Quality of Life Research 12: 721–866, 2003.
2003 Kluwer Academic Publishers. Printed in the Netherlands.

# 1566/ORAL HISTORY PROJECT: THE CONTRIBUTIONS OF JIM BUSH TO QUALITY OF LIFE RESEARCH Chair: Ivan Barofsky, The Quality of Life Institute, East Sandwich, MA, USA Health-related quality of life (HRQOL) research has now been an active process for over 30 years, making significant contributions to the development of new medical treatments through the clinical trials process, improving the nature of clinical practice, and even impacting how the quality of health services are evaluated. The Oral History Project is an effort to articulate the significant intellectual basis for these achievements. Its purpose is to provide details about how the field developed so that interested parties can be assured that the research generated by HRQOL research is based on a firm empirical foundation, a foundation that justifies its application in both a policy and clinical setting. The Project is currently focusing on persons who made significant contributions during the first decade of work – the 1970’s. Among these contributors was Jim Bush. Bush, as was true of George Torrance, applied operations research approaches to the assessment of health status and eventually HRQOL. Bush’s contributions, through collaborations with his colleagues Robert Kaplan and Donald Patrick, remains an enduring part of current HRQOL research. This symposium honors his intellectual leadership and research contributions. It also illustrates that HRQOL research originated in the application of engineering principles. # 1566-A/JIM BUSH AND THE MATURATION AND APPLICATION OF THE HEALTH POLICY MODEL Robert Kaplan, Department of Family and Preventive Medicine, University of California – San Diego, San Diego, CA, USA Bush having formulating the Health Policy Model, came to California where the model was developed and applied in a variety of settings. It was here that Bush and his colleagues developed the General Wellbeing Scale and applied it to a number of practical problems. One example of this was the Oregon experiment. These and related applications will be discussed to illustrate the contributions of Jim Bush to the quality of life, health status and health services research.

721

# 1566-B/JW BUSH – THINKER, CONTRIBUTOR, PROVOCATEUR Donald Patrick, Department of Health Services, The University of Washington, Seattle, WA, USA Jim Bush collected many persons around him in New York in 1969 to begin work on a health status index for use in resource allocation. I was a student with Jim in the MPH Program at Columbia University in 1966–1968 and then with Jim went to work at the New York State Health Department to work for Governor Nelson Rockefeller. Governor Rockefeller asked for a Plan for Health in New York State that would permit his administration to allocate resources to where they would have the greatest benefit on State Population Health. Hence was born the need for a measure of health status that would aid in health planning. Through a grant from the then National Center for Health Services Research, a team was formed consisting of Jim, Milton Chen, and myself to create what has become known as the Health Policy Model. This presentation will describe how the different elements of the model were formed and the major considerations that went into each one, including Jim’s contribution to each element. # 1566-C/BACK TO THE FUTURE Paul Kind, Centre for Health Economics, York University, York, United Kingdom The 1970 paper by Fanshel and Bush was a definitive ‘roadmap’ for health services researchers and others concerned with the development and application of methods for use in the evaluation of health care. That single paper identifies several key issues in the design and construction of what we today term as measures of health-related quality of life, including the description and valuation of health states. This presentation reviews the wider research agenda defined by the Fanshel and Bush paper and addresses two questions – how far have we come in meeting their original challenge and what is on the current agenda that was missing three decades ago?

722 # 1793/EVIDENCE REGARDING CHILDREN’S ABILITY TO SELFREPORT ON THEIR HEALTH Chair: Anne Riley, Health Policy and Management, Johns Hopkins University, Baltimore, MD, USA The value of obtaining reports from children is being increasingly recognized, with a resulting need for research defining children’s abilities and limitations, as well as optimal methodologies for ensuring reliable and valid reports from children. This Symposium brings together multiple methods describing the accuracy and understanding of children, ages 6–11 years old. The research from four international teams conducted during the development of new instruments and methods will be presented. Classic psychometric methods and cognitive interviewing studies address the accuracy of the use of recall periods in work to be presented by Dr Forrest from CHIP-CE validation research. Results from a series of qualitative research studies will be reported by Ms Serra about the work carried out by the team in Barcelona in the adaptation of the CHIP-CE. Their research further characterizes the ways in which children use specific health-related words and concepts. A series of qualitative and psychometric studies in different countries will be reported by Dr Ravens-Sieberer for the EU team developing the generic health status instrument, the KIDSCREEN. Dr Raczak will report the results of an item-response theory (IRT) approach to development of computer-adaptive testing (CAT) capability for assessing asthma in children. Finally, Dr Ware will integrate the findings from these studies in the context of the development and standardization of assessments for children and others with literacy or cognitive limitations. # 1793-A/THE SCIENCE OF CHILDREN’S SELF-REPORTS OF HEALTH: IMPLICATIONS FOR THE FUTURE AND FOR CHILDREN AND TEEN HEALTH ASSESSMENTS Christopher Forest, Health Policy and Management, Johns Hopkins University, Baltimore, MD, USA Evidence that children can provide reliable, valid health information is not well documented. Research done during the development of the Child Health and Illness Profile, Child Edition (CHIP-CE) shows strengths and limitations in children’s self-reports of health. These findings demonstrate an age-related gradient in their understanding of health-related terms, as well as the gradients in the test–retest reliability of reports of health by children associated with age, cognitive ability as measured by a short, standardized measured of verbal and quantitative ability (K-BIT), reading level as measured on a basic reading skills scale of the Woodcock–Johnson Achievement Test, and socioeconomic level of their family. Nonetheless, the reliability of even young children is acceptable and likely to prove of significant value in describing trajectories of health. Recommendations are discussed for future research to improve utility of health reports from children as well as low literacy populations from adults. # 1793-B/WHAT CAN CHILDREN AGED 6–11 REPORT ON THEIR PERCEIVED HEALTH? A QUALITATIVE APPROACH USING THE CHILD HEALTH AND ILLNESS PROFILE, CHILD EDITION (CHIPCE) MODEL Vicky Serra-Sutton, Luis Rajmil & Imma Guillamon, Cartalan Agency for Health Technology Assessment and Research; Maria Jose Fernandez-Sanmamed, Atencio Primaria; Jordi Alonso, Inst. Municipal d’Investigacio Medica (IMIM), Barcelona, Spain; Anne Riley, Health Policy and Management, Johns Hopkins University, Baltimore, MD, USA The CHIP-CE is a global health status measure developed for children aged 6–11 in the US. It includes 45 items in 5 domains: Satisfaction, Comfort, Resilience, Risks & Achievement with illustrations that help children understand the content of each item. In adapting the CHIP-CE to Spanish, we assessed what children could explain about their health with special focus on children under 8. Forty-six semi-structured probing interviews to 6–11 year olds were carried out by a sociologist

and a psychologist. Emphasis was made on the understanding of specific words, items, figures, answer categories and recall-period (e.g. if children reported feeling worried, they were asked what the word worried: meant, when did they feel worried, why, and to explain the question in their own words). After each interview the research team discussed the understanding of items and proposed alternatives. All interviews were recorded and transcribed for content analysis. Children reported on several experiences related to physical and emotional symptoms, self-esteem, family and friend support, physical activity, risk behaviors and school achievement. Older children reported with more detail their perceived health experiences and could recall on different aspects of a same construct (e.g. examples of when they felt sad, worried or were afraid). The examples and comments of children were useful to change the content of items (e.g. children described the term ‘ideas’ as what they discussed with friends or kept private; they described ‘opinions’ as what they discussed with parents). The differences between these concepts were subtler for 6– 7 year olds but confirmed that children this age can appropriately answer questions on the CHIP-CE. A second reconciled version of Spanish CHIP-CE was proposed resulting from interviews. We found this qualitative approach useful in assessing what children can understand and report on, using the CHIP-CE to assess different aspects of their health status. # 1793-C/ITEM-RESPONSE-THEORY ANALYSES OF CHILD AND ADOLESCENT SELF-REPORT QUALITY OF LIFE DATA: THE EUROPEAN CROSS CULTURAL RESEARCH INSTRUMENT KIDSCREEN Ulrike Ravens-Sieberer & Michael Erhart, Research Unit Child Health and Quality of Life, Robert Koch Institute, Berlin, Germany; Mick Power, Department of Psychiatry, University of Edinburgh, Edinburgh, United Kingdom; Pascal Auquier, University Hospital of Marseille, Marseille, France; Bernhard Cloetta, University of Bern, Bern, Switzerland; Curt Hagquist, Karlstad University, Sweden; Jeanet Bruil, TNO Prevention and Health, Leiden, The Netherlands; Wolfgang Duer, Ludwig Boltzmann-Institute for Sociology of Health and Medicine, Austria; Luis Rajmil, Catalan Agency for Health Technology Assessment & Research, Barcelona, Spain Today the concept of health related quality of life (HrQol) is a central research theme in the large field of health-sciences. However the quality of life of children and adolescents is still an under-researched area with instruments lacking to assess cross-cultural HrQoL on a high psychometric level. The European KIDSCREEN Project funded by the EC and involving 13 European countries (A, CH, D, E, F, GB, NL, HU, PL, CZ, IRL, GR, S) aims to simultaneously develop, test and implement a generic measure of HrQoL for children and adolescents which can be used in public health surveys throughout Europe. After item generation using theoretical considerations, expert delphi methods, and focus groups with children the European pilot test of the project resulted in a data set of 3019 children and adolescents (8–18 years) who filled in the pilot KIDSCREEN instruments as a basis for item reduction together with assessments of determinants of health. After examining the cross cultural acceptance and relevance of the KIDSCREEN questions by cognitive interviews the dimensionality of items was analysed using confirmatory factor analysis and the results were compared over the countries with means of structural equation modeling. In addition the application of item-response-theory techniques using Parscale, Testgraf and Winmira led to further item reduction with the focus of analyses of identifying items showing differential item functioning (DIF) cross-culturally. The control of DIF enables a comparable measurement of the identified quality of life dimensions across the different European countries. Based on these analyses an instrument with 53 items was constructed which assesses 10 relevant HrQol dimensions as well as a short monitoring version for children and adolescents. The scales proofed to be unidimensional and the items fitted the data well according to the probalitistic (Rasch) ‘Partial Credit Model’ (p ¼ 0.098–0.057).

723 # 1793-D/DYNAMIC ASSESSMENT OF PEDIATRIC FUNCTIONING John E. Ware Jr., Anastasia E. Raczek & Stephen Haley, PT, QualityMetric Inc., Lincoln, RI; Wendy Coster, Sargent College of Health & Rehabilitation Sciences, Boston University, Boston MA, USA

# 1647-A/IMPACT OF QUALITY OF LIFE MEASUREMENT IN PSYCHOGERIATRIC RESEARCH AND CLINICAL PRACTICE Monika Bullinger, Department Medical Psychology, University of Hamburg, Hamburg, Germany

This presentation will describe the development, and demonstrate the use of, an important advance in the technology of physical disability assessment for children and youth. The PEDI-CAT system, based on the Pediatric Evaluation of Disability Inventory (PEDI), is a precise, yet practical, tool for identifying disability and evaluating individual progress across multiple pediatric age groups and care settings. We used Item Response Theory (IRT) and Computer Adaptive Testing (CAT) technology to yield an assessment for clinical decisions regarding disability status, service eligibility, and monitoring intervention outcomes for individual children that can easily be used in the clinical setting. In simulations and clinical pilot studies, the adaptive PEDICAT prototype achieved comparable scoring levels, precision, and responsiveness to change as the full-length PEDI form, at 14% of the respondent burden (as measured by number of items and testing time required).

