhancement of patient’s asthma related health literacy. However, effective patient education is said to be demanding for patients. This may reduce acceptance and satisfaction with the programme. METHODS: We conduct a prospective single-centre controlled trial in an inpatient medical rehabilitation centre (duration: 3 weeks). Target group: adult asthma patients (18–65 years). Intervention: modified patient education programme (MPEP). Comparator: usual education programme. Timeline: Apr 2013-Nov 2013: recruitment of control group (CC; n = 200), Dez 2013 until Feb 2014: Implementation of MPEP, Apr 2014-Nov 2014: recruitment of intervention group (IG; n = 200). Time points of measurement: At the beginning (T0) and the end of inpatient treatment (T1), 6 (T2) and 12 (T3) months after discharge. Instruments: a.o. Comprehensibility and impact of patient education (COHEP; heiQ, at T1 only) disease specific health related quality of life (St. George’s Respiratory Questionnaire (SGRQ), socio-demographics. We calculated mean Scores for COHEP and heiQ at T0 and mean differences in change scores T0-T1 for SGRQ-Scales. We tested statistical significance with the T Test and analysis of covariance. RESULTS: 424 patients completed questionnaires at T1 (IG: 185, CC: 192, mean age 50.6 years., 75.1 % females, living a mean of

Qual Life Res (2015) 24:1–191

149 dementia (12.4 %) were the most common underlying causes of death. Nursing home (39.7 %), hospital wards (27.7 %), hospice and specialised palliative care in-patient units (11.5 %), municipality short-term facilities (8.8 %) and specialised palliative home care (6.8 %) were the most common locations of care. The most frequent symptoms were pain (70.9 %), breathing rattles (52.4 %), anxiety (46.1 %), confusion (29.0 %), and shortness of breath (25.2 %). Total symptom relief was reported for 52.4 % of patients experiencing pain, 23.7 % of those experiencing breathing rattles, 8.5 % of those experiencing anxiety, 5.1 % experiencing confusion and 9.0 % experiencing breathlessness. Preliminary results suggest 5 latent classes that represent unique symptom alleviation profiles, with class sizes ranging from 9 to 33 %. CONCLUSIONS: Symptom alleviation is a quality indicator for palliative care that is of significant importance to the wellbeing of people during their last week of life. A better understanding of profiles and predictors of symptom alleviation can inform the development of care quality for the purpose of enhanced quality of end-of-life across different diagnoses and locations of care.

Fig. 1 . 22.7 years with asthma). We observed no statistical significant differences in any of the background variable between IG and CC. With regard to health-related quality of life, we noted no differences in mean change scores (T0-T1) in any subscales of SGRQ. The intervention group rated trainer skills an average 3 points better than the control group (p = 0.005), transferability, type and amount of (medical) information was rated no different by treatment group. Luckily, health-related quality of life improved in both groups during rehabilitation substantially, but not different by treatment (Fig. 1). CONCLUSIONS: The MPEP increased patient perceived trainer skills and with no side effects on programme acceptance or healthrelated quality of life.

(2072) Symptom alleviation profiles during the last week of life: a register based quality of end-of-life study ¨ hle´n, Ersta Sko¨ndal University College, Stockholm, Joakim O ˚ restedt, Linæus University, Kalmar, Sweden; Sweden; Kristofer A Cecilia Ha˚kanson, Ersta Sko¨ndal University College, Stockholm, Sweden; Anette Henriksson, Ersta Sko¨ndal University College, Stockholm, Sweden; Carl-Johan Fu¨rst, Lund University, Lund, Sweden; Lara Russell, PhD, University of Victoria, Vancouver, Canada; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada AIMS: To examine symptom alleviation profiles as quality indicators that reflect care and wellbeing of patients during the last week of life and improve understanding of patterns. METHODS: A cross-sectional design was chosen using data from the Swedish Palliative Care Registry, a national register for all deaths that contains professionals’ retrospectively-reported data about quality indicators reflecting care and wellbeing of patients during the last week of life. All deceased adults ([18 years) with expected deaths, registered in the register in 2012 and 2013 (approximately 60 % of the total death population in Sweden each year) were included. Descriptive statistical analysis of the sample and a preliminary latent class analysis have been performed to identify symptom alleviation profiles, including alleviation of pain, nausea, anxiety, shortness of breath, confusion, and breathing rattles. Analyses of demographic and health-related characteristics predictive of latent class membership are ongoing. RESULTS: 38766 men and 49117 women were included. 65 % were 18–59 years of age, 19.0 % were 60–79 years of age, and 16.0 % were 80–110 years of age. Neoplasms (36.7 %), cardiovascular diseases (31.1 %) and

(2074) Qualitative research about the pain coping methods of spinal cord injury patients Chihyoung Son, Doctor, Korean National Rehabilitation Hospital, Seoul, South Korea, KuemJu Lee, Korean National Rehabilitation Research Institute, Seoul, South Korea, Sook-Hyun Lee, Korean National Rehabilitation Research Institute, Seoul, South Korea AIMS: This study aimed to search the pain coping methods of spinal cord injury (SCI) patients through qualitative research. METHODS: From September to December in 2014, 11 SCI inpatients in Korean National Rehabilitation Hospital (KNRH) selected by snowballsampling were interviewed indepth about pain coping methods and difficulties with pain. Qualitative research method was used to investigate and analyzed the data. The interview was recorded and transcribed. The meaningful phrases were extracted from transcription. Domains with categories about pain coping methods were organized through audit. RESULTS: Three domains about pain coping methods of SCI inpatients in KNRH were extracted. The domains were to endure the pain, to request the standard treatments and to find new treatment methods for pain. To endure the pain, they were trying to concentrate on something like exercising or talking to others, to think that pain should be accepted and to expect getting better someday. Although standard treatments were performed for pain, they were anxious about getting worse. They try to find some new treatment methods such as traditional medicine to recover the pain. CONCLUSIONS: Many spinal cord patients are struggled with pain. They don’t know how to handle it. Therefore proper pain coping strategy is needed for spinal cord patients in KNRC.

(2076) Chronic postsurgical pain (CPSP) and health related quality of Life (HRQoL) in a multicenter cohort study Antonio Montes, MD, PhD, Parc de Salut MAR, IMIM (Hospital del Mar Medical Research Institute), Universitat Auto`noma de Barcelona, Barcelona, Spain, Gemma Vilagut, Analyst, Hospital del Mar Medical Research Insitute (IMIM), Barcelona, Spain; Jordi Cantillo, BSc., Parc de Salut MAR, IMIM (Hospital del Mar Medical Research Institute), Universitat Auto`noma de Barcelona, Barcelona, Spain; Carlos G G. Forero, Researcher, CIBER in Epidemiology y SAlud Pu´blica (CIBERESP), Barcelona, Spain; Jordi Alonso, Head, Hospital del Mar Medical Research Institute, Barcelona, Spain; Gisela Roca, MD, PhD, Hospital Universitari Germans Trias i Pujol, Universitat Auto`noma de Barcelona, Badalona, Spain; Sergi Sabate, Fundacio Puigvert, Barcelona, Spain; Jaume Canet, Hospital Universitari

123

150 Germans Trias i Pujol, Universitat Auto`noma de Barcelona, Badalona, Spain AIMS: To prospectively evaluate the relationship between health related quality of life (HRQoL) pre and post intervention and chronic post-surgical pain (CPSP). METHODS: Prospective multicenter cohort study in 23 Spanish hospitals of adult patients undergoing 4 surgical procedures: abdominal (AH) and vaginal (VH) hysterectomies, herniorrhaphy (HR), or thoracotomy (TT). HRQoL was assessed with the SF-12v2 before surgery and about 4 months after surgery. Patients’ CPSP was assessed at 4-month clinical visit as: (a) pain developed after surgery; (b) at least 2 months duration; (c) other causes for pain excluded; and (d) pain continuing from a preexisting problem excluded. Pre and post-surgical SF-12 physical (PCS) and mental (MCS) scores were compared (significance and effect sizes—ES) according to CPSP, stratified by type of intervention. Linear regression models were estimated to assess post-surgical SF-12 scores as a function of baseline HRQoL, CPSP and type of intervention. Models were adjusted by age, pre-surgical pain (0–10 scale, from no pain to worst imaginable pain), body mass index, and physical status (American Society of Anesthesiologist’s classification). RESULTS: SF-12 results were obtained from 1176 patients, 33 % (388) with CPSP. Patients developing CPSP had significantly worse pre-surgical physical SF-12 scores (44.7 in CPSP vs. 48, p \ 0.001), effect sizes below 0.5 for all surgical procedures. Similarly, pre-surgical mental SF-12 scores were worse in patients developing CPSP (53.9 vs. 50.1, p \ 0.001). Post-surgical physical SF-12 scores were significantly lower for CPSP in all surgical procedures, and ES from 0.5 in TT to 1.1 in AH). Post-surgical mental SF-12 ES comparing CPSP status ranged from 0.1 (VH) to 0.6 (AH). After adjustment, CPSP was significantly associated to post-surgical physical SF12 scores (beta = -2.6; p \ 0.001), but not to mental SF12 scores (beta = 0.1, p = 0.8). No interactions were found between CPSP and surgical procedure in any SF12 scores. Pre-surgical HRQoL was found to predict post-surgical HRQoL regardless of the development of CPSP. CONCLUSIONS: Regardless of surgical procedure, development of CPSP was associated with lower postsurgical physical SF-12 scores, but not mental scores. Pre-surgical HRQoL predicted subsequent SF12 scores irrespective of appearance of post-surgical pain and surgical procedure, suggesting that HRQoL might be useful for predicting post-surgical patient course. Funding: DIUE of Generalitat de Catalunya (2014 SGR 748); Fundacio´ La Marato´ de TV3 (071210-2007)

(2078) Loss to follow-up of participants in a large, national, multisite study of long term outcomes of burn injury survivors Kara McMullen, University of Washington, Portland, OR, United States; Alyssa Bamer, University of Washington, Denver, CO, United States; Dagmar Amtmann, PhD, University of Washington, Seattle, WA, United States; Radha K. Holavanahalli, University of Texas Southwestern, Dallas, TX, United States; Gretchen J. Carrougher, University of Washington, Seattle, WA, United States; Lynne Friedlander, Spaulding Rehabilitation Hospital, Boston, MA, United States; Nicole Gibran, University of Washington, Seattle, WA, United States; David Herndon, Shriner’s Hospital, Galveston, TX, United States; James A. Fauerbach, PhD, Johns Hopkins University, Baltimore, MD, United States; Shelley Wiechman, University of Washington, Seattle, WA, United States; Kurt Johnson, University of Washington, Seattle, WA, United States AIMS: To analyze loss to follow-up of participants in the Burn Model System (BMS) longitudinal database, a national multi-center research study of long term outcomes of burn survivors. The characteristics of those lost to follow-up have been examined before1; this study

123

Qual Life Res (2015) 24:1–191 replicated the analyses using data collected since the original study to examine if rate and predictors of loss to follow-up have changed. METHODS: Participants in the BMS met American Burn Association criteria for major burn injury, received care from 1 of 5 U.S. burn centers and agreed to participate in 2-year prospective data collection between 2002 and 2015. Analyses included descriptive statistics of follow-up rate and characteristics of those who did and did not continue participation in the study, and logistic regression models to predict loss to follow-up at each time-point (6, 12, and 24 months). RESULTS: 1161 burn survivors responded to the 6 month follow-up survey and 27 % (n = 431) were lost to follow-up. 978 survivors responded to the 12 month follow-up survey and 35 % (n = 523) were lost to follow-up. 738 survivors responded to the 24 month follow-up survey and 46 % (n = 618) were lost to follow-up. Overall, the retention rate has increased by approximately 10 % compared to the initial study examining data collected up to 2002; however, only 20 % of individuals who were lost to follow-up participated in a subsequent survey time point. In the regression models, male gender, single rather than married or living with a partner, black, younger age, history of alcohol abuse, a perineum burn, lower total body surface area burned, non-job related injury, and unemployment at time of burn were associated with increased risk of loss to follow-up. CONCLUSIONS: Overall the retention rate has increased by 10 % compared to the prior study. There is a need to develop targeted approaches for people at risk for attrition. This study points to the importance of focusing on prevention of loss to follow-up by developing relationships to retain participants. Researchers using the BMS data need to be aware of the threat of selective attrition and examine the potential impact on results.

(2080) Association between nutritional status and quality of life in cirrhotic patients Eva Jua´rez-Herna´ndez, MSc., Medica Sur Clinic & Foundation, Mexico City, Mexico; Cristina Go´mez-Gutie´rrez, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Ylse Gutie´rrez-Grobe, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Brenda V. Balderas-Garce´s, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Justo Ferna´ndez-Rivero, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Liz N. Toapanta-Yanchapaxi, MD, Medica Sur Clinic & Foundation, Mexico, Mexico; Maria del Carmen Manzano-Robleda, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Sofia Ornelas-Arroyo, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Martha H. Uribe-Ramos, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Martha H. Ramos-Ostos, MD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Misael Uribe, PhD, Medica Sur Clinic & Foundation, Mexico City, Mexico; Norberto C. Cha´vez-Tapia, PhD, Medica Sur Clinic & Foundation, Mexico City, Mexico AIMS: To determine the association between nutritional status and QoL in patients with cirrhosis. METHODS: We evaluated 45 patients with cirrhosis, nutritional status was classified according to Subjective Global Assessment as well nourished, risk of malnutrition and malnutrition. A questionnaire of frequency of food consumption was applied. QoL was assessed by Spanish version of Liver Disease Quality of Life Questionnaire. Analysis of QoL domains and nutritional status was performed by one way ANOVA. RESULTS: The 55 % (n = 27) were male, mean of age and body mass index were 58 ± 10 years, and 27.7 ± 6.3 kg/m2, respectively. Most of the patients (41 %) were A Child-Pugh Score, while 31 % (n = 15) were decompensated with history of variceal bleeding. Prevalence of wellnourished patients was 31 % (n = 15) while patients in risk of malnutrition was 47 % (n = 23) and malnourished patients was 14 % (n = 7). Mean of global QoL was 4.0 ± 0.8 score, in each domain of

Qual Life Res (2015) 24:1–191

151 be replaced by women’s retrospective questionnaire reports of adverse impact? RESULTS: The percentage of days with bleeding/ spotting in the third month was 33.6 % (n = 116). The complexity/ demands of the WHO method meant that aspects of this analysis were possible for only 68 and 37 women, and this attrition contributed to bias. At 6 months after IUS, the percentages reporting BTB as ‘inconvenient’, ‘more inconvenient than before’, and ‘worrying’ were respectively 33.9, 29.3 and 9.7 %. Questionnaire responses were strongly associated with diary data, and arguably are more relevant. CONCLUSIONS: Further work is required, involving LARC–users, to develop and validate a better assessment of impact on quality of life of BTB.

(2084) What is important to cardiac arrest survivors? A qualitative exploration of outcomes

Fig. 1 . questionnaire, mean score was in abdominal symptoms 5.6 ± 1.5; fatigue 4.05 ± 1.6; systemic symptoms 4.5 ± 1.2; activity 5.1 ± 1.4; emotional functions 4.5 ± 1.3; and worry 4.1 ± 1.7. Malnourished patients presented lower score of global QoL (4.3 vs. 2.8, p = 0.01), as well as activity (5.5 vs. 3.9, p = 0.05) and fatigue (4.9 vs. 2.8, p = 0.02) domains. We observed an association between nutritional status and emotional function domain (Fig. 1). There is no association between food consumption and QoL. CONCLUSIONS: In patients with cirrhosis, QoL is associated with nutritional impairment. However frequency of food consumption was not related.

(2082) Conceptualising and assessing bleeding side-effects of hormonal contraception

Laura Whitehead, University of Warwick, Coventry, United Kingdom; Gavin Perkins, Professor of Critical Care Medicine, University of Warwick/Heart of England NHS Foundation Trust, Coventry, United Kingdom, Kirstie Haywood, Senior Research Fellow, Warwick Medical School, Warwick University, Coventry, United Kingdom AIMS: Each year approximately 326,000 people have an out-ofhospital cardiac arrest in the United States, with 10.6 % of those Table 1 Examples of outcomes contributing to a disruption to normality Overarching theme

Superordinate theme

Subordinate themes

Disruption to normality

Survival

Closeness to death

Physical function

Fatigue

Gratitude to be alive

Pamela Warner, Reader in Medical Statistics, University of Edinburgh, Edinburgh, Scotland, United Kingdom; Hilary O. Critchley, Professor of Reproductive Medicine, University of Edinburgh, Edinburgh, Scotland, United Kingdom

Breathlessness

AIMS: Highly effective and reversible hormonal contraception has transformed lives, most particularly for women. Recently-developed long-acting reversible contraceptives (LARC), with no need to remember a daily pill, have enhanced convenience and effectiveness. However, LARC often has side-effects, particularly unscheduled break-through bleeding (BTB). BTB can be distressing, imposes costs and has adverse effect on quality of life - the extent depending on degree of BTB and individual factors. BTB is often endured for the sake of contraceptive benefit, sometimes over many years. In the 1970s WHO devised a comprehensive/complex analytic framework for assessing BTB, based on research thinking and clinical experience. However: (i) This did not draw directly on women’s accounts of impact of BTB; (ii) BTB diary completion is burdensome for women participating in research; (iii) The complexity of the framework means most BTB research assesses/reports only subsets of the measures, or bespoke variants, which prevents amalgamation of results across studies. METHODS: A placebo-controlled trial was undertaken of a compound that might mitigate BTB in new users of the LARC levonorgestrel-releasing intrauterine system (IUS) [Warner et al. (2010) Human Reproduction 25, trial funded by NIH]. BTB was measured by prospective daily diary method (for 6 months after IUSinsertion), and by women’s questionnaire report (at 1, 3, 6 months). Diary BTB data have now been summarised according to both WHO method, and a (proposed) simpler plan, to address the questions: Is the simpler summary adequately informative? Can either diary summary

Pain (rib break)

Vision Muscle weakness Activities of daily living/ increased dependence Cognitive function Side effects of treatment Emotional well-being Anxiety Confidence Depression Self esteem Personality changes Frustration Social well-being and Role: job, voluntary, carer participation Leisure: hobbies, sports Social activities Family: relationships Impact on others

Increased work/care Impact on their hobbies, work Strain on relationships

123

152 treated by emergency services surviving to hospital discharge emphasizing a need for research in this field (Mozaffarian et al. 2015). However, knowledge about what is important to patients surviving a cardiac arrest is limited; their experience following hospital discharge is rarely captured. A review of cardiac arrest clinical trials highlighted the dominance of clinician-based assessment and failure to capture the patients’ perspective (Whitehead et al. 2015). We explored the lived experience of cardiac arrest survivors and health outcomes important to their recovery to incorporate the patient perspective in the development of a core outcome set for clinical trials. METHODS: Patients were recruited from a Regional Hospital in the West Midlands, England. Inclusion criteria included survivors of an out-ofhospital cardiac arrest discharged from hospital within the previous 3–12 months, aged [18 years, cognitively unimpaired and not critically ill. Semi-structured interviews were conducted with patients and where possible with patients’ partners. An interpretative phenomenological analysis was applied. Throughout the research process we worked closely with a group of patient partners. RESULTS: A convenience sample of eight patients [mean 62.8 years (SD 13.6); range 41–79; 5 men (62.5 %); mean time 6.25 months post-discharge (SD:3.1)] and three partners were independently interviewed. Several overarching themes were common across interviews: disruption to normality; coping with what has happened; moving towards normal; relationships with healthcare; and uncertainty about the future. Themes contributing to a disruption to normality included: physical functioning, emotional well-being, social well-being and participation (Table 1). Patients displayed different coping mechanisms, strategies for managing new symptoms, and moving towards a ‘new normal’. CONCLUSIONS: Through an improved understanding of the lived experience of cardiac arrest survivors this study provides evidence of what is important to their recovery journey. Patients hold their pre-arrest status as a gold standard against which they judge their current health and ‘return to normality’. The data provides patient-derived guidance for the assessment of health outcomes and a basis for a conceptual framework to inform core outcome set and patient reported outcome measurements for this important population.

(2086) ‘‘They have NO idea what it’s like to live with IBD’’ The influence of other people’s perceptions on coping with inflammatory bowel disease Hannah Shucard, Western Washington University, Bellingham, WA, United States; Fuschia Sirois, Bishop’s University, Lennoxville, QC, Canada AIMS: Living with Inflammatory Bowel Disease (IBD) is often a stressful experience due to the embarrassing and unpredictable nature of the symptoms. It is therefore important to understand the factors that affect how people cope with IBD to promote successful adjustment. How well individuals cope is influenced by their beliefs about their ability to cope, or their coping efficacy. Self-efficacy is influenced by multiple sources, one of which is verbal persuasion. The aim of this study was to investigate how verbal persuasion, or other people’s perceptions of coping, was related to how individuals coped with IBD. METHODS: A sample of people diagnosed with IBD completed an online survey on adjustment to IBD (N = 149; 74.5 % female, mean age = 36.2 ± 12.6) and answered the open-ended question, ‘‘how does it make you FEEL knowing that other people may think you should be coping better or worse with your IBD’’, as well as measures of coping efficacy, well-being, distress, and symptom severity as assessed by the IBD-Q bowel symptom subscale (Guyatt et al. 1989). RESULTS: The text analysis program Linguistic Inquiry and Word Count counted the number of positive and negative emotion words in the text-based responses. Correlational analyses

123

Qual Life Res (2015) 24:1–191 revealed that individuals who perceived that other people thought they should be coping better used more negative emotion words (r = .352, p \ .001) and reported higher psychological distress (r = .528, p \ .001) and lower well-being (r = -.443, p \ .001) than those who perceived that others thought they should be coping better, even after controlling for symptom severity. A series of regression analyses revealed that coping efficacy mediated the link between other people’s perceptions and well-being, perceived stress, and helplessness (R2s = .34, .43, .41 respectively). CONCLUSIONS: These results highlight the impact that other people’s perceptions of coping have on adjustment to IBD. Specifically, the finding that psychological distress and decreased well-being was reported by individuals who believe others think they should be coping better suggests that interventions aimed at teaching family members and friends how to provide support for individuals living with IBD may improve coping.

(2088) The relevance of signs and symptoms measured by the Psoriasis Symptom Inventory (PSI) to patients with psoriatic arthritis Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States; Alex Mutebi, PhD, Amgen, Thousand Oaks, CA, United States; Ibrahim Ali, Health Research Associates, Seattle, WA, United States; Alice Gottlieb, Tufts Medical Center, Boston, MA, United States; Philip Mease, Swedish Medical Center, Seattle, WA, United States; David Harrison, Amgen, Thousand Oaks, CA, United States; Hema N. Viswanathan, Amgen, Thousand Oaks, CA, United States AIMS: Approximately 15 % of patients with psoriatic arthritis (PsA) experience some degree of skin lesion involvement. The Psoriasis Symptom Inventory (PSI) is a patient reported outcome (PRO) measure of the severity of psoriasis signs and symptoms. The PSI was developed in psoriasis patients, with a small proportion of patients having PsA. The use of the PSI to assess the severity of skin-related signs and symptoms in a population of PsA patients requires demonstration that PSI concepts are consistent with relevant and important skin-related experiences of PsA patients. This qualitative study sought to assess the relevance of signs and symptoms measured by the PSI to patients with PsA. METHODS: Adults with a confirmed diagnosis of PsA with skin involvement were enrolled in individual concept elicitation (CE) interviews. A semi-structured interview guide was used to elicit open-ended information from PsA patients about their skin-related signs and symptoms experiences. Interviews were audio recorded, transcribed and coded using Atlas.ti software. Coded data was grouped based on similar content and compared to the 8 PSI concepts. Results were reviewed by qualitative researchers and two expert clinicians (rheumatology and dermatology) experienced in the diagnosis, treatment, and management of PsA, and assessed for alignment with all the 8 concepts represented in the PSI (itching, flaking, scaling, redness, cracking, burning, stinging, and pain). RESULTS: 31 patients participated in the CE interviews, with a mean age of 54.8 (range 29–77), 54.8 % were female and 80.6 % were white. A total of 2,346 symptom-related expressions were coded from the transcripts; 77 % of which were related to skin signs and symptoms. 23 % of the coded expressions were related to the joint and functioning component of PsA. Overall, 74.8 % of the 77 % coded expressions related to skin signs and symptoms aligned appropriately to the 8 PSI concepts. Severity and bothersome ratings reported for the aligned concepts ranged from 6.4 to 10 on an 11 point rating scale. CONCLUSIONS: Results from these qualitative interviews demonstrate that the 8 signs and symptoms measured by the PSI represent the experience of skin-related signs and symptoms in patients with PsA.

Qual Life Res (2015) 24:1–191 (2090) Self-management behaviors and psychological and social factors influencing QOL in patients with parkinson disease

153 Table 1 Methodology

Yukako Ando, Nagoya City University, Nagoya, Japan; Yoshino Ueki, Nagoya City University, Nagoya, Japan; Takemori Yamawaki, Hiroshima City Hospital, Hiroshima, Japan; Itsuko Ozaki, Nagoya City University, Nagoya, Japan; Akemi Abe, Hatsukaichi City Health Center, Hatsukaichi City, Japan; Akira Inukai, Higashi Nagoya National Hospital, Nagoya, Japan; Ikuko Aiba, Higashi Nagoya National Hospital, Nagoya, Japan; Yufuko Saito, Higashi Nagoya National Hospital, Nagoya, Japan; Katsuhiko Kawaminami, Asaka City Public Health center, Asaka, Japan; Noriyuki Matsukawa, Nagoya City University, Nagoya, Japan; Toshio Kobayashi, Hiroshima University, Hiroshima, Japan AIMS: The purpose of this study was to clarify the self-management behaviors and psychological and social factors influencing the QOL of Parkinson disease (PD) patients. METHODS: The study subjects were outpatients of PD (n = 185) who visited a hospital during the period from October 2014 to March 2015. Self-reporting questionnaires were distributed to the subjects. QOL was measured by [a scale for subjective QOL commonly to patients with intractable diseases]. For self-management behaviors, we assessed 29 items related to the self-management necessary for PD patients, and scored them. For psychological factors, depression and sense of coherence (SOC). SOC is the capability to cope with stress, was evaluated. For social factors, Social Capital (SC), which represents trusted relationships and bonds in the society or community, was evaluated. In order to reveal factors influencing QOL, Pearson correlation analysis and multiple linear regression analysis were performed. RESULTS: The response rate was 97.3 %(180 of 185). The valid response rate was 95.7 % (177 of 185). The results of the correlation analysis indicated that the age (r = -0.17, p \ 0.05), Barthel Index (r = 0.21, p \ 0.01), SOC (r = 0.59, p \ 0.001), depression (r = -0.60, p \ 0.001) and the self-management behaviors (r = 0.23, p \ 0.01) had significant associations with QOL. However, there were no statistically significant associations between SC and QOL. The variables that were proved to have a significant correlation with QOL by the multiple linear regression analysis were SOC (B = 0.35, p \ 0.001), depression (B = -0.31, p \ 0.001), Barthel Index (B = 0.13, p \ 0.05), and self-management behaviors (r = 0.13, p \ 0.05). The adjusted determination coefficient was 0.43. CONCLUSIONS: As factors influencing QOL in patients with PD, SOC, depression, Barthel Index and self-management behaviors were extracted. It was indicated that, in addition to self-managing the disease properly, enhancing SOC, i.e. the capability to cope with stress, and preventing depression might help improve the QOL of PD patients.

(2092) Getting on with your life with MS: evaluation plan for a new global self-management workbook Vanessa Bouchard, McGill University, Montreal, QC, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada AIMS: Multiple sclerosis (MS), a progressive neurological disease, varies widely from person to person. Its symptoms range from mild sensory alterations to severe disability limiting activity and restricting participation. It is the leading cause of neurological disability among young adults. Kuspinar (2013) identified 62 domains of quality of life (QoL) affected by MS through semi-qualitative interviews with 190 people with MS. Many domains are rarely addressed by healthcare professionals. A need for a global approach targeting QoL through those domains was identified. Our team developed a self-management workbook called Getting on With Your Life With MS(GETONMS). Self-management coaches patients to maintain wellness rather than illness in their foreground perspective through development of five

core skills: problem solving, decision making, resource utilization, forming patient/healthcare provider partnership, and taking action. To apply those skills, GETONMS teaches to self-assess, identify and implement strategies to improve and monitor progress. Content and usability of GETONMS will soon be examined with peer review and qualitative interviews with potential users. GETONMS, once completed, will be an online, evidence-based, understandable, action-oriented tool covering 41 topics. It will allow the users to self-tailor the content to their needs as well as provide references to further knowledge on each topic. The global aim of this pilot is to estimate the extent to which GETONMS used alone or with a partner, compared to standard educational material, improves quality of life. METHODS: The accompanying Table 1 highlights the methodology. RESULTS: To date, the content workbook has been reviewed by 4 people with MS contributing to identifying omissions and content or wording problems. Many statements expressed approval and excitement, intention of implementing strategies, and anticipation for the final product. CONCLUSIONS: This project will provide preliminary evidence towards the effectiveness of GETONMS to improve QoL.

(2094) Structural equation modeling of the quality of life for patients with Marfan syndrome Ju Ryoung Moon, PhD, RN, Samsung Medical Center, Seoul, South Korea; Yong Ae Cho, PhD, RN, Chung-Ang Univ., Seoul, South Korea; June Huh, PhD, MD, Samsung Medical Center, Sungkyunkwan Univ. School of Medicine, Seoul, South Korea; I-Seok Kang, MD, Samsung Medical Center, Sungkyunkwan Univ. School of Medicine, Seoul, South Korea; Duk-Kyung Kim, PhD, MD, Samsung Medical Center, Sungkyunkwan Univ. School of Medicine, Seoul, South Korea AIMS: We aimed to use structural equation modeling to evaluate the quality of life (QOL) for patients with Marfan syndrome (MFS). The goal of this study was to provide guidelines to facilitate the development of interventions and strategies to improve QOL in patients with MFS. METHODS: The participants comprised patients who visited the cardiology outpatient department of a tertiary hospital in Seoul, Korea, between August 17, 2013 and April 17, 2014. Demographics, social support, disease-related factors, biobehavioral factors, and QOL data were collected in one-on-one interviews. RESULTS: Final analyses included 218 patients. Anxious or depressed patients accounted for 61 and 69 % of the sample, respectively. Model fit indices for the hypothetical model were suitable for the recommended level: v2 = 151.30 (P \ 0.001, df = 45), GFI = 0.91, RMSEA = 0.05, NFI = 0.93, CFI = 0.92, TLI = 0.97, PGFNI = 0.46, and PNFI = 0.44. In patients with MFS, QOL was significantly affected by social support, disease-related factors, and biobehavioral factors.

123

154 These variables accounted for 72.4 % of QOL in patients with MFS. Biobehavioral factors had the strongest and most direct effects on QOL. CONCLUSIONS: In order to improve QOL in patients with MFS, comprehensive interventions are necessary to assess and manage biobehavioral factors, social support, and disease-related factors.

(2096) Assessment of gender bias in musculoskeletal outcome measures: methods and examples Joy MacDermid, Professor, McMaster University, Hamilton, ON, Canada AIMS: 1. To describe the conduct and integration different methods for addressing the potential for gender bias in musculoskeletal patient-reported outcome measures MSK-PRO : Rasch, cognitive interviews, ICF linking, gender-expectations assessment, 2. To provide examples of items with different forms of gender bias from a series of studies conducted on assessment of MDK-PRO 3. To describe strategies for remediation of gender bias in existing items/tools METHODS: A series of studies were conducted on MSK-PRO to assess potential sources of bias with a particular emphasis on sex and gender bias. The measures evaluated included pain and disability measures relevant to patients with musculoskeletal pathology. The techniques used included 1. Rasch analysis (differential item functioning based on se) or gendered roles i.e. working inside or outside the home), 2. cognitive interviews to understand whether items were interpreted and calibrated differently based on gender, 3. Forward and backward translation of ICF coding- forward translation by linking of items to ICF codes and comparing the distribution of codes across sex/gender and backward translation by comparing the issues of importance within these codes across genders and 4. specific tools that detect differences in a trait (e.g. pain) based on gender role expectations. RESULTS: These methods provide complementary information that can identify potential items that exhibit gender bias, or sources of bias. Rasch analysis can identify some items that exhibit differential item functioning but does not identify whether the trait is manifested differently across genders or the items structure or content is gender biased. Items that do not exhibit differential item functioning can exhibit gender bias when investigated through qualitative or ICF linking methods. Differences in responses on MSK – PROM can be due to sex-based differences in physiology (e.g. pain processing) on the difficulty of the task (i.e. strength items), in the roles assumed by different genders with respect to a particular activity (e.g. household work) or behavioural/reporting expectations that have gender bias (e.g. pain). CONCLUSIONS: Differentiation of true differences between genders with respect to a specific trait can be difficult to differentiate from gender bias in the measurement process. Explicit consideration of the potential for sex and gender bias in MSK measurement is important and should be considered a fundamental step in development and validation of measures. Gender bias assessment should be multipronged to consider the source and type of potential gender biases-preferably in measure development. PRO remediation by item removal or rehabilitation is possible; and gendered comparisons may be an alternative to reduce the potential for conclusions that are invalid due to gender bias.