A review of the literature of quality of life measurement in psychogeriatric research and clinical practice shows that area is not yet well researched. However, epidemiological studies as well as clinical studies, although not specifically focussing on elderly persons, contain information about aging and quality of life. Using both literature review and statistical analysis of available own data sets (e.g. German representative health survey using the SF-36 Health Survey instrument in over 7000 persons, as well as a cohort study on quality of life in older patients with stroke, diabetes and rheumatoid arthritis in over 700 persons), relationships between aging and quality of life were examined. In the review, key-words such as quality of life, geriatrics, aging, elderly, health state were used, searching for a recent literature from 1990. Data analysis within own data sets used statistical methods such as analysis of variance, correlation analysis and regression. The review of the literature shows very few studies on interventions in geriatric populations with regard to quality of life, more information is available from the public health perspective. In geriatrics efforts to develop quality of life measures for the older population are ongoing (e.g. WHOQOL-old-project). Results from own studies show an age specific deterioration of physical quality of life dimensions, which is especially pronounced in severe chronic disease. In conclusion, application of quality of life measures into psychogeriatric research and practice is relatively new. Tasks for the research community are to critically evaluate available measures, to specifically construct measures for differential in the geriatric population and to make them available for future research and clinical practice.

# 1647/QUALITY OF LIFE IN AGING POPULATIONS Chair: Nicole von Steinbuechel, Clinic of Geriatric Psychiatry, Center of Neurogerontopsychology, Geneva, Switzerland A review of the literature of health-related quality of life (HRQOL) assessment in psychogeriatric research and clinical practice shows that this area is not yet well investigated. Epidemiological as well as clinical studies exist, although not especially focussing on elderly persons, they contain valuable information about aging and HRQOL. However, there are some studies in the psychogeriatric context e.g. the assessment people with depression or dementia. The approach to measure a cognitive concept via a cognitive modality in elderly people who are cognitively impaired, presents a particular theoretical as well as empirical challenge, which will be discussed in this symposium. (1) A review of the concept and impact of HRQOL-measurement on gerontological, geriatric research and clinical practice, especially depression in the elderly assessed e.g. with the well known health status instrument, the SF-36 will be presented. (2) An insight in other operationalisations and validations in elder populations, e.g. in an international simultaneous developmental approach (of the WHOQOLOLD, a new instrument especially devised for elderly populations) will be offered, as well as a critical discussion of the applicability of the HRQOL-concept in dementia especially with respect to reliability and feasibility data of different generic and disease-specific instruments. (3) Methodological issues in assessing QOL in patients with dementia e.g. their ability to self-report, the role of proxies in the evaluation and the development and application of a fixed disease specific self-rating instrument (DQOL) will be discussed. (4) A brief overview of individualized assessment of QOL-assessment in people with dementia and their carers (with the QOLAS) will be provided. Intensive discussion of theoretical, methodological and empirical questions, characteristic for the assessment of HRQOL in the elderly will be stimulated after each presentation as well as after all four here in form of penal discussion with the plenum.

# 1647-B/CONCEPTUAL AND METHODOLOGICAL ISSUES OF ASSESSING QUALITY OF LIFE IN DEMENTIA Meryl Brod, The Brod Group, San Francisco, CA, USA Assessing Quality of Life (QoL) in Dementia poses both conceptual and methodological challenges. This presentation will present a robust conceptual framework for QoL in dementia that was developed from clinical experience, the literature, expert and caregiver interviews and patient focus groups. The framework includes domains of Self Esteem, Positive Affect, Negative Affect, Feelings of Belonging and Sense of Aesthetics. The process for developing the DQoL, the first patient reported measure of QoL for this population that is based on the conceptual model will be outlined. Additionally, the psychometric qualities of the DQoL that supports the reliability and validity of the 29item instrument will be presented. Methodological issues in assessing QoL including dementia patients ability to self report, the use of proxies, questionnaire formatting, logistical issues in assessment and comparisons between healthy aging and demented patients will be discussed.

724 # 1647-C/INDIVIDUALIZED ASSESSMENT OF QUALITY OF LIFE (QOL) Caroline Selai, The National Hospital for Neurology and Neurosurgery, Institute of Neurology, London, England, United Kingdom This presentation will begin with a brief overview of individualized assessment of Quality of Life (QOL), as distinct from other approaches e.g. the fixed questionnaire. I shall then describe a study conducted at the National Hospital, Queen Square, in which we used such an approach to assess the QOL of patients with dementia and their carers. We adapted a generic, individualized, patient-centered quality of life (QOL) assessment technique, the Quality of Life Assessment Schedule (QOLAS) for use with patients with dementia. The QOLAS was administered to a group of patients with mild-to-moderate dementia alongside a number of other measures of well-being to assess its psychometric properties. Each patient’s main carer also completed the QOLAS, giving a proxy rating of the QOL of the patient. The patients understood the interview and were able to describe their quality of life both qualitatively and quantitatively. In this preliminary study the QOLAS was demonstrated to have good validity, (content, construct and criterion) and good internal reliability. The carers rated the patients as having a worse QOL than did the patients themselves on all subdomains of the QOLAS. The results suggest that patients with mild-tomoderate dementia can rate their own QOL and that the QOLAS is a promising method for assessing QOL in this patient group. The discrepancy between the patients’ own views and the views of their carers raises important issues about whether the patient or a proxy is the best judge of QOL in patients with dementia. #1647-D HEATLH RELATED QUALITY OF LIFE IN AGING POPULATIONS: CONCEPTS AND IMPLEMENTATIONS Nicole von Steinbuechel & Nicole Weber, Clinic of Geriatric Psychiatry, Center of Neurogerontopsychology, Geneva, Switzerland This presentation will give a critical insight in theoretical considerations, measurement models and different fields of application of health-related quality of life (HRQOL) assessment in the ageing population with and without dementia. Widely accepted components of HRQOL are physical, psychological and social factors, which are assessed on the subjective as well as on the behavioural level via selfand/or observer rating. Especially in geriatrics and psychogeriatrics some authors add facets as spiritual, mental, environmental, ethical and aesthetical aspects. In this paper answers to questions as ‘what, why, how, in whom and by whom HRQOL should be measured’ will be addressed. In the past, in elderly populations – with or without cognitive impairment – most often only proxy measures of objective pathology, e.g., limitations in activities of daily living or cognitive impairment have been used to indicate HRQOL. Yet, for older people, an assessment of health status and HRQOL may be a particularly relevant possibility for doctors and patients to decide whether an intervention is beneficial, as opposed to only improving a particular set of symptoms or prolonging life, however with less quality. A special overview of the development of HRQOL research in dementia will be given. Here, some authors question the reliability of HRQOL self-ratings, yet I will show that this endeavour is valid and reliable. International and national aspects of the most recent advancements concerning developments in generic (the WHOQOL-OLD and the MLDL) and disease-specific questionnaires (the D-QOL and QOL-AD) in elderly population with and without dementia will be considered.

# 1632/THE VALUE OF HEALTH-RELATED QUALITY OF LIFE ASSESSMENTS IN ROUTINE ONCOLOGY CLINICAL PRACTICE: ISSUES AND EVIDENCE Chair: Neil Aaronson, Psch. Research, The Netherlands Cancer Institute, Amsterdam, The Netherlands The use of standardized measures of health-related quality of life (HRQL) in routine clinical practice may influence doctor–patient communication, clinical decision making and improve the process of care. This theory is not consistently substantiated by randomized trials and many barriers prevent routine inclusion of HRQL in oncology. This symposium will synthesize current clinical and theoretical evidence for the use of HRQL assessments in clinical practice. The first paper reviews the literature on the barriers to the use of HRQL measures in clinical practice and the theoretical basis of the intervention to provide guidance for future research. The next three papers present empirical evidence from recent studies that have examined the feasibility and impact of using HRQoL measures within routine oncology practice. The second paper summarizes three related studies examining the attitudes and behavior of physicians and patients toward HRQL-related issues, the role of HRQL factors in treatment decision-making, and the effect of standardized HRQL assessments on physician-patient communication and physicians awareness of patients problems. The third paper presents the results of a study investigating the acceptability and effect of standardized HRQL assessments on physician-patient communication and patient well being. The fourth paper reports on clinicians and patients views of the utility and feasibility of using telephone and computer-assisted technology to integrate weekly HRQL information into the routine management of patients. Finally, the discussant will review the symposium from a clinical perspective and raise a number of issues for further discussion and future research. These include approaches to implementation, linking HRQL and clinical data with treatment guidelines and methodological issues in examining the impact of HRQL measures on decision making. # 1632-A/HRQOL ASSESSMENTS IN CLINICAL PRACTICE: A REVIEW OF BARRIERS TO THEIR USE AND GUIDANCE FOR FUTURE RESEARCH Joanne Greenhalgh, Andrew F. Long & Rob Flynn, Health Care Practice R&D Unit, University of Salford, Salford, United Kingdom Systematic reviews have concluded that the use of HRQoL measures within routine clinical practice improves doctor patient communication and the detection of patients’ problems but has little impact on the subsequent processes or outcomes of care. This paper will critically review the empirical explanations for their lack of effect and further develop the theoretical basis of the intervention to provide guidance for future research. The empirical literature on the barriers to the use of HRQoL measurement within clinical practice will be reviewed. The impact of HRQoL measures on clinical decision making depends on a number of factors including the aims of treatment, patient characteristics, the training of clinicians and their attitudes to health status measures. Trials have rarely addressed these issues sufficiently in their design and there is wide variation in the nature of the intervention across the trials. The theoretical basis of the trials that have evaluated the impact of this intervention will be examined. This impact has been operationalized in a number of different ways, ranging from proximal outcomes relating to the dynamics of the consultation, to outcomes of the decision making process, to more distal outcomes such as patient satisfaction. Until recently, the majority of trials have focused on the measurement of the middle set of outcomes and have paid less attention to understanding how health status measures influence the process of clinical decision making. No overall theoretical framework to describe the relationships between different sets of outcomes has been adequately developed. The trials have focused on an assessment of whether the intervention ‘works’ but have paid less attention to understanding ‘how it works.’ This paper will begin to address these issues by highlighting the factors that may contribute to the impact of this intervention and presenting a theoretical framework to guide future research.