(2098) Self-rated health after stroke: a follow-up study with multiple measurements over a 2 years period Louise Pape Larsen, MHSc, PhD-student, Herning, Denmark; Karin Biering, MHSc, PhD, Herning, Denmark; Soeren P. Johnsen, MD, PhD, Aarhus, Denmark; Grethe Andersen, MD, DMSc, Aarhus, Denmark; Niels Henrik Hjollund, MD, PhD, AmbuFlex, Herning, Denmark

123

Qual Life Res (2015) 24:1–191 AIMS: Self-rated health is an important aspect of life after stroke, and patient-reported outcomes (PRO) are required to adequately describe prognosis. We aimed to provide a detailed description of the course and determinants of self-rated health from 3 to 24 months after stroke in a large population-based cohort of first time stroke patients, and to compare self-rated health to the general Danish population. METHODS: We followed 2414 stroke patients, admitted to any hospital in the Central Denmark Region between October 1, 2008 and December 31, 2011, with 5 questionnaires over a 2 years period. Self-rated health was assessed by the SF-12. Information on possible clinical and patient-related determinants of self-rated health was obtained from questionnaires and population-based national registers. RESULTS: The mean values of all SF-12 subscales were at all measuring times lower than that of the general Danish population. Mean values of the majority of SF-12 scales increased slightly (0.8–2.1 points per year), but statistically significant during the follow-up period. No improvement was, however, found in bodily pain, physical function or physical component score. The level of self-rated health differed across subgroups of all determinants, and for age, comorbidity and stroke-severity there were differences in the course of self-rated health between subgroups. CONCLUSIONS: In a population-based cohort of stroke patients, we found minor improvements in mental, but not physical health over time. Overall, the change over time in self-rated health after stroke was limited.

(2100) Associations between disease severity and health-related quality-of-life (HRQoL) outcomes in psoriasis patients Hilde Randa, MSc, PhD-fellow, Unit for Psychooncology and Health Psychology, Department of Oncology, Aarhus University Hospital and Department of Psychology and Behavioural Science, Aarhus University, Aarhus C, Denmark; Peter Jensen, DMSc, Department of Dermato-Allergology, Gentofte Hospital, Copenhagen University Hospital, Hellerup, Denmark; Lone Skov, Professor, DMSc, Department of Dermato-Allergology, Gentofte Hospital, Copenhagen University Hospital, Hellerup, Denmark; Claus Zachariae, Professor, DMSc, Department of Dermato-Allergology, Gentofte Hospital, Copenhagen University Hospital, Hellerup, Denmark; Robert Zachariae, Professor, DMSc., MSc, Unit for Psychooncology and Health Psychology, Department of Oncology, Aarhus University Hospital and Department of Psychology and Behavioural Science, Aarhus University, Aarhus C, Denmark AIMS: Psoriasis negatively affects quality-of-life (QoL), but little is known about how the different disease characteristics contribute to this effect. The aim was therefore to explore the associations of different aspects of disease severity with psoriasis-related QoL, sleep impairment, and depressive symptoms. METHODS: As a part of a larger study, 116 out-patients with psoriasis (66.4 % men, mean age: 47.8 years) in treatment at the Department of Dermato-Allergology, Gentofte Hospital, completed the Dermatology Life Quality Index (DLQI), the Insomnia Severity Index (ISI), Beck’s Depression Inventory (BDI), and the frequency, sensory, affective, and intensity subscales of the Itch Severity Scale (ISS) questionnaire. Psoriasis severity was evaluated by dermatologists with the Psoriasis Area and Severity Index (PASI). RESULTS: The mean PASI score was 5.8 (SD = 5.6; range 0–30) and mean DLQI score was 6.6 (7.2). Twenty-seven patients (23.3 %) reported clinical insomnia and 12 (10.3 %) had moderatesevere depression. When examining the relative contribution of the different PASI area scores to QoL, sleep impairment, and depression with multiple regressions, only the association between scores for the head region and DLQI scores reached statistical significance (Beta: 0.23; p = 0.012). When examined with univariate regression, total PASI scores were significantly associated with all three outcomes (DLQI: 0.38, p \ 0.001; ISI: 0.23, p = 0.011; BDI: 0.19, p = 0.046)

Qual Life Res (2015) 24:1–191

155

explaining 3–14 % of the variance. When adding itch severity (ISS) scores to the model, they emerged as significantly associated with all three outcomes (DLQI (omitting item 1 on itch): 0.45, p \ 0.001; ISI: 0.44, p \ 0.001; BDI: 0.52, p \ 0.001), explaining additional 14–19 % of the variance. Total PASI scores no longer reached statistical significance (p: 0.16–0.94). Boot-strapping procedures for assessing direct and indirect effects showed that itch severity mediated the association between PASI scores and all three outcomes (Effect: 0.52–0.28; p \ 0.001), while no directs effects of PASI scores reached statistical significance. CONCLUSIONS: The results indicate that itch severity, an aspect not covered by PASI, the commonly used psoriasis severity outcome, explains a significant proportion of the variance in psoriasisrelated QoL, sleep quality, and depressive symptoms. To better evaluate treatment effectiveness, dermatologists are recommended to include a validated measure of itch severity as a standard outcome assessment.

(2102) The quality of life in trauma patients at 6 months after injury: a prospective cohort study Nobuichiro Tamura, MD, Kurashiki central hospital, Kurashiki, Okayama, Japan; Hiroshi Okamoto, MD, Kurashiki central hospital, Kurashiki, Okayama, Japan; Toshie Kaihara, Kurashiki central hospital, Kurashiki, Okayama, Japan; Hayaki Uchino, MD, Kurashiki central hospital, Kurashiki, Okayama, Japan; Toshio Fukuoka, PhD, Kurashiki central hospital, Kurashiki, Okayama, Japan AIMS: Health-related quality of life (HRQOL) has gained widespread acceptance as an important measure of efficacy in clinical trials. However, there is paucity of prospective study regarding HRQOL after trauma. The aim of our study was to assess HRQOL improvement of trauma patients using SF-36 Health Survey at 6 months after injury. METHODS: A prospective cohort study was performed in our tertiary care hospital from November 2013 to November 2014. All consecutive trauma patients who were admitted to our department were included. Patients under age of 18, with cognitive impairment and who deceased were excluded. SF-36 score were obtained by direct interviews based on a standardized protocol at discharge and by mails at 6 months after injury. RESULTS: During the study period, complete data collection was achieved in 69 patients. Overall rate of completion of questionnaire was 76.7 % (69/90). Median age was 67 years (interquartile range; 47–75), and 46 (68 %) were male. Median Injury Severity Score was 17 (IQR; 13–24), length of stay at our department was 11 days (IQR; 4–42), and total length of hospital stay was 32 days (IQR; 6–3). While four of the eight SF 36 domains (Physical Functioning, Role Physical, Bodily Pain and Social Functioning) improved significantly over the 6 months follow up period, there was no significant increase (p \ 0.05) in the rest of the domains (General Health, Vitality, Role Emotional and Mental Health) between at discharge and at 6 months after Injury. Table 1 . At discharge 6 months after P value Median injury (IQR) Median(IQR) Physical functioning (SF-36) 21 (8, 41)

42 (33, 53)

0.00

Role-physical (SF-36)

20 (4, 33)

36 (23, 48)

0.00

Bodily pain (SF-36)

36 (30, 45)

42 (35, 49)

0.02

General health (SF-36)

46 (40, 54)

46 (40, 53)

0.98

Vitality (SF-36)

42 (33, 49)

43 (34, 51)

0.85

Social functioning (SF-36)

35 (21, 45)

44 (30, 56)

0.01

Role-emotional (SF-36)

36 (20, 54)

39 (28, 55)

0.06

Mental health (SF-36)

43 (32, 50)

43 (37, 56)

0.27

CONCLUSIONS: In trauma patients at 6 months after injury, all domains of SF-36 were below average of healthy Japanese individuals. Especially, domains of mental and role-social component had not improved during the study period. Our results suggest that trauma patients need not only physical rehabilitation but social and mental support. Further research is needed to elucidate the long term impact of trauma in HRQOL (Table 1).

Poster Session 3001: Saturday Poster Session 1 Clinical Practice (3003) PRO feedback in a clinical practice setting: What patientreported information do pediatric oncology providers find helpful? Lotte Haverman, PhD, Emma Children’s Hospital, Academic Medical Center, Amsterdam, Netherlands; Sasja Schepers, Emma Children’s Hospital, Academic Medical Center, Amsterdam, Netherlands; Sima Zadeh, International Cancer Institute at the National Institute of Health, Bethesda, MD, United States; Martha Grootenhuis, Prof, Emma Children’s Hospital, Academic Medical Center, Amsterdam, Netherlands; Wiener Lori, International Cancer Institute at the National Institute of Health, Bethesda, MD, United States AIMS: This study aimed to explore what patient-reported outcomes (PROs) pediatric oncology providers (POPs) find helpful in caring for their patients (8–21 years). METHODS: 352 (43 % male) POPs from 52 countries participated in an online self-administered 28-item questionnaire. POPs were recruited from different oncology societies. The questionnaire assessed POPs preferences regarding the use of psychosocial PRO measures. Responses on a 5-point scale were dichotomized into categories 1–2 ‘not useful at all’ and ‘hardly useful’ and 3–4–5 ‘somewhat useful’, ‘useful’, ‘very useful’. RESULTS: Almost all POPs (94 %) reported standardized use of PROs to be useful. POPs emphasized that PROs should include specific information on physical symptoms, and emotional, practical, social, school/cognitive, and spiritual difficulties. Perceived barriers to obtaining PRO data were: time (57 %), inadequate staff to address psychosocial issues raised (48 %) and PRO measures not being available at their organization (33 %). Most preferred to collect PRO data within a patient’s electronic record (97 %), printed on paper (85 %) and online (73 %). Preferred time-points for assessment were: during treatment (87 %), at diagnosis (86 %), and follow-up (81 %). Majority of POPs preferred child and parent report at ages 8–12 (92 %) and 13–17 (88 %); less than half felt parent report was necessary for ages 18–21 (47 %). CONCLUSIONS: To our knowledge, this is the first study to assess what information is most valuable to pediatric oncology providers in a pediatric screening setting. Our study showed that PRO data could be clinically useful for children with cancer. In most organizations this information is not integrated in standard care. To successfully incorporate PROs in clinical pediatric practice, the needs of pediatric oncology providers should be met by developing strategies to overcome the barriers.

(3005) The use of PRO measures in the clinical setting from a clinician perspective Michele Halyard, MD, Mayo Clinic, Scottsdale, AZ, United States; Amylou C. Dueck, PhD, Mayo Clinic, Scottsdale, AZ, United States; Louise Humphrey, PhD, Clinical Outcomes Assessment Abacus International, Manchester, United Kingdom; Anna Beurskens, PhD, PT, Maastricht University, Maastricht, Netherlands; Jeff A. Sloan, PhD, Mayo Clinic, Rochester, MN, United States

123

156 AIMS: The aim is to present the clinician experience and challenges of utilizing patient reported outcome (PRO) measures in the clinical practice setting to help inform clinicians contemplating embarking on PRO data collection, to articulate the potential value and utility of PRO data for clinical decision-making and patient management, and provide guidance to researchers wanting to implement PRO initiatives in clinical practice. METHODS: Literature and results from a previous pilot study were reviewed. The pilot study enrolled radiation oncology patients receiving 5–7 weeks of radiotherapy. Enrolled patients completed the Linear Analogue Self-Assessment (LASA) electronically at baseline and biweekly prior to seeing clinicians. LASA scores were not provided to clinicians during the treatment visit for the first 79 patients. For the subsequent 69 patients, LASA score were systematically provided to clinicians. Clinical staff underwent system training prior to the study, and completed a semistructured interview following completion of the study. RESULTS: Based on experience of this pilot trial in the radiation oncology setting, and drawing upon experience in the literature, key discussion points related to how PRO data can facilitate communication from a clinician standpoint will be shared. In addition, the barriers to PRO use from the clinician standpoint will also be presented including: (1) methods and timing of data collection that integrate efficiently in the clinical workflow, (2) need for clinician education on PRO use, and (3) additional resource requirements to support PRO use at the time of the patient encounter. CONCLUSIONS: By understanding the benefits and challenges of PROs from a clinician’s real-world usage of these measures, clinicians and other stakeholders will have increased understanding of how to implement PRO data collection in the clinical setting and avoid identified barriers that could derail success. While clinicians aim to take the best care possible of patients, the added value of PRO data collection in clinical practice should not detract from efficient clinical workflows. Understanding of pitfalls in PRO data collection will enhance chance of their successful use and facilitate clinician care and management of patients.

(3007) Current practices in patient-reported outcome (PRO) data collection in trials: a survey of UK trial staff and management Derek G. Kyte, PhD, University of Birmingham, Birmingham, United Kingdom; Jonathan Ives, PhD, University of Birmingham, Birmingham, United Kingdom; Heather Draper, PhD, University of Birmingham, Birmingham, United Kingdom; Melanie J. Calvert, PhD, University of Birmingham, Birmingham, United Kingdom AIMS: Patient-reported outcomes (PROs) are an important measure of effectiveness, increasingly used in clinical trials. PRO questionnaires should be administered in a standardized way across trial sites and routinely screened for avoidable missing data, in order to maximize data quality and minimize risk of bias. Our previous qualitative study, however, identified concerns about the consistency of PRO administration in trials. The purpose of the current study was to determine the extent to which these qualitative findings could be generalized to the wider community of trial personnel. METHODS: We conducted a national web-based cross-sectional survey of UK-based research nurses, data managers/coordinators, trial managers and chief/principal investigators involved in clinical trials that used PRO measures. Participants were recruited from all 55 UK Clinical Research Collaboration Registered Clinical Trials Units and 19 Comprehensive Local Research Networks. We undertook descriptive analyses of the quantitative data and directed thematic analysis of free-text comments. Factors associated with the management of missing PRO data were explored using logistic regression. RESULTS: 767 respondents completed the survey. The survey data supported the generalizability of previous qualitative study findings, suggesting inconsistencies in PRO administration with regard to the level of assistance given to trial participants, the timing of

123

Qual Life Res (2015) 24:1–191 PRO completion in relation to the clinical consultation and the way missing PRO data was managed. Having 10 years or more experience in the research role was significantly (p = .035) associated with trial personnel managing missing PRO data according to existing recommendations [odds ratio 2.26 (95 % CI 1.06–4.82)]. There were conflicting reports concerning the current level of PRO-specific guidance provided in trials. There was a consensus, however, that more guidance was needed in future trials and agreement between professional groups about the necessary components. CONCLUSIONS: There are inconsistencies in the way PROs are administered by trial staff. Such inconsistencies may reduce the quality of PRO data and have the potential to introduce bias. There is a need for improved PRO guidance in future trials that support trial personnel in conducting optimal PRO data collection.

(3009) Payer perspectives of patient reported outcomes data: an online assessment Susan L. Hogue, RTI Health Solutions, Research Triangle Park, NC, United States; Andrew P. Brogan, RTI Health Solutions, Research Triangle Park, NC, United States; Carla DeMuro, RTI Health Solutions, Research Triangle Park, NC, United States; Amy Barrett, RTI Health Solutions, Research Triangle Park, NC, United States, Lori McLeod, PhD, RTI Health Solutions, Research Triangle Park, NC, United States; Denise D’Alessio, Novartis Pharmaceuticals Corporation, East Hanover, NJ, United States; Vasudha Bal, Novartis Pharmaceuticals Corporation, East Hanover, NJ, United States AIMS: To examine the usefulness and future impact of patient-reported outcome (PRO) data on payer decision making on reimbursement and access of drugs. METHODS: An online survey was developed following a targeted literature review and one-on-one interviews conducted with a group of payer/payer advisors. The assessment was launched on December 8, 2014, with invitations sent to 197 payer and payer advisor e-mail addresses from the RTI Health Solutions Global Payer Advisory Panel. The survey closed on March 4, 2015, with 20 completed surveys (China, France, Germany, Spain, Taiwan, the United Kingdom [UK], and the United States [US]) and 7 partially completed surveys (Australia, South Korea, the US). RESULTS: Payers and payer advisors indicated that oncology is the disease area where PRO data is most influential. Key characteristics of PRO data that maximize influence on payer decision making were (1) content and psychometric validation of the PRO measure in targeted disease populations and (2) publication of PRO data in a peer-reviewed journals. PRO data from registration trials and postmarketing studies are both useful in payer decision making. While some payers preferred PRO data that was included in a regulatory label, most payers indicated that PRO data included in peer-reviewed publications have a greater influence on decision making. Payer responses revealed that differences in health care systems influence payer preferences for PRO data: 47 % of US payer respondents versus 78 % of ex-US payer respondents indicated that PRO data impact their decision making when considering one therapy over another. The consensus among payers surveyed was that PRO data, particularly in oncology, will continue to gain importance in decision making over the next 5 years. CONCLUSIONS: The results of this survey showed that the patient’s perspective is an important and growing aspect of payer decision making on reimbursement and access of drugs.

(3011) Challenges and opportunities for PROs and big data: applications for the pharmaceutical industry William R. Lenderking, PhD, Evidera, Lexington, MA, United States AIMS: This presentation will consider the pharmaceutical industry context and how PROs included in big databases could be helpful in

Qual Life Res (2015) 24:1–191 drug development, post-marketing surveillance, comparative effectiveness research, and validation. Big databases are increasingly being created and utilized to answer important questions about healthcare in populations. What is the burden of illness? What are the predominant treatment patterns? How do side effects impact adherence? What are the long-term safety effects of a given medication? METHODS: Literature review. RESULTS: These questions are all of importance to pharma. When the impacts of a disease or treatment are at least in part subjective, all of the preceding questions can be informed by PRO data. CONCLUSIONS: Drug development in an increasingly competitive marketplace requires more than evidence of safety and efficacy, but also requires evidence of value, which is based on demonstrations that the drug has positive benefits across an expanded set of outcomes, including those which are subjective to patients. Although analyses of big databases are not likely to replace standard clinical trials in the registration process, certain types of questions can be addressed robustly using big data. Post-marketing surveillance could benefit from the use of PROs in certain disease areas. Several points will be addressed, including: PROs are not routinely collected in big data sets; just because you have a lot of data, does not mean the data are unbiased, and this also applies to PRO data. Collecting additional evidence of validity could be a powerful use of big data for PROs.

Measure Development (3013) Using cognitive interviews to assess and refine the Coeliac Disease Assessment Questionnaire (CDAQ) Helen Crocker, MSc, University of Oxford, Oxford, United Kingdom; Crispin Jenkinson, University of Oxford, Oxford, United Kingdom, Michele Peters, PhD, University of Oxford, Oxford, United Kingdom AIMS: The Coeliac Disease Assessment Questionnaire (CDAQ) is a new patient-reported outcome measure developed to assess quality of life in adults with coeliac disease. The aim of this study was to identify and resolve possible sources of response error in the questionnaire through the conduct of cognitive interviews, as part of the measure’s development. METHODS: Cognitive interviews (n = 10) were conducted with adults with coeliac disease in two rounds, with revisions to the questionnaire occurring after each round. Participants were recruited through online social networking groups and from a pool of those who had expressed an interest in participating in the qualitative interview phase of the study. Interviews were conducted over the telephone and lasted for approximately 1 h. Participants completed the CDAQ during the interview and answered generic and item-specific questions about their experience of doing so retrospectively. RESULTS: Cognitive interview participants (male = 3; female = 7) were aged between 24 and 80 years old and had been diagnosed with the condition for between 3 and 16 years. Sources of response error identified in the cognitive interviews were due to participants answering items in general rather than in relation to their coeliac disease; interpreting words or phrases in different ways to each other; and interpreting items not as intended. Amendments were made to address the issues identified. For example, the phrase ‘away from your home’ was amended to ‘out of the house’ as participants interpreted the phrase as holidaying, rather than any instance of being outside the home as intended. The questionnaire’s instructions and formatting were also revised. For example, the item stem was amended to instruct participants to think about each item in relation to their coeliac disease. In total, 50 items were reviewed, of which nine items were revised, one item was deleted, and two items added. The additional items were alternative wordings of existing items. CONCLUSIONS: Conducting cognitive interviews as part of the development of the CDAQ allowed potentially problematic items to be identified and resolved. The questionnaire’s instructions and

157 formatting were also improved. The psychometric properties of the final version of the questionnaire are currently being assessed.

(3015) How much change really matters? Development and comparison of two novel approaches to defining clinically important differences in fatigue scores Karon F. Cook, PhD, Department of Medical Social Sciences, Northwestern University Feinberg School of Medicine, Chicago, IL, United States; Michael A. Kallen, PhD, MPH, Northwestern University, Feinberg School of Medicine, Missouri City, TX, United States; David Victorson, Northwestern University, Evanston, IL, United States; Deborah Miller, PhD, Cleveland Clinic, Cleveland, OH, United States AIMS: To test novel methods for establishing clinically important differences METHODS: T-score-anchored, clinical vignettes were constructed for the NeuroQoL Fatigue scale (T = 36–70; 2 pt. increments). Vignettes included 5 items and the most likely response to each item given the vignette’s T-score location. 500 individuals with multiple sclerosis responded to the fatigue short form, then were presented 7 vignettes that were in the range of participant’s own levels of fatigue (e.g. those with highest fatigue were presented vignettes anchored at T-score of 58–70). Respondents compared their fatigue to that represented by each vignette and rated their fatigue as greater, the same, or less. If greater/lesser, they reported if the change would be enough ‘‘to make a difference in [their] daily life’’. Clinically important improvement/decrement values were defined as score differences endorsed as enough to make a difference by [50 %. In a second question set, participants were shown 6 paired score reports representing baseline and one month change. Each had 5 items and showed improvements of 2, 4, and 6 points. Presented comparisons were targeted to participants’ own fatigue levels. Instructions read, ‘‘This individual’s fatigue IMPROVED over the past month…. How would you describe this improvement: (1) not enough to make a difference in one’s daily life; (2) enough to make a difference in one’s daily life (day-to-day activities would be easier)?’’ Respondents rated their confidence that their responses to the vignettes and to score comparisons reflected ‘‘how you would judge actual changes in your fatigue’’ (0–3; not at all, a little bit, moderately, very confident). (Example survey questions in Fig. 1). RESULTS: Based on the vignettes, clinically important improvement and decrement were defined as -3.7 and 3.9, respectively. In the score report comparisons, 30, 59 %, and 79 % indicated that 2, 4, and 6 point improvements, respectively, were ‘‘enough to make a difference’’. Mean confidence

Fig. 1 Sample questions from different sections of survey

123

158 ratings were [3.3 (moderate to very confident). CONCLUSIONS: Both methods contextualized score changes in the context of differences that would matter in daily life. Participants made judgments in the range of their own fatigue levels. The similar results and high confidence ratings suggest the potential of these novel methods.

(3017) Development of novel patient-reported outcomes (PRO) instruments to assess atopic dermatitis signs, symptoms and impacts Christina Curry, MSc., Genentech, Inc., South San Francisco, CA, United States; Martha Gauthier, Endpoint Outcomes, Boston, MA, United States; Mary Lynn Cala, Endpoint Outcomes, Boston, MA, United States; Shayna Egan, Endpoint Outcomes, Boston, MA, United States; Brooke Witherspoon, Endpoint Outcomes, Boston, MA, United States; Christopher Evans, Endpoint Outcomes, Boston, MA, United States; Kristina Fitzgerald, MPH, Genentech Inc., South San Francisco, CA, United States AIMS: The aim of this study was to develop patient-reported outcome (PRO) instruments that assess atopic dermatitis (AD) signs, symptoms and impacts according to the U.S. Food and Drug Administration’s (FDA) PRO Guidance. METHODS: Initially a literature review was conducted to assess the AD PRO landscape. Concept elicitation (CE) interviews with adolescent and adult subjects with moderate or severe AD were then conducted to elicit signs, symptoms and impacts of AD that are important to patients. Interview transcripts were qualitatively analyzed using Atlas.ti and an item generation meeting was held to review the data. Preliminary PRO instruments were drafted based on patient feedback from the CE interviews. RESULTS: Based on the literature review, preliminary conceptual frameworks (CFs) were developed for both AD signs and symptoms and AD impacts. Thirty eligible subjects participated in the CE interviews; 50 % were adolescents (ages 12–17) and 50 % were adults (ages 18–75). The most frequently reported signs and symptoms of AD were itchy skin (n = 30, 100.0 %), dry skin (n = 27, 90.0 %), redness of skin (n = 21, 70.0 %), sensitive skin (n = 18, 60.0 %), bumps on the skin (n = 17, 56.7 %), flaking skin (n = 16, 53.3 %), and skin pain (n = 16, 53.3 %). The most frequently reported impacts of AD were feeling self-conscious (n = 23, 76.7 %), skin damage (bleeding) (n = 22, 73.3 %), avoiding triggers (n = 19, 63.3 %), clothing choice to conceal AD (n = 18, 60.0 %), AD was noticeable (n = 18, 60.0 %), waking up (n = 17, 56.7 %), feeling frustrated (n = 17, 56.7 %), difficulty falling asleep (n = 16, 53.5 %), burning sensation (n = 16, 53.3 %), feeling embarrassed (n = 16, 53.3 %), negative impact on appearance (n = 16, 53.3 %), difficulty with sports and exercise (n = 16, 53.3 %), and unwanted attention (n = 15, 50.0 %). The CE interviews supported the development of one signs and symptoms instrument and one impacts instrument for both adolescents and adults. Based on the CE interview data and the item generation meeting, a 12-item instrument to assess the symptoms of AD and a 14-item instrument to assess the impacts of AD were drafted and the preliminary CFs were updated. CONCLUSIONS: This is the first step in developing two novel PRO instruments measuring the symptoms and impacts of AD. These instruments will be refined following cognitive interviews and psychometric validation.

(3019) Psychometric properties of the Patient-Reported Outcomes Measure of Pharmaceutical Therapy for Quality of Life (PROMPT-QoL) Phantipa Sakthong, Associate Professor, Faculty of Pharmaceutical Sciences, Chulalongkorn University, Bangkok, Thailand; Pattarin Sukarnjanaset, Faculty of Pharmaceutical Sciences, Chulalongkorn

123

Qual Life Res (2015) 24:1–191 University, Bangkok, Thailand; Nontapat Sonsa-ardjit, Faculty of Pharmaceutical Sciences, Chulalongkorn University, Bangkok, Thailand; Wipaporn Munpan, Faculty of Pharmaceutical Sciences, Chulalongkorn University, Bangkok, Thailand AIMS: To evaluate the psychometric properties of a novel instrument: the Patient-Reported Outcomes Measure of Pharmaceutical Therapy for Quality of Life (PROMPT-QoL) including practicality, reliability, and validity. METHODS: Eleven hundreds and fifty-six adult outpatients continuously taking any medicines at least three months were purposively sampled from three university hospitals in Bangkok, Thailand, between July 2014 and March 2015. The PROMPT-QoL consisted of 43 items including nine domains: General Attitude toward Medication Use, Medicine and Disease Information, Medication Effectiveness, Impacts of Medicine Side-effects, Psychological Impacts of Medication Use, Convenience, Availability and Accessibility, Therapeutic Relationships with Healthcare Providers, and Overall QoL. Practicality was assessed by administration times and percentages of missing data. Internal consistency and test–retest reliability were evaluated employing Cronbach’s alpha and intraclass correlation coefficients (ICCs), respectively. Validity was evaluated with construct, convergent, and known-groups validity. A confirmatory factory analysis (CFA) was employed to check the construct validity of the PROMPT-QoL. For convergent validity, a multiple regression was utilized to assess the relationship between the Overall QoL domain of the PROMPT-QoL and its other domains. For knowngroups validity, independent t-tests and ANOVA were performed to assess the differences between PROMPT-QoL domain scores and patient characteristics. RESULTS: The average administration time was 13.4 ± 5.8 min, lower than an unacceptable time of 20 min. Only two items had missing data of 0.1–0.2 %. All PROMPT-QoL domains provided high Cronbach’s alpha values between 0.77 and 0.89, except the Availability and Accessibility (0.58), and excellent ICCs between 0.80 and 0.91. The CFA showed that the PROMPTQoL had a good fit with comparative fit index of 0.958 and root meansquare error of approximation of 0.032, while v2/df was 2.189 and adjusted goodness-of-fit index was 0.923 showing an adequate fit. The multivariate regression showed that Medication Effectiveness was the strongest predictor of the Overall QoL of PROMPT-QoL, followed by Therapeutic Relationships, Psychological Impacts, Convenience, and Availability/Accessibility. As expected, higher PROMPT-QoL domain scores were associated with younger, high-educated, lower number of medicines, patients’ perceptions of better disease control, having no adverse drug reactions, and medicine preference. CONCLUSIONS: The PROMPT-QoL was practical, reliable, and valid in Thai patients with chronic diseases.

(3021) Title: Reliability of the Aboriginal Children’s Health and Well-Being Measure (ACHWM) Nancy L. Young, PhD, Laurentian University, Sudbury, ON, Canada; Mary Jo Wabano, MHK, Wikwemikong Health Centre, Wikwemikong, ON, Canada; Koyo Usuba, MHK, Laurentian University, Sudbury, ON, Canada; Debbie Mishibinijima, BA, Wikwemikong Health Centre, Wikwemikong, ON, Canada; Tricia A. Burke, BA, Laurentian University, Sudbury, ON, Canada AIMS: This presentation reports on the internal consistency and test– retest reliability of the Aboriginal Children’s Health and Well-Being Measure (ACHWM). This research builds on the ACHWM development research presented at the ISOQOL 2013 meeting and the validity of the ACHWM presented at the 2014 meeting. METHODS: Three cohorts of children (8–18 years of age) from Wikwemikong Unceded Indian Reserve were recruited for this study. Each child completed the ACHWM independently on a computer tablet running a customized survey app. The first cohort completed the ACHWM in

Qual Life Res (2015) 24:1–191 2014 to estimate the internal consistency of the items within the summary score, using Cronbach’s alpha. The second cohort completed the ACHWM in 2015 and this new data was used to confirm the internal consistency. A third cohort completed the ACHWM later in 2015 to determine the test–retest reliability. The survey was completed on 2 separate occasions, within the same day. Completion on the same day was important to reduce the chance of intervening life events between the 2 measurement periods. The combination of an intervening time period (ranging from one to several hours) and an activity between the 2 survey completions has been used in previous research to effectively minimize recall. The reliability data were analysed using a random effects Intra-class Correlation Coefficient (ICC) to assess concordance. RESULTS: The sample characteristics for each of the 3 cohorts were: 124 participants in the 1st cohort (mean age 14.5 years; mean ACHWM = 72.5); 80 participants in the 2nd cohort (age 11.7 years; ACHWM = 74.1); and 29 participants in the 3rd cohort (age 16.8 years; ACHWM = 78.5). The internal consistency was 0.93 for the 1st sample and 0.91 for the 2nd sample. The test–retest reliability was 0.94 (95 % CI 0.86–0.97) for the ACHWM summary scores based on the 3rd sample. CONCLUSIONS: These results establish the internal consistency and the test– retest validity of the ACHWM. This is an important finding that will enable Aboriginal communities to use this measure with confidence and promote the voices of their children in reporting their health. This is a critical piece to enable evidence-based health care for Aboriginal communities.