725 # 1632-B/THE ROLE OF HEALTH-RELATED QUALITY OF LIFE (HRQL) ASSESSMENTS IN DAILY CLINICAL ONCOLOGY PRACTICE: NATURALISTIC FINDINGS AND EXPERIMENTAL RESULTS Symone Detmar, TNO-Prevention and Health, Leiden, The Netherlands; Martin Muller, Lidwina Wever, Jan Schornagel & Neil Aaronson Although there is increasing interest in the use of HRQL assessments in daily clinical oncology practice, relatively little empirical work has been carried out. We will summarize the results of a series of related studies: –patients’ and physicians’ preferences for discussing HRQL issues, –the ‘natural’ communication regarding HRQL issues, –the role of HRQL issues in treatment decisions, and –the value of standardized HRQL assessments in patient care. The study sample included 10 oncologists and 240 patients receiving outpatient palliative chemotherapy. The data were collected via questionnaire, audiotapes and medical chart audits. Patients and oncologists expressed a willingness to discuss HRQL issues. However, the expectation of both was that the other should initiate such discussions. The actual amount of doctor–patient communication devoted to HRQL issues remained limited. In 20 to 54% of the consultations in which patients had serious HRQL problems, the problems were not discussed. In the presence of tumor progression or treatment toxicity, HRQL considerations played little role in treatment decisions. Seventy percent of those patients without tumor progression or treatment toxicity, but with seriously impaired HRQL, continued to receive their treatment as planned. Providing a HRQL summary during consecutive consultations resulted in a significant increase in the frequency with which HRQL issues were discussed, and significantly increased physician awareness of patients’ problems. Minimal effects were found on satisfaction, patient management, and patients’ self-reported HRQL over time. Future efforts should be directed at improving the flexibility and precision of HRQL assessments and linking patients’ HRQL scores to specific treatment and care strategies. # 1632-C/A RANDOMIZED STUDY OF QUALITY OF LIFE (QL) ASSESSMENT IN ONCOLOGY PRACTICE: CLINICAL USEFULNESS OF THE INTERVENTION J. Brown, L. Booth, A.B. Smith, P. Brown, P. Lynch & P. Selby Routine assessment of QL in oncology may improve process of care and patient well being. This study randomized 286 patients to (1) Intervention group: touch-screen completion of EORTC QLQ-C30 and HADS with results reported to physicians; (2) Attention-control group: touch-screen completion of questionnaires alone; (3) Control group: no clinic QL measurement. Primary outcomes were patient QL (FACT-G) and doctor–patient communication (audio taped consultations). Doctors’ and patients’ perceptions of clinical usefulness of QL data were assessed with a checklist after each consultation and with end-of study questionnaires. In the intervention group (n ¼ 113), a positive effect on doctor–patient communication and patient well being was found. Improvement in patient well being was associated with explicit use of QL data during consultations (p ¼ 0.02) and discussion of pain (p ¼ 0.03). Doctors referred to QL data in 64% of intervention consultations and 96% stated that they used the information. Fifty-five percent said this resulted in management changes and 68% expressed interest in using the system in the future. Eighty-six percent of patients in the intervention group felt that QL questionnaires helped during consultations and 92% supported their use in routine practice. Routine assessment of QL was well accepted by patients and oncologists and improved communication and patient well being. Further research will seek to maximize the benefits using strategies for changing medical practice and training of physicians in how to work with QL data.

# 1632-D/REAL-TIME SYMPTOM MONITORING OF PATIENTS WITH ADVANCED LUNG CANCER Kimberly Davis, Shaheen Khan, Michael Bass, David Eton, Greg Masters & David Cella For patients with advanced lung cancer, information on health-related quality of life (HRQL) and other patient-reported outcomes could facilitate treatment decision-making. One barrier is the availability of information that is summarized in an interpretable manner and available during a clinic visit. Computer technology can facilitate this but has rarely been applied to individual decision making. The objective of the Advanced Lung Cancer Management Program-2 (ALCaMP-2) 1year pilot is to assess the feasibility and acceptability of a computerized telephone assessment program to monitor symptoms in lung cancer patients and to examine its usefulness in treatment decisionmaking. Patients call every week into a telephone assessment of a 14item FACT-Lung Symptom Index (FLSI-14) and complete the FACTLung (FACT-L) and treatment satisfaction questionnaire at weeks 6 and 12. Patient symptom responses are monitored weekly; patients with a 2-point worsening of any symptom are contacted within 24 hours and the treating physician is notified. Summary reports displaying symptom and HRQL information are generated for discussion with patients at each clinic visit, and patients and clinicians provide feedback on the acceptability of the program and report. The feasibility and acceptability by patients and clinicians will be presented along with symptom, HRQL and treatment satisfaction data. To date, patient compliance with weekly assessments is 97%. Sixty four percent of patients have identified mid-cycle symptoms warranting clinician notification. Several patients have spontaneously requested to continue weekly assessments beyond the scheduled 12 weeks. The ultimate goal is to integrate symptom and HRQL assessments into clinical practice to identify potential problems early, prevent serious problems from developing, and improve patient-provider discussions about treatment decisions. # 1249/CHALLENGES IN TRANSLATION METHODOLOGY: COMPARING APPROACHES AND A DISCUSSION OF GOOD PRACTICES Chair: Sonya Eremenco, Evanston Northwestern Healthcare, Center on Outcomes, Chicago, IL, USA Synopsis: This symposium seeks to bring together different practitioners of translation for use in quality of life (QOL) research to share expertise with the larger body of researchers on good practices in translation methodology. The symposium will include a presentation on an effort to achieve consensus on good practices in translation across disparate methodologies currently in use; a presentation on four aspects which, besides the generally accepted Linguistic Validation methodology are essential to producing conceptually equivalent translations (translatability assessment, list of concepts, work in the target country, international harmonisation); a presentation on the use of a universal language approach to the translation of patient reported outcomes (PRO) measures; and a presentation of different approaches to cultural adaptation and its role in the translation of PROs for use in multinational studies. Finally, a discussant will comment on the issues raised in each presentation and raise questions for discussion by all participants. The goal of this symposium is to promote rigorous translation methodology to ensure validity of translated patient reported outcomes measures intended for data-gathering purposes.

726 # 1249-A/THE DEVELOPMENT OF A SET OF PRINCIPLES OF GOOD PRACTICE: PROGRESS OF THE ISPOR SPECIAL INTEREST GROUP Diane J. Wild, Oxford Outcomes, Ltd., Oxford, England, United Kingdom In 1999 a series of special interest groups were setup by ISPOR in order to try and address the expressed needs of the ISPOR membership. The translation and cultural adaptation group has met five times and has been working towards the development of a document which describes principle of good practice in this area. In this effort we have been interpreting principles of good practice to represent a standardized criteria by which new translations can be developed in order to support the rigorous requirements of the PRO-linked research process. A working group was established early on in this process which includes representatives from the CRO, industry, and academic communities. The approach taken by the group builds upon current practice, taking account of existing guidelines and theoretical perspectives. These diverse perspectives have been evaluated with regard to the issues and challenges facing industry and the broader outcomes research community. Thirteen sets of guidelines were identified and critiqued. The translation and cultural adaptation process was divided into its requisite components (forward translation and reconciliation, back translation etc.) and a consensus of guidelines was taken by the working group. Based on the knowledge and expertise of the working group the existing guidelines were summarized and criteria for achieving success were suggested. The document produced by the group is tabular in format and includes four columns: name of step, key components, rationale, and risks of failing to complete the step. This presentation will focus on the nature of the principles of good practice document, on the challenges of achieving consensus in an area where there are a number of methodologies used, and on the intended uses of the document once finalized. # 1249-B/4 ASPECTS OF LINGUISTIC VALIDATION METHODOLOGY Isabelle Mear, MAPI Research Institute, Lyon, France Our presentation will focus on four aspects which – besides the generally accepted forward/backward/cognitive debriefing steps – we regard as essential to producing conceptually equivalent language versions which will allow the international pooling of data: (1) the international review of the original questionnaire to check the translatability of concepts prior to finalizing the instrument (translatability assessment), (2) the establishment of a list of concepts as a basis of the translation process with the author, (3) the importance of conducting the translation process in the target countries and (4) the necessity of an international comparison of language versions when the process is conducted in more than one country (international harmonization). Examples for the four aspects will be presented and discussed in the presentation.

# 1249-C/UNIVERSAL LANGUAGE APPROACH TO TRANSLATION Sonya Eremenco, Evanston Northwestern Healthcare, Center on Outcomes, Chicago, IL, USA This presentation will focus on the issue of the ‘universal’ language approach to translation that is essential to the FACIT translation methodology. In this approach, a ‘universal’ translation is developed for languages which are spoken in multiple countries. This approach focuses on finding commonalities and consensus between the different regional dialects so that a single version is appropriate and valid in multiple cultures. This approach requires certain steps to be taken during the translation phase and it is necessary to test the same version of a translation in multiple countries to ensure that the universal version works both qualitatively and quantitatively. Examples of FACIT Translations in Spanish and Portuguese will be presented to show how the universal version is achieved through translation and then the testing results in each case. Finally, we will examine different types of patient reported outcomes measures which may not be suitable for the universal approach, such as health resource utilization questionnaires. The conclusion is that the suitability of the universal language approach depends a great deal on the type of source measure that is being translated, and when performed appropriately, this approach can yield a valid measure that can be used across regions and minimizes heterogeneity in clinical trials. # 1249-D/CULTURAL ADAPTATION: SHOULD WE DO IT? PROS AND CONS David U. Himmelberger, Health Outcomes Group, Palo Alto, CA, USA Cultural adaptation or localization, as it is called in the marketing world, is an attempt to homogenize the intrinsic differences in culture and language throughout the world. For the objective of developing standardized questionnaires in multiple languages that are evaluating common concerns, this issue is best addressed by minimizing these differences in choosing content that is universal. Nevertheless when one is trapped with a source document that has a content that is not universal one must make compromises in achieving commonality. The most common approach is to make substitutions that are locally relevant while maintaining the intent of the developer in the source language. This approach relies on having the developer articulate the ‘intent’ and defining what are acceptable substitutes. It is very uncommon to define the ‘intent’ a priori because the developer usually does not perceive a need for it. Furthermore there are not well established standards for what constitutes an acceptable substitution. This process of substitution generally results in a ‘new’ questionnaire with a content that is understandable in a new culture and language, but is not equivalent to the source. On the other extreme, is the approach where no cultural substitutions are allowed, except for measurement conversions. Measurement conversions are relatively noncontroversial, for example 0 C ¼ 32 F, however the degree of accuracy is frequently debated. For example, is 5 lb ¼ 2.27 kg or about 2 kg? In this approach linguistic equivalence is given priority at the possible sacrifice of cultural relevance. There are many issues around idomatic and colloquial expressions that are intimately intertwined with language and culture. These issues are even more problematic. Examples will be presented to amplify the pros and cons of the different approaches to cultural adaptation and challenge its role in translation of standardized questionnaires in multiple languages. Do we want translation or interpretation?

727 # 1281/EVALUATING MEASUREMENT EQUIVALENCE: A CONCERN FOR QUALITY OF LIFE RESEARCH Karen H. Sousa, College of Nursing; Fang Fang Chen, College of Psychology; Sheila Tann, College of Nursing, Arizona State University, Tempe, AZ, USA When quality of life measures are used to examine differences in diverse groups evidence of the equivalence of conceptual and measurement properties should be evaluated. The risks of misinterpreting results of quality of life research without assurances of equivalence are great. The purpose of this paper is to provide an example of a statistical procedure, invariance analysis, which is ideally suited to address conceptual and psychometric equivalence across groups. The sample for this analysis (n ¼ 924) is persons living with HIV. The sample was divided into two groups based on work status. QOL was measured using a second-order factor model with four dimensions: cognition, vitality, mental health, and health worry. Five items represented cognition and the item-scale correlations ranged from 0.59 to 0.75. Vitality was measured by four items and in the Health Insurance Experiment demonstrated adequate internal-consistency (a ¼ 0.81). The five mental health items in the Medical Outcome Study had itemscale correlations ranging from 0.65 to 0.81. The item-scale correlations for health worry (three items) from the Medical Outcome Study ranged between 0.76 to 0.87. Two statistical queries were addressed: Do the measured variables operate equivalently across the groups? and is the factorial structure of the theoretical concepts equivalent across the groups? The adequacy of conceptual and measurement properties across the two groups were tested within the framework of confirmatory factor analysis using structural equation modeling. A series of nested models were tested for baseline/configural invariance, factor loadings invariance, and invariance of factor loadings and intercepts of the observed variables. The v2 likelihood ratio statistic and fit indices (CFI, RMSEA, SRMR) were used to assess model fit. The analysis indicates that the factor structure between the groups was equivalent and the invariance in the factor loadings will be discussed. The use of invariance analysis is an imperative step in the identification and development of instruments that are intended to be used across diverse groups. # 1575/COPARABILITY AND EQUIVALANCE OF EORTC QLQ-C30 AND FACT-G Bernhard Holzner, Department of Biological Psychiatry, University Hospital Innsbruck, Innsbruck, Tyrol, Austria; Stacie Hudgens & David Cella, Center on Outcomes, Research and Education, Evanston Northwestern Healthcare, Evanston, IL, USA; Georg Kemmler & Martin Kopp, Department of General Psychiatry; Barbara SpernerUnterweger, Department of Biological Psychiatry, University Hospital Innsbruck, Innsbruck, Tyrol, Austria Background: There are numerous well accepted and widely used quality of life (QOL) instruments in oncology currently available. Each of these typically consists of different rating scales and items; however, the underlying domain-specific constructs measure the same trait(s). Hence, a model for equating these instruments would be quite useful for health care providers, as well as outcomes researchers. Methods: For the purpose of this analysis, internationally used QOL questionnaires, namely the EORTC QLQ-C30 and the FACT-G were evaluated. Subjects included 487 cancer patients and 776 healthy controls; each completed both instruments. Item response theory- and