(3023) An early look: a mixed methods study to examine the feasibility, preliminary psychometric performance, and early construct validity of the CLEFT-Q Karen Wong, MD MSc FRCSC, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; Elena Tsangaris, MSc, McMaster University, Hamilton, ON, Canada; Christopher Forrest, MD MSc FRCSC, Hospital for Sick Children, University of Toronto, Toronto, ON, Canada; Karen Harman, MD, McMaster University, Hamilton, ON, Canada; Andrea Pusic, MD MPH FACS, Memorial Sloan-Kettering Cancer Center, New York City, NY, United States; Stefan Cano, PhD, Modus Outcomes, Stotfold, United Kingdom; Timothy Goodacre, MBBS BSc FRCS, University of Oxford, Oxford, United Kingdom; Anne Klassen, DPhil, McMaster University, Hamilton, ON, Canada AIMS: The CLEFT-Q is a new PRO instrument for individuals with cleft lip and/or palate (CLP) aged between 8 and 29 years. It includes 13 independently-functioning scales that assess appearance, quality of life, and physical function. It is intended that the CLEFT-Q be used internationally, but large-scale international field tests are resource-intensive. Therefore, prior to embarking on a large multi-national study, we carried out a pilot field test to attempt to troubleshoot any potential feasibility and psychometric issues. METHODS: Individuals with CLP from two centers completed the 13 CLEFT-Q scales between November 2014 and April 2015. Time to completion was recorded. Participants rated the length of the entire CLEFT-Q. We then used Rasch Measurement Theory to analyze the responses to the 13 scales to examine the ordering of thresholds and targeting. A descriptive post hoc analysis of the Facial Appearance scores was then performed as a preliminary assessment of construct validity. RESULTS: Participants (N = 271; RR 87 %) took a mean time of 20 min to complete the CLEFT-Q. Time to completion decreased with increasing age. The majority of patients rated the length of the CLEFT-Q as ‘about right’. The Rasch analysis showed that the appearance scales had ordered thresholds. The proportion of extreme scores in the appearance scales ranged from 8 to 21 %. A total of 14 items were dropped from the scales when they were found not to fit the model and 2 items were added to try and reduce

159 ceiling effects. Individuals’ appraisal of their facial appearance decreased as age increased (p \ 0.001). Participants with visible facial difference had lower scores compared to those with no visible difference (p = 0.013). Participants requiring future treatment for their nose, lip, or jaw had lower appraisal of their facial appearance compared to those that did not require further treatment (p = 0.003, p = 0.004, p = 0.001, respectively). CONCLUSIONS: The pilot field test showed that the CLEFT-Q is feasible and the preliminary Rasch analysis led us to further revise the scales to address ceiling effects. The descriptive analysis showed that the CLEFT-Q has early construct validity. Employing mixed methods in early scale development can optimize scales prior to field testing.

(3025) Establishing content validity for the National Eye Institute’s Visual Function Questionnaire (VFQ-25) in intermediate, posterior, and panuveitis Jacob Devine, ICON Plc, San Francisco, CA, United States; Kellee Howard, MSc, MA, ICON Plc, San Francisco, CA, United States; Martha Skup, PhD, AbbVie Inc, Chicago, IL, United States; Jingdong Chao, PhD, AbbVie Inc, Chicago, IL, United States; Arijit Ganguli, MBA, PhD, AbbVie Inc, Chicago, IL, United States; John Sheppard, MD, Departments of Ophthalmology & Microbiology, Eastern Virginia Medical School and Virginia Eye Consultants, Norfolk, VA, United States AIMS: Despite adequate performance as a measure of visual-relating functioning in clinical trials, the VFQ-25 lacks evidence of content validity in uveitis patient populations. The goal of this study is to determine the content validity of the VFQ-25 and its constituent domains among non-infectious intermediate, posterior and panuveitis patients. METHODS: Open-ended, qualitative interviews were conducted to spontaneously elicit concepts relevant to uveitis and the resulting data evaluated for comparability to the VFQ-25 domain/subscale structure. Clinical experts were consulted to assess the congruence between the instrument structure and patient data. Cognitive debriefing interviews scheduled to follow will systematically review the instrument for relevance and comprehensibility to demonstrate further evidence of content validity per FDA PRO guidance. RESULTS: Seventeen patients participated in concept elicitation interviews: nine (52.9 %) with panuveitis, six (35.3 %) with posterior, and two (11.8 %) with intermediate. The mean age of patients was 44.8 years (SD = 11.6, range 21–61), and twelve (70.6 %) were female. The following domains correspond to concepts elicited by the majority of respondents: driving (n = 15, 88.2 %), general vision (n = 14, 82.4 %), role difficulties (n = 11, 64.7 %), near activities (n = 11, 64.7 %), and distance activities (n = 9, 52.9 %). The remaining domains had some degree of correspondence to patient responses (response rates ranging from 47.1 to 11.8 %), except general health which wasn’t elicited through the interview process. There was little variation in terms of concepts elicited across patient subtypes. Similarly, clinical experts reported the VFQ-25 to be comprehensive of the symptoms and impacts experienced by uveitis patients and an adequate measure of visual function, with variation dependent on severity rather than disease subtype. Subsequent cognitive debriefing will evaluate the instrument’s instructions, wording, recall period, and administration procedure to ensure the instrument is understandable and ready for use in practical application. CONCLUSIONS: The VFQ25 instrument shows promising evidence of content validity as a measure of visual-related functioning in intermediate, posterior, and panuveitis. Qualitative data from patients and clinicians suggests the instrument’s domain structure is comprehensive of the disease experience in terms of symptoms and impacts. When considered alongside existing psychometric evidence, the VFQ-25 is likely to be an adequate metric for establishing treatment benefit in uveitis.

123

160 (3027) Initial development of patient-reported outcome measures for children, adolescents and adults with primary ciliary dyskinesia: the QOL-PCD Alexandra L. Quittner, PhD, University of Miami, Miami, FL, United States; Margaret Leigh, MD, University of North Carolina - Chapel Hill, Chapel Hill, NC, United States; Tom Ferkol, MD, Washington University in St. Louis, St. Louis, MO, United States; Sharon Dell, MD, University of Toronto, Toronto, ON, Canada; Anjana M. Morris, PhD, MPH, University of Miami, Miami, FL, United States; Laura Behan, MSc, University of Southampton, Southampton, United Kingdom; Jane Lucas, BM, PhD, University of Southampton, Southampton, United Kingdom AIMS: Primary Ciliary Dyskinesia (PCD) is a rare genetic disease (*1:15,000) characterized by pulmonary infections, bronchiectasis, rhinosinusitis, infertility, and hearing impairment, caused by impaired mucociliary clearance. Our aim was to create age-appropriate, quality of life questionnaires (QOL-PCD) for adults, children and adolescents with PCD and parents of younger children. Uniquely, this process was conducted cross-culturally and harmonized using a protocol-driven approach between the UK and North America. METHODS: Following a literature review and expert panel meeting, open-ended interviews evaluated the impact of PCD on HRQoL in the UK and North America. They were conducted separately with adults (n = 21), children (n = 20), adolescents (n = 20), and parents of younger children (n = 29). Transcripts were content-analyzed using Atlas.ti/Nvivo to derive saturation matrices for each group. Concordance of symptoms between parents and children were examined. Adult patients (n = 49) rated items for relevance. The questionnaire was refined following cognitive interviews with adults (n = 15), children (n = 14), adolescents (n = 16) and parents (n = 17). RESULTS: Saturation matrices confirmed comprehensive coverage of content. The QOL-PCD for each patient/parent group covered 7 domains: Physical Functioning, Emotional Functioning, Treatment Burden, Respiratory & Sinus Symptoms, Ears & Hearing, Social Functioning, Vitality and Health Perceptions. Body Image, Eating, and Digestive Issues were not highly endorsed by participants; thus, we eliminated these scales from the questionnaires. A greater proportion of parents than children reported key symptoms: Ear infection (66 vs. 30 %); coughing up mucus (83 vs. 75 %); sinus headaches/congestion (55 vs. 50 %); stuffy nose (45 vs. 40 %); and ear pain (34 vs. 30 %). In contrast, a greater proportion of children reported trouble breathing/shortness of breath (70 vs. 52 %) and runny nose (95 vs. 83 %). Cognitive testing confirmed that remaining content was comprehensive and items were well-understood by respondents. The QOL-PCD included 48 items for adults, 37 items for children, 43 items for adolescents, and 41 items for parents. CONCLUSIONS: Content validity and cognitive testing supported the items and structure of these new PROs. Differential endorsement of symptoms by parents vs. children highlighted the need to include both respondents in HRQoL assessments for PCD. These measures have been translated into several European languages and are undergoing psychometric testing.

Qual Life Res (2015) 24:1–191 related symptoms and treatment-related side effects, none specifically address radiotherapy (RT) inconvenience. Overall, our project aims to develop a valid and reliable patient-reported measure of RT inconvenience: the RT inconvenience questionnaire (RIQ). Following the guidelines of the European Organisation for Research and Treatment of Cancer for the development of PROMS, the aims of the first two phases are to develop an RT-specific conceptual framework and populate it with patient-centred issues. METHODS: We reviewed the literature and existing PROMS, and conducted focus groups, interviews and surveys of consumers and health professionals to generate an exhaustive list of patient-centred issues pertaining to RT related inconvenience. RESULTS: Qualitative data were provided by 26 consumers and 30 health professionals, and quantitative data by 1191 consumers and 253 health professionals, generated 38 issues, which we grouped into the five conceptual domains: (1) opportunity convenience, e.g. waiting time for consultation, clinic frequency; (2) decision convenience, e.g. gaining information regarding RT, seeking information on support services; (3) treatment convenience, e.g. daily travel to RT, parking; (4) benefit convenience, e.g. side-effects of RT, management of side effects; (5) post-benefit convenience, e.g. need for follow-up investigations and appointments. CONCLUSIONS: The range of RT inconvenience issues collated here will be operationalized, in future

Opportunity convenience: time & effort in gaining entry to RT service

• gaining information regarding treatment options • seeking information on support services • assistance with logistics of treatment • choosing RT service provider Decision convenience: • availability of additional advice or recommendation(eg GP) time & effort in in making • fear and anxiety about RT treatment decision • fear and anxiety about relocating away from home

• locally available RT service • need for pre-RT appointments and interventions • requirement for tattoos • requirment for immobilization during RT • wait for RT start • long treatment duration (fractionation schedule) • daily travel to attend RT appointments • parking • requirement to be away from home Treatment convenience: • finding accommodation when relocation needed time & effort in receiving • organising family commitments (care of dependants) around RT times • Disruption to family life RT • impact on family relationships • ability to work • organising work times and commitments • impact on work productivity • availability of a support person

Benefit convenience: time & effort in gaining treatment benefit

(3029) Development of a patient reported measure to assess perceived RT inconvenience: Phase I (conceptual framework) and Phase II (issue generation) Puma Sundaresan, Radiation Oncologist, The University of Sydney, Sydney, NSW, Australia; Christopher Milross, The University of Sydney, Sydney, NSW, Australia; Martin Stockler, NHMRC Clinical Trials Centre, Sydney, NSW, Australia; Daniel Costa, Dr, The University of Sydney, Sydney, NSW, Australia, Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia AIMS: Although validated, cancer-specific patient reported outcome measures (PROMS) are widely used, and include a range of cancer-

123

• gaining referral to RT service provider • contacting & obtaining appointment with radiation oncologist • waiting time for initial consult with radiation oncologist • clinic frequency and hours of operation of clinic • locally available RT service • travel to attend initial consult with radiation oncologist • gaining RO opinion

Follow-up convenience: time & effort with ongoing contact with service

Fig. 1 .

• impact on existing health status / coping with other medical conditions • acute (early) side effects • issues related to managing acute side effects • late (long term) side effects • stigma of having RT • impact on existing health status

• travel to attend follow-up appointments • need for specific follow-up investigations and the associated travel

Qual Life Res (2015) 24:1–191 phases, into a set of items which will then undergo pre-testing and large scale field testing. The final RIQ may have useful applications in examining the impact of perceived RT inconvenience on the uptake of RT services, and the utility of interventions and modifications to service provision to make RT service delivery more patient-centred and thereby improve uptake. It may also have clinical application in understanding patient preferences during treatment decision-making and in the setting of clinical trials comparing different RT schedules (Fig. 1).

(3031) Health related quality of life in narcolepsy: validation of the DISABKIDS-narcolepsy module John Chaplin, PhD, AFBPsS, CPsychol., Inst. Clinical Sciences, Sahlgrenska Academy at the University Gothenburg, Sweden, Gothenburg, Sweden; Attila Szaka´cs, MD, Institute of Clinical Sciences, Sahlgrenska Academy, Halmstad County Hospital, Halmstad, Sweden; Niklas Darin, MD, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden; Tove Hallbo¨o¨k, MD, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg. 3Lund Child and Youth Habilitation Center, Lund, Sweden AIMS: To construct a valid and reliable narcolepsy-specific quality of life (QOL) DISABKIDS module for young people (DCSM-Narc). Narcolepsy is a lifelong, sleep disorder with a prevalence of 0.06 % with onset at around 14 years. Narcolepsy is associated with psychiatric comorbidities and cognitive difficulties. Symptoms are often more severe in children than in adults. METHODS: An extended DISABKIDS approach to questionnaire development was followed. Focus groups were conducted and a patient panel identified the statements most salient to QoL. A pilot questionnaire was sent to young people via a patient organisation formed following the Pandemrix influenza vaccination in Europe. Psychometric properties were tested and a series of exploratory factor analyses determined domain structures and a reduced item set. RESULTS: Twenty young people (ages 8–17 years, mean 13.5 years) and their parents participated in focus groups. A set 125 items were produced and reviewed by the patient panel resulting in a subset of 40 items with the greatest relevance to QoL. A pilot questionnaire was sent to member of a patient organisation. 41 questionnaires were returned (ages 8–20 years, mean 14.4; 22 girls). Following psychometric evaluation one item was rejected due to asymmetry (skewness above 2.0). Two cases were rejected due to missing data. A correlation analysis revealed three distinct non-overlapping concepts in the item list: (1) QoL; (2) future expectations, and (3) symptoms. A factor analysis found eleven QoL items with loadings of 0.6 or greater, in three domains (emotional reaction, being alert, being social). Test–retest reliability was 0.785 (mean elapsed time 15 days). There was good item intercorrelation (Cronbach’s alpha: 0.788) and convergent validity with Kidscreen (coefficient 0.736). Results were confirmed in the retest sample. Additionally a two-domain ‘future module’ (8 items/Cronbach’s alpha 0.806); and a two-domain ‘symptoms module’ (7 items/Cronbach’s alpha 0.664) were derived. CONCLUSIONS: Developing questionnaires for populations of people with rare diseases poses challenges for the developers. The relationship between the concept of ‘future expectations’ and QoL needs to be explored further. Further testing is required but the 11-item DCSM-narc appears to have an excellent domain structure and promises to be useful for future studies of narcolepsy.

(3033) Integrating the child’s voice in adverse event reporting in oncology trials: cognitive interview findings from the multisite pediatric PRO-CTCAE initiative Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Molly McFatrich, University of North

161 Carolina at Chapel Hill, Chapel Hill, NC, United States; Justin N. Baker, St. Jude Children’s Research Hospital, Memphis, TN, United States; David R. Freyer, Children’s Hospital Los Angeles, Los Angeles, CA, United States; Joshua Gagne, Dana-Farber Cancer Institute/Boston Children’s Hospital, Boston, MA, United States; Jami S. Gattuso, St. Jude Children’s Research Hospital, Memphis, TN, United States; Deborah Gibson, St. Jude Children’s Research Hospital, Memphis, TN, United States; Racquel E. Kohler, Univeristy of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Alyssa Levine, Dana-Farber Cancer Institute/Boston Children’s Hospital, Boston, MA, United States; Selena Lollar, Palmetto Health Children’s Hospital, Columbia, SC, United States; Jane Lukshis, Palmetto Health Children’s Hospital, Columbia, SC, United States; Jennifer W. Mack, Dana-Farber Cancer Institute/Boston Children’s Hospital, Boston, MA, United States; Christa Martens, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Catriona Mowbray, Children’s National Health System, Washington, DC, United States; Diana Palma, Children’s Hospital Los Angeles, Los Angeles, CA, United States; Laura C. Pinheiro, MPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Lillian Sung, The Hospital for Sick Children, Toronto, ON, Canada; Sarah C. Taddei, Dana-Farber Cancer Institute/Boston Children’s Hospital, Boston, MA, United States; Deborah Tomlinson, The Hospital for Sick Children, Toronto, ON, Canada; Mia Waldron, Children’s National Health System, Washington, DC, United States; Meaghann Weaver, Children’s National Health System/St. Jude Children’s Research Hospital, Memphis, TN, United States; Jennifer Wind, Dana-Farber Cancer Institute/Boston Children’s Hospital, Boston, MA, United States; Janice Withycombe, Palmetto Health Children’s Hospital, Columbia, SC, United States; Pamela S. Hinds, Children’s National Health System, Washington, DC, United States AIMS: Over 60 % of children with cancer in the U.S. participate in a clinical trial. Collection of adverse events (AEs) is legally mandated and the current standard for AE collection is clinician report, even though many AEs are subjective (e.g., fatigue). Our overarching goal is to design a child self-report measure of subjective AEs to inform AE reporting for the National Cancer Institute’s Common Terminology Criteria for Adverse Events (CTCAE), the standard lexicon for grading AEs in oncology trials. This cognitive interview study evaluated children’s understanding and ability to provide valid responses to the PRO-CTCAE measures to inform questionnaire refinements. METHODS: The newly-developed pediatric PROCTCAE (and parent-proxy version) was evaluated separately in 7–8 and 9–12 year olds, and the previously-developed adult PROCTCAE was evaluated in 13–15 and 16–20 year olds. Because of the length of the PRO-CTCAE item library (130 items measuring 63 AEs), multiple forms were created to reduce participant burden. Across seven pediatric hospitals, trained interviewers conducted cognitive interviews with the child (and parent separately) after the child/parent completed the assigned PRO-CTCAE forms independently. RESULTS: ninety-six children (59 % female) with cancer (and their parents) participated in Round 1. The adult PRO-CTCAE performed well in 16–20 year olds; however comprehension of some question stems (e.g., using the word ‘‘severity’’) and certain AE terms (e.g., anxiety) was difficult in 13–15 year olds, resulting in our recommendation to use the pediatric PRO-CTCAE in that age group. With younger age groups, problems with items from the pediatric PRO-CTCAE increased, including describing ‘‘past 7 days’’. Most seven year olds required the interviewer to read the items aloud. The AE severity stem and 14 AE terms were revised for further evaluation in Round 2, which will involve interviewing 36 children. CONCLUSIONS: Overall, most participants had little to no problem completing PRO-CTCAE on their own. Round 2 of interviews is expected to be completed by the ISOQOL conference and results will be shared. The pediatric PRO-CTCAE will next be

123

162 longitudinally validated. The availability of a valid and reliable system for children to self-report symptom toxicities will enhance AE reporting and improve care for children with cancer.

(3035) Design of the HIV dependent quality of life (HIVDQoL) questionnaire and HIV Symptom Rating Questionnaire (HIVSRQ) Jacquelyn Romaine, Health Psychology Research Unit, Royal Holloway, University of London, Egham, Surrey, United Kingdom; Janet Bayfield, Health Psychology Research Unit, Royal Holloway, University of London, Egham, Surrey, United Kingdom; Rosalind Plowright, Health Psychology Research Ltd, Royal Holloway, University of London; Health Psychology Research Unit, Royal Holloway, University of London, Egham, Surrey, United Kingdom; Miranda Murray, ViiV Healthcare, London, United Kingdom; Clare Bradley, Health Psychology Research Unit, Royal Holloway, University of London; Health Psychology Research Ltd, Royal Holloway, University of London, Egham, Surrey, United Kingdom AIMS: An HIV-specific QoL measure is needed in addition to widely used health-status tools to evaluate treatments for people living with HIV. An up-to-date symptom measure is also required, given recent innovations in antiretroviral treatments. The present research aimed to design two instruments, each in UK and US English: the HIV-Dependent Quality of Life (HIVDQoL) questionnaire and HIV Symptom Rating Questionnaire (HIVSRQ). METHODS: The format of the HIVDQoL and HIVSRQ are based on established -DQoL and -SRQ measures (e.g. ADDQoL for diabetes; ThySRQ for hypothyroidism, with copyright owned by co-author, Bradley). Items were selected from existing—DQoL and—SRQ item banks and newly drafted as needed, based on a review of relevant research, HIV specialist websites, and transcript analysis of 68 interviews conducted in UK, Germany and Spain (Murray et al. 2013). The initial HIVSRQ item list was revised following consultation with three HIV specialist clinicians. Participants, recruited via the internet by Opinion Health, were paid for their time. In-depth telephone interviews lasted approximately 1 h, with questionnaires revised as needed between interviews, until no new content/changes were required. Following 10 UK interviews, a US linguist adapted the English for the US version. US interviews (11) followed, with four additional UK interviews. RESULTS: Participants (n = 25) included 14 from the UK (including 3 women, mean age = 48 [SD = 9.89], HIV mean duration = 16 years [SD = 7.43]), 11 from the USA (including 3 women, mean age = 53 [SD = 11.87], HIV mean duration = 19 years [SD = 9.71]). The HIVDQoL has 26 items including 10 existing, 12 modified and 5 new items. The HIVSRQ has 64 items including 31 existing, 24 modified and 9 new Items. CONCLUSIONS: Internet recruitment and telephone interviews worked well, eliciting information from a range of participants. Designing two language versions simultaneously avoided subsequent linguistic validation, reduced unnecessary UK/US wording differences, and helped simplify the wording. Existing templates and item banks from other condition-specific -DQoLs and -SRQs facilitated the design process, thus enabling participants and researchers to focus on the content. Comparisons across questionnaires for different conditions have now become possible. Psychometric development of the HIVDQoL and HIVSRQ will follow shortly.

(3037) Item generation for the Long-Term Conditions Questionnaire (LTCQ): qualitative interviews with patients Caroline Potter, University of Oxford, Oxford, United Kingdom; Cheryl Hunter, University of Leeds, Leeds, United Kingdom; Laura

123

Qual Life Res (2015) 24:1–191 Kelly, University of Oxford, Oxford, United Kingdom; Elizabeth Gibbons, University of Oxford, Oxford, United Kingdom; Crispin Jenkinson, University of Oxford, Oxford, United Kingdom; Angela Coulter, University of Oxford, Oxford, United Kingdom; Julien Forder, University of Kent and London School of Economics and Political Science, Canterbury, United Kingdom; Ann-Marie Towers, University of Kent, Canterbury, United Kingdom; Ray Fitzpatrick, University of Oxford, Oxford, United Kingdom; Michele Peters, PhD, University of Oxford, Oxford, United Kingdom AIMS: Building on a conceptual framework developed through previous literature reviews and professional stakeholder interviews, this research sought to elucidate the concerns of people living with at least one long-term condition (LTC). Domains identified through this research will inform item content of the Long-Term Conditions Questionnaire (LTCQ), a patient-reported outcome measure (PRO) designed to work across mental and physical health and multi-morbidity. METHODS: Forty-two participants with LTCs were recruited through primary care practices in London and Oxfordshire, UK. Patients were identified on the basis of having one of ten selected conditions (chronic obstructive pulmonary disease, diabetes, cancer, ischaemic heart disease, depression, schizophrenia, inflammatory bowel disease, stroke, multiple sclerosis, osteoarthritis). Semi-structured interviews averaging 60 min were conducted, audio-recorded, and transcribed. Interviews were analysed with NVivo using an initial framework derived from the prior research; the framework and interview guide were then revised concurrently with data collection to reflect emerging themes. To supplement the 42 primary interviews this analytic framework was applied to six transcripts from previous research with schizophrenia patients, who proved difficult to recruit via primary care. RESULTS: More than half the sample reported multiple LTCs. Several broad themes emerged consistently across all participants: the effects of feeling supported (or not) by health and other services, the variable impact of LTCs on day-to-day living, the importance of social support and participating in social life, the impact of LTCs on mental health, and the importance of self-management strategies for coping with longterm illness. Burden of treatment (medications/therapies themselves and the practical aspects of engaging with services) emerged more prominently among those with multiple LTCs. The data thus reflected both established concerns of PROs (e.g. daily functioning, emotional well-being) and less traditional domains such as personal control, treatment burden or self-management. Together with the findings from the literature reviews and professional stakeholder interviews, these findings informed the generation of 23 items for the LTCQ. CONCLUSIONS: The broad agreement of participants around core aspects of life with LTCs was used to generate items that are relevant for a range of conditions and multi-morbidity. Future steps will include further testing of the 23 items for comprehensiveness and acceptability.

(3039) Validity and responsiveness of Short Version of Western Ontario Rotator cuff Index (Short-WORC) in patients with rotator cuff pathology Neha Dewan, PhD student, McMaster University, Hamilton, ON, Canada; Joy MacDermid, Professor, McMaster University, Hamilton, ON, Canada; Norma MacIntyre, Associate professor, McMaster University, Hamilton, ON, Canada; Ruby Grewal, Associate Professor, Western University, Roth | McFarlane Clinical Research Laboratory, Hand and Upper Limb Centre, London, ON, Canada AIMS: To estimate the validity and responsiveness of the shorter version of Western Ontario Rotator Cuff Index (Short-WORC) in comparison to the full length WORC and other joint specific (Shoulder Pain and Disability Index; SPADI, Simple Shoulder Test; SST), region specific (Disabilities of the Arm, Shoulder and Hand; DASH) or quality of life (12 item short form health survey; SF-12v2)

Qual Life Res (2015) 24:1–191

163

Table 1 Baseline characteristics of the study population (n = 223)

Table 1 continued

Variable

N/percentage of participants

Variable

Age in years (mean ± SD)

223/100 % (56.70 ± 11.08)

Occasionally

103/49 %

1–6 drinks/week

32/15 %

Sex

223/100 %

7–14 drinks/week

23/11 %

Male Female

151/68 % 72/32 %

15+ drinks/week

12/5 %

Affected Shoulder

222/99 %

Left

93/42 %

Right

128/58 %

Both

1/0.4 %

Hand dominance

220/99 %

Left

30/13 %

0–3 month

Right

190/85 %

N

Medication used before surgery

209/94 %

N/percentage of participants

SD standard deviation, N number of participants, % percentage

Table 2 Percentage of missing data Measure

Time point 0–6 month %

N

%

Yes

104/49 %

ShortWORC

52

23

55

No

105/50 %

WORC

41

18

43

19

Injections used before surgery

212/95 %

SPADI

148

66

145

65

Yes

99/47 %

SST

27

12

37

16

No

113/53 %

DASH

47

21

42

18

Employment status at enrolment Full time regular duties

213/95 % 62/29 %

SF12-PCS

57

25

53

23

SF12-MCS

53

24

53

23

Part-time regular duties

10/5 %

Full-time light duties

19/9 %

Part-time light duties

8/4 %

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SPADI Shoulder Pain and Disability Index, SST simple shoulder test, DASH disabilities of the arm, shoulder and hand, SF12-PCS Physical component scale of Short Form 12 item (version 2) Health Survey, SF12-MCS Mental component scale of Short Form 12 item (version 2) Health Survey, N number of participants, % percentage

Unable to work because of injury

47/22 %

Unable to work for other medical reasons

9/4 %

Home maker

2/1 %

Retired

56/26 %

Worker compensation involved

211/95 %

No

136/64 %

Yes

68/32 %

Pending

7/3 %

Highest education level

208/93 %

Some grade school

6/3 %

Finished grade school

20/10 %

Some high school Finished high school

45/22 % 36/17 %

Some college/technical/diploma program

33/16 %

Finished college/technical/diploma program

30/14 %

Some university

18/9 %

Finished University

11/5 %

Some graduate work at university

4/2 %

Finished graduate work at university

5/2 %

Smoker

207/93 %

No

82/39 %

Yes

45/22 %

I quit

80/39 %

Alcohol

210/94 %

Never

40/19 %

24

measures in the patients with rotator cuff pathology (RCP). METHODS: A cohort of patients (n = 223, men 151; women 72, mean age 56.70 ± 11.08) who completed the Short-WORC, WORC, SPADI, SST, DASH and SF12 at 1–2 weeks prior to surgery, were re-tested at 3 and 6 months after the rotator cuff repair. The Short-WORC scores were extracted from the full length of WORC questionnaire. Known group validity and construct validity of Short-WORC and WORC were evaluated using independent-t test and Pearson correlation coefficient (r) respectively. Responsiveness was evaluated using paired-t test/relative efficiency (RE), effect size (ES) and standardized response mean (SRM). RESULTS: Short-WORC scores were significantly different (p \ 0.05) among patients with respect to their worker compensation and gender when assessed at baseline and 6 months of follow up. The Short-WORC demonstrated high correlation (r = 0.89–0.96) with the WORC and moderate to strong correlation (r = 0.51–0.92) with other measures assessing related constructs at both preoperative and postoperative assessments. The disease specific measures (Short-WORC and WORC) were found to be highly responsive (0–3 month: ESShort-WORC = 0.72, ESWORC = 0.92, SRMShortWORC = 0.75, SRMWORC = 0.81; 0–6 month: ESShort-WORC = 1.05, ESWORC = 1.12, SRMShort-WORC = 0.89, SRMWORC = 0.89) than other upper extremity measures (0–3 month: ES = 0.07–0.55, SRM = 0.09–0.49; 0–6 month: ES = 0.05–0.78, SRM = 0.07–0.78) at both 3 and 6 months postoperatively. Overall, both the Short-WORC and the WORC were equally responsive (RE, Short-WORC/WORC = 1) to

123

164

Qual Life Res (2015) 24:1–191

Table 3 Floor and ceiling effects for the ShortWORC and WORC Time point Baseline Month 6

Measure

Mean (SD)

Median (IQR)

%Floor effects

%Ceiling effects

ShortWORC (n = 207)

30.2 (20.3)

28.0 (29.3)

39/207 = 18.8

1/207 = 0.004

WORC (n = 213)

32.3 (18.0)

30.7 (25.5)

25/207 = 12.0

0/207 = 0

ShortWORC (n = 183)

56.8 (28.5)

60.4 (48.9)

11/183 = 6.0

26/183 = 14.2

WORC (n = 190)

56.8 (27.9)

59.9 (46.9)

10/190 = 5.2

26/190 = 13.7

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SD standard deviation, IQR inter quartile range, % percentage

Table 4 Cross-sectional known group validity: with and without worker compensation (WC) Time point

Measure

With worker compensation

Without worker compensation

Baseline (with WC, n = 66; without WC, n = 129)

ShortWORC mean (SD)

24.6 (18.5)

32.97 (21.1)

8.4

WORC mean (SD)

25.5 (16.2)

35.49 (18.7)

9.9

.00**

ShortWORC mean (SD)

40.4 (25.3)

66.43 (26.4)

26.0

.00**

WORC mean (SD)

40.7 (25.1)

66.10 (25.2)

25.3

.00**

Month 6 (with WC, n = 60; without WC, n = 109)

Mean difference

P value (2-tailed) .00**

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SD standard deviation, WC worker compensation, p value level of significance ** Statistically significant difference at 0.001 level

Table 5 Longitudinal known group validity: with and without worker compensation Time point

Measure

With worker compensation

Without worker compensation

Mean difference

P value (2-tailed)

0–3 Month (with WC, N = 54; without WC, N = 107)

ShortWORC mean (SD)

0–6 Month (with WC, N = 58; without WC, N = 103)

WORC mean (SD) ShortWORC mean (SD)

-6.7 (15.4)

-18.2 (23.6)

-11.4

0.00**

-11.3 (14.9) -15.4 (22.6)

-19.1 (24.2) -33.9 (26.6)

-7.8 -18.5

0.01* 0.00**

WORC mean (SD)

-14.9 (21.4)

-30.7 (25.6)

-15.8

0.00**

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SD standard deviation, WC worker compensation, p value level of significance, N number of participants * Statistically significant difference at 0.05 level ** Statistically significant difference at 0.001 level

Table 6 Cross sectional known group validity: females and males Time point Baseline (males = 138; females = 69) Month 6 (males = 120; females = 62)

Measure

Female

Male

Mean difference

P value (2-tailed)

ShortWORC mean (SD)

19.26 (15.4)

35.65 (20.3)

16.4

.00**

WORC mean (SD)

24.01 (15.6)

36.41 (18.1)

12.4

.00**

ShortWORC mean (SD)

49.05 (32.8)

60.71 (25.4)

11.6

.01*

WORC mean (SD)

52.13 (31.6)

59.12 (25.3)

6.9

.13

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SD standard deviation, p value level of significance * Statistically significant difference at 0.05 level ** Statistically significant difference at 0.001 level

123

Qual Life Res (2015) 24:1–191

165

Table 7 Longitudinal known group validity: females and males Time point

Measure

Female

Male

0–3 Month (males = 113; females = 58)

ShortWORC mean (SD)

-20.7 (20.5)

-11.9 (22.2)

8.8

WORC mean (SD)

-23.8 (22.3)

-13.2 (21.0)

10.5

.00*

ShortWORC mean (SD)

-30.7 (28.1)

-25.2 (25.7)

5.5

.20

WORC mean (SD)

-28.1 (27.7)

-23.0 (23.7)

5.1

.21

0–6 Month (males = 108; females = 59)

Mean difference

P value (2-tailed) .00*

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SD standard deviation, p value level of significance * Statistically significant difference at 0.05 level ** Statistically significant difference at 0.001 level

Table 8

Cross-sectional convergent construct validity: Pearson r ± 95 % CI

Time point

Measure

Short WORC

WORC

SPADI

SST

DASH

SF12-PCS

SF12-MCS

Baseline (N = 88)

Short WORC

1

.92** (0.88, 0.94)

-.63** (-0.73, -0.51)

.69** (0.58, 0.77)

-.77** (-0.83, -0.68)

.51** (0.36, 0.63)

.41** (0.25, 0.55)

WORC

.92** (0.88, 0.94)

1

-.63** (-0.73, -0.51)

.68** (0.57, 0.76)

-.82** (-0.87, -0.75)

.53** (0.39, 0.64)

.48** (0.33, 0.60)

Short WORC

1

.89** (0.84, 0.92)

-.82** (-0.87, -0.75)

.58** (0.44, 0.68)

-.73** (-0.80, -0.63)

.61** (0.48, 0.71)

.38** (0.21, 0.53)

WORC

.89** (0.84, 0.92)

1

-.80** (-0.85, -0.72)

.53** (0.38, 0.67)

-.69** (-0.77, -0.58)

.60** (0.47, 0.70)

.35** (0.18, 0.50)

Short WORC

1

.96** (0.94, 0.97)

-.92** (-0.94, -0.89)

.87** (0.82, 0.90)

-.86** (-0.89, -0.81)

.77** (0.69, 0.83)

.52** (0.38, 0.63)

WORC

.96** (0.94, 0.97)

1

-.89** (-0.92, -0.85)

.84** (0.78, 0.88)

-.84** (-0.88, -0.78)

.76** (0.68, 0.82)

.58** (0.45, 0.68)

Month 3 (N = 83)

Month 6 (N = 99)

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SPADI Shoulder Pain and Disability Index, SST simple shoulder test, DASH disabilities of the arm, shoulder and hand, SF12-PCS Physical component scale of Short Form 12 item (version 2) Health Survey, SF12-MCS Mental component scale of Short Form 12 item (version 2) Health Survey, N number of participants, Pearson r correlation coefficient, CI confidence interval ** Statistically significant correlation at 0.001 level

Table 9 Longitudinal convergent construct validity (external responsiveness statistics): Pearson r ± 95 % CI Time point

Measure

Short WORC

WORC

SPADI

SST

DASH

SF12-PCS

SF12-MCS

0–3 Month (N = 53)

Short WORC

1

.86** (0.79, 0.91)

-.62** (-0.74, -0.46)

.58** (0.40, 0.71)

-.70** (-0.80, 0.56)

.53** (0.34, 0.67)

.23 (-0.00, 0.43)

WORC

.86** (0.79, 0.91)

1

-.51** (-0.66, -0.31)

.53** (0.34, 0.67)

-.60** (-0.73, 0.43)

.55** (0.37, 0.69)

.32* (0.09, 0.51)

Short WORC

1

.92** (0.88, 0.95)

-.69** (-0.79, -0.56)

.77** (0.67, 0.84)

-.64** (-0.79, 0.49)

.63** (0.48, 0.74)

.29* (0.08, 0.47)

WORC

.92** (0.88, 0.95)

1

-.67** (-0.77, -0.53)

.75** (0.63, 0.83)

-.63** (-0.74, 0.48)

.56** (0.39, 0.69)

.39** (0.19, 0.56)

0–6 Month (N = 61)

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SPADI, Shoulder Pain and Disability Index, SST simple shoulder test, DASH disabilities of the arm, shoulder and hand; SF12-PCS Physical component scale of Short Form 12 item (version 2) Health Survey, SF12-MCS Mental component scale of Short Form 12 item (version 2) Health Survey, N number of participants, Pearson r correlation coefficient, CI confidence interval * Statistically significant difference at 0.05 level ** Statistically significant correlation at 0.001 level

evaluate change at 6 months of follow up. CONCLUSIONS: Both the Short-WORC and WORC appear to be valid and responsive disease specific measures to evaluate HRQoL in patients with RCP. However, the brevity of Short-WORC makes it more attractive substitute to

WORC for clinicians, researchers as well as patients. Future researchers should consider evaluating minimal clinically important difference of Short-WORC among patients with RCP (Tables 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11).

123

166

Qual Life Res (2015) 24:1–191

Table 10 Internal responsiveness indices: mean change, SRM (bootstrapped), ES Time point 0–3 Month (N = 53)

0–6 Month (N = 61)

Measure

Mean change (D) (95 % CI)

t0

SD D

SDbaseline

SRM (D/SDD) (95 %CI)

ES(D/SDbaseline)

Short WORC

5.52

16.94** (10.79, 23.10)

22.54

23.40

0.75 (0.42, 1.20)

0.72

WORC

5.93

19.36** (12.82, 25.91)

23.97

20.90

0.81 (0.48, 1.22)

0.92

SPADI

3.54

13.25** (5.75, 20.75)

27.48

26.25

0.48 (0.21, 0.81)

0.50

SST

3.41

10.96** (4.51, 17.40)

23.61

24.49

0.46 (0.17, 0.84)

0.45

DASH

3.49

10.44** (4.62, 16.25)

21.31

18.75

0.49 (0.20, 0.79)

0.55

SF 12-PCS

1.42

1.66 (-0.68,4.01)

8.59

9.21

0.19 (0.08, 0.50)

0.18

SF 12-MCS

0.69

0.97 (-1.84, 3.78)

10.30

14.05

0.09 (0.18, 0.33)

0.07

Short WORC

6.96

24.40** (17.39, 31.42)

27.38

23.03

0.89 (0.67, 1.17)

1.05

WORC

6.95

22.97** (16.36, 29.59)

25.82

20.37

0.89 (0.66, 1.17)

1.12

SPADI

6.07

20.69** (13.88, 27.51)

26.59

26.26

0.77 (0.54, 1.05)

0.78

SST DASH

6.11 4.58

21.99** (14.79, 29.19) 11.98** (6.75, 17.20)

28.09 20.40

28.22 21.18

0.78 (0.54, 1.08) 0.58 (0.37, 0.82)

0.78 0.56

SF 12-PCS

5.26

4.91** (3.04, 6.78)

7.28

8.19

0.67 (0.48, 0.91)

0.59

SF 12-MCS

0.54

.66 (-1.8, 3.14)

9.68

12.87

0.07 (0.19, 0.32)

0.05

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SPADI Shoulder Pain and Disability Index, SST simple shoulder test, DASH disabilities of the arm, shoulder and hand, SF12-PCS Physical component scale of Short Form 12 item (version 2) Health Survey, SF12-MCS Mental component scale of Short Form 12 item (version 2) Health Survey, N number of participants, SRM standardised response mean, ES effect size, CI confidence interval, t0 the observed value of t on paired t test ** Statistically significant mean change at 0.001 level

Table 11 Relative efficacy (t1/t2)2 of short-WORC versus WORC to detect change over time REShortWORC/

REShortWORC/

REShortWORC/

REShortWORC/

REShortWORC/SF12-

REShort

WORC

SPADI

SST

DASH

PCS

MCS

0–3 Month (N = 53) 0.9

1.6

1.6

1.5

3.9

8.0

0–6 Month (N = 61) 1.00

1.1

1.1

1.5

1.3

12.9

Time point

WORC/SF12-

ShortWORC Short version of Western Ontario Rotator Cuff Index, WORC Western Ontario Rotator Cuff Index, SPADI Shoulder Pain and Disability Index, SST simple shoulder test, DASH disabilities of the arm, shoulder and hand, SF12-PCS Physical component scale of Short Form 12 item (version 2) Health Survey, SF12-MCS Mental component scale of Short Form 12 item (version 2) Health Survey, RE relative efficiency, N number of participants, t paired t test statistics

(3041) Using a personalized measure to capture QOL in people with HIV Nancy Mayo, PhD, McGill University, Montreal, Canada; Carolina Moriello, McGill University Health Centre - RI, Montreal, QC, Canada; Lesley K. Fellows, Dr., McGill University, Montreal, QC, Canada; Marie-Josee Brouillette, Dr., McGill University, Montreal, QC, Canada; Susan C. Scott, McGill University Heath center, Montreal, QC, Canada AIMS: The impact of HIV on quality of life (QOL) has been measured using generic and HIV specific indices and profiles. Only one study has used a personalized measure, Patient Generated Index (PGI), in HIV but in a predominately female sample from Thailand. The purpose of our study is to compare the value on health-related QOL obtained from the PGI with values obtained from the EQ-5D and the SF-6D, in a population of predominantly men HIV+. METHODS: The data arose from a Canadian, multi-site, longitudinal study of persons HIV+, Brain Health Now cohort, targeting people C40 years. Generalized estimating equations were used to estimate differences and confidence bounds between the measures (all scaled to be out of 100). RESULTS: 275 people contributed

123

data at visit 1; mean age 52, 80 % men. EQindex averaged 81 (SD:16), range 2.5–100; mean SF-6D was 68 (SD: 12), range 38–100. PGI mean value was 50 (SD: 2), range 23–100. Correlation between EQindex and SF-6D was 0.60 but only 0.19 for PGI vs. EQindex and 0.12 for PGI vs. SF-6D. The difference between measures ranged from 13 between the EQindex and the SF-6D (95 % CI 11–14); to -44 (95 % CI -46–41) between the PGI and EQindex and -32 (95 % CI -35–29). EQindex was higher than SF6D likely because EQ-5D includes walking but not fatigue but the opposite is true for SF-6D. Only 13 % of the sample identified any difficulty with walking but vitality was lower for the HIV sample than age matched norms (mean 52 vs. 68). The most common domains identified from PGI where HIV impacted on QOL were: work, intimacy, relationships, mood, worry, health, stigma, appearance, regret, and fatigue. One of the reasons for discrepant QOL ratings is that, of the important domains from PGI, only mood and fatigue are represented in the generic measures. CONCLUSIONS: The PGI is easy to use, short, and provides a single value for QOL. It likely more accurately reflects the impact of HIV on QOL than ‘‘one size fits all’’ measures with fixed domains as it captures individual heterogeneity.

Qual Life Res (2015) 24:1–191 (3043) Preliminary measurement properties of the Pain Assessment for Lower Back-Impacts Questionnaire (PAL-I): development using a mixed-methods approach Donald M. Bushnell, MA, Health Research Associates, Inc., Seattle, WA, United States; Kelly P. McCarrier, PhD, Health Research Associates, Inc, Seattle, WA, United States; Abhilasha Ramasamy, Forest Research Institute, Jersey City, NJ, United States; Hiltrud Liedgens, Grunenthal GmbH, Aachen, Germany; Lauren A. Braam, MIH, Health Research Associates, Inc., Seattle, WA, United States; Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States; Steven I. Blum, GlaxoSmithKline (with Forest Research Institute at time of study), Collegeville, PA, United States AIMS: The Pain Assessment for Lower Back-Impacts (PAL-I) is a patient-reported outcome (PRO) instrument developed to assess key impacts, e.g. walking, sitting, standing, etc. associated with chronic low back pain (cLBP). Following a mixed methods approach, the PAL-I development included qualitative work (both concept elicitation and cognitive interviews), an initial wave of quantitative data to evaluate items, additional qualitative interviews, and now a second wave of quantitative data to confirm item and scale performance and to evaluate preliminary measurement properties. METHODS: Two US-based samples were recruited to test the revised questionnaire: (1) commercial panel: 401 qualifying adults self-reporting a clinical diagnosis of cLBP completed a web-based survey consisting of the 9-item PAL-I analyzed to confirm item- and scale-level performance of the PAL-I using Rasch Measurement Theory analyses; and (2) clinic sample: 45 adults identified by patient records completed the PAL-I and painDETECT by web, and the MOS-36 and RolandMorris Disability Questionnaire (RMDQ) on paper. The clinic sample was used to assess test–retest reliability at 1 week, convergent and known-groups validity. RESULTS: The 401 subjects in the webbased sample reported having cLBP (between 1 and 10 on 11-point numerical rating scale [NRS], 0 = no pain), were 55.3 years old (SD 12.6), 67.8 % female, 87.8 % white and 65.6 % married. The Rasch item threshold map showed all items having an ordered threshold. Category probability curves indicated subjects had no difficulty endorsing items. The 45 subjects in the clinic sample reported having cLBP (NRS scores between 4 and 10), were 53.0 years old (SD 12.4), 52.3 % female, 79.5 % white, and 50.0 % married. Test–retest was acceptable (ICC = 0.88). As predicted, strong associations were seen between PAL-I and RMDQ (r = 0.82, p \ 0.001), MOS-36 Physical Functioning (r = -0.71, p \ 0.001), and MOS Bodily Pain (r = -0.71, p \ 0.001). PAL-I scores were able to discriminate by painDETECT (F = 14.5, p \ 0.001). CONCLUSIONS: The mixedmethods approach provides valuable support in the development of a fit-for-purpose instrument assessing impacts of cLBP. Along with prior qualitative and quantitative work, the results from this second wave provide confirmation of the content validity and early acceptable measurement properties of a newly developed PRO instrument.

(3045) Preliminary measurement properties of the Pain Assessment for Lower Back-Symptoms (PAL-S): development and evaluation through a mixed-methods approach Kelly P. McCarrier, PhD, Health Research Associates, Inc, Seattle, WA, United States; Donald M. Bushnell, MA, Health Research Associates, Inc., Seattle, WA, United States; Abhilasha Ramasamy, Forest Research Institute, Jersey City, NJ, United States; Hiltrud Liedgens, Grunenthal GmbH, Aachen, Germany; Lauren A. Braam, MIH, Health Research Associates, Inc., Seattle, WA, United States; Mona L. Martin, RN, MPA, Health Research Associates, Inc., Seattle, WA, United States; Steven I. Blum, GlaxoSmithKline (with Forest Research Institute at time of study), Collegeville, PA, United States

167 AIMS: The Pain Assessment for Lower Back Symptoms (PAL-S) is a Patient-Reported Outcome (PRO) instrument designed to assess the key symptoms of chronic low back pain (cLBP). The instrument’s initial development included qualitative concept elicitation and cognitive interviews. Further evaluation and refinement of the initial draft has used a mixed-methods approach involving an initial wave of quantitative data to evaluate items, additional cognitive interviews, and the current (second) wave of quantitative testing to confirm item and scale performance and evaluate preliminary measurement properties. METHODS: The PAL-S was evaluated in two US-based samples: (1) a web-based commercial panel of 401 adults with cLBP completed an online version of the PAL-S to confirm item- and scalelevel performance using Rasch Measurement Theory analyses; and known-groups validity with the painDETECT. (2) a clinic-based sample of 45 adult cLBP patients completed PAL-S and painDETECT assessments via web and the MOS-36, Roland-Morris Disability Questionnaire (RMDQ), and Neuropathic Pain Scale Inventory (NPSI) on paper to assess the PRO’s one-week test–retest reliability and convergent and known-groups validity. RESULTS: All 401 subjects in the web-based sample reported having cLBP of 1 or greater on an 11-point numerical rating scale (NRS; 0 = no pain), were 55.3 years old (SD = 12.6), 67.8 % female, 87.8 % white and 65.6 % married. The Rasch item threshold map showed all items having an ordered threshold. One item (‘‘aching pain’’) displayed evidence of a ceiling effect with only 2 % of the population indicating a response of ‘‘no aching at all’’. The 45 subjects in the clinic sample reported pain between 4 and 10 on the NRS, were 53.0 years old (SD 12.4), 52.3 % female, 79.5 % white, and 50.0 % married. Test–retest was acceptable (ICC = 0.81). As predicted, strong associations were seen between PAL-S and RMDQ (r = 0.67, p \ 0.001), NPSI (r = 0.73, p \ 0.001), and MOS-36 Bodily Pain (r = -0.79, p \ 0.001) scales. PAL-S scores significantly discriminated among painDETECT groups in both samples (p \ 0.001). CONCLUSIONS: The mixed-methods approach provides valuable support in the development of a fit-for-purpose instrument assessing cLBP symptoms. The results from this second wave provide confirmation of the content validity and early acceptable measurement properties of a newly developed PRO instrument.

(3047) Measurement properties of a new claudication symptom instrument (CSI) Todd Edwards, PhD, University of Washington, Seattle, WA, United States; Danielle C. Lavallee, PharmD, PhD, University of Washington, Seattle, WA, United States; Alexander Clowes, MD, University of Washington, Seattle, WA, United States; Beth Devine, PhD, PharmD, University of Washington, Seattle, WA, United States; Ellen Thomason, MD, MPH, Providence Health Services, Seattle, WA, United States; David R. Flum, MD, MPH, University of Washington, Seattle, WA, United States; Mark Meissner, MD, University of Washington, Seattle, WA, United States; Skye P. Barbic, Post-Doctoral Fellow, Faculty of Medicine, Vancouver, BC, Canada; Donald Patrick, PhD, MSPH, University of Washington, Seattle, WA, United States AIMS: Describe development of the Claudication Symptom Instrument (CSI) and assess its cross-sectional and longitudinal measurement properties for evaluating symptom experience of patients diagnosed with intermittent claudication. METHODS: We conducted concept elicitation interviews with claudication patients from Seattle, Washington area for item development and cognitive interviews for assessing item comprehension. Cross-sectional and longitudinal data were collected/analyzed in the context of a comparative effectiveness study of surgical and medical treatments for peripheral artery disease. Items were developed and self-administered on a 5-point response

123

168 scale ranging from 0 to 4, coded such that 4 indicated the worst symptom intensity during the last 7 days. RESULTS: We conducted 11 concept elicitation interviews to saturation: 7 male, 4 female; age range 54–78 years (median 62); 9 Caucasian, 2 African-American. Five symptom items important to patients were developed for assessing Pain, Cramping, Numbness, Heaviness, and Tingling. We analyzed complete data from 327 adults: mean age 71 years (9.5 SD); 70 % male; 87 % Caucasian, 11 % African-American. Item means ranged from 2.3 (Pain) to 1.1 (Tingling) and standard deviations from 1.09 (Pain) to 1.37 (Cramping). Percentage of responses in the lowest/ floor (0) and highest/ceiling categories (4) ranged from 10.1 % (Pain) to 54.7 % (Heaviness) and from 1.2 % (Tingling) to 9.8 % (Pain), respectively. Mixed methods including exploratory factor and Rasch analysis yielded support for an overall unidimensional factor (Cronbach’s alpha = 0.73). Total CSI score differed significantly by clinician-rated severity categories (mild vs. moderate/severe only) at p \ 0.01. As hypothesized, Walking Impairment Questionnaire (WIQ) scores were significantly positively correlated with CSI total score (Distance r = 0.46, Stairs r = 0.37) and Impact on Daily Activities (r = 0.51). Re-administration of the CSI 5–10 days after baseline yielded an intra-class correlation coefficient of 0.86. Changes in CSI total score and WIQ Distance score between baseline and 6-months post-treatment were correlated at -0.29. CONCLUSIONS: The CSI meets accepted measurement standards for reliability, content and psychometric validity, and ability to detect change. It assesses five distinct claudication symptoms identified as important by patients, and is sensitive to symptom changes associated with treatment. Further testing is needed to evaluate how the CSI performs in different contexts of use.

(3049) Content development of a patient-reported outcome instrument for the measurement of the primary signs and symptoms chronic plaque psoriasis April Armstrong, University of Colorado, Denver, CO, United States; Murali Sundaram, AbbVie Inc., North Chicago, IL, United States; Catherine Foley, Adelphi Values, Boston, MA, United States; Farrah Pompilus, Adelphi Values, Boston, MA, United States; Jonathan Stokes, Adelphi Values, Boston, MA, United States; Alan L. Shields, Adelphi Values, Boston, MA, United States AIMS: To report the content development of the Self-Assessment of Psoriasis Symptoms (SAPS), a patient-reported outcome (PRO) questionnaire to evaluate primary signs and symptoms of chronic plaque psoriasis. METHODS: A review of the empirical literature, advice meetings with dermatological experts specializing in psoriasis (n = 3), and qualitative concept elicitation patient interviews (CEIs; N = 60) were conducted to identify, document, and describe primary signs and symptoms of chronic plaque psoriasis. Conceptual models were developed based on each perspective and results were harmonized to inform the draft SAPS. Once developed, cognitive debriefing interviews (CDIs; N = 20) were conducted to evaluate subjects’ ability to read, comprehend, and complete the SAPS. Based on CDI results, the SAPS was revised and finalized. RESULTS: Eleven articles were reviewed and the following 12 chronic plaque psoriasis symptoms were identified in the literature: plaques, pain, itching, flaking, scaling, cracking, dry skin, burning/stinging, bleeding, redness, nail changes, and fatigue. Additionally, two expert advice meetings were held, in which a total of 20 chronic plaque psoriasis signs and symptoms were reported. Next, 60 CEIs were conducted with chronic plaque psoriasis subjects in two stages. During the first stage (N = 40), the most frequently reported signs and symptoms were itching (n = 37), redness (n = 30), flaking (n = 29), pain (n = 26), bleeding (n = 25), scaling (n = 23), and burning/stinging (n = 21). During the second stage (N = 20), the most commonly

123

Qual Life Res (2015) 24:1–191 reported signs and symptoms were itching (n = 11), flaking (n = 9), pain (n = 8), and bleeding (n = 8). Substantiated based on evidence from the three perspectives, symptoms selected for measurement in the SAPS included pain, itching, redness, scaling, flaking, dry skin, bleeding, burning, stinging, tenderness, pain due to skin cracking, and joint pain and instructions, items, and response options were generated to assess these concepts. During CDIs, 20 subjects provided feedback on the interpretability and relevance of the SAPS, and minor revisions were made to better assess flaking and joint pain, and an item assessing dry skin was deleted. A final expert meeting confirmed the SAPS to be comprehensive. CONCLUSIONS: Results from these research activities support the content validity of the SAPS and its readiness for use in clinical settings.

Kids and Young Adults (3051) Application of the EQ-5D-Y in a clinical study: how children and adolescents with chronic conditions assess health-related quality of life Kathrin I. Fischer, M.Sc.med., Charite - Universitaetsmedizin Berlin, Berlin, Germany; Dana Barthel, Dipl.-Psych., University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Christiane Otto, PhD, University Medical Center Hamburg-Eppendorf, Hamburg, Germany; Ute Thyen, MD, University Medical Center Schleswig– Holstein, Lu¨beck, Germany; Marcus Klein, MD, University Medical Center Schleswig–Holstein, Kiel, Germany; Otto Walter, PhD, MD, Charite - Universitaetsmedizin Berlin, Berlin, Germany; Silke Schmidt, Prof., Ernst-Moritz-Arndt University Greifswald, Greifswald, Germany; Holger Muehlan, PhD, Ernst-Moritz-Arndt University, Greifswald, Germany; Matthias Rose, MD PhD, Charite Universitaetsmedizin Berlin, Berlin, Germany; Ulrike RavensSieberer, PhD, MPH, University Medical Center HamburgEppendorf, Hamburg, Germany; Sandra Nolte, PhD, Charite Universitaetsmedizin Berlin, Berlin, Germany AIMS: This paper describes the assessment of health-related quality of life (HRQoL) employing EQ-5D-Y in a clinical sample of children and adolescents with chronic conditions. The focus of this investigation is the comparison of the self-reported health status comprising five health dimensions among children and adolescents diagnosed with asthma diabetes or rheumatoid arthritis. METHODS: EQ-5D-Y was completed by 304 children and adolescents consecutively recruited at two pediatric clinics in Germany. Descriptive analyses of the health status were performed with regard to the five dimensions and three levels comprising the EQ-5D-Y and the EQ-VAS score stratified by chronic condition and age. Kruskal–Wallis Tests were performed to compare the three different disease groups in terms of the five EQ-5D-Y dimensions. Mann–Whitney U Test were run to examine age group dependent differences among the three different chronic conditions with regard to the five health dimensions. RESULTS: 115 children (7–11 years) and 189 adolescents (12–17 years) with chronic conditions reported their HRQoL using EQ-5D-Y. The mean EQ VAS score was 79.17 for children/adolescents diagnosed with asthma, 83.21 in the group of diabetic patients and 81.28 for children/adolescents with rheumatoid arthritis. Among the total sample of 304 children/adolescents 12.5 % reported problems for the dimension mobility, 1.9 % stated to have problems washing or dressing themselves, 9.9 % had problems doing usual activities, 30.9 % reported pain/discomfort and 27.3 % felt worried, sad or unhappy. Further, significant differences revealed between the three disease groups related to the dimensions mobility, usual activities and pain/discomfort, while asthmatic children were more likely to report problems. No significant difference was found between children and adolescents with regard to the five health dimensions within the respective disease group. CONCLUSIONS: Despite the

Qual Life Res (2015) 24:1–191 ceiling effects, which could be found for all EQ-5D-Y dimensions, the assessment of HRQoL among a clinical sample of children and adolescents revealed that 53.0 % of all participants stated no problems at all in any health dimension. Compared to previous studies we could not find differences in HRQoL between a German norm population and our clinical sample. However, our results indicate that children with chronic diseases experience difficulties in terms of mobility, pain/discomfort and anxiety/depression.

(3053) Chronic conditions and health-related quality of life in school-aged children: findings from a large population-based study in the Netherlands Guannan Bai, MPH, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands; Marieke H. Herten, PhD, Statistics Netherlands, Heerlen, Limburg Province, Netherlands; Jeanne M. Landgraf, MA, HealthActCHQ Inc, Boston, MA, United States; Ida Korfage, PhD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands; Hein Raat, PhD, MD, Erasmus MC, University Medical Center Rotterdam, Rotterdam, South Holland Province, Netherlands AIMS: The aim of this study was to assess the impact of prevalent chronic conditions on health-related quality of life (HRQOL) in childhood, and to identify which HRQOL domains were specifically affected. METHODS: Statistics Netherlands included 10,651 children aged 4–11 years old and their parents in a national cross-sectional health survey from 2001 to 2009. In-house interviewers administered a parent questionnaire regarding the presence of chronic conditions (19 items), and the Child Health Questionnaire Parent Form 28 (CHQ-PF28) to assess HRQOL. Yearly response rate was approximately 75 %. The five most prevalent conditions were asthma (9.4 %), eczema (6.2 %), attention-deficit/hyperactivity disorder (ADHD) (4.8 %), dyslexia (3.9 %) and severe headache (3.2 %). Data was weighted given the child’s probability of selection, and to compensate for (selective) non-response. Population-for-analysis for this study consisted of children with no reported chronic condition (n = 6062), and children with only one condition: asthma (n = 521), eczema (n = 308), ADHD (n = 207), dyslexia (n = 207), or severe headache (n = 139) (see Table 1). Mean differences in CHQ scales and Component Summaries - Physical (PhS) and Psychosocial (PsS) were calculated using independent-T tests; Cohen’s effect sizes (d) were calculated (mean differences divided by the largest standard

Table 1 CHQ-PF28 scale scores and summary scale scores for 5 common childhood chronic conditions compared to a subgroup with no parent-reported chronic condition (N = 7445)#

169 deviation; 0.2 B d \ 0.5 indicates small difference, 0.5 B d \ 0.8 moderate, dC0.8 large). RESULTS: Mean age of children was 7.5(SD2.3) years; 49.1 % were girls. Compared to children with no reported chronic condition, HRQOL scores for PhS and/or PsS were lower (p \ 0.05) across all conditions except eczema (Table 1). A large effect size was observed in ADHD (PsS d = 0.89). Small effects were observed for both PhS (d = 0.44) and PsS (d = 0.22) and across many CHQ-scales for severe headache, including a moderate effect for bodily pain (d = 0.53). The subgroup with ADHD showed a specific HRQOL-profile with large and moderate effects for behavior (d = 1.38), and mental health (d = 0.58). With regard to asthma, a moderate effect was observed for general health (d = 0.65) and small effects for other CHQ-scales related to the physical domain. CONCLUSIONS: Significant negative impacts on physical and psychosocial aspects of HRQOL were observed across prevalent childhood chronic conditions, depending on the specific nature of the condition. Health professionals should take into account specific negative HRQOL-impacts of chronic conditions in daily practice.

(3055) Quality of life and alcohol consumption behaviors by gender among first-year students of a Chilean university Carlos A. Hidalgo-Rasmussen, Researcher, Universidad de Guadalajara, Universidad de Playa Ancha, Cd. Guzma´n, Jalisco, Mexico; Marı´a Jacqueline Rojas, Researcher, Universidad de Playa Ancha, Vin˜a del Mar, Chile; Fabiola Vilugro´n, Professor, Universidad de Playa Ancha, Quilpue´, Chile; Vero´nica Paste´n, Professor, Universidad de Playa Ancha, Vin˜a del Mar, Chile AIMS: To analyze differences in quality of life according to alcohol use variables by gender, among first-year university students. METHODS: A cross-sectional study was carried out in the frame of the Research Observatory of risks and quality of life of students of the University of Playa Ancha. From 1345 students matriculated and 1154 were assessed on March 2014, for this study 735 students of 17–18 years were selected. To measure alcohol use, items from Youth Risk Behavior Survey (YRBS) were used.The KIDSCREEN-52 Chilean version was used (Cronbach alpha was .94). Independent samples t test were used, considering p \ .05. Informed consent was obtained. RESULTS: Participants were 66.9 % women (W). W who abstained to consume alcohol in her life had a lower mean in Friends and Social Support (FSS) dimension. W who started drinking from 8 to 14 years old, had a lower mean than 15 or more years in SelfPerception (SP) and Parents Relation and Home Life (PHL); while men (M) in Physical Well-Being (PWB) and PHL. W who reporting at least one drink of alcohol in the past 30 days, had a lower mean in School Environment and W who did not reported it, had a lower mean in Friends and Social Support (FSS) while in M, those who reporting hadn´t drunk in the past 30 days, had a lower mean in FSS. W who reporting not drunk 5 or more drinks of alcohol (like wine, beer, pisco or other) within a couple of hours in the past 30 days had a lower mean in Autonomy (A), FSS and Economical Resources (ER). While M who not reporting had a lower mean in FSS and those who drunk 5 o more had a lower PHL. Finally, W who reporting an imperative necessity to drink had a lower mean in PHL and among M those who doesn’t have the imperious necessity had a lower mean in FSS. CONCLUSIONS: Seven of ten QoL dimensions had a significant difference in some alcohol variables. On average W had more QoL dimensions associated to alcohol use variables than M. In some dimensions, the alcohol use remain associated with a higher QoL mean.

123

170

Qual Life Res (2015) 24:1–191

(3059) Longitudinal associations among asthma control, sleep problems, and health-related quality of life in children with asthma: a report from the PROMIS Pediatric Asthma Study

Coventry, United Kingdom; Sophie Staniszewska, Senior Research Fellow, Warwick Medical School, Warwick University, Coventry, United Kingdom

Zheng Li, University of Florida, Gainesville, FL, United States; Heather Gross, University of North Carolina at Chapel Hill, Chapel Hill, United States; Bryce B. Reeve, PhD, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; Darren DeWalt, MD MPH, University of North Carolina at Chapel Hill, Chapel Hill, NC, United States; I-Chan Huang, PhD, St. Jude Children’s Research Hospital, Memphis, TN, United States

AIMS: HRQoL-research involves the development, use and evaluation of questionnaires in research and patient care settings to assess how patients feel, function and live their lives in relation to their health and associated healthcare. The active engagement of patients as research partners is increasingly viewed as essential to ensuring that patient-driven views are considered throughout the research process. However, guidance for active PE in HRQoL research does not exist and the evidence-base is limited. Moreover, the values held by different stakeholders (e.g., why do we do PE?; how should we do PE?; and what’s important?) may affect the approaches to PE being adopted and its likely impact. We explored the values that should underpin PE in HRQoL-research which would be helpful in informing future good practice guidance. METHODS: A modified ‘World Cafe´’ was hosted as a collaborative activity between patient partners, clinicians and researchers: self-nominated ISOQOL (2014) conference delegates participated in small group discussions to explore values associated with the conduct and consequences of PE in HRQLresearch. Values were captured individually via post-it notes and per table by nominated note-takers. Four reviewers (KH, SS, AL, SBM) completed independent inductive thematic analyses of the data: emergent themes were coded and agreement checked. Association between the data, emergent themes, values and the Public Involvement Impact Assessment Framework (PiiAF) were explored. RESULTS: Eighty participants, including 12 patient partners, participated in the 90-minute event. Three core values were underpinned by six over-riding themes: (1) Building relationships (genuine relationships; PE challenges); (2) Improving research quality and impact (research quality; research impact); and (3) Developing best practice (PE approaches; evidence-base). CONCLUSIONS: An explicit statement of values seeks to align all stakeholders on the purpose, practice and credibility of PE activities. Participants valued the importance of building genuine, collaborative and deliberative relationships between all stakeholders, underpinned by honesty, respect, co-learning and equity. Also valued was the impact of effective PE on the quality and relevance of research. An innovative, flexible and transparent research environment was also valued as essential to developing a trustworthy evidence-base with which to underpin future guidance for Good PE Practice (GPEP).