classical test theory methods were used to evaluate/compare the psychometric properties of both instruments. The EORTC QLQ-C30 items shared one rating scale while FACT-G items shared another. Additionally, the items were anchored at the calibrations obtained from the combined analysis and separate analyses were performed to produce FACT-G and EORTC QLQ-C30 measures for each patient. Furthermore, correlations between the measures were done. Results: Both instruments demonstrate acceptable internal consistency with Cronbach’s a ranging from 0.76 to 0.88, 0.69 to 0.89 for the FACT-G and the EORTC QLQ-C30 respectively, which is an essential prerequisite for comparison. The resulting common 57-item scale had good summary statistics: person separation (PS) 3.5 (reliability 0.91), item separation 12.9 (reliability 0.99). Independent PS statistics indicate that the instruments are moderately comparable (2.08 FACT-G; 2.60 EORTC QLQ-C30). Test efficiency (T.E.: defined as PS divided by the square root of the number of items) showed the EORTC QLQC30 to have a slightly higher T.E. rating (0.47) than the FACT-G (0.40). The combined T.E. was 0.42. Correlation between the FACT-G and EORTC QLQ-C30 patient measures was moderate (0.72). Conclusions: Despite some discrepancies in the social and emotional domain, the FACT-G and the EORTC QLQ-C30 seem to measure a similar QOL construct. When combined, these instruments sufficiently cover the continuum of QOL, which was indicated by the item distribution. # 1486/AN ITEM RESPONSE THEORY BASED PAIN ITEM BANK CAN ENHANCE MEASUREMENT PRECISION Jin-shei Lai, Center on Outcomes, Research and Education (CORE), Evanston Northwestern Healthcare & Northwestern University, Evanston, Illinois; Kelly Dineen & David Cella, CORE, ENH & NU, Evanston, Illinois; Jamie von Roenn, MED-Hematology Oncology, NU, Chicago, IL, USA Many cancer patients experience pain, which is often under-recognized and under-treated. Accurate pain assessment requires comprehensive and precise measurement that is easily integrated into clinical practice. Single item scales (e.g., 0–10 ratings) are popular but can be unreliable. Modern computerized adaptive testing (CAT) can enable the development of a highly reliable and clinically useful assessment by selecting well-targeted items from a calibrated item bank. The purpose of this study was to create such a bank. The sample included 400 cancer patients from the metropolitan Chicago area. A total of 61 pain items were administered: 43 existing questions and 18 newly written by clinicians with experience in oncology pain. An Item Response Theory (IRT) model was used to analyze data. The final bank consisted of 43 items that did not exhibit differential item functioning (DIF) across gender, education, cancer types and races. These 43 items satisfied IRT (Rasch) measurement requirements, demonstrated high internal consistency (Cronbach’s a ¼ 0.97) and reasonable item-total correlations (range 0.30–0.90). When we divided all patients into five different groups based on their response to item I have pain, significant mean differences across groups were found F(4, 288) ¼ 198.98, p < 0.0001. All but one pairwise comparison was statistically significant in post hoc analysis (Tukey HSD). Patients who had ‘very much pain’ did not significantly different from patients who had ‘quite a bit’ of pain. In conclusion, this bank demonstrates good psychometric properties and is sensitive to patients’ pain. This bank can be used as the foundation for a CAT pain-testing platform to use in clinical practice.

728 # 1513/INCREASING EFFICIENCY OF THE MULTIPLE SCLEROSIS QUALITY OF LIFE INVENTORY: USING CLINICAL TRIAL DATA TO IMPROVE MEASUREMENT PRECISION Deborah M. Miller, Mellen Center/Neurology, Cleveland Clinic Foundation, Cleveland, OH; Monika Baier, Cooper Institute, Golden, CO; Gary Cutter, Applied Research Facility, University of Neveda, Reno, NV, USA The IMPACT trial showed benefit of IFN-beta 1a on disease progression and HRQoL using the Multiple Sclerosis Quality of Life Inventory (MSQLI) in a randomized trial of secondary-progressive multiple sclerosis (S-PMS) patients. The MSQLI includes the SF-36 and 9 symptom-specific scales. We reviewed IMPACT results and examined MSQLI ceiling/floor effects and change over time to begin a process improving psychometric properties and reducing the number of items in the MSQLI. The MSQLI was administered only within the English speaking Clinics yielding 324 subjects with HRQoL data of the 436 study participants. MSQLI scores were comparable for both study groups at baseline. At month-12 subjects in the IFN group was significantly better on 2 scales and trended to better functioning on the remaining scales. By month-24 IFN subjects had statistically better HRQoL on 8 of the scales and tended towards improvement on the remaining scales. Baseline MSQLI measures had few missing data point and Subjects endorsed the range of responses. Ceiling/floor effects > than 15% were noted at baseline for the following: SF-36 scales Role Physical (46% floor), Role Emotional (49% ceiling), Bodily Pain (20% ceiling) and Social Functioning (20% ceiling) Similar ceiling effects were noted for 3 of 9 disease specific scales: Bladder Control (16% no symptoms), Bowel Control (21% no symptoms) and Impact of Visual Impairment (44% no symptoms). Over the 24-month period, floor/ceiling effects remained as follows: SF-36 RP (61% no change), SF-36 RE (68% no change), SF-36 BP (88% no change), Bladder Control (49% no change), Bowel Control (55% no change) and Impact of Visual Impairment (73% no change). Given the significant disease of these subjects at baseline, the impact of floor/ceiling effects across all MS patients could become problematic. Thus these data indicate that the MSQLI is sensitive to change in patients with advanced MS disability but also suggest that revisions to the MSQLI should be evaluated based on available cross-sectional and longitudinal data. # 1508/ITEM RESPONSE THEORY AND CONFIRMATORY FACTOR ANALYSES TO ASSESS THE VALIDITY OF THE PRELIMINARY VERSION OF THE EUROPEAN CROSS CULTURAL KIDSCREEN PROXY MEASURE S. Robitail, Perceived Health Research Unit, Public Health Department, School of Medecine, Marseille, France; M. Erhart, Research Unit Child Health and Quality of Life, Kidscreen Group, Berlin, Germany; B. Cloetta, Bern University, Kidscreen Group, Bern; Switzerland; M. Power, Edinburgh University, Kidscreen Group, Edinburgh, UK; D. Orbicini, Perceived Health Research Unit, Public Health Department; School of Medecine, Marseille, France; A. Gosch, Research Unit Child Health and Quality of Life, Kidscreen Group, Berlin, Germany; M.C. Simeoni, Perceived Health Research Unit, Public Health Department, School of Medecine, Marseille; Ulrike Ravens-Sieberer, Research Unit Child Health and Quality of Life, Robert Koch Institute, Berlin; P. Auquier, Perceived Health Research Unit, Public Health Department, School of Medecine, Marseille, France The KIDSCREEN project aims to develop a European cross-cultural generic self-administered HRQoL instrument for children and adoles-

cents. Proxy measures HRQoL are an useful alternative to assess children’s HRQoL. In this context the problem of structure comparability between children’s and parents measures must be discussed in order to allow comparisons of reports. The KIDSCREEN pilot study involved 3834 children and 2356 parents in seven European countries (A,CH,D,E,F,GB,NL) and led to a preliminary research instrument with 10 dimensions. The proxy measure instrument is based on the model of the preliminary. KIDSCREEN research instrument. The aim of this study is to assess the validity of the proxy measure instrument in terms of multidimensional structure and the construct of every dimension. Structural equation modelling of the parent’s data allows to test the multidimensional structure of the proxy measure. Item response theory analyses assess the internal construct-validity of every dimension and enables comparable calibrations of corresponding children’s and parents items. Confirmatory factor analyses (LISREL) indicates satisfactory fit of the children’s model to the parents data (RMSEA ¼ 0.053). For every dimension, Rasch Partial Credit Modelling indicate that items form a unidimensional continuum (Infit and Outfit Mean Square ranged between 0.6 and 1.4). Parent-children DIF analyses show interesting results to understand differences between a self report and proxy measure. The internal consistency is satisfactory (0.78–0.87). This new instrument provides a promising proxy measure for assessing HRQoL in European children. # 1051/PARENTAL PSYCHOLOGICAL WELLBEING IN FAMILES OF PRESCHOOL CHILDREN ADMITTED TO A NICU AT BIRTH Anne F. Klassen, Herbert W. Chan & Shoo K. Lee, Pediatrics, University of British Columbia, Vancouver, British Columbia, Canada Background: Neonatal intensive care is associated with a range of long-term health problems such as cerebral palsy, mental retardation, deafness, blindness and milder but more common problems such as learning disabilities and behavioral problems. These health problems can create challenges for the parent responsible for the day-to-day provision of care to their child at home. Objectives: To compare psychological well-being in parents of NICU children and parents of healthy full-term children; and to investigate factors associated with poorer psychological well-being in both samples. Methods: A questionnaire booklet was sent to the caregiver of all babies admitted to any level III NICU in the province of British Columbia (Canada) over a 16 month period (n ¼ 2221) and a group of healthy full-term infants (n ¼ 718). The SF-36 mental health score was our main outcome measure. Bivariate analysis was used to compare both samples, and to identify predictor variables associated with the primary outcome measure. Staged multiple regression analysis was used to examine the independent effects of, and proportion of variance explained by, predictor variables. Results: The response rate to our survey was 68%. Caregivers of the NICU did not differ from caregivers of healthy children on any SF-36 psychological domains, nor in the mental component summary score. In the final stage of the regression analysis, the following predictor variables were significantly associated with poorer caregiver mental health: female gender (caregiver); marital status (unmarried); pain/discomfort(child); general health (child); behavior problems; caregiver demand; and poorer family function. These variables explained 24% of variance in scores. Conclusions: Studies of psychological stress in parents beyond the neonatal period, comparing NICU children with healthy children are few and tend to report inconsistent results. We did not find any differences in psychological wellbeing between parents of NICU and healthy children but we did identify certain variables (esp child behavior) associated with poor parental outcome.