AIMS: Few studies have compared the complex relationship among asthma control, sleep problems, and health-related quality of life (HRQOL). This study tested the longitudinal associations of asthma control status with HRQOL through the mechanism of nighttime sleep quality and daytime sleepiness. METHODS: A total of 238 asthmatic children engaged in the Patient-Reported Outcomes Measurement Information System (PROMIS) Pediatric Asthma Study to assess the change of asthma control, sleep problems, and HRQOL from baseline (T1) to 3 follow-ups (T2 through T4), occurring over 2 years. Asthma control was measured using the Asthma Control and Communication Instrument. Nighttime sleep quality was measured using a brief scale developed in this study that captures difficulty falling asleep and getting up, and sleep disturbance. Daytime sleepiness was measured using the Iowa Pediatric Daytime Sleeping Scale. Pediatric HRQOL was measured using the PROMIS Asthma Impact Scale. Randomintercept models were conducted to investigate the multivariate associations of HRQOL with asthma control and sleep problems. Multilevel structural equation modeling (MSEM) was performed to quantify the direct and indirect effects of asthma control status on HRQOL through the influence of nighttime sleep quality and daytime sleepiness at within-subject (repeated measures of an individual) and between-subject (population average) levels. RESULTS: Bivariate analyses showed asthma control status was associated with nighttime sleep quality and HRQOL across 4 time points (p’s \ 0.001); asthma control status was associated with daytime sleepiness at T1 and T4 (p’s \ 0.01). Multivariate analyses suggested HRQOL was associated with asthma control status and daytime sleepiness (p’s \ 0.001) after controlling for socio-demographic factors. In MSEM, poorer asthma control status was directly associated with lower HRQOL at the within-subject and between-subject levels (p’s \ 0.05); however, effects of asthma control on HRQOL were indirectly affected through daytime sleepiness at the within-subject levels (p’s \ 0.05), and through nighttime sleep quality and daytime sleepiness at the between-subject levels (p’s \ 0.05). CONCLUSIONS: Poorer asthma control status is associated with worse HRQOL, and this association is mediated by poorer nighttime sleep quality and greater daytime sleepiness. Clinicians should address sleep problems related to poor asthma control in order to improve HRQOL for children with poor asthma control.

International (3061) Defining the values for patient engagement (PE) in healthrelated quality of life (HRQoL) research: an international perspective Kirstie Haywood, Senior Research Fellow, Warwick Medical School, Warwick University, Coventry, United Kingdom; Sam Salek, PhD RPh FFPM MRPSGB MCMS FESCP, School of Life & Medical Sciences, University of Hertfordshire, Hatfield and Director - Institute for Medicines Development, Cardiff, United Kingdom; Anne Lyddiatt, Patient Research Partner, OMERACT, Cochrance Collaboration., London, Canada; Samantha Brace-McDonnell, Patient Research Partner, Warwick Medical School, Warwick University,

123

(3067) Quality of life in residents for earthquake, tsunami and radiation disaster in Fukushima, Japan Rika Hayashida, University of Nagasaki, Siebold, Nishisonogigun, Nagasaki, Japan; Takashi Mandai, Japanese Society of Quality of Life Research, Kobe, Japan; Ayako Tsukihashi, Japanese Society of Quality of Life Research, Kobe, Japan; Hiroshi Shimagami, Japanese Society of Quality of Life Research, Kobe, Japan; Akihiro Yamamoto, Japanese Society of Quality of Life Research, Kobe, Japan; Li-sa Chang, Japanese Society of Quality of Life Research, Kobe, Japan; Taiji Omura, Japanese Society of Quality of Life Research, Kobe, Japan; Yoshihiro Tairako, Tounou Sangyou Sinkou Jigyou Cooperation, Iwaki, Fukushima, Japan; Shinichi Suzuki, Kashima Hospital, Iwaki, Fukushima, Japan; Yoshie Tada, Narahamachi, Iwaki Office, Iwaki, Fukushima, Japan; Ichirou Nagano, Japanese Society of Quality of Life Research, Kobe, Japan; Kouki Kaketaka, Japanese Society of Quality of Life Research, Kobe, Japan; Shuuto Yamashita, Japanese Society of Quality of Life Research, Kobe, Japan; Masaya Kusuda, Japanese Society of Quality of Life Research, Kobe, Japan; Atsuhiko Fukuoka, Japanese Society of Quality of Life Research, Kobe, Japan; Haruyasu Fujita, Japanese Society of Quality of Life Research, Kobe, Japan; Souichirou

Qual Life Res (2015) 24:1–191 Maruyama, Japanese Society of Quality of Life Research, Kobe, Japan; Michiko Kobayashi, Japanese Society of Quality of Life Research, Kobe, Japan; Kouzaburou Adachi, Japanese Society of Quality of Life Research, Kobe, Japan; Tomotaka Sobue, Graduate School of Medicine, Osaka University, Osaka, Japan AIMS: The purpose of this study was to investigate the quality of life (QOL) deteriorations in residents who had to move to another city for 2011 earthquake, tsunami and radiation disaster in Fukushima, Japan. METHODS: Thirty-four residents participated in this study. Our new original self-administered QOL questionnaire including 37 questions divided into 11 categories and the Life Satisfaction Index questionnaire by Neugarten were used. RESULTS: Cronbach’s alpha coefficients of our new original QOL questionnaire were excellent enough to accept for clinical use: 0.90 in sense of happiness, 0.89 in dietary problems, 0.87 in higher cerebral function, 0.84 in health status, 0.83 in economy, etc., respectively. Our new original QOL questionnaire contained 11 main factors and cumulative contribution was 0.73. There was a significantly positive correlation on mean QOL score between our new original QOL questionnaire and the Life Satisfaction Index questionnaire by Neugarten (r = 0.77), p \ 0.01). Compared with before disaster, almost all (31 of 34) residents showed deteriorated QOLs after disaster. Compared with before disaster, there was a significant deterioration of mean QOL score after one (p \ 0.01). Compared with before disaster, significant deteriorations of QOLs were demonstrated in sense of happiness (p \ 0.01), in health status (p \ 0.01), in dietary problems (p \ 0.01), in social participation (p \ 0.01), in economy (p \ 0.01), in environmental problems (p \ 0.01) and in medical service (p \ 0.05) after disaster, respectively. There were also significantly positive correlations between total QOL and sense of happiness (r = 0.84, p \ 0.01), total QOL and environmental problems (r = 0.83, p \ 0.01), total QOL and social participation (r = 0.76, p \ 0.01), total QOL and economy (r = 0.76, p \ 0.01) etc. after disaster, respectively. CONCLUSIONS: These findings indicate that our new original QOL questionnaire has excellent enough reliability and potency of validity to investigate the QOLs in residents who had to move to another city for earthquake, tsunami and radiation disaster. Though both almost all residents and 7 of 11 categories of QOLs showed the deterioration on QOL after disaster compared with before one, we must reconstruct individual lives, homes and the brilliant societies after the disaster again in the future. In order to accomplish our duty, the QOL evaluation system must play the most important role as scientific evidence.

(3069) Purpose in life and personal growth: predictors of quality of life in Mexican elders Ana L. Gonzalez-Celis, Psychology, National Autonomous University of Mexico (UNAM), Tlalnepantla, Estado de Mexico, Mexico; Juana Gomez-Benito, Psychology, University of Barcelona, Barcelona, Spain; Margarita Maldonado-Saucedo, Psychology, ITESO, University Western Institute of Technology and Higher Education, Tlaquepaque, Jalisco, Mexico; Maria E. Vidan˜a-Gaytan, Psychology, Autonomous University of Ciudad Jua´rez, Ciudad Juarez, Chihuahua, Mexico; Ana G. Magallanes-Rodriguez, Psychology, Autonomous University of Baja California, Tijuana, Baja California, Mexico; Margarita Chavez-Becerra, Psychology, National Autonomous University of Mexico (UNAM),, Tlalnepantla, Estado de Mexico, Mexico; Edgardo Ruiz-Carrillo, Psychology, National Autonomous University of Mexico (UNAM), Tlalnepantla, Estado de Mexico, Mexico AIMS: The aim of this study was to determine whether subjective well-being (SW) and two of its dimensions, purpose in life and

171 personal growth, are predictors of quality of life in Mexican elders. METHODS: A sample of 582 older adults (mean age = 70.3, SD = 7.5) gave their informed consent to an individual interview in which two instruments were administered: the WHOQOL-OLD (Power, Quinn, Schmidt & the WHOQOL-OLD Group, 2005), a selfreport measure of quality of life (QoL) in older adults that comprises 24 items distributed across six domains (sensory abilities; autonomy; past, present and future activities; social participation; fear of death and dying; and intimacy); and the Ryff Scales of Psychological Wellbeing (Ryff, 1995), the focus here being on two of its dimensions, purpose in life (PiL) and personal growth (PG). RESULTS: Significant differences were found between the younger (60–74 years) and older ([75 years) groups in total quality of life (t = 3.942, p \ .0001) and on four QoL domains, namely sensory abilities (t = 2.428, p \ .015), autonomy (t = 3.988, p \ .0001), social participation (t = 2.567, p \ .011) and intimacy (t = 4.969, p \ .0001). No differences were observed for the remaining domains, and neither was there statistical evidence of differences in quality of life (total and domain scores) between men and women or between elders who reported health problems and those who did not. Statistically significant correlations were observed between QoL and PiL (r = .565, p \ .0001), QoL and PG (r = .442, p \ .0001), and QoL and SW (r = .633, p \ .0001). The significant coefficients in the linear regression model were as follows: PiL (b = 0.134, p \ .038), PG (b = -0.138, p \ .025) and SW (b = 0.632, p \ .0001), with F(3,423) = 99.28, p \ .0001, and R2 = .413. CONCLUSIONS: The results indicate that purpose in life, personal growth and subjective well-being predict quality of life in Mexican elders. Seeking to achieve one’s goals and dreams across the years is important for remaining active and for enjoying life.

(3071) Evaluating a methodology for providing individualized feedback on quality of life and its importance in community and primary care, using the WHOQOL-BREF Suzanne M. Skevington, PhD, CPsychol, FBPsS, University of Manchester, Manchester, United Kingdom; Alison Llewellyn, Consultant Health Psychologist, Univ of Bath, Bristol, United Kingdom AIMS: Wellbeing is at its very lowest when individuals’ rate quality of life (QoL) dimension as both poor and very important (Skevington and O’Connell, 2004). We aimed to find out how people evaluated guided individualised feedback when a graphical profile of subjective quality of life (QoL) dimensions and perceived importance was presented. The study which is underpinned by Carver & Scheier’s SelfRegulation Theory (SRT) examined whether this novel feedback method was relevant and valued by healthy and sick people. METHODS: A mixed-methods cross-sectional pilot study recruited samples of adults from the community and primary care, targeting 70 % with a chronic illness and 30 % healthy. The WHOQOL-BREF and WHOOL-BREF Importance measures were completed by 129 participants. Graphed results from both measures were inspected together. Directed guidance was used to identify differences on the 25 QoL dimensions, and interpret good and poor QoL. Evaluative ratings and written comments appraised the feedback. At follow up, relevance to healthcare was explored with a subgroup who found feedback helpful. RESULTS: Following feedback, 65 % of participants reported changes in thoughts and perceptions of QoL, describing new insights that were interpreted as self-affirming. Goals or expectations changed for 34 %, and increased motivation to change reported. Over 50 % evaluated the feedback as helpful in the short or long term. Follow up in-depth interviews showed participants valued the feedback process; 92 % acknowledged its usefulness in sharing information with healthcare professionals. It would facilitate

123

172 communications and help health professionals to treat them with greater understanding and relevance. They wanted health professionals to use these understandings to provide targeted health advice and support. Comparisons between samples are reported. CONCLUSIONS: To capitalize on the benefits of individualised, comprehensive guided feedback, this pilot complex intervention needs testing in a fully randomized controlled trial. Our novel feedback process that presents self-rated QoL alongside its importance has value beyond self-monitoring. Potentially it could promote individual behavior change and patient self-management, and support shared clinical decision-making about priorities in health and social care settings.

(3073) Parent’s experiences of neonatal care in England: a large scale postal survey Jenny King, Picker Institute Europe, Oxford, United Kingdom; Amy Tallett, Picker Institute Europe, Oxford, United Kingdom; Sarah-Ann Burger, Picker Institute Europe, Oxford, United Kingdom; Tamara van Doorn, Picker Institute Europe, Oxford, United Kingdom AIMS: The aim of the survey was to enable neonatal units in England to better understand the experiences of the parents’ who use their services. METHODS: A large scale survey of parents’ experiences of neonatal care was carried out in England in 2014. The survey covered themes such as interactions with staff, knowledge and information sharing, and active care. The paper-based self-completion survey was mailed out to 15,944 parents from 88 Neonatal units. Fieldwork was carried out between May 2014 and December 2014 and was conducted in two waves. RESULTS: Overall, responses were received from 6,000 parents, and after taking into account undelivered questionnaires and those parents ineligible for inclusion, a response rate of 37.6 % was achieved. The highest response rate achieved by a participating neonatal unit was 59 % and the lowest response rate was 9 %. Looking at results at an overall level, a large variation in scores across neonatal units was evident for a number of questions. Eleven questions had a range in scores greater than 50. For example for question A2 ‘Before your baby was born, did a member of staff from the neonatal unit talk to you about what to expect after the birth?’ the best performing neonatal unit on this question achieved a score of 97 out of 100 whereas the worst performing neonatal unit on this question only achieved a score of 12 out of 100. This demonstrates that for many of the themes covered by the questionnaire, performance is incredibly varied across neonatal units. The survey also identified room for improvement in information provision and emotional support. CONCLUSIONS: Evaluating experience is a key component to monitoring the quality of health care. This survey not only provided a national picture of parents’ experiences of neonatal care but produced findings that could be used by individual neonatal units to drive improvement at a local level.

(3083) Spiritual and religious coping and quality of life among older adults: Brazilian study Luciano Vitorino, Paulista School of Nursing, Federal University of Sa˜o Paulo - UNIFESP., Itajuba, Minas Gerais, Brazil; Gail Low, Associate Professor, University of Alberta, Edmonton, AB, Canada; Lucila Amaral Vianna, Full Professor, Federal University of Sa˜o Paulo - UNIFESP, Sao Paulo, Sao Paulo, Brazil AIMS: We investigated whether positive and negative spiritual and religious coping (SRCOPE) were important to the quality of life (QOL) of institutionalized and community-dwelling older Brazilians. We were also interested in whether SRCOPE was differently importance to the QOL of either study group. METHODS: We

123

Qual Life Res (2015) 24:1–191 analyzed two sets of cross-sectional survey data collected from 77 institutionalized and 326 community-dwelling older adults. Study measures were: The Brief SRCOPE, and the WHOQOL-BREF and WHOQOL-OLD. Demographic data were also collected. We used a MANOVA to analyze the effects SRCOPE upon QOL. We compared the effect size of SRCOPE upon QOL using AMOS. RESULTS: Institutionalized older adults had higher scores across all domains and facets but these differences were not significant. Positive SRCOPE alone had a significant main effect upon QOL of both study samples (F = 13.205, df = 4, p \ .001). SRCOPE positive enhanced QOL across 4 WHOQOL-BREF domains. This was also the case with respect to the WHOQOL-OLD (F = 13.502, df = 6, p \ 001) across five facets. The enhancing effect of SRCOPE on QOL significantly depended upon place of residence for the WHOQOL-BREF social relationships domain (F = 6.253, df = 4, p \ .05) and autonomy (F = 11.876, df = 6, p \ .05), Past, Present and Future Activities (F = 21.096, df = 6, p \ .01), Social Participation (F = 7.841, df = 6, p \ .05) and intimacy (F = 9.230, df = 6, p \ .05) WHOQOL-OLD facets. The effect of positive SRCOPE across these 5 aspects of QOL was far larger for the institutionalized group. CONCLUSIONS: Positive SRCOPE was important to the social relationships, living environment, autonomy, and physical health of 403 older Brazilians. Positive SRCOPE more strongly influenced the institutionalized group’s social relationships, autonomy, past, present and future, social participation, and intimacy. In future, collecting data over time would help us determine whether the QOL enhancing effects of positive SRCOPE is consistent across the same domains and facets of life. Collecting data from larger groups of nursing home residents would allow for more robust comparisons with communitydwelling older Brazilians.

(3085) Adapting to aging among Norwegian elders Mary Kalfoss, Professor, Diakonova University College, Oslo, Norway; Liv Halvorsrud, Associate Professor, Oslo & Akershus University College of Applied Sciences, Oslo, Norway, Gail Low, Associate Professor, Faculty of Nursing, University of Alberta, Edmonton, AB, Canada AIMS: Very little is known about how older Norwegians psychologically adapt to their own process of aging. Our work is guided by the Identity Process Theory. This theory outlines three psychological attribution processes for adapting to one’s own process of aging. We were particularly interested in older Norwegian propensities to attribute difficulties in physical functioning to their age versus their health. In theory, acknowledging age-related difficulties enhances negative affect and negativity toward one’s own process of aging. METHODS: We analyzed cross-sectional data collected in the 2003 WHOQOL-OLD Norwegian field study. These data were collected by Statistics Norway from older people residing in 20 randomly selected and stratified communities throughout Norway. The measures used in this study were the 3 domain Attitudes to Ageing Questionnaire (AAQ), the SF-12 Physical Functioning and General Health subscales, and the short-form Geriatric Depression Scale. These were completed using a variety of culturally appropriate methods, including mail-outs, self-administration, and interviews. We had complete data on 382 community-dwelling Norwegians between 60 and 91 years of age. In our analyses we controlled for gender, education, and living arrangements. Significant predictors of attitudes to aging and depressive symptoms observed in a GLM regression analysis were further scrutinized using path analysis. RESULTS: The interaction between physical functioning and age was not statistically significant in all three path models. Accordingly our participants attributed difficulties in physical functioning to a more variable general state of health as opposed to age. This attribution process

Qual Life Res (2015) 24:1–191 moderately enhanced participants’ symptoms of depression and negativity toward psychosocial loss, physical change, and psychological growth. Use of the identity assimilation process was significantly associated with managing everyday activities without the support of adult children and/or a partner. Women exhibited far more negativity about psychosocial loss and psychological growth. CONCLUSIONS: Our findings are theoretically contradictory but speak to the Likhet-based norm of not dwelling upon the self. The psychological costs of denying age-related functional losses speak to denying a core aspect of social identity – family cohesiveness. Comparisons with frailer Norwegians and those managing at home with formal support are warranted.

(3087) Impact of socio-economic factors on oral health related quality of life in Japanese population Mariko Naito, DDS, PhD, Nagoya University Graduate School of Medicine, Nagoya, Japan; Yoshimi Suzukamo, PhD, Tohoku University Graduate School of Medicine, Sendai, Japan; Tatsuro Miyake, DDS, PhD, Baika Woman’s University, Ibaraki, Japan; Kenji Wakai, MD, PhD, Nagoya University Graduate School of Medicine, Nagoya, Japan AIMS: The impact of socioeconomic status on health has been investigated in relation to morbidity and mortality. Some studies reported that low educational level and low income were profound predictors of negative health-related quality of life. This study aimed to examine relationships between socio-economic factors and oral health-related quality of life (OHQOL) in Japanese population. METHODS: A cross-sectional study was conducted in 2006 using a representative sample in Japan. Respondents were selected through stratified multi-stage random sampling. A questionnaire survey was performed in 2,400 sampled males and females aged 15–79 years. The self-administered questionnaire included the General Oral Health Assessment Index (GOHAI) for measuring OHQOL. A higher GOHAI score indicates better OHQOL. Information was also collected on socio-economic factors such as education and household income. The mean GOHAI score was computed by education or household income-level groups after adjusting for age and gender by analysis of covariance. RESULTS: Replies were obtained from 1,170 of the 2,400 subjects, of whom 1,111 (539 males and 562 females) were analyzed (range 20–79 years). The GOHAI scores were inversely associated with age (trend P \ 0.001). Analysis according to characteristics showed no significant differences in the GOHAI score among residential areas, city scales or gender. The positive association between the educational level and the GOHAI score was observed among the subjects aged 40–59 years (trend P = 0.02). There were positive correlations between the household income and the GOHAI score among the subjects aged 40–59 years (trend P = 0.03) and females aged 20–39 years (trend P = 0.06). No significant association was found in subjects aged 60 or over. CONCLUSIONS: Our findings indicate that the socioeconomic status may contribute to impairment of OHQOL in middle-aged Japanese males and females.

(3089) Health related quality of life in a community dwelling elderly sample in Brazil Andre´a Libo´rio Monteiro, HTA Ms MSc, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil; Bruna Veras, MSc, Instituto Nacional de Cardiologia, Rio de Janeiro, Brazil; Juliana Peixoto, MSc, Instituto Nacional de Cardiologia, Rio de Janeiro, Brazil; Sandra Furtado, MSc, Prefeitura do Rio de Janeiro – SESQV, Rio de Janeiro, Brazil; Helio Furtado, MSc, Prefeitura do Rio de Janeiro –

173 SESQV, Rio de Janeiro, Brazil; Marisa Santos, MD, PhD, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil AIMS: To explore and characterize Health Related Quality of Life (HRQoL) in a community dwelling elderly sample, to investigate the determinants of HRQoL assessed by EQ-5D-3L. METHODS: Data was collected from a sample of newly-admitted elder participants of a social program that takes place in Rio de Janeiro, Brazil. This dataset includes data from 149 individuals older than 60 years, interviewed from May 2014 to November 2014. The interview comprised information on sociodemographic characteristics, life-style and general health. The HRQoL was assessed using the EQ-5D-3L, cognitive impairment was assessed using the Mini Mental State Examination (MMSE) and social support was measured using the MOS Social Support Survey (MSSS). Participants with cognitive impairment (MMSE \ 18) were excluded from this analysis. Exploratory analysis and literature research were used to identify potential HRQoL predictors to be included in the model. The Mann–Whitney and Kruskall–Wallis tests were applied to examine the differences in means between categories. Ceiling and floor effects, represented by the proportion of participants with the best and worst theoretical scores respectively, were also examined. EQ-5D Index Values were estimated using the weights from the Brazilian valuation study. RESULTS: The sample (n = 144) is composed by 84 % women, with a mean age of 70.5 years (SD = 6.8). The overall utility mean was 0.76(SD = .114) and the EQ-VAS was 79.7(SD = 17.83). While floor effects found for both EQ-5D descriptive system and EQ-VAS were negligible (less than 1 % in both measures), the ceiling effects were considerable (12.5 % on EQ-VAS and 21.5 % on Descriptive System). Men (0.81, SD = 0.09) presented higher HRQoL (Z = -2.581, p = 0.009) than and women (0.75, SD = 0.11). Several variables were tested, but only diabetes (b = -0.05), depression (b = -0.06), back problems (b = -0.08), and cancer (b = -0.12), integrated the final model [F = 16.35 (p \ 0.001); df1 = 4, df2 = 139; adj.R2 = 0.3004)]. CONCLUSIONS: Chronical diseases were identified as predictors of HRQoL in our sample. This results are consistent with the literature.

(3091) Disparities in quality of life and mental distress: a comparison of Chinese heterosexual and homosexual populations Janet Y. Wong, PhD, RN, The University of Hong Kong, Hong Kong SAR, China; Daniel Y. Fong, PhD, The University of Hong Kong, Hong Kong SAR, China AIMS: Considerable evidence showed that homosexual people have increased risk of poor quality of life and negative mental health outcomes. However, relevant research in Chinese population is very limited. With the traditional Chinese culture of general role and central value of filial piety to continue the family name via marriage, homosexual people faced many obstacles in family and society. We investigated the disparities in quality of life and mental distress in Chinese heterosexual and homosexual populations and their associated factors. METHODS: We compared the quality of life and mental distress (including anxiety, depression and perceived stress) in a sample of 400 Chinese heterosexual and homosexual people in dating relationships or being separated in the past year. We recruited participants from university campuses and collected data by using either self-administered questionnaires either by online platform or pencil-and-paper format and then used structured multiphase regression to examine their risk factors. RESULTS: The participants aged between 18 and 62, with 16.3 % of them being homosexual people. Majority of them are currently in relationship (74.8 %). There were significant difference between two groups in quality of life, anxiety, depression and perceived stress (p \ 0.05).

123

174 Also, results showed that dating violence was much more severe in homosexual people than heterosexual people. In addition, we found that (1) hiding sexual orientation to family members and (2) dating violence were the consistent significant risk factors of quality of life (Z1 = -2.1, p = 0.03; Z2 = -1.2, p \ 0.0001), anxiety (Z1 = .9, p = 0.02; Z2 = .6, p \ 0.0001), depression (Z1 = 1.1, p \ 0.0001; Z2 = .4, p \ 0.0001) and perceived stress (Z1 = 1.6, p = 0.001; Z2 = 1.1, p \ 0.0001) after controlling for other sociodemographic factors. CONCLUSIONS: Homosexual Chinese people have unique experiences and challenges in family and society. This study helps in understanding the specific risk factors for their quality of life and mental distress. Helping in identifying filial conflicts and providing access to knowledge about dating violence will enable homosexual people to re-allocate their resources for their mental distress which in turn enhancing quality of life.

Utilities (3093) Validity of the EQ-5D-Y in a cohort of asthmatic children in Europe Gimena Herna´ndez, MD, MPH, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Spain, Universitat Auto`noma de Barcelona, Bellaterra, Barcelona, Spain; Monica Avila, BPharm, MPH, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Spain, Universitat Pompeu Fabra, Barcelona, Spain; Olatz Garin, PhD, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Spain, Universitat Pompeu Fabra, Barcelona, Spain; Angels Pont, BSc, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP)., Barcelona, Spain; Alexandra Dima, PhD, Dept of Communication Science, ASCoR, University of Amsterdam, Amsterdam, Netherlands; Laurent Laforest, MD, PhD, Unite´ de Pharmaco e´pide´miologie, Faculte´ d’Odontologie, UMR 5558 CNRS - Universite´ Claude-Bernard Lyon, CHU, Lyon, France; Montse Ferrer, MD, PhD, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain, CIBER en Epidemiologı´a y Salud Pu´blica (CIBERESP), Spain; Universitat Auto`noma de Barcelona, Bellaterra, Barcelona, Spain AIMS: Up to date, metric properties of EQ-5D-Y have not been assessed in children with asthma. Our objective was to examine construct validity and score distribution of the EQ-5D-Y in an European sample of asthmatic children. METHODS: A subgroup of 69 patients between 6 and 11 years included in the AstroLab cohort who completed EQ-5D-Y via online questionnaire was analyzed. The EQ-5D-Y is a brief, multi-attribute, generic, preference-based health status measure consisting of five dimensions with three response levels. The index value ranges from 1 (best health possible) to -0.594 (negative values indicate health state worse than death). Index values were calculated using the EQ-5D-3L preference value sets for France. To examine the distribution of the index measures, we calculated measures of central tendency, dispersion, ceiling and floor effects, and observed range. Construct validity was assessed by comparing means of known groups defined by the Royal College of Physicians 3-Questions (RCP3Q) for children that measures the presence of asthma symptoms in the last month in 3 Likert scale response options, with a score ranging from 0 to 9 (lower score better asthma control). Two groups of asthma control were defined: well-controlled (RCP3Q 0–1), and not well-controlled (RCP3Q [ 1). ANOVA tests and effect

123

Qual Life Res (2015) 24:1–191 sizes (ES) between groups were calculated. RESULTS: Mean EQ-5DY index was 0.878 (0.211).The observed range was -0.12 to 1. Floor and ceiling effects were 0 and 50 %, respectively. Mean EQ-5D-Y index for children with well-controlled asthma was 0.938 (95 % CI 0.912–0.964); and 0.815 (95 % CI 0.715–0.914) for children not wellcontrolled. These indexes were significantly different (p = 0.015), and the effect size between both groups was large (ES: 1.09). CONCLUSIONS: The new EQ-5D-Y provided an adequate index distribution in children with asthma, with a considerable ceiling effect. Statistical and substantive differences across known asthma control groups indicates good construct validity for assessing health related quality of life in these patients.

(3095) Using the Health Utilities Index to improve chronic-care management Maria Jose Santana, PhD, O’Brien Institute for Public Health, Calgary, AB, Canada AIMS: Health Utilities Index (HUI) is a family of generic preferencebased measures of health status and health-related quality of life. HUI has been used widely in clinical and population studies. Recently, there is emerging evidence on the use of HUI in the care of chronically-ill patients at the individual level. Our objective is to describe the role of HUI in chronic-care management. METHODS: HUI has been used to support patient management at the out-patient lung transplant clinic in a tertiary care hospital in Canada.Upon arrival to the clinic, patients complete the HUI questionnaire electronically and results are graphically presented on the HUI score card (HUIsc) to patients and healthcare providers prior to the consultation. The HUIsc displays severity for each attribute and the overall score using colorcodes that range from green (no problem) to red (severe problem). The clinician team used both clinical parameters (e.g., FEV1 %) and HUI scores in patient management. The potential effects of using HUI in clinical settings include communication, patient management, adherence, satisfaction, and patient outcome. Furthermore, we describe how the use of the measures provided a platform of communication among healthcare providers. We examine clinical cases for evidence of these effects. RESULTS: Examples of the effects of using HUI in clinical care of chronic diseased patients are provided. For instance, a 72 year old man with advanced heart failure who was receiving a very complex pharmacologic treatment appeared unresponsive to clinician’s advice. The HUIsc identified a hearing deficit. The patient was referred to the audiology clinic and weeks later got two hearing aids. In a number of cases HUI scores declined before clinical makers did. As evidence accumulates, it may be possible to use a decline in health as recorded by the HUI as an early warning sign. The use of the HUIsc affected communication among healthcare providers, reflecting the influence of the HUI color-coding system. For instance, a stable patient (with green a HUIsc indicated no problems) was referred as a ‘green patient’. CONCLUSIONS: The use of HUI in clinical care can screen for problems, identify under or unrecognized burdens and support patient management. Communication among members of the healthcare team level is also affected.

(3097) Further evidence on EQ-5D-5L preference inversion: a Brazil/US collaboration Benjamin M. Craig, PhD, Moffitt Cancer Center and University of South Florida, Tampa, FL, United States; Marisa Santos, MD, PhD, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil; Andre´a Libo´rio Monteiro, HTA Ms MSc, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil

Qual Life Res (2015) 24:1–191 AIMS: In a recent United States (US) study, most respondents preferred ‘‘extremely’’ over ‘‘severely anxious or depressed’’ (30 out of 59; 51 %), which contradicts the labeling of the 5-level EQ-5D (EQ5D-5L). Anecdotal evidence further suggests that the Portuguese version may have greater problems, where the translation of ‘‘severely’’ (gravemente) signifies ‘‘gravely.’’ METHODS: Through a US infusion clinic and Brazil (BR) adult community centers, 570 respondents completed a tablet-based survey, including paired comparisons designed to assess EQ-5D-5L preference inversion. Specifically, the surveys included the same five pairs trading of Levels 4 and 5 from the original study (Mobility [MO], Self-Care [SC], Usual Activities [UA], Pain/Discomfort [PD], Anxiety/Depression [AD]) as well as three new pairs: only A, only D, and A+D. For example, the English-versions of the D pair asked ‘‘Which do you prefer: Starting today, 30 days with health problems: Extremely depressed or Severely depressed?’’ RESULTS: The samples ranged from 165 to 194 per pair. Similar to the previous study, 63 out of 140 Englishspeaking respondents preferred ‘‘Extremely anxious or depressed’’ over ‘‘Severely anxious or depressed’’ (US AD 45 %), but preference inversions were more common in Brazil (18 out of 25; BR AD 72 %). For the other four EQ-5D-5L domains, preference inversion was less common (US MO 8 %, SC 14 % %, UA 14 %, PD 20 %; BR MO 18 %, SC 38 %, UA 24 %, PD 57 %). Separating out the two components of AD had little or no effect on the prevalence of preference inversion (US only A 46 %, only D 45 %, A+D 41 %; BR only A 41 %, only D 64 %, A+D 54 %). CONCLUSIONS: The EQ-5D-5L preference inversion was confirmed in clinical populations, shown to be worse in the Portuguese translation, and was not reduced by separating out anxiety from depression or replacing the ‘‘or’’ with an ‘‘and’’ in the AD description. This evidence confirms the prediction that valuation studies of the English and Portuguese versions may find that Levels 4 and 5 have the same value. To mitigate such inversions, the labels could be revised or a 4-level version could be considered.

Poster Session 3002: Saturday Poster Session 2 Health Economics (3004) Face to face comparison between EQ-5D-5L and the traditional EQ-5D-3L in Catalan general population Marc Marti-Pastor, Resident physician in Preventive Medicine and Public Health, Teaching Unit of Preventive Medicine and Public Health, PSMAR-UPF-ASPB, Barcelona, Spain; Angels Pont, BSc, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain; CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Barcelona, Spain; Olatz Garin, PhD, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain; CIBER en Epidemiologı´a y Salu´d Pu´blica (CIBERESP), Spain; Universitat Pompeu Fabra, Barcelona, Spain; Ricard Tresserras, Direccio´ General de Planificacio´ i Recerca en Salut. Departament de Salut. Generalitat de Catalunya., Barcelona, Spain; Antonia Medina, Direccio´ General de Planificacio´ i Recerca en Salut. Departament de Salut. Generalitat de Catalunya., Barcelona, Spain; Oriol Garcia, Direccio´ General de Planificacio´ i Recerca en Salut. Departament de Salut. Generalitat de Catalunya., Barcelona, Spain; Montse Ferrer, MD, PhD, Health Services Research Unit, IMIM (Hospital del Mar Medical Research Institute), Barcelona, Spain;CIBER en Epidemiologı´a y Salud Pu´blica (CIBERESP), Spain;Universitat Auto`noma de Barcelona, Bellaterra, Spain., Barcelona, Spain AIMS: To assess to what extent expanding the number of levels would improve the index distribution and its ability to discriminate among health related known groups METHODS: Data were obtained

175 from 4 waves (January 2011 to July 2014) of the Catalan Health Interview Survey (Spain), including both EQ-5D versions (n = 7834). UK Time Trade Off original preference values were used to calculate the EQ-5D-3L index. For the EQ-5D-5L index, the crosswalk and the definitive UK value set were used. Indexes ranged from 1 (best health status) to negative values, -0.59 for 3L and its crosswalk 5L and, -0.21 for definitive 5L. EQ-5D versions were compared in terms of distribution (ceiling - floor effects and observed range), and discriminative capacity measured by effect size (ES, 95 % CI) to distinguish between known groups based on age, perceived health, self-reported chronic conditions, and limitation of activity. RESULTS: Ceiling effect of EQ-5D-3L and EQ-5D-5L was 62.6 and 61.6 %. Effect sizes (ES) between extreme categories of known groups for EQ-5D indexes generated with 3L and the two 5L versions were similar. For 15–44 vs C75 years old the ES was 0.23 (95 % CI 0.18, 0.27), 0.22 (95 % CI 0.18, 0.26) and 0.22 (95 % CI 0.17, 0.26); for excellent vs poor perceived health was 2.47 (95 % CI 2.3, 2.64), 2.68 (95 % CI 2.5, 2.85) and 2.67 (95 % CI 2.5, 2.85); for no selfreported chronic conditions vs 5 or more was 0.85 (95 % CI 0.78, 0.91), 1.21 (95 % CI 1.14, 1.28) and 1.03 (95 % CI 0.97, 1.1); for no limitation vs seriously affected on activities was 1.68 (95 % CI 1.79, 1.58), 1.91 (95 % CI 2.01, 1.8) and 2.36 (95 % CI 2.47, 2.25). CONCLUSIONS: Our results showed good discriminative capacity for both EQ-5D versions (3L and 5L) in the Catalan general population. Future studies that further examine the comparison between EQ-5D-3L and 5L in clinical samples are encouraged.

(3006) An examination of generic health status estimates in the context of spinal cord injury: is there a case for standardizing the framing of mobility-related questions? David G. Whitehurst, PhD, Simon Fraser University, Vancouver, BC, Canada; Nicole Mittmann, PhD, Sunnybrook Health Sciences Centre, Toronto, Canada; Vanessa K. Noonan, PT, PhD, Rick Hansen Institute, University of British Columbia, Vancouver, BC, Canada; Marcel Dvorak, MD, University of British Columbia, Vancouver, Canada; Stirling Bryan, PhD, University of British Columbia, Vancouver, BC, Canada AIMS: To explore the comparability of four generic preference-based health-related quality of life (HRQoL) instruments in a sample of individuals living with a spinal cord injury (SCI). METHODS: Community-dwelling individuals living with SCI were invited to complete a web-based cross-sectional survey. The survey comprised questions that were split into five sections; (i) demographics, (ii) SCI classifications and characteristics, (iii) secondary health complications and conditions, (iv) quality of life, and (v) functioning in activities of daily living. Four generic preference-based health status classification systems were included in the survey (randomized to prevent order-effect bias); Assessment of Quality of Life 8-dimension questionnaire (AQoL-8D), EQ-5D-5L, Health Utilities Index (HUI) and SF-6D (derived from the SF-36v2). In addition to descriptive comparisons of index scores and dimension responses, pairwise empirical analysis [that is, comparison of two instruments at a time) explored dimension-level correlation (Spearman correlation matrix) and absolute agreement (two-way, mixed model intraclass correlation coefficient (ICC)]. RESULTS: Three hundred sixty-four individuals (59 % of all invitees) completed the survey. Mean index scores were 0.164 (EQ-5D-5L), 0.248 (HUI-3), 0.573 (AQoL-8D) and 0.605 (SF6D); the Fig. 1 illustrates index score variation across the sample. Agreement analysis identified two distinct pairs of instruments with ‘moderate’ agreement; ‘EQ-5D-5L & HUI-3’ (ICC = 0.670) and ‘AQoL-8D & SF-6D’ (ICC = 0.634). All remaining pairwise ICCs were below 0.220. A distinguishing feature between these two instrument pairs is any reference to respondents’ ability to walk (i.e.,

123

176

Qual Life Res (2015) 24:1–191 as hypothesized moderate correlated with physical aspects of SF-36 (PCS, role-physical, vitality, health transition), number of restricting symptoms reported, and EQ-VAS (qS = 0.31–0.39, p \ 0.001).SG was low correlated with mental aspect of SF-36, and EQ-5D (qS = 0.17–0.28, p \ 0.001). There was no evidence for an association between SG and HADS, echocardiographic measures, age, gender and educational level (qS = 0.01–0.06). CONCLUSIONS: SG is a practical and valid method of assessing utilities in patients with severe AS. SG takes into account something more than standard approached utilities.

(3010) A new valuation method: direct elicitation of personal utility functions for EQ-5D Nancy J. Devlin, Professor, Office of Health Economics, London, United Kingdom; Koonal K. Shah, Office of Health Economics, London, United Kingdom; Brendan Mulhern, University of Technology Sydney, Sydney, NSW, Australia; Ben van Hout, University of Sheffield, Sheffield, United Kingdom

Fig. 1 Ranked scatter graph of individual-level scores for the AQoL8D, EQ-5D-5L, HUI-3 and SF-6D (n = 364) present in the EQ-5D-5L and HUI-3, not present in the AQoL-8D and SF-6D). Spearman correlations identified a low (high) level of correlation between mobility-related items across (within) these two instrument pairs. CONCLUSIONS: Variation in the way that instruments permit individuals to describe their health state is not unique to the SCI context. However, in practice, for health care decision making, it is the health state values (rather than health state descriptions) that are most important. Further research is necessary to understand the sizeable differences in index scores observed in this study and, in particular, the implications of framing mobility questions in the context of respondents’ ability to walk.

(3008) Validity of standard gamble utilities in patients referred for aortic valve replacement Amjad I. Hussain, Medical doctor, Oslo University Hospital, Oslo, Norway; Andrew Garratt, Institute of Health Management and Health Economics, Oslo, Norway; Lars Gullestad, prof, Oslo University Hospital, Oslo, Norway; Kjell I. Pettersen, OUS, Oslo, Norway AIMS: From a theoretical perspective, the Standard Gamble (SG) is the most valid direct approach for assessing preferences in situations with uncertainty and risk. Hence SG has potential relevance for patients considered for aortic valve replacement (AVR). The aim of the study was to validate SG utilities in patients with severe aortic stenosis. METHODS: We enrolled all patients [ 18 years old referred for AVR and report the finding of 439 patients. In addition to routine clinical examination including echocardiography the Mini Mental Status (MMS) and SG were assessed through interviews and PROs (SF-36, EQ-5D, Hospital Anxiety and depression Scale (HADS), symptoms) were assessed by self-completed questionnaire. Practicality was assessed by means of completion rates. The construct validity of the SG was tested through comparisons with clinical markers of disease progression and scores for other instruments. We hypothesized that SG utility would have moderate correlations with physical and low correlations with mental aspects of health due to pathophysiological understanding of severe AS. No correlations were expected with echocardiographic measures of the aortic valve. RESULTS: The response rate for SG was 98 %. SG correlations were

123

AIMS: The end product of EQ-5D valuation studies is an algorithm describing, for a given population, the average utility decrements associated with each dimension and level within the EQ-5D. Standard methods for eliciting the preference data upon which these algorithms are based (TTO, discrete choice experiment, standard gamble, visual analogue scale) vary considerably in underlying approach but have in common an aim to ‘uncover’ people’s preferences by asking them evaluate a sub-set of EQ-5D states, then using their responses to infer their preferences over all dimensions and levels. An alternative approach, which has never been explored, is to ask people to construct their own personal utility functions. Instead of asking people to value a selection of states, this approach involves directly asking people about the relative importance to them of dimensions, levels and interactions between them. The aim of this study is to develop a new stated preference approach for directly eliciting personal utility functions, and to test its feasibility and acceptability in valuing EQ5D. This presentation will outline the concept underpinning our approach, and report early results from piloting. METHODS: We developed a questionnaire designed to directly elicit personal utility functions from general public respondents via face-to-face interviews. The approach is grounded in utility theory, and informed by swingweighting techniques in the multi-criteria decision analysis literature. The anchoring of dead = 0 is achieved by an innovative trading task which locates the individual’s region of 0 in the descriptive system by an iterative trading process using the dimensions/levels. Respondents are encouraged to reflect on their answers, using validation tasks, and to change any previous responses. The approach is based on the premise that individuals construct their preferences in response to stated preference studies. The questionnaire was piloted using convenience samples in England, Netherlands and Australia in 2015. RESULTS: Piloting work demonstrated the acceptability of the questions and has highlighted ways of improving the method to reduce interviewer/respondent burden. CONCLUSIONS: The approach is a feasible way of eliciting health state preferences, and arguably has some important advantages over current approaches. We outline the research now required to further refine and operationalise the method.

(3012) Anchoring bias in EQ-5D valuation based on face-to-face time trade-off interviews Kim Rand-Hendriksen, Cand.psychol, PhD, Akershus University Hospital; University of Oslo, Oslo, Norway; Liv Ariane Augestad, MD, PhD, Univeristy of Oslo; Akershus University Hospital, Oslo,

Qual Life Res (2015) 24:1–191 Norway; Nan Luo, PhD, National University of Singapore, Singapore, Singapore AIMS: EQ-5D is the most used preference-based instrument worldwide for measuring health-related quality of life for the estimation of quality-adjusted life-years (QALYs). Values for EQ-5D health states are determined by modeling data collected from general population surveys using the time trade-off (TTO) method. In TTO, respondents are given a sequence of choice tasks, in which a fixed length of life in the target state is compared to varying lengths of full health, until preferential indifference is reached. In a web-based TTO performed in 2010, we found that the starting point for the variation of the shorter life greatly influenced resulting values, an effect typically described as anchoring bias. Anchoring bias has been found to be greater in situations with low engagement, low familiarity, and difficult questions. The objective of this study was to assess the influence of the starting point and the subsequent indifference procedure on elicited TTO values in face-to-face interviews. METHODS: 200 Norwegian and 200 Singaporean respondents were interviewed face-toface and asked to value 10 EQ-5D-5L health states using TTO. Five starting points, corresponding to TTO values 0.8, 0.6, 0.4, 0.2, and 0.0, and two increments, 1 and 2 years, were used to present the varying lives. Anchoring bias was assessed using multiple linear regression with the end value as dependent variable and starting point value as the only predictor. Influence from increment size was assessed by comparing the standard deviations of TTO values by health state for the respondents with 1 vs. 2 year increment sizes. Value clustering was assessed visually. RESULTS: The impact of starting point on mean state values was non-significant in both the Norwegian and Singaporean study arms. Location and magnitude of value clustering varied by starting point and increment size. Larger increment sizes increased the spread of elicited values. CONCLUSIONS: The lack of observed anchoring bias is promising, as it suggests that face-to-face interviews are substantially better than webbased interviews in preventing premature indication of indifference. Different distributions by starting points and increment size suggest that the search procedure influences TTO-based EQ-5D values.

177 between the two TTO procedures, but the ‘non-stopping’ procedure was faster than the standard procedure (mean time per task: 0.79 min vs. 1.12 min, p \ 0.0001). Among subjects interviewed with the ‘non-stopping’ procedure, [40 % did not indicate any indifference point in the valuation tasks, and only a few exhibited response patterns without clear interpretation. In Singapore, the standard TTO procedure was found to be easy by a slightly larger proportion of subjects than the ‘non-stopping procedure (86 vs. 75 %, p = 0.0028), and subjects interviewed with the ‘non-stopping’ procedure gave more internally inconsistent values (mean occurrence per subject: 1.07 vs. 0.52, p = 0.0094) than those receiving the standard procedure. The ‘non-stopping’ procedure generated lower health-state utility values than the standard procedure in both Norway (mean: 0.358 vs. 0.418, p = 0.0049) and Singapore (mean: –0.014 vs. 0.033, p = 0.1015). CONCLUSIONS: The ‘non-stopping’ TTO is a feasible valuation procedure for EQ-5D-5L health states that could be more efficient than the standard procedure. Health-state utility values are sensitive to elicitation procedure and culture.

(3016) Determinants of health preferences in the Canadian EQ5D-5L Valuation Study Fatima Al Sayah, PhD, University of Alberta, Edmonton, AB, Canada; Nick Bansback, PhD, University of British Columbia and Centre for Health Evaluation and Outcomes Sciences, Vancouver, BC, Canada; Stirling Bryan, PhD, University of British Columbia, Vancouver, BC, Canada; Arto Ohinmaa, University of Alberta,

Table 1 Beta (b) coefficients from adjusted  multivariable linear regression random effects models for the associations of explanatory variables with TTO scores (N=1209)

(3014) Testing the ‘non-stopping’ time trade-off procedure for valuation of the EQ-5D-5L health states: a study of two countries Nan Luo, PhD, National University of Singapore, Singapore, Singapore; Kim Rand-Hendriksen, Cand.psychol, PhD, Akershus University Hospital; University of Oslo, Oslo, Norway; Liv Ariane Augestad, MD, PhD, Univeristy of Oslo; Akershus University Hospital, Oslo, Norway AIMS: In the standard time trade-off (TTO) task for valuation of EQ5D-5L health states, the respondent must state indifference at one and only one pair of the alternatives provided. We assessed the feasibility and acceptability of a new procedure in which the respondent is allowed to indicate no, one, or multiple indifferences. Because the procedure does not stop at the first indicated point of indifference, it is referred to as ‘non-stopping’ TTO. METHODS: The ‘non-stopping’ and the standard EQ-5D-5L TTO procedures were used to elicit the values of selected EQ-5D-5L health states from two different groups of subjects (n = 200 each) in both Norway and Singapore. A similar set of alternatives was used in the two procedures. In the standard TTO, only relevant alternatives were provided and the task stopped at the first stated indifference; in the ‘non-stopping’, all alternatives were provided in a standard order regardless of indifference statement. Propensity score matched samples were used to compare the performance of the two TTO procedures in both countries. RESULTS: In Norway, self-reported understanding and difficulty of the TTO questions and internal consistency in valuation were similar

123

178 Edmonton, Canada; Eleanor Pullenayegum, PhD, Toronto Hospital for Sick Children, Toronto, ON, Canada; Feng Xie, PhD, McMaster Univerity, Father Sean O’Sullivan Research Centre, St. Joseph’s Healthcare, Hamilton, ON, Canada; Jeffrey A. Johnson, PhD, University of Alberta, Edmonton, AB, Canada AIMS: To identify determinants of preferences for hypothetical health states elicited using time-trade off (TTO) in the Canadian 5 level EuroQol 5-dimensions (EQ-5D-5L) Valuation Study. METHODS: Data were taken from the Canadian EQ-5D-5L Valuation Study, which took place in Hamilton, Vancouver, Edmonton, and Montreal. Using the TTO approach, each respondent valued 10 health states out of 86. TTO score of the valued health states was the outcome measure in this analysis. Other data collected included age, sex, marital status, education, employment, annual household income, ethnicity, country of birth, dwelling, study site, self-reported height and weight (used to calculate body mass index), health literacy, multimorbidity (presence of C 2 of 15 self-reported chronic conditions), previous experience with illness, and the EQ visual analogue scale (VAS). To examine the independent associations of the explanatory variables with the TTO score, we fitted multivariable linear regression models with random effects adjusted for the severity level of health states and for inconsistencies in valuations. RESULTS: Average age of respondents (n = 1209) was 48 (SD = 17) years, 45 % were male, 53 % married, 50 % employed, 63 % had more than high school education, 29 % had an income [$75,000, 82 % lived in urban settlements, and 75 % were born in Canada. Only 6.9 % had inadequate health literacy, 31 % reported [ = 2 chronic conditions, and 83 % had previous illness experience. The mean EQ-VAS score was 82.3 (SD = 14). In adjusted models, older age (beta = -0.077), male sex (beta = 0.042), being married (beta = 0.069), and urban dwelling (beta = -0.055) were significantly associated with preferences for hypothetical health states (Table 1). Additionally, participants from Edmonton (beta = -0.124) and Vancouver (beta = -0.156), but not those from Hamilton, had significantly lower preferences than those from Montreal. CONCLUSIONS: Overall, socio-demographic characteristics were the main determinants of preferences for hypothetical health states in the Canadian EQ-5D-5L valuation study; while, factors such as multimorbidity, health literacy and previous illness experience had no impact on these preferences. Additionally, preferences were significantly lower in western Canadian cities compared to eastern ones. These results bring into question whether a single utility algorithm is suitable for use in all parts of Canada, or region (western, eastern) specific algorithms estimated from region-specific preferences need to be developed.

(3018) Exploring the role of disease labels on general population preferences Helen McTaggart-Cowan, Canadian Centre for Applied Research in Cancer Control, Vancouver, BC, Canada; Dean A Regier, PhD, Canadian Centre for Applied Research in Cancer Control, Vancouver, BC, Canada; Stuart J Peacock, DPhil, Canadian Centre for Applied Research in Cancer Control, Vancouver, BC, Canada AIMS: Difficult policy decisions are increasingly being based on general population preferences for hypothetical health states. For generic instruments, the states are generally not labelled by disease types to avoid negative perceptions being incorporated in the preferences. Disease-specific utility instruments tend to identify the disease, thereby incorporating any societal perceptions about that disease into the valuation. This study explores whether labelling a health state affects general population valuations. METHODS: An online panel of general population respondents (n = 800) were recruited to complete a discrete choice experiment (DCE). The DCE

123

Qual Life Res (2015) 24:1–191 was designed to evaluate individuals’ preferences for different health scenarios pertaining to colorectal cancer, type 2 diabetes, and rheumatoid arthritis (RA). Half the participants were randomly selected to complete a four-attribute DCE (health state before treatment, health state after treatment, duration of life, and disease type). The remaining participants completed an identical DCE except that disease type was excluded. For these respondents, after they completed each choice set they were asked if they would change their answers if the corresponding disease label was applied. For each DCE version, a conditional logit model was used to determine the marginal utility weight for the attribute levels, from which trade-offs and marginal rates of substitution were calculated. RESULTS: Relationships between personal utility and all attributes were in the hypothesized direction. There were no statistically significant differences in demographic characteristics between those completing either DCE version. When adjusted for the two versions, the labels associated with colorectal cancer (beta = -0.40, SE = 0.04, p \ 0) and type 2 diabetes (beta = 0.28, SE = 0.04, p \ 0) had an effect on the responses relative to RA. On average, the respondents were willing to forego 1.58 life-years for a health state to be labelled as diabetes rather than RA. Similarly, they would be willing to forego 2.25 life-years for RA rather than colorectal cancer. CONCLUSIONS: The provision of disease labels influenced general population respondent’s valuations for a given health state. There was a clear ordering effect such that respondents preferred to have diabetes over RA, and for RA over colorectal cancer. This raises concerns regarding the use of disease-specific utility instruments for guiding difficult policy decisions.

(3024) Quality of life in diabetes mellitus type I adults in Brazil Monica Cintra, Medical coordinator, HC-FMRP- USP, Ribeira˜o Preto, Brazil; Stella Carletti, MD, INC, Rio de Janeiro, Brazil; Andre´a Libo´rio Monteiro, HTA Ms MSc, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil; Marcelo Goulart, INC, Rio de Janeiro, Brazil; Bernardo Tura, MD PhD, Instituto Nacional de Cardiologia, Rio de Janeiro, Brazil; Marisa Santos, MD, PhD, Instituto Nacional de Cardiologia, Rio de Jeneiro Area, Brazil AIMS: Diabetes Mellitus Type I (DMI) is a chronic, highly prevalent condition, often associated with complications that may cause a loss in quality of life (QoL) and increase mortality among this population. The aim of this study was to assess Qol of DMI patients using EQ-5D3L. METHODS: A multicentric cross-sectional study was conducted in patients, older than 18 years, followed in secondary care centers in Rio de Janeiro, Fortaleza and Porto Alegre, Brazil. They answered a clinical sociodemographic questionnaire and the EQ-5D-3L, between January 2012 and May 2013. Spearman correlation test, Mann– Whitney U and Kruskall-Wallis were performed in order to verify the association between the variables and EQ-5D index values. Variables that displayed an association with p value lower than 0.25 were included a linear regression model using forward stepwise method. Statistical analysis was performed using R 3.1.1. RESULTS: Of 220 patients included 113(51 %) were female, with median age 27 years, median time of disease 13 years and median body mass index 23.8. Regarding treatment, 73 % of the patients used NPH insulin, 25 % used an insulin analogue, 81 % followed a restrict diet and 28 % had hypertension. In the last 90 days before the interview, 94 % of the patients reported hypoglycemic episodes, among them 35 % needed assistance to recover from the episode. They present some complication: neuropathy 42 %, retinopathy 38 %, nephropathy 35 %, hypothyroidism 33 % and depression 33 %. The utility scores derived from the descriptive system were 0.79 ± 0.104 in females and 0.845 ± 0.069 in males. The Visual analogue scale (VAS) were 70.71

Qual Life Res (2015) 24:1–191

179

Table 1 Scores in EQ-5D-3L according to complications or not

Utilities

Comorbities

(3028) Time Trends in the Health Utilities Index Mark 3 attributes in the general population of Canada, 2000–2010

Nephropathy

EQ-5D-3L index

VAS

No

Yes

P

No

Yes

p

0.88

0.82

\0.001

75

70

[0.05

Neuropathy

0.88

0.77

\0.001

80

60

\0.001

Retinopathy

0.88

0.80

\0.001

75

70

\0.001

in females and 72.92 in males. No significant differences in the utility values were found between patients that reported and didn´t report to have had hypoglycemic episodes (0.833 and 70 in VAS, in both groups). Mean VAS scores varied according the frequency of hypoglycemic episodes, 60.0 in weekly episodes and 70.0 in monthly episodes, with utility values of 0.77 and 0.83 respectively. Patients that presented any complications had lower QoL, as showed in Table 1. CONCLUSIONS: Complications presented by DMI patients’ impact in QoL and hypoglycemic episodes don´t. This can help physicians to choose the best way to treat patients.

(3026) Does EQ-5D-5L account for the impact of limitations in activities of daily livings (ADLS) on social care outcomes, capabilities and well-being? Clara Mukuria, PhD, University of Sheffield, Sheffield, United Kingdom; John Brazier, PhD, University of Sheffield, Sheffield, South Yorkshire, United Kingdom; Tessa Peasgood, PhD, University of Sheffield, Sheffield, United Kingdom AIMS: There have been concerns about using EQ-5D in those most in need of social care (as opposed to health interventions). This study examines the extent to which ADLs explain variation in social care outcomes (ASCOT), capabilities (ICECAP-O) and well-being (life satisfaction and WEMWBS) over and above EQ-5D-5L. METHODS: Individuals (n = 1749) who were over 65 completed a survey including EQ-5D-5L, ASCOT, ICECAP-O, WEMWBS and life satisfaction as well as problems with activities of daily living (ADL) in 7 areas (getting around indoors, getting up, feeding, dressing, bathing, using the toilet and managing finances). Regression analysis was undertaken to test the relationship between these ADLs and the wellbeing, capability and need measures while controlling for: health as measured by the EQ-5D-5L, age, gender and marital status. RESULTS: Mean (SD) age was 72.9 (5.84), 51 % were female and 73.3 % had a long-term condition (73.3 %) yet only 16.6 % reported problems in one or more ADL. Mean (SD) EQ-5D-5L scores were (0.81 (0.19); IQR: 0.744–0.941). Problems with finances were negatively associated with all the measures (p \ 0.001). Respondents with problems in the other ADLs also had lower scores in ICECAP-O, ASCOT and WEMWBS apart from problems with getting in/out of bed and using the toilet. When health was taken into account, managing finances remained negative and statistically significant in all measures. Problems with getting around indoors, feeding, dressing and bathing remained negative in the ASCOT regressions while this was not the case for ICECAP-O or WEMWBS. Life satisfaction and WEMWBS had weaker associations with the ADLs with some counter-intuitive results. CONCLUSIONS: These results indicate that alternative measures such as the ASCOT may be able to capture the impact of ADLs over and above EQ-5D-5L including activities related to mobility and self-care. This may indicate that such problems have an impact other aspects of life such as independence, control or need that are not reflected by EQ-5D-5L.

Jacek A. Kopec, University of British Columbia, Vancouver, Canada; Eric C. Sayre, Arthritis Research Canada, Vancouver, Canada; Philippe Fines, Statistics Canada, Montreal, Canada; Anya Okhmatovskaia, McGill University, Montreal, Canada; William M. Flanagan, Statistics Canada, Ottawa, Canada; Michael C. Wolfson, University of Ottawa, Ottawa, Canada AIMS: In modeling changes in health-related quality of life in a population, an important question is how to deal with a potential time effect that is not explained by other variables (residual effect). The aim of this study was to determine if significant time effects were present when modeling the evolution of the Health Utilities Index Mark 3 (HUI3) in the general population of Canada in the period 2000–2010 after adjusting for key socio-demographic and health variables. METHODS: The HUI3 is comprised of 8 components (health attributes): vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain, each measured on a 5 or 6-level scale. We developed a prediction model for each attribute from the longitudinal National Population Health Survey (1994–2010, n = 17,276) using ordinal logistic regression with proportional odds. Predictors included concurrent values of selected HUI3 attributes, age, sex, education, income, smoking, BMI, and selected self-reported chronic conditions. Odds ratios reflect the effect on the odds of being in a worse level of a given attribute over a period of 10 years. The time variable was log-transformed. Predictive utility was assessed with the AUC statistics. RESULTS: We observed significant, unexplained trends (OR \ 1 indicates improvement over time) for cognition (OR = 1.76, 95 % CI 1.63–1.89), speech (OR = 2.39, 1.29–4.43), pain (OR = 1.16, 1.05–1.28), and dexterity (OR = 0.61, 0.40–0.94), borderline significant for hearing (0.85, 0.72–1.00) and no effect for vision (OR = 1.00, 0.93–1.08), emotion (OR = 1.06, 0.97–1.16), and mobility (OR = 1.19, 0.96–1.47). For all attributes, the most important determinant of health at a given point in time was previous levels of the attribute in question. Adjusting for the previous levels of the attribute (2 and 4 years previously) did not result in substantive changes in the size of the effects. CONCLUSIONS: During the period 2000–2010, we observed residual (unexplained) trends for improvement in hearing and dexterity, deterioration in cognitive function, speech, and pain and no residual time effect for mobility, vision and emotional health. These effects may need to be taken into account when modeling future levels of health in Canada.

(3030) Performance of the EQ-5D-5L in a large community-based survey Yash J. Jalundhwala, MS, University of Illinois at Chicago, Chicago, IL, United States, A. Simon Pickard, PhD, University of Illinois at Chicago, Chicago, IL, United States AIMS: The aims of this study were: (1) to assess the performance of the EQ-5D-5L in a large community-based survey; (2) to compare EQ-5D-5L summary scores based on a UK value set (5L-UK), 5L crosswalk based on 3L values (5L-XW) and equal weighting (5LEWS). METHODS: A secondary analysis was conducted using crosssectional data from a mail-in wellness survey in community residents of Kirklees, UK (n = 10,422). The discriminative ability of the EQ5D-5L was examined using known groups comparisons based on selfrated health status, age and medical conditions. Construct validity of the EQ-5D-5L was examined using strength of correlation (strong defined as r = [0.50) between other measures of related health constructs and EQ-5D-5L dimensions and summary scores. The relative efficiency (RE) of summary scores was compared using

123

180 F-statistic ratios from known group comparisons, with 5L-XW as the referent. RESULTS: All the levels of each dimension of the EQ-5D5L were utilized by the respondents, although few respondents reported severe (1.2–6.6 %) or extreme (0.4–1.8 %) problems in any dimension, and 31.7 % reported no problems in any dimension. As hypothesized, significantly lower 5L-UK scores were observed in respondents who were: (i) older age (ii) obese, (iii) current smokers, (iv) had multiple chronic conditions, (v) poorer health status (p \ 0.001). All EQ-5D dimensions showed strong correlations (r = 0.56–0.73) with related health measures. Health states were valued as negative (worse than dead) in twice as many respondents using 5L-XW scores (1.8 %) compared to 5L-UK scores (0.6 %). Compared to the 5L-XW, the relative efficiency of 5L-UK and EWS was equal or higher (Age: RE5L-UK = 0.96, REEWS = 1.36; BMI: RE5L-UK = 1.00, REEWS = 1.0; Number of prescription medications: RE5L-UK = 1.00, REEWS = 1.18; Number of Chronic conditions: RE5L-UK = 1.00, REEWS = 1.61; General Health: RE5LUK = 1.13, REEWS = 1.20) CONCLUSIONS: These results provide evidence to support the validity of EQ-5D-5L in large community based surveys. An EQ-5D-5L UK value set appears to have similar or greater statistical efficiency than the crosswalk to detect group differences, and for non-QALY purposes, use of EWS or a ‘‘misery index’’ provides greater statistical efficiency than value set weights.

(3032) Developing and testing national tariff for EQ-5D-5L in Japan with special reference to statistical methods to obtain QOL score Shinya Saito, Medicine, Okayama University, Okayama, Japan; Takeru Shiroiwa, Pharmacy, National Institute of Public Health, Wako, Saitama, Japan; Shunya Ikeda, Medicine, International University of Health and Welfare, Ohtawara, Tochigi, Japan; Shinichi Noto, Health Care, Niigata University of Health and Welfare, Niigata, Japan; Ataru Igarashi, Pharmacy, The University of Tokyo, Tokyo, Japan; Takashi Fukuda, Health Care, National Institute of Public Health, Wako, Saitama, Japan; Kojiro Shimozuma, Medicine, Ritsumeikan University, Kusatsu, Shiga, Japan AIMS: Since health economic evaluation is supposed to be introduced to the Japanese heath care system in 2016, we have developed and tested national tariff in Japan to meet with the immediate demand for EQ-5D-5L and corresponding value set. The EQ-5D-5L valuation study involved following two models to create a scoring algorithm; composite time-trade off (cTTO) and a discrete choice experiment (DCE). To obtain the utility (QOL scores), the scores derived from DCE must be anchored to the scale from 0 (death) to 1 (complete health). However, the statistical methods and models for converting the DCE scores to the utility are not yet standardized. In this study, we present the development of Japanese DCE value set of EQ-5D-5L and comparison among three methods for converting latent DCE values to the QOL scores. METHODS: 1,000 people from Japanese general public, calibrated by age, education and living area (rural/urban, east/ west) were recruited for interviews. Data collection was conducted through computer assisted face-to-face interviews, using Japanesetranslated version of EQ-VT software. To obtain and compare the value sets of EQ-5D-5L, the cTTO and DCE data were analyzed by a linear mixed model and conditional logit, respectively. The DCE scores were converted to the QOL scale by (i) anchoring to the worst state using cTTO, (ii) mapping DCE onto cTTO, and (iii) a hybrid model. RESULTS: The data from 1,026 respondents were analyzed which collectively yielded 14,364 cTTO data and 7,182 DCE data. For EQ-5D-5L based on cTTO, the mapping and hybrid methods yielded very similar scoring coefficients. The DCE anchored to the worst state overestimated the cTTO scores of almost all health states.