729 # 1709/IS HRQOL IN CHILDREN WITH CHRONIC CONDITIONS ASSOCIATED WITH PARENTAL HRQOL? Ute Thyen, Esther Mueller-Godeffroy & Verena Wagner, Klinik fuer Kinder- und Jugendmedizin, Universitaetsklinikum Schleswig-Holstein, Luebeck; Andreas Hahn, Abteilung fuer Neuropaediatrie, Universitaetskinderklinik Giessen, Giessen; Ulrich Seidel, SPZ Bereich Neuropaediatrie, Otto-Heubner-Centrum, Virchow-Klinikum, Berlin; Hartwig Lehmann, Klinik fuer Kinder- und Jugendmedizin, Saechsisches Krankenhaus, Wermsdorf; Johannes Oepen, Rehabilitationsklinik fuer Kinder und Jugendliche, Viktoriastift, Bad Kreuznach; Ulrike RavensSieberer, Child and Adolescent Health, Robert-Koch-Institut, Berlin, Germany We sought to measure the impact of various chronic conditions in childhood on self reported health related quality of life (HRQOL) in children and their parents and to determine the association among mothers’, fathers’ and childrens’ HRQOL. We included 272 children age 8–16 years with diabetes type 1 (67), juvenile chronic arthritis (JCA,72), epilepsy (50), spina bifida (MMC,50), muscular dystrophy (DMD,33), and attention deficite hyperactivity disorder (ADHD,64) and both parents. We used the KINDL questionnaire, a German HRQOL instrument for children with six dimensions and an added modul for children with chronic conditions and the SF-36 for parents. Children with conditions involving the central nervous system (CNS) were much less likely to attend regular classrooms. Family status was not different among groups but mothers of children with diabetes, spina bifida and muscular dystrophy were much less likely to be employed outside the home. Except for the family dimension children reported significant different HRQOL in most dimensions: compared to the other groups children with JCA and DMD reported lower physical well-being, those with ADHD reduced well-being in psychological health and in school, those with JCA and MMC low self-esteem, those with MMC and DMD low scores in the friends dimension. Children with diabetes and epilepsy ranked highest in overall and in condition related QOL, not signicant different from the norm population. Mothers of children with MMC and DMD reported significantly lower HRQOL in the area of physical well-being, bodily pain and social role functioning. Fathers did not show relevant differences among groups. We found no clinically important correlations (all <0.25) between children’s and mothers’ and fathers’ HRQOL. The results suggest that certain illness characteristics such as CNS involvement and mobility impairment affects both children’s and carers’ HRQOL. The effects appear not to be moderated by child or parental well-being but functioning has a direct effect. # 1351/PARENT AND SELF-REPORTED QUALITY OF LIFE IN SIBLINGS OF A CRITICALLY ILL CHILD, CORRESPONDENCES AND DIFFERENCES Bregje A. Houtzager, Martha A. Grootenhuis & Bob F. Last, Paediatric Psychosocial Department; Huib N. Caron, Department of Paediatric Oncology, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands Assessment of quality of life (QoL) has thus far been a neglected approach in studies on psychological adaptation in siblings of seriously ill children. In the present study, correspondences and differences between parent- and child-reported QoL were assessed in siblings of paediatric cancer patients, at one and 24 months after the diagnosis of cancer in the ill child as well as the course of QoL over time. The study group consisted of 83 siblings aged 7–18 at 1 month (M1), and 57 siblings (69%) at 24 months after diagnosis (M4). Parent and child completed the TNO-AZL Children’s Quality of Life questionnaire (TACQOL) and the Child Behaviour Check List (CBCL). Parents

completed the General Health Questionnaire (GHQ). Differences and concordance were assessed, and concordance was compared to reference data. Results showed low to moderate correspondence between parent and child for most domains. Children (7–11) reported significantly more physical and motor problems (M1, M4) and lower positive emotions (M4) than parents. Adolescents reported more physical problems (M1) and more emotional/behaviour problems (CBCL, M1, M4). Adolescents reported better social QoL than parents (M4). Lower agreement was observed in our study group compared to reference data on several QoL domains. Parent- and child-reported emotional and total behavioral functioning and parent-reported physical functioning improved. No significant change in agreement was observed between M1 and M4. Parent mental health problems correlated negatively with parent-reported physical QoL at M1. In conclusion, results imply that siblings experience more burden from the illness than parents can observe. Physical complaints and emotional problems remain most unnoticed, although distressed parents are more focused on the child’s physical health. Further studies on sibling QoL are needed. The need to obtain self-reported information from children or adolescents on their subjective psychosocial functioning is especially relevant in siblings of a critically ill child. # 1454/PERCEIVED HEALTH AND USE OF HEALTH SERVICES IN CHILDREN POPULATION IN BARCELONA, SPAIN Silvina Berra, Luis Rajmil & Maria D. Estrada, Research Area, Catalan Agency for Health Tech. Assess. and Research; Carme Borrell, Health Information Service, Agency for Public Health of Barcelona, Barcelona; Vicky Serra-Sutton, Research Area, Catalan Agency for Health Tech. Assess. and Research, Barcelona; Maica Rodrı´guez, Health Information Service, Agency for Public Health of Barcelona, Barcelona Spain; Anne Riley & Barbara Starfield, School of Public Health, Johns Hopkins University, Baltimore, Maryland, USA The aim of this study was to analyse the relationship between use of health services and health care needs in children population. Data were obtained from the Barcelona Health Interview Survey 2000. 5– 14 year-old children were randomly selected (n ¼ 836). Proxy respondents were the mother, the father or other responsible party. Health services utilization variables were: visits to a health care professional in the previous 15 days, and visits to the emergency service and hospitalisation during the year prior to the interview. Children’s needs were assessed by means of the ‘discomfort’ domain, and ‘satisfaction with health’ domain of the parent version of the Child Health and Illness Profile (CHIP-PE), a comprehensive measure of children’s health. Both domain scores were categorized in three levels (lowest, middle and highest). The worse health profile (WHP) was defined when both domains were at the lowest level, and the best health profile (BHP) was defined when they were at the highest level. Sociodemographic and family characteristics were also collected, including self-perceived health and mental health of the proxy-respondent (General Health Questionnaire, GHQ-12). Logistic regression models were adjusted. The factor associated with visits to a health care professional was being in the WHP (Odds Ratio OR ¼ 2.32; 95% confidence interval 95% CI: 1.29–4.17). Some family factors, as well as a poor mental health of the proxy-respondent (scoring 3 or more in the GHQ-12), were associated with visits to the emergency department. The probability of hospitalisation was higher in children reporting a chronic condition (OR: 3.57; 95% CI: 1.74–7.33). We did not find inequalities by social class in the use of health services. The results suggest an equitable use of health services in relation to need. The CHIP-PE seems to be a suitable measure for health care needs in children population.

730 # 1213/LIVING WITH CLASSICAL GALACTOSEMIA Martha A. Grootenhuis, Psycho Social Department, Emma Child Hospital AMC, Amsterdam; Annet M. Bosch, Pediatrics, Emma Child Hospital, Amsterdam, Holland Classical galactosemia is an autosomal recessive disorder of galactose metabolism. Treatment, consisting of a severe restriction of dietary galactose, is life saving. While reports of late complications are a result of medical research, the patients’ experience of their disease has not been investigated. Thus, we studied the influence of galactosemia on health related quality of life (HRQoL) and on educational levels. We used generic questionnaires for different age-levels. Disease specific concerns were also studied. All 75 members of the Dutch Galactosemia Society and their families received questionnaires and were asked to return them by mail. To measure HRQoL we used the TAPQOL for preschool children, the TACQOL for 6–15 year old children (CF and PF) and the TAAQOL for 16 years and older patients. T-tests were performed to test the differences between patients and Dutch norms on HRQoL. Sixty-three classical galactosemia patients from 58 families returned the questionnaire (84%). For preschool children differences were found on the domain of abdominal pain and communication. Children 6-15 years differed on the domain of cognitive functioning. Their parents reported a significantly lower HRQoL on the domains of motor and cognitive function. Older patients reported lower HRQoL for cognitive and social functioning. According to mothers 44% of the younger children attend special schools as opposed to 3% of the general population. For the older patients we found that their educational attainment is also much lower. Many parents feel the galactosemia influences their contact with the child and they are concerned about infertility of their daughters. Sixty percent of the parents feel it is a burden to take care of a galactosemic child. In conclusion, this is the first study to describe the HRQoL of patients with classical galactosemia using well developed and validated instruments in different age groups. In all age groups we find a lower reported HRQol, especially on the cognitive domain. While patient numbers are small, the severe effects of classical galactosemia on the cognitive skills of the patients demand further research and early intervention. # 1274/THE INFLUENCE OF DROPOUTS ON ESTIMATED RATES OF CHANGE FOR COMPLETERS IN A LONGTUDINAL STUDY Sonia Gavriel, Paul W. Jones, Respiratory Medicine, St. George’s Hospital Medical School, London, UK Hierarchical linear models (HLM) are widely used for analysing unbalanced longitudinal data. However the HLM estimates are biased towards patients who complete the study. We have tested the influence of dropouts on patients who complete a study using HLM estimates of change in health status. We used St George’s Respiratory Questionnaire (SGRQ) scores from 298 patients who completed a 3year trial in chronic obstructive pulmonary disease. Seven observations were available per patient (baseline then 6-monthly assessments for 3 years). HLM models contained the entire study population. Each patient was labelled randomly, a priori, to be either a ‘completer’ or ‘dropout’. Two analyses were performed. The first analysis used the orthogonal dataset in which each patient contributed seven observations regardless of whether they were a ‘completer’ or ‘dropout’. The slope of change in SGRQ score in the ‘completers’ was 2.08 SD 2.73 units/year and 1.77 SD 2.60 in the ‘dropouts’. For the second analysis we removed the last four data points from each ‘dropout’ patient and re-ran the analysis. The slope in the ‘completers’ was now 2.02 SD 2.86 units per year and 1.43 SD 1.50 units/year in the ‘dropout’ group. The slope of the ‘dropout’ group changed as expected, however in the ‘completer’ group the difference between HLM estimates of change between the two analyses were statistically significant (paired t test p ¼ 0.0001). Thus the number of observations contributed by patients that dropped out early influenced HLM estimates for patients that completed the study, even though their data set was completely unchanged.

# 1444/IMPUTATION OF MISSING DATA AND SENSITIVITY ANALYSIS IN AN OUTCOME STUDY OF INPATIENT PULMONARY REHABILITATION Henk F. van Stel, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht, Netherlands Introduction: Long-term outcome studies on inpatient pulmonary rehabilitation (IPR) frequently suffer from extensive dropout. The robustness of change in health-related quality of life (HRQL) and hospital admissions after IPR was tested by sensitivity analysis, comprising four scenarios for imputation of missing data, ranging from optimistic to pessimistic. Methods: 56 patients with asthma and 84 patients with chronic obstructive pulmonary disease (COPD) participated in an IPR program. HRQL was measured with the Quality of Life for Respiratory Illness Questionnaire (QoLRIQ). Dropout was divided in treatmentrelated (TD) and study-related dropout (SD). Imputation was done using multiple imputation (with NORM) and using four scenarios with simple imputation methods: Scenario 1, Optimistic: TD and SD show similar improvement as study completers; Scenario 2, Realistic: TD have not improved; SD show similar improvement as study completers; Scenario 3, Sombre: both TD and SD have not improved; Scenario 4, Pessimistic: TD have deteriorated twice minimal important difference (MID ¼ 0.5); SD have not improved. Results: About 20% of the patients dropped out at each follow-up moment (discharge, 6month and 12-month post-IPR). There were no baseline differences between completers and dropouts. Improvement in QoLRIQ-total score at discharge remained significant (p < 0.0001) and clinically relevant (1.09–0.46) using multiple imputation and in all scenarios in both astma and COPD. Multiple imputation gave counter-intuitive results for hospital admissions. Decrease in hospital admissions at 12 months post-IPR (compared to pre-IPR) was large in the completecase analysis, and remained significant (p < 0.01) in all scenarios for COPD, and up to the sombre scenario in asthma. Conclusion: Imputation of missing data with several scenarios, reflecting different assumptions about the possible outcome in dropouts, is a useful method to test the robustness of the change in studies with high dropout. # 1580/INTERMITTENT AND MONOTONE MISSING DATA IN QUALITY OF LIFE RESEARCH: HOW SHOULD WE PROCEED? Graeme Hawthorne & Peter Elliott, Psychiatry, Melbourne University, Heidelberg Heights, Victoria, Australia Although there are many missing data imputation methods in QoL research there is little advice concerning which methods should be used under different circumstances. This paper compares widely reported methods. Six intermittent missing data (e.g. cross-sectional surveys) and six monotone missing data (e.g. longitudinal studies) methods were studied. Using two complete datasets each method was studied under a variety of sample sizes and with differing levels of missing data; the criteria were the complete dataset true t-values. Intermittent missing data methods were listwise deletion, vertical mean substitution, horizontal mean substitution (two levels), regression, and hot deck imputation. The results show important differences. If missing data are from a scale where half the items are present hot deck or horizontal mean substitution are best. Because horizontal mean substitution is simpler than hot deck, similar in its efficiency, and is an option on most statistical software packages, it is the method of choice under these circumstances. If missing data are from a scale where more than half the items are missing, or with single-item measures, then hot deck is recommended. Listwise deletion and vertical mean substitution performed poorly. Monotone missing data methods were listwise deletion, last value carried forward (LVCF), standardized score imputation, regression, and closest match (two methods). The results suggest a reliable and efficacious method is closest match, where a missing datum is substituted with a value from the individual with the closest scores on the same variable measured at other timepoints. LVCF and standardized score performed relatively poorly, which is of concern since these are often recommended. Listwise deletion was also an inefficient missing data method. Some of the intrinsic reasons for these findings are explored. That in both studies the most popular methods (also the default methods in many statistical software packages) – listwise deletion, vertical mean substitution and LVCF – were the worst performing methods, the findings have broad implications for QoL resear.

731 # 1752/LATENT TRAIT MODELLING OF ITEMS WITH MISSING DATA IN QUALITY OF LIFE RESEARCH Thomas Kohlmann, Institute for Community Medicine, University of Greifswald, Greifswald; Heiner Raspe, Institute for Social Medicine, Medical University of Luebeck, Luebeck, Germany

of change were quite different for the treatment arms. This study presented major methodological challenges to QOL data analysis, all of which were addressed using state-of-the-art modeling techniques. The analysis plan and results should be of interest to statisticians, researchers and clinicians who analyze and interpret QOL data.

Item nonresponse (INR) can be a serious problem in quality of life studies when the proportion of missing data is not small and data are not missing at random. Despite the widespread use of multi-item instruments in quality of life assessment, only few of the methods for dealing with missing data are based on models for scaling multiple items. This paper presents a comparison of results obtained from a latent trait model for scaling binary data subject to INR with results from standard methods for missing data handling. The two-factor logitprobit model of Albanese and Knott assumes existence of an underlying trait and of a second latent variable representing an individual’s ‘tendency to respond’. INR may depend on the latter but also on the latent trait of interest. Thus, the model is supposed to recover information about the underlying trait from patterns of INR. 610 patients with various medical conditions (mainly musculoskeletal, cardiovascular, and respiratory) completed a questionnaire before and after inpatient rehabilitation treatment. Four items addressing sleep problems produced unusual high proportions of INR (9–16%) resulting in only 72% cases with complete data. Scale scores were calculated after dichotomization of items using complete case analysis, mean value substitution, and estimation of conditional means based on the Albanese–Knott model (program TWOMISS, Knott 1996). Scale scores produced by these methods were compared with respect to correlation with SF-36 subscales, differences between diagnostic groups, and responsiveness. Despite the observed dependence of INR on the latent trait, our analyses demonstrated that all three methods for obtaining scale scores produced very similar results. Average correlation with SF-36 subscales was in the range of )0.26– 0.29, discrimination between diagnostic groups and responsiveness was almost equal (effect size: 0.43–0.48, standardized response mean: 0.41–0.43). It can be concluded that, in this sample, application of a versatile scaling model for data with non-random missing data provided valuable information about the admissibility of simple scoring methods.

# 1737/ARE THE DAYS OF UTILITY MEASUREMENT NUMBERED? Paul Kind, Outcomes Research Group, Centre for Health Economics, University of York, York, England

# 1090/ADDRESSING METHODOLOGICAL CHALLENGES IN ANALYZING QOL DATA IN A CLINICAL TRIAL Elizabeth A. Hahn, CORE, Evanston Northwestern Healthcare, Evanston, IL; G. A. Glendenning, Oncology, Novartis Pharmaceuticals Corp., East Hanover, NJ; Mark V. Sorensen, Stacie A. Hudgens & David Cella, CORE, Evanston Northwestern Healthcare, Evanston, IL, USA

# 1720/MULTI-ATTRIBUTE UTILITY FUNCTIONS FOR THE REVISED HUI2 CLASSIFICATION SYSTEM: RESULTS FROM A UK SURVEY Katherine J. Stevens, Chris J. McCabe & John E. Brazier, School of Health and Related Research, The University of Sheffield, Sheffield, South Yorkshire, United Kingdom

The analysis of QOL data in an international clinical trial presents methodological, statistical and interpretive challenges. For the International Randomized IFN vs. STI571 (IRIS) Study, we developed strategies regarding psychometric measurement properties of selfreport instruments, cross-cultural measurement equivalence, definition of clinical significance, missing data, longitudinal modeling, crossover, and description of clinically interpretable results. 1106 newly diagnosed patients with chronic phase chronic myeloid leukemia were enrolled in IRIS from 16 countries. The primary endpoint was the duration of progression-free survival; QOL was a secondary endpoint. Crossover to the other treatment was permitted because of intolerance or lack of efficacy. The Functional Assessment of Cancer TherapyBiologic Response Modifiers (FACT-BRM) was completed at baseline, months 1–6, 9, 12, 18, and 24 in the patient’s preferred language. The table summarizes methodological issues and the strategies developed to address them. The Rasch measurement model was used to evaluate psychometrics, including cross-cultural comparability (eight languages), and to aid in interpretation of treatment differences. A mixedeffects model was chosen for longitudinal analyses, with a patternmixture technique to adjust for nonignorable missing data. Crossover effects were added as a time-dependent covariate. To better understand the meaning of the QOL scores, a clinically significant treatment effect was prespecified, and a modified forest plot was used to summarize item responses. 1049 patients (95%) participated in the assessments. More patients discontinued treatment in the IFN arm (35%) compared with the STI571 arm (11%; p < 0.001). The patterns

Conventional teaching and practice in health economics gives prominence to the quality-adjusted life year (QALY) as the unit of measure in quantifying the benefits of interventions. Central to the computation of QALY benefits is the capacity to express health status on a scale that assigns values of 1 and 0 to the respective states of full-health and dead. The value associated with intermediate health states is typically unknown. Since the QALY product results from the combination of data on quality and quantity of life, it is clear that the choice of qualityadjustment factor is critical to the estimation of benefit. Economic theory tells us that individuals have well-defined preferences that can be represented by a utility function. Expected utility theory continues to provide a dominant methodological underpinning. The major practical consequence of this is that a utility-weighted scale is required when making quality-adjustments in computing QALYs. However, no single standard method for establishing utility weights commands general acceptance. Preferences are revealed through one of several elicitation methods in widespread usage. It is universally recognised that Time Trade-Off (TTO) and Standard Gamble (SG) yield dissimilar results when used to value identical health states. We lack tests that distinguish true ‘utility’ from a degraded approximation. The practical solution to the fruitless task of defending the continued use of utility measurement is to be found in the preference elicitation methods that are commonplace in the practice of market research and in the academic discipline of psychology. This paper reviews the case for adopting non-utility preference measurement methods for use in valuing quality of life. It concludes that non-metric choice-based methods such as paired comparisons have much to offer economists and clinicians alike.

This paper reports the results of a study to estimate a multi-attribute utility function (MAUF) for a revised version of the Health Utilities Index Mark 2, from Standard Gamble (SG) and Visual Analogue Scale data (VAS). Two hundred and one members of the UK general population undertook VAS valuations of all single attribute health states and 13 multi-attribute health states. Four of the 13 multi-attribute health states were also valued using SG. Power and Cubic VAS-SG transformation curves were estimated and used to map the single and multi-attribute VAS values to utilities. Two MAUF valuation models were constructed and their predictive performance compared in a validation sample of 65 health states valued using SG (n ¼ 252). The predictive performance of the two models were compared using the Mean Absolute Error, (MAE), the proportion of health states predicted to within 0.05 (absolute) of the observed mean value, and the Ljung Box test for systematic errors. The restricted Cubic MAUF has a lower MAE than the Power MAUF (0.082 vs. 0.097) and it predicts 10% more health states to within 0.05 of the observed mean (44.6% vs. 33.8%). There was no evidence of systematic errors for either MAUF. Both models had superior predictive performance compared to the original MAUF developed in Canada. The predictive performance of the two new models and the Canadian model does not compare well with that reported for statistical valuation models such as the SF-6D and the EQ5D.

732 # 1522/SIDE EFFECTS OF ANTI-EPILEPTIC DRUGS: A DISCRETE CHOICE EXPERIMENT OF PATIENTS PREFERENCES Andrew J. Lloyd, CHOR, MEDTAP International, London; Emma McIntosh, Health Economics Research Centre, University of Oxford, Oxford; Martin Price, Janssen Cilag UK, High Wycombe, UK Main purpose: Parallel versions of a discrete choice experiment (DCE) were employed to estimate benefit either in terms of willingness to pay (WTP) or utility (in a time trade off model-(TTO)) to explore patients preferences regarding the side effects and seizure control of anti-epilepsy drugs (AEDs). Methods: The main side effects of AEDs were identified through literature search and expert consultation. The DCE included seven attributes (alopecia, nausea, skin rash, concentration effects, and weight change, plus seizure control & TTO or WTP) with suitable levels which were selected with input from a national epilepsy patient advocacy group. The cost attribute ranged from 25 to 75 (UK pounds). The TTO version included months (6, 12, or 18) given up for every 30 years of life. Preferences were elicited from patients with epilepsy either via the post or the internet. Separate analyses using random effects probit models were conducted. Results: 145 WTP and 138 TTO survey forms were returned. Interim analysis indicated that all attributes had the expected polarity i.e. participants showed a preference for less severe side effects, greater seizure control and less cost or life lost. Participants were willing to trade seizure control, cost and months of life lost against a better side effect profile. Conclusions: This study represents a novel method for estimating utility within the context of a DCE. Discussion will focus on the interpretation of the utility information with respect to other published utility studies in epilepsy. We will also discuss whether this approach can be used to estimate the value of a QALY. # 1889/PREFERENCE VALUES FOR EQ-5D HEALTH STATES IN PATIENTS WITH RHEUMATOID ARTHRITIS COMPARED TO SPANISH GENERAL POPULATION Agata Carren˜o, Montserrat Roset & Xavier Badia, Health Outcomes Statistics & Modelling, Health Outcomes Research Europe, Barcelona, Spain; Xavier Carbonell, Rheumatology Department, Hospital del Mar, Barcelona, Spain Characteristic symptoms of Rheumatoid Arthritis (RA) are joint pain and mobility reduction, affecting Health Related Quality of Life (HRQoL) of patients. One of the most used instruments to measure HRQoL and preference values for health states is the EQ-5D. EQ-5D measures health in five dimensions (mobility, self-care, usual activities, pain and anxiety) using three levels of severity: no problems (1), some/moderate problems (2) and severe/extreme problems (3). Respondents are classified into any one of 243 (35) possible health states (HS). The aim was to compare preference values for HS between RA patients and Spanish general population . 62 RA patients rated their actual HS using EQ-5D, and then assessed the preference value for it and two more extreme hypothetical HS (11111 and 33333) using the Visual Analogue Scale (VAS). Patients were asked to imagine that they were in the HS for 10 years without change, followed by death. In order to obtain preference values, worst and best HS (33333 and 11111) were rescored to 0 and 100, respectively, and actual HS was escalated by this values. Preference values obtained in patients with RA were compared with reference values from Spanish general population in the 12 HS observed in the patients sample. Statistical significant differences (p < 0.05) were only obtained for two HS (22231 and 22232), assigning RA patients higher preference values than general population. Patients with RA tended to assign slightly higher preference values to HS with more problems, specially pain, than the general population. Results of the present study confirm the findings obtained in previous studies, in which patients with health problems assign higher preference to actual states than the value assigned by general population as hypothetical HS. Study sponsored by Shering Plough Spain. # 1122/COST-UTILITY ANALYSIS OF POPULATION-BASED PROSTATE CANCER SCREENING – HOW ABOUT THE QUALITY OF LIFE EFFECTS? Marie-Louise Essink-Bot, Ida J. Korfage & Harry J. de Koning, Department of Public Health, Erasmus MC/University Hospital Rotterdam, Rotterdam, The Netherlands