123

Qual Life Res (2015) 24:1–191 The coefficients were entirely consistent throughout the analyses. CONCLUSIONS: We developed Japanese value set based on DCE. Comparing the predicted and observed cTTO scores, the hybrid model was slightly superior to the simpler methods. In practice, however, the value sets derived from the mapping methods were very similar. Although the standard method cannot be decided at this time, we demonstrated the properties and feasibility of each model.

(3034) German utility weights for the EORTC QLQ-C30: first results of a general population survey using discrete choice experiments Georg Kemmler, Statistician, Innsbruck Medical University, Innsbruck, Tyrol, Austria; Eva-Maria Gamper, Department for Nuclear Medicine, Medical University Innsbruck, Innsbruck, Austria; Virginie Nerich, Besanc¸on University Hospital, Besanc¸on, France; Richard Norman, Curtin University, Perth, Australia; Madeleine King, PhD, University of Sydney, Sydney, NSW, Australia; Bernhard Holzner, Medical University of Innsbruck, Innsbru¨ck, Austria AIMS: In view of the scarcity of financial resources in the health sector there is a growing demand to complement randomized clinical trials by health economic analyses (HEA), e.g., cost-utility analyses. To make the EORTC QLQ-C30, a widely used cancer-specific QOL instrument, applicable for HEA, utility weights for this instrument have to be determined. This is presently being done on a multi-national level within a cooperation between Australia (MAUCa Consortium) and Europe (EORTC). Aim of the present study was the determination of utility weights for the QLQ-C30 in Germany, the first EU country for which such data are available. METHODS: Discrete choice experiments (DCE) were used to value health states within a utility instrument derived from the QLQ-C30, the QLUC10D, comprising ten dimensions. The valuation task was run in an online panel of German adults. Recruitment and assessment were contracted to a company specialized in the conduct of surveys. Statistical analysis was performed using conditional logistic regression. Representativity of the sample was tested by comparison with the German national statistics. RESULTS: 1062 subjects were recruited. 1002 (94.3 %) completed all 16 choice sets of the DCE. Mean age was 46.5 years (range 18–82), 60.9 % were male. There were significant deviations from the German national statistics with respect to gender and education, but results were not strongly affected by these two variables. Generally the DCE approach was well accepted by the respondents. Logistic regression revealed that mobility had the largest impact on the estimated utilities, followed by pain and role functioning. Among cancer-specific domains, nausea showed the largest effect. CONCLUSIONS: Findings indicate that DCEs are a feasible tool for the determination of utility weights for the QLQ-C30 and yield meaningful results. Ongoing valuation studies for the QLQ-C30 in other European countries will allow comparisons across countries and a more detailed assessment of the validity and usefulness of the DCE approach for the determination of cancer-specific utility weights.

(3036) Valuing the SF-6D Version 2 in Canada Nick Bansback, PhD, University of British Columbia and Centre for Health Evaluation and Outcomes Sciences, Vancouver, BC, Canada; Brendan Mulhern, University of Technology Sydney, Sydney, NSW, Australia; Richard Sawatzky, PhD RN, Trinity Western University, Langley, BC, Canada; John Brazier, PhD, University of Sheffield, Sheffield, South Yorkshire, United Kingdom; David G. Whitehurst, PhD, Simon Fraser University, Vancouver, BC, Canada

Qual Life Res (2015) 24:1–191 AIMS: To estimate Canadian population-based preference weights for the SF-6D Version 2 (SF-6Dv2), a revised health state classification system derived from the SF-36. METHODS: Members of the Canadian general population were recruited to complete a web survey in English or French. All participants completed a set of 10 discrete choice experiment (DCE) questions, which comprised two SF-6Dv2 health states descriptions (with the addition of a duration attribute for both states). The SF-6Dv2 covers 6 dimensions of health, each with 5 or 6 levels of severity: physical, role and social functioning, pain, mental health and vitality. The descriptive system defines 18750 unique health states. A total of 300 unique DCE pairs were valued. Results were modelled using conditional logit regression, where each dimension level was interacted with duration, with the coefficients used to estimate a preference index such that full health and death were valued at 1 and 0, respectively. RESULTS: One thousand and seventeen respondents have completed the first phase of data collection, with 16 % completing in French. Respondents were broadly representative of the Canadian population in terms of age, gender and geographical location, and feedback from respondents was generally positive. Preliminary models find slight disordering of coefficients in levels 2 and 3 for some dimensions, although none of these disorderings were statistically significant. The models find pain to be the largest driver of utility weights (level 6 = 0.207, p \ 0.001) followed by mental health (level 5 = 0.095, p \ 0.001) and physical functioning (level 5 = 0.086, p \ 0.001). In comparison to previous scoring algorithms for the original SF-6D, the model predicts a wider range of utility values. CONCLUSIONS: Our first phase of data collection for the SF-6Dv2 in Canada provides evidence that the survey is working properly but identifies a requirement for a larger sample size to estimate values for mild impairment levels within dimensions. Further data collection is planned, and the resulting preference weight estimates will provide the first time Canadian utility values can be obtained from the SF-36 in Canada. How Canadian preference weights from the VR-12 and SF-12 can be estimated using results of this survey will also be described.

(3038) Instrument-defined estimates of minimally important difference in EQ-5D-5L index score for Canadian, Chinese and Spanish scoring algorithms Nathan S. McClure, MSc, University of Alberta, Edmonton, AB, Canada; Fatima Al Sayah, PhD, University of Alberta, Edmonton, AB, Canada; Feng Xie, PhD, McMaster Univerity, Father Sean O’Sullivan Research Centre, St. Joseph’s Healthcare, Hamilton, ON, Canada; Nan Luo, PhD, National University of Singapore, Singapore, Singapore; Jeffrey A. Johnson, PhD, University of Alberta, Edmonton, AB, Canada AIMS: The EQ-5D-5L is a generic indirect measure of health state utility developed by the EuroQol Group. The aim of this study is to estimate the minimally important difference (MID) for the EQ-5D5L index score using instrument-defined single state transitions for recently introduced scoring algorithms including Canada, China and Spain. METHODS: Estimates of the minimally important difference are presented for all 3125 possible unique health states as the difference in the index score for the new state resulting from a single state transition compared to the index score for the baseline EQ-5D5L state. For comparison, results are also provided with maximum valued scoring coefficients for each dimension excluded. The maximum valued scoring coefficient for a dimension was identified for each country specific scoring algorithm as the largest index score decrement/increment associated with a single transition to a worse/better level (e.g., for the Canadian scoring algorithm, transitions between level 3 and level 4 on any dimension were excluded).

181 Table 1 Summary statistics of instrument defined minimally important difference (MID) estimates for EQ-5D-5L country-specific scoring algorithms

Fig. 1 Instrument-defined estimates of minimally important difference (MID) as a function of baseline EQ-5D-5L index score for country-specific scoring algorithms. Blue open circles and solid lines are for MID estimates obtained from all 25,000 single state transitions, while orange stars (*) and dashed lines are for MID estimates with maximum values transitions within a dimension excluded. Red lines represent the summarized mean MID estimate, and black lines are generated from a loess curve RESULTS: Estimates of the minimally important difference for the EQ-5D-5L index score are generally between 0.03 and 0.07 (see Table 1). Excluding maximum valued scoring coefficients within each dimension decreases the summarized mean and median of minimally important difference estimates, and can also affect the shape of the distribution as a function of baseline index score (see Fig. 1). CONCLUSIONS: Estimates of the minimally important difference in EQ-5D-5L index score provided by this study are in general agreement with estimates of minimally important differences for other preference-based measures of health related quality of life. This work supports the development of minimally important difference estimates for interpretation of EQ-5D-5L health state utility scores.

Health Care Providers (3040) Development of a feedback form for routine clinical use of PROs in primary care for patients with multimorbidity Ian Porter, Dr, University of Exeter Medical School, Exeter, Devon, United Kingdom; Jose M Valderas, MD, PhD, MD, University of Exeter, Exeter, United Kingdom AIMS: Patient Reported Outcome (PRO) measures have the potential to improve quality of life and promote patient centred care, but are currently under-utilised in Primary Care. For patients with

123

182 multimorbidity they could be helpful in setting priorities and planning care. In order to do this individualized PRO feedback needs to be developed which is relevant and acceptable to both patients and clinicians, encompassing multiple conditions. The aim of this project is to develop such feedback and incorporate it into existing routine Primary Care appointments, such as annual reviews, for patients with chronic conditions. METHODS: Separate individualized feedback forms for patients and health professionals in Primary Care were developed (following ISOQOL guidelines) encompassing 6 chronic conditions (asthma, COPD, diabetes, heart failure, depression, and osteoarthritis of the hip and/or knee) incorporating 3 different types of PRO measure, including both standardized (condition specific and generic) and individualized measures. Development of the feedback forms was informed by a systematic review of clinical guidance available in the UK for each of the conditions in relation to PROs. Information extracted from this guidance was incorporated into the feedback form. RESULTS: Two separate feedback forms have been produced, informed by UK clinical guidance, for 6 chronic conditions encompassing standardized and individualized PROs: (1) For patients: a summary incorporating multiple PRO scores using a visual analogue scale with colour graded background (worst to best) for standardized measures, plus a list of 5 health priorities identified by patients using an individualized PRO measure. (2) For health professions: 3 level feedback incorporating (a) an explanation of what the scores mean (b) how these scores relate to population level data (c) clinical guidance linking PRO scores to specific management options or quality of life information. Examples of the both types of feedback form will be included on the poster. CONCLUSIONS: This marks the first time individualized PRO feedback has been developed and implemented for patients with multiple chronic conditions in Primary Care. A mixed methods pilot evaluation of the feasibility of an intervention centred around the use of the feedback form is currently underway in family practices in England.

(3042) Presenting PRO results from comparative research studies in clinical practice to inform treatment decision-making: making pictures speak more clearly to patients and clinicians Michael Brundage, MD, MSc, Kingston General Hospital, Queen’s University, Kingston, ON, Canada; Elissa Bantug, MHS, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States; Elliott Tolbert, Mr., Johns Hopkins School of Medicine, Baltimore, MD, United States; Katherine C. Smith, PhD, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States; Claire Snyder, PhD, Division of General Internal Medicine, Johns Hopkins School of Medicine, Baltimore, MD, United States, PRO Data Presentation Stakeholder Advisory Board AIMS: Patient-reported outcomes (PROs) from comparative research studies (e.g., clinical trials) provide data on the patient-centered impacts of different treatment options, thereby informing decisionmaking. Graphical presentation of PRO results (e.g., line graphs of mean scores, bar charts of proportion responding to treatment) may facilitate patient and clinician understanding of PRO data; however, our previous research showed that patients’ and clinicians’ preferences differ for how the data are summarized. Clinicians value more details (e.g., p values, confidence intervals), but patients find this information confusing. Participants suggested improvements to (1) clarify the meaning of the scores (whether they are good or poor, what scores are ‘normal’), (2) convey whether between-group differences are important (statistically, clinically), and (3) simplify terminology. We engaged with stakeholders to improve formats for presenting comparative study PRO results to patients and clinicians, respectively. METHODS: Using the previous study’s results and partnering with Stakeholder Work Groups (3 patients and 5 clinicians who participated in the previous study), we developed candidate formats to

123

Qual Life Res (2015) 24:1–191 address the issues noted above. To clarify score meaning, we tested shading the range of normal scores and descriptive labeling of the y-axis (e.g., ‘‘none, mild, moderate, severe’’). To convey important differences between treatments, we tested approaches using asterisks and shading (and confidence limits for clinicians). We also replaced confusing terms with simpler ones (e.g., ‘‘start treatment’’ for ‘‘baseline’’). We then conducted one-on-one in-depth interviews to obtain feedback on these candidate approaches. RESULTS: For patient-focused presentation approaches, we purposively sampled 20 patients (mean age 66; 55 % female; 25 % high school graduate) and 10 clinicians (median: age 41; years practicing 14). Key findings included: (1) descriptive y-axis labels, but not shading the normal range, helped participants interpret score meaning; (2) participants were split on the importance of understanding clinical/statistical significance, but asterisks were preferred for this purpose; and (3) identification of terms to effectively explain scientific language. Findings for clinician-focused presentation approaches were comparable and will be presented. CONCLUSIONS: The improved formats that emerged from this stakeholder-driven process will be further evaluated in larger studies to inform best practices for presenting PRO data to patients and clinicians to inform treatment decision-making.

(3044) Presenting individual patient PRO data in clinical practice: making line graphs ‘‘in Line’’ with patient and clinician understanding Claire Snyder, PhD, Division of General Internal Medicine, Johns Hopkins School of Medicine, Baltimore, MD, United States; Katherine C. Smith, PhD, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States; Emily Little, MPH, Johns Hopkins School of Medicine, Baltimore, MD, United States; Elissa Bantug, MHS, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States; Michael Brundage, MD, MSc, Kingston General Hospital, Queen’s University, Kingston, ON, Canada, PRO Data Presentation Stakeholder Advisory Board AIMS: Using patient-reported outcomes (PROs) in clinical practice for individual-patient monitoring requires that patients and clinicians understand the PRO scores’ meaning. In previous research, we found that patients and clinicians prefer line graphs depicting a patient’s PRO scores over time, but several issues needed to be addressed: (1) the variation in scaling and scoring (e.g., whether higher scores are better or worse) and (2) how to highlight potentially concerning scores for clinical attention. We undertook an iterative, stakeholderengaged process to develop candidate approaches for presenting PRO scores to address these issues. METHODS: Based on the previous study’s results, the research team developed candidate presentation formats, which were refined based on feedback from a Stakeholder Work Group (6 patients, 2 clinicians) who participated in the previous study and volunteered to collaborate with the research team in developing improved presentation formats. The formats tested to address inconsistent scoring directionality included descriptive labeling of the y-axis (e.g., none, mild, moderate, severe), shading areas of the graph (e.g., green for good, red for bad), and presenting results such that higher scores were always better. To highlight scores for clinical attention, we tested various line graph markers (circles, exclamation points), a threshold line for potentially concerning scores, and shading. We then conducted one-on-one in-depth interviews to obtain feedback on these candidate approaches. RESULTS: We purposively sampled 19 patients (median age 62; 42 % female; 79 % White; 21 % Bhigh school graduate) and 10 clinicians (median age 56; 30 % female; median 30 years in practice). Key findings included (1) descriptive y-axis labels helped participants understand the meaning of the scores; (2) shading with a single color (red or green) was preferred over multi-color formats (red, yellow, green); (3)

Qual Life Res (2015) 24:1–191 participants were split on whether presenting scores with consistent direction (e.g., higher always better) was helpful; and (4) using circles/lines to highlight scores for clinical attention was preferred over shading options. CONCLUSIONS: Engaging with stakeholders, we developed and tested approaches to improve the presentation of individual-patient PRO data. These approaches will be tested in larger samples to identify best practices for presenting PRO data to promote understanding and use.

(3046) Evaluation of electronic patient-reported outcome data used in real-time with nurses and patients Kara Schick-Makaroff, PhD, University of Alberta, Edmonton, AB, Canada; Anita Molzahn, RN, PhD, FCAHS, University of Alberta, Edmonton, AB, Canada AIMS: There has been international movement towards patient-reported outcomes (PROs) through which a person may report on her/ his health-related quality of life. Electronic PROs (e-PROs) offer immediate access of such reports for healthcare providers. The purpose of this study was to evaluate the usefulness and impact of ePRO collection for patients and nurses. METHODS: A mixed methods study was conducted over 6 months during home dialysis outpatient clinics in two cities. Patients (n = 99) provided ePROs using tablets. Results were scored, printed, and given to nurses immediately before patient appointments. All clinic nurses (n = 11) participated and they were interviewed twice. Patients completed satisfaction items from the Comox Valley Nursing Centre questionnaire following their appointments. RESULTS: Nurses reported that receiving ePRO data in real-time informed their clinical decision-making and interdisciplinary team follow-up. The visual display of ePRO results impacted their interpretation of data. The nurses believed that the ePRO assessments helped identify patient issues that otherwise may have been missed, and that patients ‘‘felt listened to’’. Patients believed that they received excellent care (85.4 %), and that the nurses understood their needs completely (88.5 %). However, their satisfaction of care and health perceptions did not change when ePRO data was repeatedly provided to their nurses. CONCLUSIONS: Sharing ePRO data in real-time informed nurses’ practice, yet it did not impact patients’ perceptions of health or satisfaction of care. As PROs are adopted as routine practice, further work may be needed to provide guidance about how this data can enhance patient-centered care.

(3048) Weight bias among healthcare providers: quality of life implications for patients with obesity Mary Forhan, OT Reg(AB) PhD, University of Alberta, Edmonton, AB, Canada; Jacqueline Torti, MHP, University of Alberta, Edmonton, AB, Canada; Allyson Jones, PhD, University of Alberta, Edmonton, AB, Canada AIMS: Negative attitudes toward patients with obesity are associated with below standard quality health care that puts patients with obesity at risk for poorer quality of life compared to patients without obesity. Negative attitudes toward and beliefs about persons with obesity amongst health care professionals have been reported but little work has examined physiotherapist’s perception of patients with obesity. The objective was to determine the attitudes physiotherapists have toward patients with obesity and therapists’ beliefs about obesity as a health condition. METHODS: Physiotherapists registered with a provincial regulatory college responded to a request to complete an on-line version of the Attitudes Toward Obese Persons (ATOP) and Beliefs About Obese Persons (BAOP) questionnaires which are 20-item and 8-item companion questionnaires that use a likert-type response scale. The higher the ATOP score (range 0–120), the more

183 positive the attributes a person ascribes to obese persons. Higher BAOP scores (range 0–48) indicate a belief that the person is not solely responsible for their obesity. RESULTS: Three-hundred and ninety-one physiotherapists responded to the survey, which represent 17 % of physiotherapists licensed to practice in Alberta Canada. Ninety-five percent reported working with patients who had obesity. Participants reported the following types of practice: acute care (20 %), rehabilitation (19 %), homecare (8 %), private practice (33 %), other (20 %) with the majority (70 %) having 10 years or more of practice. The mean ATOP score was 72.52 ± 15.307; range (30–120) and BAOP mean score was 19.39 ± 7.639; range (1–48). CONCLUSIONS: Most physiotherapists reported working with patients who have obesity yet attitude and belief scores were indicative of weight bias amongst physiotherapists toward their patients who have obesity. Although a provincial strategy on obesity treatment has been implemented, discipline-specific education about the complex causes of obesity targeting beliefs about the controllability of obesity is recommended to improve overall attitudes toward patients with obesity and potentially improving the quality of care and associated quality of life for this population.

(3050) Developing tools for use in clinical practice: a methodological approach Chloe Tolley, Adelphi Values Ltd, Cheshire, United Kingdom; Rebecca Hall, Adelphi Values Ltd, Cheshire, United Kingdom; Kevin Mills, Adelphi Values Ltd, Cheshire, United Kingdom; Bryan Bennett, Adelphi Values Ltd, Cheshire, United Kingdom, Chris Marshall, Adelphi Values Ltd, Cheshire, United Kingdom AIMS: As patient-centric care becomes increasingly important, so do tools that can improve disease awareness, aid diagnosis and treatment decisions, and help with monitoring clinical outcomes. Tools used in clinical practice can include patient-reported outcomes (PROs), clinicianreported outcomes (ClinROs) and observer-reported outcomes (ObsROs) as well as screeners completed by patients, clinicians or observers. Historically, many tools currently used in clinical practice have been developed by clinicians based on their clinical experience, without much input from patients. However, well developed and validated tools can help ensure diseases are better identified, treated and managed from the outset. METHODS: Here, we provide an overview of the many uses of tools for clinical practice and a guide to developing such tools - based on a literature review and author experience. RESULTS: There are a number of important considerations when developing tools for use in clinical practice. Prior to development, an unmet need for the tool needs to be identified; features of a tool which end users (both clinicians and patients) would value most need to be considered; target end users need to be defined; and an appropriate mode of administration should be selected. Once drafted, tools should be pilot tested with end users and should be thoroughly debriefed. Validation of these tools should include assessment of inter-rater (and intra-rater) reliability and other measurement properties where appropriate. Receiver Operator Characteristic (ROC) analysis should be conducted to evaluate the discriminative properties of different iterations of the tool, aiding identification of the most discriminative items to support item reduction, development of scoring and identification of the most discriminative cut-points. CONCLUSIONS: Tools can be used in clinical practice to help identify specific conditions or co-morbidities early - allowing for prompt sign-posting and referral, confirmatory testing or intervention. They can help drive treatment decisions based on factors such as symptom severity, adherence drivers, preference, lifestyle and medication tolerability. Finally, when used collaboratively with the patient they can facilitate communication, allowing patients to provide a full overview of how they feel and function, how the disease affects their life and what is important to them – which can potentially improve adherence and treatment satisfaction.

123

184 (3054) The Distress Assessment and Response Tool (DART) at the Princess Margaret Cancer Centre: how to engage clinicians

Qual Life Res (2015) 24:1–191 Table 1

Correspondency betwen quality of work life with results of General Health Questionnaire of Goldberg in nursing staff

Estatistic

Alyssa Macedo, MScOT, University Health Network - Princess Margaret Cancer Centre, Toronto, ON, Canada; Madeline Li, MD PhD FRCP(C), University Health Network, and the University of Toronto, Toronto, ON, Canada; Gary Rodin, MD, University Health Network, and University of Toronto, Toronto, ON, Canada; Sabira Bagha, MSW, University Health Network, Toronto, ON, Canada; Yvonne Leung, PhD, University Health Network, Toronto, ON, Canada; Lindsay McColl, Medical Student, University of Toronto, Toronto, ON, Canada; Alexis Twiddy, Medical Student, University of Toronto, Toronto, ON, Canada; Kathy Davison, RN, BAS, MHS, University Health Network, Toronto, ON, Canada; Terri StuartMcEwan, University Health Network - Princess Margaret Cancer Centre; University of Toronto, Toronto, Canada AIMS: The routine use of patient reported outcome (PRO) measures for distress screening has been advocated by many cancer care organizations. Princess Margaret (PM), a leading Canadian comprehensive cancer centre has established the Distress Assessment and Response Tool (DART), as a clinical standard of care. DART is an electronic screening tool that detects physical and emotional distress and practical concerns and is linked to triaged inter-professional collaborative care pathways. With over 250,000 screens completed to date, DART is the largest single institution distress screening program in the world. Key factors to successful implementation of DART were institutional support and clinician engagement. This abstract outlines several change management strategies to support clinician engagement in DART. METHODS: DART implementation began with a pilot in 2010 with graded expansion to all 15 disease sites by 2012. Several clinician engagement strategies were utilized; creation of Monthly Screening Rate Performance Report Cards, Symptom Prevalence Update Reports (SPURs), creation of a Responding to Distress Toolkit, customization of DART with disease specific PROs and launch of a DART Star program to recognize site based champions. Integrated knowledge translation (KT) approaches were used to feedback program evaluation data, including screening rates, patient satisfaction data and health outcomes associated with clinician response to identified distress. RESULTS: Since 2013, PM has maintained screening rates above the provincial 70 % target, with an average rate of 75 %. The DART program is associated with high levels of site based ownership and accountability, seen with engagement in the Monthly Screening Rate Performance Report Cards. Patient satisfaction with clinician response to emotional and physical distress remains high, at 93 and 97 %, respectively. Preliminary findings suggest that clinician response to threshold distress was associated with increased patient completion of planned chemotherapy cycles (p \ 0.01, n = 66), and increased psychosocial service use, regardless of age and gender (p \ 0.001, n = 349). CONCLUSIONS: Clinician engagement in responding to distress is essential to improve the patient experience and health outcomes. Critical success factors to support clinician response include provision of screening and symptom reports, establishment of local champion programs, educational toolkits, integrated KT feedback and customizations based on disease site needs.

(3056) Quality of work life perception and mental health in nursing staff Raquel Gonzalez-Baltazar, Researcher, University of Guadalajara, Guadalajara, Jalisco, Mexico; Gustavo Hidalgo-Santacruz, Researcher, University of Guadalajara, Guadalajara, Jalisco, Mexico; Silvia G. Leon-Cortes, Researcher, University of Guadalajara, Guadalajara, Jalisco, Mexico; Monica I. Contreras-Estrada, Researcher, University of Guadalajara, Guadalajara, Jalisco, Mexico;

123

Dimensions Institutional work support (1)

Job reliability (2)

Integration to the job (3)

Satisfaction with the work (4)

Well-being obtained through the job (5)

Personal development (6)

Free time administration (7)

Prevalence of satisfaction with Goldberg’s non case

80.2 %

81.6 %

81.6 %

80.9 %

82.3 %

79.7 %

80.9 %

Prevalence of non satisfaction with Goldberg’s case

33.3 %

48.7 %

47.5 %

33.3 %

42.9 %

29.5 %

36.4 %

Square Chi

4.037

16.587

15.514

5.335

15.593

5.924

6.913

Significance

0.045

0.000

0.000

0.021

0.000

0.024

0.009

Source: direct

Brenda J. Hidalgo-Gonzalez, Researcher, University of Guadalajara, Guadalajara, Jalisco, Mexico AIMS: The aim of this study was to find the relationship between the perception of Quality of Work Life (QWL) and the mental health of nursing staff of a public health institution in Guadalajara, Mexico. METHODS: The study was analytical and was carried out with nurses of a health institution in Guadalajara, Mexico. The universe was of 1048 nurses from 98 health units and the probability sample of 197 people. In all of them, CVT-GOHISALO and Goldberg GHQ-28 questionnaires were applied in order to measure their perception of QWL and the risk of mental illness respectively. Both instruments were selected because of their high reliability. We measured the satisfaction of the people interviewed for each one of the seven dimensions of CVT-GOHISALO and related to the points assigned to GHQ-28 case or non-case to consider of mental illness. RESULTS: We interviewed 100 % of those selected. The percentages of satisfaction with the QWL were high for all dimensions, 79.5–87.9 %. However highlights dissatisfaction level greater than 12 % in all dimensions of the instrument, being the highest value of 20.5 % dissatisfaction for the Job satisfaction dimension. In the classification of the case or non-case, according to the Goldberg GHQ-28, 22 % were considered cases and 78 % were considered non-cases. We sought correspondence of satisfaction with the QWL by dimensions and non-case of Goldberg’s, as well as dissatisfaction with the QWL and Goldberg’s case, finding a clear link between being satisfied and having mental health. All dimensions have a correspondence that goes from 79.7 to 82.3 % between satisfaction and non-case, being in all the statistical significance value less than 0.05. CONCLUSIONS: For the nursing staff in the studied institution, there is a clear relationship between being satisfied with the QWL and presenting less risk of mental illness. At least a fifth of the nurses are dissatisfied with the work done, since the dimension with the lowest satisfaction is Job satisfaction. The risk of mental illness in nurses is equal or greater than the general population. It’s necessary to provide preventive programs among health care workers in order to improve their QWL and mental health (Table 1).

(3058) Future rehabilitation professionals’ knowledge, skills and attitudes towards self-management support: helping future clinicians to help patients Sabrina Figueiredo, PhD Candidate, McGill University, Montreal, Canada; Nancy Mayo, PhD, McGill University, Montreal, Canada; Aliki Thomas, PhD, McGill University; Jewish Rehabilitation Hospital, Montreal, Canada AIMS: Self-management support (SMS) was developed in the late 1980’s as a new way of delivering care for those with chronic conditions. This type of support aims to empower patients to cope with diseases and to foster productive interactions between patients and

Qual Life Res (2015) 24:1–191 Table 1 Themes and sub-categories emerged during Focus Group

providers, which in turn can positively impact patients’ quality of life. In order to promote the necessary skills to be a good self-manager, specific courses are offered to patients; however, there is a lack of similar educational programs for providers. To address such gap, education regarding this aspect of care should be incorporated into professional educational programs. This project aims to estimate the extent to which educational modules on SMS increases future clinicians’ knowledge, skills and attitudes. METHODS: Mixed-method design with pre-post focus groups and questionnaires. Inductive analysis of common themes on students’ knowledge, skills and attitudes. Deductive mapping of themes to the Theory of Planned Behavior model. A convenience sample of 10 Masters’ students received a 5-week theoretical course on SMS followed by observation of patients’ assessment and 6 weeks of tele-monitoring. RESULTS: As shown in the accompanying Table 1, five themes and 22 subcategories emerged from the inductive-deductive focus group content analysis. The questionnaires showed that 90 % of participants increased knowledge and skills, 10 % of participants had a better perception about SMS, and 40 % of them reported feeling pressured to adopt SMS. After the educational modules, 70 % of participants did not intend in adopting SMS. These participants claimed that the short duration of the educational modules did not allow them to see the results of such intervention. CONCLUSIONS: These findings suggest that there is a gap regarding formal training of SMS for future clinicians. The educational and practical modules increased knowledge and skills of future clinicians; whereas attitudes and intention to adopt a behavior did not increase as these constructs may require longer and more intense exposure to the educational intervention. This study can contribute in delineating courses/training regarding SMS for clinicians and future clinicians, which in turn can improve the way patients adopt and experience SMS.

Vulnerable Populations (3060) Content validation of a quality of life measure for individuals who are homeless and vulnerably housed Lara B. Russell, PhD, University of Victoria & Centre for Health Evaluation and Outcome Sciences, Vancouver, BC, Canada; Anita M. Hubley, PhD, University of British Columbia, Vancouver, Canada AIMS: The goal of this study was to collect content-related validity evidence for the Health Impact section from the Quality of Life of Homeless and Hard-to-House Individuals (QoLHHI) instrument. The Health Impact section assesses the subjective impact on quality of life (QoL) of various aspects of health (e.g., physical health, mental health, pain, substance use). METHODS: Content experts consisted of 7 Subject Matter Experts (SMEs) with research or clinical experience with individuals who are homeless or vulnerably housed (HVH), 8 Experiential Experts (EEs) who were HVH, and 7 Practical Experts

185 (PEs) who administered the scale in previous research. Content experts rated elements such as the relevance and clarity of items, instructions, and response scales. The SMEs and PEs rated a total of 82 content elements; the EEs rated 63 elements. The Content Validity Index (CVI), which indicates the proportion of experts that rate an item a 3 or 4 (out of 4), was computed. Experts also provided openended feedback in the form of comments and suggestions for improvements to the instrument. RESULTS: Results: Items were considered endorsed if their CVI was at least .80. Overall, SMEs endorsed 89 %, PEs endorsed 84 %, and EEs endorsed 79 % of the content elements. The SMEs and PEs endorsed 100 and 93 %, respectively, of the content for relevance and 86 and 91 %, respectively, of the content for clarity. EEs, however, endorsed a much higher percentage of elements for clarity (98 %) than for relevance (57 %). Experts suggested simplifying the language for some items and adding examples. It was also suggested that some items had too narrow a focus (e.g., asking about ‘medically recommended diet’) or were not necessarily relevant to individuals who are HVH (e.g., asking about physical activity or exercise). CONCLUSIONS: This study highlights the different perspectives of SMEs, PEs, and EEs. Overall, there was strong support for the relevance and the clarity of the QoLHHI Health Impact items, instructions, and response scales, providing one piece of validity evidence supporting the usefulness of this measure with individuals who are HVH. Feedback was used to develop recommendations for improvements to the measure.

(3062) Measuring health-related quality of life in homeless and vulnerably housed adults: a validation study Anita M. Hubley, PhD, University of British Columbia, Vancouver, Canada; Bruce MacLaurin, MSW, MSc, University of Calgary, Calgary, AB, Canada; Susan Farrell, PhD, Royal Ottawa Mental Health Centre & University of Ottawa, Ottawa, ON, Canada; Anne M. Gadermann, PhD, University of British Columbia, Vancouver, BC, Canada; Alina Turner, PhD, Turner Research and Strategy & University of Calgary, Calgary, AB, Canada; Katrina Milaney, PhD, University of Calgary, Calgary, AB, Canada; Wendy Muckle, RN, BScN, MHA, Ottawa Inner City Health, Inc., Ottawa, ON, Canada AIMS: The purpose of this study is to provide validation evidence for the health-related scales from the Quality of Life in Homeless and Hard-to-House Individuals (QoLHHI) inventory by examining convergent, discriminant, and known-groups validity evidence. METHODS: The sample consisted of 240 homeless and vulnerably housed (HVH) men and women ages 19–72 recruited from sites in Vancouver, Calgary, and Ottawa. Study participants completed measures in a single session with an interviewer. Support for convergent and discriminant validity is shown if QoLHHI Health Impact (QoLHHI-HI) and Health Satisfaction (QoLHHI-HS) are relatively highly correlated with, and more highly correlated with, convergent measures of similar or related constructs (i.e., life satisfaction, depression, overall subjective quality of life, self-esteem, and self-reported health) than with discriminant measures of dissimilar or less related constructs (i.e., social desirability, age identity, and cognitive measures of attention and figural fluency). Support for known-groups validity is shown if HVH adults with many chronic (mental or physical) conditions show significantly poorer QoLHHI-HI and QoLHHI-HS scores than HVH adults with few chronic conditions. RESULTS: Correlational analyses to examine convergent and discriminant validity showed that both QoLHHI-HI and QoLHHI-HS were correlated more highly with convergent measures than with discriminant measures: QoLHHI-HI convergent: |r| = .50 to .74; discriminant: |r| = .02 to .24 and QoLHHI-HS convergent: |r| = .42 to .68; discriminant: |r| = .02 to .33. Moreover, the convergent validity coefficients were of a moderate to large magnitude and the discriminant validity coefficients,

123

186 with one minor exception, were of a small magnitude. The knowngroups t test analysis showed that a subgroup of 65 HVH adults with 9–19 chronic conditions reported statistically significantly poorer QoLHHI-HI (M = 3.2, SD = 1.27) and QoLHHI-HS (M = 2.9, SD = 1.59) scores than a subgroup of 59 HVH adults with 0–2 chronic conditions (QoLHHI-HI: M = 4.2, SD = 1.24; QoLHHI-HS: M = 5.0, SD = 1.58). CONCLUSIONS: The convergent, discriminant, and known-groups evidence provides strong evidence to support the intended inferences for the health-related scales of the QoLHHI inventory.

(3064) Effect of health promotion seminars using traditional drum performance on the QOL and physical status of Japanese community elderly Toshio Kobayashi, Hiroshima University, Hiroshima, Japan; Akemi Furuyashiki, Hiroshima University, Hiroshima, Japan; Suzumi Okuda, Hiroshima University, Hiroshima, Japan; Kousuke Yabase, Hiroshima University, Hiroshima, Japan AIMS: For Japanese local elderly with slight abnormalities in blood pressure, lipid, and/or glucose level determined by health checkups, health promotion seminars using traditional drum performance was executed to improve these risks and the patients health-related QOL (HRQOL). METHODS: The health promotion seminars were carried out over a period of seven months. They included exercise using music and the drum performance of local traditional arts, in addition to lectures on health and nutrition. The participants were 22 elderly community dwellers (three men and 19 women, mean age 64.7 ± 4.4 years). Physical function measurements and blood testing were carried out three times at the beginning of the seminar period, after three months, and at the end of the seminars (after seven months). In addition, a questionnaire survey was carried out to measure changes in HRQOL (SF8), physical symptoms, life satisfaction, and the feeling of attachment to the local community. RESULTS: Weight, waist circumference, and systolic blood pressure were all significantly decreased after the seminars as compared to the beginning. HDL cholesterol significantly increased after three months, and triglycerides and HbA1c decreased. For HRQOL (SF8), the standardized score of general health feeling at the beginning was 48.9, which was slightly lower than the standard value (50). It increased to 51.2 after three months, and significantly increased to 52.0 at the end. The physical component summary (PCS) score of SF8 at the beginning was 45.8, which significantly increased to 49.8 at the end. The mental component summary (MCS) score was 51.3 at the beginning, and showed no change during the seminar period. Also, the degree of feeling of attachment to the local community increased and finally became significantly higher than at the beginning. Life satisfaction also increased, but this was not significant. Regarding the change of SF8, the PCS showed a significant negative association with the symptoms of joint pain. CONCLUSIONS: After the health promotion seminars using traditional drum performance, the general health feeling and HRQOL of the participants increased as well as their physical status, and their blood data improved. Furthermore, the increase of the feeling of attachment to the community might influence the community QOL.

(3068) Subjective age, age satisfaction, and self-rated health in adults who are homeless or vulnerably housed Ayumi Sasaki, University of British Columbia, Vancouver, Canada; Anita M. Hubley, PhD, University of British Columbia, Vancouver, Canada; Anne M. Gadermann, PhD, University of British Columbia, Vancouver, BC, Canada; Lara Russell, PhD, University of Victoria,

123

Qual Life Res (2015) 24:1–191 Vancouver, Canada; Anita Palepu, Professor, University of British Columbia, Vancouver, BC, Canada AIMS: The concept of subjective age (SA) refers to the age that one feels or looks relative to one’s chronological age. A key variable that contributes to SA in North America is health, with poorer health associated with older SA. Only a few studies, however, have focused on individuals from vulnerable populations who frequently experience health challenges. The purpose of this study was to examine levels of SA and age satisfaction, as well as their correlations with health variables, in a sample of homeless or vulnerably housed (HVH) adults. METHODS: The sample consisted of 240 HVH adults (50.4 % male) aged 19–72 years (M = 42.4, SD = 11.79). Study participants completed measures with an interviewer. RESULTS: Results showed that HVH adults, on average, felt and thought they looked slightly younger than their chronological ages. They felt slightly older physically, but mentally the same, as their chronological ages. While HVH adults reported wanting to be slightly younger than their actual ages, generally, they were slightly satisfied with being their age. Participants overall were ‘‘neither dissatisfied nor satisfied’’ with their current health, but reported their health had a slight negative impact on them. Notably, their SF-12 physical (M = 42.7, SD = 11.6) and mental (M = 39.5, SD = 12.1) health composite scores were below average relative to normative values of 50. The average EQ-5D health thermometer score was also low (M = 55 out of 100; SD = 21.2). HVH adults’ overall and physically felt SA were significantly and negatively associated with all health variables (r = -.17 to -.48). Mentally felt SA and the age one looks showed low significant correlations with only a few health ratings. The age one wants to be ideally showed a low positive and significant correlation only with the SF-12 physical health composite score. Finally, age satisfaction showed low to moderate positive correlations with all health variables (r = .22 to .43) except the SF-12 physical health composite score (r = .03). CONCLUSIONS: HVH adults reported SAs close to their chronological ages. Despite generally low health scores, those who reported higher physically felt SA tended to have worse health and those who reported higher age satisfaction tended to report better health.

(3070) Recommendations from the field: interviewing and retaining homeless and vulnerably housed participants in longitudinal health research Sophie Ma Zhu, University of British Columbia, Vancouver, BC, Canada; Roxanne Joyce, Providence Health Care Research Institute & University of British Columbia, Vancouver, BC, Canada; Anne M. Gadermann, PhD, University of British Columbia, Vancouver, BC, Canada; Julia Gerlitz, University of British Columbia & Providence Health Care Research Institute, Vancouver, BC, Canada; Meagan Stowe, Providence Health Care Research Institute, Vancouver, BC, Canada; Julie Parrell, Providence Health Care Research Institute, Vancouver, BC, Canada; Anita M. Hubley, PhD, University of British Columbia, Vancouver, Canada; Anita Palepu, Professor, University of British Columbia, Vancouver, BC, Canada AIMS: The Health and Housing in Transition (HHiT) study is a 5-year longitudinal study on health, housing, and quality of life for homeless and vulnerably housed (HVH) people in three Canadian cities. In Vancouver, 396 participants were recruited in 2009 and the study retained 81.3 % of this group over the next 4 years. Study interviewers were invited to take part in a series of focus groups and self-reflection to discuss the unique challenges and subsequent successes of conducting longitudinal health research with this vulnerable and hard-to-reach population. METHODS: Four female research assistant interviewers ages 26–30 years took part in 6 one-hour focus

Qual Life Res (2015) 24:1–191 groups. RESULTS: Content analysis identified 5 overarching themes related to conducting research with HVH individuals: (1) recruitment and retention of participants over time, (2) safety concerns and risks, (3) participant mental health, (4) training and support, and (5) impacts on interviewers. In this presentation, we will focus on practical recommendations stemming from the first 3 themes. Under recruitment and retention, we will discuss recruitment strategies that improved follow-up rates, importance of tracking through local agencies, creative use of social media, persuasive methods for eliciting information via phone, and pointers on collecting participant contact information. Under safety concerns, we will discuss the importance of teamwork, monitoring safety, setting boundaries, and health safety practice. Under participant mental health, we will discuss participants’ capacity to give informed consent and respond to the survey questions accurately and consistently, and unpredictability in the interview dynamic. CONCLUSIONS: This study extracted themes shared by interviewers of their experience working on a longitudinal study with a HVH sample. It is important that interviewers have a background or training in mental health issues. Understanding various service providers and building long-term collaborative relationships with them are crucial in tracking participants. Maintaining organized records and using creative, individualized methods to track participants are essential. Safety protocols must be discussed and implemented consistently. The findings of this study will be of benefit to health researchers conducting longitudinal research with individuals who are homeless, vulnerably housed, or experiencing mental health challenges.

(3072) Does socioeconomic inequality moderate the reduction in vitality, life satisfaction and increased psychological distress in those suffering from chronic illness? A population-based study on possible pathways Jocelyne Clench-Aas, Senior Scientist, Norwegian Institute of Public Health, Oslo, Norway; Neha Agnihotri, Norwegian University of Life ˚ s, Norway; Ragnhild B. Nes, Research Scientist, Science, A Norwegian Institute of Public Health, Oslo, Norway AIMS: Coping with a chronic illness (CI) will depend on individual differences in psychological and socioeconomic resources as well as the physical and psychological effects of the illness. CI is associated with low well-being and higher prevalence of psychological distress (PD), but the impact of socioeconomy on these associations is scarcely studied. This presentation presents findings on the associations between chronic illness, well-being (vitality, life satisfaction) and psychological distress across socio-economic groups. Additionally, potential mediating effects of sense of mastery and social support are examined. METHODS: Data were obtained from a large (N = 13,729), nationally representative sample of Norwegians aged 16 and older collected in 2002, 2005, 2008 and 2012. Two subscales from the SF36 were used to measure vitality and psychological distress (MHI-5). Additional measures included sense of mastery (SM), chronic illness (CI,) and social support (SS). Socio-economic status (SES) was assessed using information on occupation. The PROCESS macro for SPSS was used for mediation analysis. RESULTS: Having a CI was directly associated with reduced vitality, reduced LS and increased PD. SM, but not SS, was shown to significantly mediate all three outcomes. In addition to constituting a significant moderator of these direct effects, SES was found to significantly moderate both pathways for the mediator SM for PD and LS, whereas only for the pathway CI to SM for vitality. For the mediator SS, SES significantly moderated both pathways for vitality, none of the pathways for LS and for PD, only for the pathway CI to SS. In the lowest SES group, CI was associated with a 10 % reduction of vitality (5 % considered clinically relevant); an 8 % reduction in LS; and a 5 % increase in

187 PD. CONCLUSIONS: As SES decreases, the decline of vitality and LS and increase in PD that results from chronic illness is intensified, with the reduction of vitality reaching a clinically relevant magnitude. The associations are partially mediated through sense of mastery.

(3074) The relationship between physical disability and overall health-related quality of life of residents in long-term care facilities Allyson Jones, PhD, University of Alberta, Edmonton, AB, Canada; David Feeny, PhD, Department of Economics, McMaster University, Hamilton, ON, Canada, Hamilton, ON, Canada; Susan Slaughter, University of Alberta, Edmonton, AB, Canada; Adrian Wagg, University of Alberta, Edmonton, AB, Canada AIMS: Mobility and health of older adults with dementia decline once they are in long-term care facilities. We explored the inter-relationship of 2 measures which include constructs of physical disability and associated required hours of care in this frail cohort. Each of these indicators is intended to measure a different underlying construct. METHODS: We examined baseline physical disability in 86 residents enrolled in a RCT for a mobility intervention in terms of the amount of assistance required to perform functional activities (Functional Independence Measure, FIM) and overall health-related quality of life (HUI2 and HUI3). Residents’ function and health-related quality of life were evaluated by the health care aide who was responsible for their care prior to the implementation of the 6-month intervention. RESULTS: All residents had dementia, 61 (71 %) were female, of mean age 87 (sd 7) years. Mean FIM total score was 62.5+19.3 (range 26–113) with higher scores reflecting the requirement for less assistance to perform the activity. The mean overall HUI2 and HUI3 scores were 0.49 (sd 0.16) and 0.31 (sd 0.27), respectively. Activities measured with the FIM that required the most assistance were bowel/ bladder management and locomotion. HUI3 identified moderate to severe problems in ambulation and severe problems in cognition. HUI2 identified severe problems in self-care as well as moderate to severe problems in mobility and severe problems in cognition. The median number of hours of care which was derived from the FIM was 4 h per day; 9 residents were classified as requiring less than 1 h of care and 7 residents required 7 h per day. The overall HUI2 score adequately explained the number of hours of care in this cohort with substantial physical disability. CONCLUSIONS: A fuller understanding of clinical outcomes for older adults with physical disability may be improved by enhancing our ability to differentiate between function and the complexities of health status and health-related quality of life.

Family (3076) Developing a measure of hemophilia caregiver impact Carolyn Schwartz, ScD, Tufts University Medical School; DeltaQuest Foundation, Inc., Oslo and Akershus University College,, Concord, MA, United States; Adi Eldar-Lissai, M.S., MBA, PhD, Biogen, Cambridge, MA, United States AIMS: Caring for someone with hemophilia is a demanding role. Usually diagnosed at birth or soon thereafter, hemophilia is a disabling and life-threatening condition that requires constant vigilance to prevent and/or identify bleeds, and continual proactive medical management. Although the impact of hemophilia on the patient and family is clearly important, there is currently no validated measure of hemophilia caregiver burden. We aim to develop an item bank of questions measuring hemophilia caregiver impact, and to establish the content validity of this item bank. METHODS: We implemented a

123

188 review of the literature on caregiver burden to identify a useful conceptual model of the construct, and then reviewed existing caregiver-burden measures for other conditions (e.g., cancer, asthma). We drafted items drawing from these existing measures, with an emphasis on maintaining a consistent response-option format for ease of use. We then pretested the item bank for clarity and content coverage in interviews with current hemophilia caregivers. RESULTS: Our conceptual model identified the following caregiver impact dimensions as relevant: practical support, symptoms of hemophilia, lifestyle, social, physical, emotional, financial, and positive emotions. Three rounds of pretesting interviews (n = 22) included 20 female and 2 male caregivers, age 29–58. Our initial item bank of 126 questions was modified, expanded, and contracted to yield a final item bank of 105 questions, with subsets of items intended for subgroups of caregivers for the newly diagnosed, and for infants/young children, and adolescents/young adults. Feedback from interviewees supports the content validity and clarity of the item bank, which is estimated to take 15 min to complete. CONCLUSIONS: The Hemophilia Caregiver Impact Measure covers content relevant to hemophilia caregiver burden, and is transparent and feasible. Further testing is required to evaluate the measure’s reliability, validity, and responsiveness.

(3078) Psychometric properties of the Diabetes Self-care Efficacy Scale Sravanti Ghosh, Senior Research Fellow, Indian Statistical Institute, Kolkata, West Bengal, India; Debdulal Dutta Roy, Assistant Professor, Indian Statistical Institute, Kolkata, West Bengal, India AIMS: Epidemiological studies indicate that the incidence of diabetes is on the rise and will further increase in India. Diabetes self careefficacy (DSCE) plays pivotal role in management of diabetes as it affects both cognitive and motivational levels of the patient. Following Bandura’s definition of self-efficacy, Diabetes self care-efficacy is defined as the diabetic patient’s confidence in their capabilities to perform relevant behaviors to maintain and control their blood glucose level. It determines how the patients feel, think, motivate themselves and behave about their diabetes. There is dearth of research about how different domains of diabetes self-efficacy affect the symptom severity due to lack of any assessment tool. This study aims at development of psychological instrument for assessment of diabetes self-care efficacy. METHODS: Following the interviews and review of literature, one 44-item Likert type scale was constructed to assess four areas of self-care efficacy namely, self-monitoring, vicarious experience, controlling physiological and emotional response and social persuasion. It was administered to 200 patients with either Type I (n = 36) or Type II (n = 164) diabetes mellitus. Besides, data about HbA1C and postprandial measure of blood sugar were collected from the patient. RESULTS: Cronbach’s alpha of the 44-items of the diabetes self-care efficacy was 0.88 indicating high internal consistency. Alpha values for controlling physiological and emotional responses were lower than others. The highest value was for self-monitoring. The scale was independent of age but was negatively correlated with HbA1C and post prandial measure of blood sugar. CONCLUSIONS: The scale is found to be a valid measure of diabetes self-care efficacy and is very useful for psychosocial rehabilitation of diabetic patients.

(3080) Cultural influences on the caregiver role, caregiving experiences and perceptions of QOL of Hispanic family caregivers of patients with advanced cancer Gloria Juarez, PhD, RN, Consulting Services, San Pedro, CA, United States; Joan J. Branin, PhD, University of La Verne, Pasadena, CA,

123

Qual Life Res (2015) 24:1–191 United States; Lina Mayorga, PhD MPH CHES, City of Hope, Duarte, CA, United States AIMS: Although Hispanics are the fastest growing ethnic group and cancer the second leading cause of death among Hispanics in the United States (US), there is a substantial gap in the literature about the influence of culture on the quality of life (QOL) and caregiving experiences of Hispanic family caregivers. The influence of culture has been seen by healthcare and community-based providers as important in the development of culturally congruent care and support services for cancer patients and their caregivers. This study examines the influence of culture on the caregiver role, caregiving experience, and perceptions of QOL of Hispanic family caregivers of adults with advanced cancer. METHODS: This qualitative descriptive study was conducted in the ambulatory care setting of an NCI-Comprehensive cancer care center in the US. Twenty Hispanic family caregivers of patients with advanced cancer participated in qualitative interviews. Thematic interview analysis and the conceptual model of Quality of Life developed by Ferrell et al. (1989) were used to identify themes that characterized the caregiving experience, caregiver perceptions of QOL, and the role of caregiving of this underserved population. RESULTS: In general, the caregiving experience and the meaning of QOL of these caregivers centered on spiritual or religious beliefs and practices, concern for providing quality home care, strong family values and extended family support, and desire to fulfill cultural role and responsibility expectations. Overall, Hispanic caregivers emphasized that their caregiving experience had more positive than negative outcomes. The caregiver’s QOL was primarily affected by the patient’s advanced disease, cultural background, the consequences of the cancer treatment and related side effects, and the patient’s emotional/psychological needs. Secondarily, the caregiver’s overall perception of QOL was affected by their physical, social, emotional and spiritual needs. CONCLUSIONS: These findings provide support for the development of culturally relevant interventions for this underserved population and the need for the inclusion of culture in QOL research. A better understanding of the role of culture on the perceptions of QOL and experiences of Hispanic caregivers can improve their caregiving experience, reduce health care disparities, improve patient and caregiver health outcome, and better inform the design of public policies.

(3082) Elderly patient and caregiver involvement in the quality improvement of cancer care pathways Asiong Jie, MD, PhD, Atrium-Orbis Medical Centre Heerlen, and Zuyd University of Applied Sciences, Heerlen, Netherlands; Inge Melchior, PhD, Zuyd University of Applied Sciences and AtriumOrbis Medical Centre Heerlen, Heerlen, Netherlands; Albine Moser, PhD, MPH, RN, Zuyd University of Applied Sciences, Heerlen, Netherlands; Marja Veenstra, PhD, House of Care, Sittard, Netherlands; Esther Stoffers, PhD, House of Care, Sittard, Netherlands AIMS: In recent years, cancer care has been improved by the implementation of cancer care pathways that standardise organisational procedures. However, in these predetermined pathways, patient involvement is in its infancy, especially regarding elderly patients. The aim of this study is to identify and implement improvements of quality of care for elderly cancer patients by involving patients, their caregivers and their health care professionals into a co-design process. METHODS: Experience-Based Co-Design (EBCD) is used to involve patients and their caregivers in a structural quality improvement process. Open discovery interviews with 24 patients and their caregivers, and focus group discussions with the involved professionals were conducted to gather the data. The data were analysed according to the constant comparative method. In the next phase the patients’,

Qual Life Res (2015) 24:1–191 Phase 1 Elderly cancer patients Discovery interviews

Phase 1 Caregivers Discovery interviews

189 Phase 2 Healthcare professionals Focus group discussions

Table 1 Comparison of Attitudes to Disability among PWD, caregivers and Public

Phase 3 Prioritization in co-design group First in the three groups separately, then as a joint group

Phase 4 Co-design quality – improvement teams Elderly cancer patients and caregivers, health care professionals and project employee

Phase 5 Patients, caregivers and health care professionals Review and celebrate gains

Fig. 1 EBCD research process

caregivers’ and health care professionals’ points of improvement were first prioritized in the three separate groups, and subsequently in a joint event. In various co-design meetings, ‘quality improvement teams’ (4 professionals, 2 patients and 2 caregivers) work as partners on the improvement process (see Fig. 1 for an overview). This research is being conducted in the colorectal and breast cancer care pathway of the Atrium-Orbis Medical Centre in Heerlen, The Netherlands. RESULTS: Bringing patients, carers and professionals together to share their experiences proved to be very challenging. Especially the involvement of clinicians appeared to be an important hurdle. Obviously we will share our experiences with this process. The outcomes of the joint events for the elderly colorectal cancer patients, their caregivers and health care professionals are (1) the need for a patient navigator, (2) improvement of communication between hospital departments, (3) the need for one principle healthcare provider, (4) improvement of the communication between the hospital health care professionals and general practitioner and (5) more respect to patients’ needs. The quality improvement teams are currently in the co-design process. CONCLUSIONS: Working in partnership with elderly cancer patients, their caregivers and healthcare professionals proved to be very successful in identifying common points of quality improvement and their implementation in cancer care pathways. We collaborate with the Quality and Safety department of the hospital to eventually implement EBCD in other pathways.

(3086) Empowerment for whom? Comparison of attitudes towards disability among people with disability, caregivers and the public Chun Hao, Sun Yat-sen University, Guangzhou, China; Qiaolan Zheng, Sun Yat-sen University, Guangzhou, China; Jing Gu, PhD, Sun Yat-sen University, Guangzhou, Guangdong, China; Yuantao Hao, PhD Sun Yat-sen University, Guangzhou, Guangdong, China AIMS: Disability has become a natural part of the human condition due to population ageing, increasing of chronic diseases, and medical advances that preserve and prolong life. A negative attitude towards disability is one of the potential barriers for people with disability (PWD) to achieve their social equality and the better quality of life. This study is to investigate and compare the level of attitudes towards disability among PWD, caregivers and the public in China. The potential influential factors of the attitudes about disability were

examined within each group as well. METHODS: A cross-sectional study was conducted among 2,912 PWD, 507 caregivers and 354 members of the public in Guangzhou, China. Data were collected on participants’ socio-demographic information and personal attitude towards disability by using the global scale (Attitude to Disability Scale, ADS). ANOVA and ANCOVA were applied to compare the level of ADS among three groups; simple linear regression analysis was used to investigate the relationship between each background factor and ADS within each group. RESULTS: Over 90 % caregivers were PWD’s family members. After controlling the socio-demographic characteristics, caregivers had the lowest total scores of ADS (caregivers: 47.7; PWD: 52.3; the public: 50.5). Caregivers who had taken care of PWD for longer durations of time had a more negative attitude towards disability; in contrast, PWD who had been disabled for longer times had a more positive attitude towards disability (50.0–50.5). CONCLUSIONS: Under the current social security system in China, PWD’s family member caregivers may need more empowerments than PWDs themselves. The PWD-family as the unit for social subsidizes may undermine PWD’s independent living capacity, and also increases caregivers’ burdens for the entirety of their lives. The development of a new health-care model for PWD is needed in China.

(3090) Quality of Life (Qol) Questionnaire for mothers of schoolchildren Rika Hayashida, University of Nagasaki, Siebold, Nishisonogigun, Nagasaki, Japan; Michiko Kobayashi, Japanese Society of Quality of Life Research, Kobe, Japan; Takashi Mandai, Japanese Society of Quality of Life Research, Kobe, Japan AIMS: The purpose of this study was to develop a desirable original QOL questionnaire for mothers of schoolchildren. METHODS: Seventy-two mothers of schoolchildren over 10 years of age participated in this study. The former QOL questionnaires completed in 2009 and 2013 (30 questions divided into seven categories) had a sufficient reliability and potency of validity to use for schoolchildren. We have developed an original, self-administered questionnaire for mothers of schoolchildren consisting of 48 questions, divided into eleven categories this time. RESULTS: The Cronbach’s alpha coefficients of our questionnaire were sufficient to accept for clinical use for mothers of schoolchildren: 0.93 in living environment, 0.92 in eating habits, 0.92 in sleeping habits, 0.91 in mother–child interaction, 0.88 in psychological conditions, 0.86 in financial circumstances, 0.85 in social circumstances etc., respectively. Our questionnaire contains twelve main factors which matched the eleven

123

190 categories precisely in this study. Its cumulative contribution was 0.72. There were significant positive correlations between eating habits and well-being (r = 0.82, p \ 0.01), eating habits and psychological conditions (r = 0.74, p \ 0.01), psychological conditions and living environment (r = 0.82, p \ 0.01), psychological conditions and social circumstances (r = 0.73, p \ 0.01), psychological conditions and well-being (r = 0.71, p \ 0.01), etc., respectively. Average QOL levels of mothers of schoolchildren were higher for the mothers who had more contact with other parents compared with mothers who had less contact with others (p \ 0.05). And average QOL levels of mothers of schoolchildren were higher for the mothers participating in community activities compared with mothers who did not (p \ 0.05). CONCLUSIONS: These findings indicate that our original QOL questionnaire has somewhat sufficient reliability and potency of validity to use for mothers of schoolchildren. The most significant finding of this study is that intimacy with other parents and support for mothers of schoolchildren is important. We think it’s important for mothers of schoolchildren to have a good environment (living environment, time and place to talk with their friends and social circumstances), since the support of neighboring people (community) will have a positive effect on mothers. In future, we must pay more attention to supplying full childcare service and teaching mothers how to take care of children, including permanent affection.

(3096) Living well? Examining quality-of-life and associated care needs among families of children with medical complexity Janet McCabe, RN, MEd, PhD, University of Saskatchewan, Prince Albert, SK, Canada; Jennifer Kryworuchko, PhD, RN, CNCC(C), University of Saskatchewan, Saskatoon, SK, Canada; Janet Gaunt, RN, University of Saskatchewan, Saskatoon, SK, Canada AIMS: This project explored Family Quality of Life (FQoL) of families living with children with complex medical needs with a focus on identifying areas for improved service provision and supports of families across Saskatchewan. Children with complex medical and their families engage with a network of intersectoral services in their home and community (e.g. health, education, social services). When services are not available to effectively predict, meet and manage the needs of family members and children with medical complexity FQoL may suffer. METHODS: Participants (n = 126) were recruited from three partner community agencies. Interviews (in-person and telephone) were used to collect data from participants, who were a primary caregiver of a child with complex medical needs. Participants, aged 18 or over, had a child currently receiving or waitlisted to receive services from a partnered community agency and resided in Saskatchewan, Canada. Participants completed a short demographic form and the Family Quality of Life Survey 2006 (FQOLS-2006) which measures FQOL across nine domains (health of the family, financial well-being, family relationships, support from other people, support from disability related services, influence of values, careers/ prepping for careers, leisure and recreation, and community interactions). RESULTS: Satisfaction was highest in the domains of family relationships and values, and lowest in the domains of financial wellbeing, support from others and support from services. Over half of the participants indicated that they had unmet service needs in the community. Among the top barriers to service identified were long waits, lack of services, available services not helping enough, and not knowing where to go for service and support. 31 % of families surveyed experienced discrimination in the community. CONCLUSIONS: Despite the majority of families identifying as being satisfied with their FQoL, this study highlights ongoing areas of focus for service providers and researchers with special attention to barriers that families encounter in obtaining satisfaction in FQoL. Social

123

Qual Life Res (2015) 24:1–191 exclusion and difficulty accessing health related services still deeply affects families with children with complex medical needs. This paper will address the signals for research, service provision, and health care that surfaced from the data.

(3098) The quality of life and experience of the family caregivers for stroke patients admitted on Korean National Rehabilitation Hospital Chihyoung Son, Doctor, Korean National Rehabilitation Hospital, Seoul, South Korea; Jung Hwan Kim, Korean National Rehabilitation Hospital, Seoul, South Korea; Hyun-Jeong Im, Korean National Rehabilitation Center, Seoul, South Korea; Jeong-A Yu, Korean National Rehabilitation Hospital, Seoul, South Korea; KuemJu Lee, Korean National Rehabilitation Research Institute, Seoul, South Korea; JiEun Lee, Korean National Rehabilitation Research Institute, Seoul, South Korea; Bum Suk Lee, Korean National Rehabilitation Hospital, Seoul, South Korea; Hyun Choi, Korean National Rehabilitation Research Institute, Seoul, South Korea AIMS: The aim of this study is to investigate the quality of life and experience of the family caregivers caring for inpatients with stroke in KNRC through Consensual Qualitative Research (CQR) method. METHODS: From April to November in 2013, eight family caregivers caring for stroke inpatients over 3 months in KNRC were interviewed in depth. The CQR method was used to investigate and analyze their experiences. The questionnaires were as follows; daily schedule, the change aspects in their life after caring for patients, the relationship with other family members, the reason for caring, and the way to relieve their stress. WHOQOL-BREF questionnaire was also investigated. RESULTS: Family caregivers started caring life for financial burden on family, responsibility for their family and dissatisfaction with non-family caregivers. They had a hard time with bad night’s sleep and little spare time. Their interpersonal relationship, health state and financial condition got worse. The recovery of patient, intimate relationship with patient and reliable relationship with others encouraged them. They regarded caring as sacrificing themselves for family. The mean (±SD) scores of each domains in the WHOQOL-BREF were as follows; the physical health score was 47.9 ± 9.3 points, psychological score was 45.4 ± 23.3, social relationships score was 43.0 ± 19.2, and environment score was 11.8 ± 10.8 which was lowest of 4 domains. CONCLUSIONS: Family caregivers in KNRC had lowest mean score in the environment domain of WHOQOL-BREF. They were caring for the patient sacrificing with themselves because of financial burden on family and responsibility for family. The program for improvement of family caregivers’ life in KNRC is necessary.

(3100) Predictors of quality of life of family caregivers of drug addicts Susane F. Cosentino, Assistant Professor, Federal University of Santa Maria/CESNORS. Brazil., Palmeira das Misso˜es/RS, Rio Grande do Sul, Brazil, Luciano Vitorino, Paulista School of Nursing, Federal University of Sa˜o Paulo - UNIFESP., Itajuba, Minas Gerais, Brazil; Leila G. de Brizola Perdonssini, Nurse in Family Health Strategy, Specialist, Palmeira das Misso˜es, Rio Grande do Sul, Brazil; Maria Helena Souza, Adjunct Professor, Federal University of Rio de Janeiro., Palmeira das Misso˜es, Rio Grande do Sul, Brazil; Lucila Amaral Vianna, Full Professor, Federal University of Sa˜o Paulo UNIFESP, Sao Paulo, Brazil AIMS: To identify the importance of sociodemographic characteristics and work overload to the quality of life (QOL) of family caregivers of drug addicts (FCDA). METHODS: A cross-sectional study

Qual Life Res (2015) 24:1–191 was developed in South/Brazil with 95 of FCDA providing care for drug addicts cared for in public health services. Study measures were: The WHOQOL-BREF and Caregiver Burden Scale (CBS). Demographic data were also collected. We used multiple linear regression to determine the predicting of QOL. RESULTS: The caregivers (85.3 %) were women; 53.7 % had between 40 and 59 years of age and 89.5 % had perceived health between good to regular. The higher score of QOL was in physical domain (14.58; SD = .24) and lower score was environment domain (13.35; SD = .16). Multiple linear regression analysis showed the hierarchical order of variables that most affected the scores of the WHOQOL-BREF were: physical QOL was explained in 48.3 % by perceived health (ß = .488; p \ .000), use of medicine daily (ß = .305; p \ .001), environment of CBS (ß = -.162; p \ .033). Psychological QOL was explained in 49.3 % by perceived health (ß = .396; p = .000), disappointment of CBS (ß = -399; p = .000), gender (ß = -204; p = .007) and

191 relationship quality with drug addict (ß = .191; p \ .014). Isolation of CBS (ß = -312; p = 0.002) and perceived health (ß = .23; p \.014) explained 14.7 % of the social relationships QOL. The environmental QOL was explained by 39.8 % by environmental of CBS (ß = - 416; p \ .000) and disappointment of CBS (ß = -.231; p \ .006) and the variables have a job (ß = .261; p \ .002) and perceived health (ß = .253; p = .003). CONCLUSIONS: We identified that family caregivers of drug addicts who had better perceived health, that taking your medicines daily, gender male, with good quality relationship with the drug addict and formal employment were predictors of a better of QOL of FCDA. In another hand we also observed that the caregivers with higher work overload were predictors impairment of QOL. With these results, we reinforce the need for public health managers and health professionals should have attention with work overload affecting the QOL of FCDA.

123