Three empirical HRQoL studies relating to the short term effects of prostate cancer screening, primary treatment of localized prostate cancer and to advanced disease, respectively, have been conducted within the Rotterdam centre of the European Randomised study on Screening for Prostate Cancer since 1995. The results are used to estimate utilities of the favourable and unfavourable HRQoL effects of prostate cancer screening, to be included in cost-utility analysis. We employed the common framework for estimating utilities, including empirical health status description by patients with SF-36 and EQ-5D; linking these descriptions to utility estimates using the standard preference-based algorithm of the EQ-5D; combining them with the appropriate durations into QALYs and sensitivity analysis (in the MISCAN computer simulation model programme). The longitudinal study of screening and diagnostic follow-up (n ¼ 626) did not show effects of prostate cancer screening on SF-36 scores or on anxiety levels, despite short-lasting pain and discomfort. Men reported perceived benefit of screening even after initially false-positive screening tests. In the primary therapy study (n ¼ 314, median 4.3 years of follow-up), changes in generic health status (SF-36, EQ-5D)were small, despite lasting physical symptoms and discomfort (UCLA Prostate Cancer Index). A cross-sectional study among 82 men with advanced prostate cancer showed significantly impaired HRQoL. Employing the results of the three empirical studies to estimate utilities for the relevant stages of prostate cancer screening and therapeutic follow-up underestimates especially the unfavourable HRQoL effects that are inevitably associated with prostate cancer screening. Explanations include non-health effects of the screening itself (information, uncertainty reduction) and adaptive changes in perception of patients’ own situation after curative primary treatment (‘response shift’). Possible directions for adapting CUA methodology in the evaluation of cancer screening will be indicated # 1039/THE MEANING OF ADVANCED BREAST CANCER ON PATIENTS QUALITY OF LIFE: A QUALITATIVE STUDY OF WOMENS’ EXPERIENCE. Minna-Liisa Luoma & Liisa Hakamies-Blomqvist, Psychology, University of Helsinki, Helsinki, Finland The purpose of the present study was to investigate the meaning of advanced breast cancer patients quality of life. The sample consisted of 25 patients who were randomized to receive Methotrexate-Fluorouracil or docetaxel. A semi-structured interview was used. The structure of the interviews followed broadly the structure of the EORTC QLQ-C30, encompassing global ratings of QoL, physical, role, social, emotional, and cognitive functioning. Phenomenological research method was chosen to analyse the data. The findings show that limitations in physical functioning meant dependency on others. Dependency meant also decreased autonomy. Patients tended ensure their autonomy by generating easier actions or changing their internal standards. Changes in appearance and decreased condition affected patients’ ability to carry out roles and responsibilities. These limitations in role functioning were uneasy to accept and feelings of uselessness were dominant. Patients tried to compensate the limitations by substitute actions. For patients who were still active in professional life it was important to keep on working. Changes in lifestyle and appearance affected patients’ social functioning, often meaning isolation. The meaning of being able to sustain reciprocal relationships was often highlighted. Patients kept control of to what extent their illness was recognised in social relationships, this was important to their feelings of autonomy. Close family was a main source of social support, but also a major source of concern. Meaning of emotional functioning was often expressed in terms of ability to enjoy life day by day and of commitment to treatment. Patients’ cognitive functioning was largely intact and patients’ troubles in cognitive functioning were more characteristic to anxiety. Global QoL was dependent on patients’ ability to live normal life, maintain reciprocal relationships and control illness experience. These themes of different key areas of patients QoL can be seen as separate entities, mirroring the various dimensions of the meaning of patients QoL.

733 # 1323/DEVELOPMENT OF A COMPREHENSIVE QUESTIONNAIRE IN POST-MENOPAUSAL WOMEN: RESULTS FROM MULTINATIONAL QUALITATIVE STUDIES Mauro Niero, Education, University of Verona, Verona, Italy; Linda Abetz, Mapi Values, Bollington, UK; Hanne Thorsen, General Practice, University of Copenhagen, Copenhagen, Denmark; Anne Marciniak, Pfizer, Sandwich, UK; Thomas Kohlmann, Social Medicine, University of Greifswald, Greifswald, Germany Ageing brings a myriad of positive and negative outcomes, yet little is known about how post-menopausal women view changes in their lives related to ageing. A 5-stage approach was taken to develop the questionnaire prior to validation: (1) Focus groups (10 women per country, aged 60–85) in France, Italy, Denmark and US to assess worries about aging. (2) Development of a test questionnaire (3) Qualitative testing of the questionnaire in focus groups (20 women per country, aged 50–80, healthy or having osteoporosis or cardiovascular disease/risk) in Germany, Italy, Denmark (4) Revision of questionnaire (5) Cognitive debriefing interviews with 5–10 women held in Germany, Italy, Denmark, Spain, US. Results of qualitative phases suggested that women were generally positive about this time of their life, specifically stating that they felt more liberated and relaxed, with no need to worry about periods, contraception, or conception. However women also reported a number of negative aspects of ageing including: increase in physical symptoms, limited mobility, depression, loss of energy, sleep problems, anxiety, worries about the future, feeling less needed, feeling less attractive, mourning the loss of fertility and fear of dependence and death. Cross-cultural differences were noted particularly in the 1st phase of the study (e.g. ageing was seen more positively in Denmark and US than in Italy and France). In the 3rd phase, all women felt that the test questionnaire was relevant and suggested the addition of concepts including pain, mobility, and vasomotor symptoms. The questionnaire was accordingly revised. Cognitive debriefing suggested rewording several items to enhance comprehension. The 118-item Specific Measure for women In Mature life (SMILE) was developed based on the input of post-menopausal women in Europe and the USA. Results suggest the importance of developing an instrument internationally from the item generation stage, as cultural bias can be minimized. The SMILE is currently undergoing international and longitudinal psychometric evaluation, including item reduction. # 1835/QUALITY OF LIFE OF OLDER ADULTS Laura Schwartzmann, Bettina Caporale, Adriana Suarez & Rossana Lucero, Medical Psychology Department, School of Medicine, University of Uruguay, Montevideo, Montevideo, Uruguay Quality of life (QL) of older adults is a main health priority in our country as well as for the World Health Organization (WHO). Higher survival rates have not always been accompanied by positive QL outcomes. It is of utmost importance for health programs and resources allocation to be able to evaluate QL in elderly to promote healthy ageing. In the framework of a European project (WHOQOL-OLD) a qualitative and cuantitative study was undertaken. Focus groups with older adults and relevant professionals (N:5) were run to identify priority areas to be included in a transcultural older adults module for the WHOQOL-100. In our center, the field trial included 250 subjects of 60 or more years of age, 60% female, 45% elementary school, 40% married, 70% healthy.

Terminal and dementia patients were excluded. Data were collected using WHOQOL-100 and age-dependent variables included in the Older Adults Module, currently being studied. Analysis was performed by linear regression models. Focus groups identified importance of emotional support, feeling valued, independence, positive feelings and spirituality as well as negative impact of loneliness, bad financial situation and pain. Multivariate analysis showed five predictors of overall QL explaining 63% of variance: the quality of environment, independence, psychological domain, mainly negative feelings and a specific question referring to ability to do things you would like to do. Considering healthy subjects, fear of death and dependency on medical treatments enter into the model (64% explained variance). Predictors are stronger for unhealthy people (85% variance explained) showing the impact of the disease (fear of being in pain at death, sleep disturbances and worries about the future). The study showed a high correlation between priority areas resulting from focus groups and initial results of local field trial. Relevance of items included in WHOQOL-100 for older adults was confirmed. Results also point to the importance of psychosocial interventions promoting well being through social support and self-esteem enhancing, detecting and adequately treating depression and stimulating pleasant activities. # 1819/UNDERSTANDING GLOBAL TRANSITION ASSESSMENTS Kathleen W. Wyrwich, Research Methodology and Health Services Research; Vicki M. Staebler Tardino, Organizational Psychology, Saint Louis University, St. Louis, MO, USA Purpose of research: Several methods currently used for determining meaningful or important change and establish clinically important difference standards in HRQoL measures center on patients’ responses to global transition (GT) items, which are questions administered during HRQoL follow-up interviews to determine patient-perceived change in a domain of interest. It can be empirically demonstrated that patients’ responses to these gold standards for change do not adequately incorporate the prior condition. Nonetheless, these GT items play an influential role and cannot be disregarded. This is the first known follow-up study where patients have been queried regarding how they ascertained their GT responses. Subject sample and methods: We used face-to-face cognitive interviews to provide quantitative clues for understanding HRQoL measurement scores to patient global transition assessments, and a more enlightened understanding of patient-perceived reports of HRQoL changes over time. Participants had completed 1 year of enrollment in a clinical study with bi-monthly HRQoL telephone interviews that included the SF-36 and a diseasespecific (asthma, COPD, or heart disease) HRQoL instrument, as well as GT assessments for each domain on these instruments, prior to their qualitative interviews. Summary of results: Patient explanations for how they determined their bi-monthly GT assessments in HRQoL domains included: self-assessment of their current health state, results from recent conversations with their clinicians, comparisons of their current states with the prior health conditions realized by a memory marker or some other salient life event, and spiritual convictions that primarily reflected a resistance to voicing complaints. Conclusion: Quantitative studies that incorporate GT assessments to anchor the interpretations of HRQoL changes should also consider additional relevant items identified by this qualitative investigation to inform researchers of the potential framework that patients use to assess and report their transitions.

734 # 1650/ASSESSING THE FEASIBILITY AND UTILITY OF HRQOL INSTRUMENTS: THE VALUE OF QUALITATIVE METHODS Joanne Greenhalgh, Health Care Practice R&D Unit, University of Salford, Salford, UK When a new instrument is developed, priority is given to assessing its psychometric properties. An assessment of its feasibility or utility is usually limited to an examination of item non-response. Qualitative methods may gain a more comprehensive picture of respondents’ views. Semi structured interviews and questionnaires were used to examine respondents’ views of the feasibility and utility of a new HRQoL instrument for multiple sclerosis (MS): The MS Symptom and Impact Diary (MSSID). In study 1, 13 people with MS were interviewed after completing the MSSID for two weeks to explore their views on its ease of completion and personal utility. Interviews were taped, transcribed and analysed using a grid based method called framework. Main themes were used to develop 10 attitude statements to the feasibility and utility of the MSSID. These statements were completed by 63 people with MS after they had completed the MSSID for 12 weeks in study 2. The majority of interviewees in study 1 found the MSSID feasible to complete and 81% of those in study 2 reported they had got into the habit of completing it. However, 31% of those in study 2 admitted that they had difficulty remembering to complete the MSSID everyday. Study 1 identified two threats to the validity of the MSSID: retrospective completion and not reporting symptoms that were expected to improve. Respondents in studies 1 and 2 felt completing the MSSID had made them more aware of their MS. For some it had enabled them to understand the relationship between their MS and daily activities and manage their lives more effectively or had highlighted when a relapse had occurred. Others did not find it useful, as it would not predict their MS in the future. Perceptions of utility in study 2 were not related to clinical or demographic characteristics but in study 1 appeared to be related to coping strategies. The use of qualitative methods can provide more comprehensive information about respondents’ views of the feasibility and utility of HRQoL measures and can be used to make these instruments more relevant to respondents. # 1565/ARE THERE ANY OBJECTIVE HEMODYNAMIC PREDICTORS OF THE QUALITY OF LIFE IN ESSENTIAL HYPERTENSIVES? Sergey A. Golubev, Vitebsk Regional Center for Refractory Hypertension, Vitebsk Cardiology Center; Maxim N. Mily, Clinical Pharmacology Unit, Vitebsk City Emergency Hospital, Vitebsk, Belarus Relationships between objective hemodynamic parameters and quality of life (QL) patterns in hypertensive patients may be important for planning therapeutic interventions. Our aim was to estimate existence, direction and magnitude of QL parameters associations with an objective parameters set consisted of ambulatory blood pressure monitoring (ABPM) variables, systolic (SBP)and diastolic (DBP) blood pressure (BP) reactivity to laboratory stress and systemic vascular compliance (SVC) in untreated essential hypertensives (EH). A noninvasive 24 hours ABPM (DynaPulse 5000A; Pulse Metric, Inc., USA), standard mental stress test (MST) and cold pressor test (CPT) were carried out in 35 untreated verified ambulatory EH (24 males; aged 43.6 ± 10.0 years; 24-hour SBP 155.7 ± 15.4 and 24-hour DBP 89.4 ± 9.1 mm Hg). At each monitoring session conventional ABPM variables were calculated, including mean BP, time indexes (TI,%) of BP load, daytime and nighttime SD, as well as SVC, which was derived by previously validated pulse dynamics analysis technology. The

QL assessment was performed using previously tested the General Well-Being Adjustment Scale and the Giessen Somatic Complaints Questionnaire. The Total Well-Being Index (TWBI) was related to both 24-hour SBP and 24-hour TI for SBP load (r ¼ 0.39 and 0.4, respectively, p < 0.05 for both).The Total Complaints Index (TCI) were observed to correlate negatively with nighttime SD for DBP (r ¼ )0.35, p < 0.05) and DBP response to CPT (r ¼ )0.38, p < 0.05), while TWBI correlated positively with the same variables (r ¼ 0.4 and 0.35, respectively, p < 0.05 for both). The opposite SVC correlations with TWBI and TCI (r ¼ )0.34 and 0.35, respectively, p < 0.05 for both) were revealed. In multiple linear regressions the hemodynamics variables mentioned explained 29.6% of TCI and 46.5% of TWBI variation. Thus, the stage of the disease progression characterized by stable SBP elevation due to the arterial tree stiffening is accompanied with safe feeling. It may be a baseline cause of extremely low patients’ compliance documented worldwide in general practice of hypertension management. # 1422/A COMPARISON OF DEPRESSIVE SYMPTOMS AMONG CAD PATIENTS TREATED FOR HYPERTENSION WITH A BETABLOCKER- OR CALCIUM ANTAGONIST-LED TREATMENT STRATEGY L. Douglas Ried, Pharmacy Health Care Administration, University of Florida; Michael J. Tueth, Department of Psychiatry, Malcom Randall Veterans Affairs Medical Center, Gainesville, FL; Wolfgang Wittenberg, Health Economics and Outcome Research, Abbott GmbH & Co. KG, Ludwigshafen, Germany; Carl J. Pepine, Division of Cardiovascular Medicine, University of Florida, Gainesville, FL, USA Effective hypertension treatment for patients with coronary artery disease (CAD) is an important issue. According to JNC-VI guidelines, a beta-blocker (BB) based strategy is the standard of care for hypertension treatment among CAD patients. Given treatment parity for adverse outcomes (death, MI or stroke) and blood pressure control, as demonstrated by the International Verapamil/Trandolapril Study (INVEST), cost and quality of life are relevant treatment choice issues. Beta-blockers and calcium antagonists (CAs) both are alleged to be associated with depression or depressive symptoms. The purpose of this study was to compare the level of depressive symptoms of patients assigned to BB- or CA-led hypertension treatment strategies among a subgroup of INVEST patients enrolled between April 1, 1999 and October 31, 1999. Patients living in the United States (n ¼ 2318) were mailed a questionnaire containing items from the Center for Epidemiologic Studies – Depression (CES-D) Scale within 24 hours of randomization. Baseline surveys were returned by 1675 patients (72.3%). In addition to the baseline survey, surveys were mailed 6weeks, 6-months, and 1-year after the index date. The 6-week, 6month, and 1-year surveys were returned by 1459, 1303, and 1233 study patients, respectively. At baseline, females and non-Caucasians reported more depressive symptoms and 37% were classified as ‘atrisk’ of depression. The average CES-D score was 13.9 for patients assigned to the CA-led group and 14.2 for patients in the BB-led group (t-test ¼ 0.61, p ¼ 0.54). After one-year, the average CES-D scores were 12.4 and 13.8, respectively (t-test ¼ 2.04, p ¼ 0.04). After controlling for age, gender, race and baseline CES-D, the average adjusted 1-year CES-D score was 11.4 for patients assigned to the CAled group and 12.5 for patients assigned to the BB-led group (F ¼ 4.86, p ¼ 0.03, Adjusted R 2 ¼ 0.52). Among patients with preexisting CAD, guideline-based, hypertension treatment choice may influence patients’ quality of life.

735 # 1161/EARLY EVOLUTION, DETERMINANTS AND PREDICTIVE POWER OF HEALTH RELATED QUALITY OF LIFE IN A MULTICENTRIC POPULATION OF PATIENTS WITH HEART FAILURE N. Soriano, G. Permanyer-Miralda, A. Ribera, P. Cascant, I. Ferreira & C. Brotons, Cardiology Department, Hosp. Vall d’Hebron, Barcelona, Spain Although HRQL is known to be impaired in heart failure (CHF), its profile and determinants and its power to predict outcome are not sufficiently understood. The early results of a prospective 1 year follow up study of HRQL of patients (pts) admitted to 32 general hospitals during a 10 month period for CHF are reported. Diagnosis was established based on the European Society of Cardiology criteria and HRQL was measured with the SF-36 health profile. Of the 849 already recruited pts, information of HRQL is available for 605 pts at baseline and for 506 at one month. Forty two percent were women, mean age was 70 (from 17 to 96), 62% had previous diagnosis of heart failure, 42% had diabetes mellitus, 40% had at least one important comorbid condition, 49% had biventricular heart failure at admission and 37% had functional class IV. Heart failure pts had low scores in all dimensions of the SF-36 and improvement after admission was only modest. At one month evaluation the Physical Component Summary (PCS) score of the SF-36 increased from 33 to 35, and its determinants were PCS at baseline and biventricular heart failure. The Mental Component Summary (MCS) increased from 41 to 44 and its determinants were PCS (b coef. of 1.7 for a 10 units increase) and MCS (b coef. 4.2 for a 10 units increase) at baseline, gender (b coef. of )3.4 for women) and comorbidity (b coef. of )2.47). Eight percent of pts died during the first month, being PCS at baseline (odds ratio of 0.96 for one unit increase) and age (odds ratio of 1.09 for each year) independent predictors of this short term mortality. Compared with pts with coronary syndromes these patients had worse scores of general health perceptions and mental health and also less improvement. In conclusion: (1) All dimensions of HRQL are significantly reduced in CHF and improve only modestly after hospital admission; (2) More severe CHF, comorbid conditions and baseline HRQL independently predict poorer HRQL after admission; (3) Baseline HRQL is an independent predictor of death.

# 1479/VO2max/kg AND VO2p/kg AS INDICATORS OF THE PHYSICAL STATUS OF PATIENTS’ HEALTH RELATED QUALITY OF LIFE IN HEART TRANSPLANT WAITING LIST Teresa Ortega, Nephrology I; Maria A. Montoliu, Juan C. Alameda & Blanca Rodriguez, Ergonomics Unit; Covadonga Valdes, Pablo Rebollo, Francisco Ortega, Nephrology I, Hospital Central de Asturias, Oviedo, Asturias, Spain Previous studies have shown that a relation exists between variables of the effort test and the health related quality of life (HRQoL) of patients with chronic cardiac insufficiency. The most useful indicator to include a patient in a heart transplant waiting list is the maximal oxygen consumption (VO2max). The aim of this study was to find out a possible association between sociodemographic variables and functional parameters during a progressive effort test and the HRQoL of patients included in heart transplant waiting list. Prospective study of 103 patients waiting for a heart transplantation was carried out. Clinical and sociodemographic variables were collected. HRQoL questionnaire: Spanish version of the Sickness Impact Profile (Physical-PHYS, Psychosocial-PSY dimension and the total score-TSIP). Functional parameters of the effort test: the VO2max and peak oxygen consumption (VO2p), both related with the body weight, heart rate, and ventilation. A logistic regression model (enter) was constructed with the statistical significant variables of the bivariant analysis. Interestingly, female patients, subjects with no active working life and those without school studies expressed worse HRQoL. VO2max, VO2p, VO2max/kg and VO2p/kg were associated with the PHYS and PSY dimensions and with the TSIP (p < 0.01). The only independently associated variables in the regression model were the VO2max/kg body weight with PHYS, 1.28 (95% CI: 1.02–1.61), p ¼ 0.03, and R 2 ¼ 0.22; and VO2p/kg with PHYS, 1.07(95% CI: 1.01–1.13), p ¼ 0.04, and R 2 ¼ 0.21. Conclusions: Female, patients with no active working life and those without school studies in a heart transplant waiting list showed worse HRQoL. The aerobic capacity (oxygen

consumption) keeps a significant relation with the HRQoL. The parameters of the eort test, VO2max/kg and VO2p/kg, are indicators of the physical dimension status of the patients in the heart transplant waiting list. This is the first time that it is found that HRQoL is related with both parameters. # 1451/QUALITY OF LIFE AND LONG TERM SURVIVAL AFTER OPEN HEART SURGERY Zoran R. Potic, Marijela M. Potic, Djordje D. Jakovljevic & Ninoslav D. Radovanovic, Clinic of cardiovascular surgery, Institute for cardiovascular diseases, Sremska Kamenica, Vojvodina, Serbia & Montenegro The aim of this study was to estimate long term benefit of open heart surgery in 7 year period, using quality of life and cumulative survival as outcomes. For assessment of quality of life, we used diseases and treatment specific health related quality of life questionnaire and derived integral numeric QOL index QOLi-NS (0–100). Zero value represents complete inability for any functioning and value 100 means perfect health. The cumulative survival rate was calculated using actuarial method. The prospective study involved 787 consecutive patients. The survival adjusted QOL (QOLadj) was calculated according the following formulae QOLadj(t) ¼ QOLi-NS(t)
Survival(t). The benefit of operation was calculated as difference between postoperative survival adjusted QOL and preoperative QOL value (QOLiNS ¼ 40.9). Patients’ benefit, concerning survival adjusted QOL, is the greatest at 1 year after the surgery and declines during the follow-up period, but is very clinically and statistically significant. The mean QOL benefit in 7 year follow-up period is 29. The survival adjusted QOL could be the most appropriate measure for the evaluation of open heart surgery outcome. # 1033/INITIAL DEVELOPMENT AND VALIDATION OF A NEW SELF-REPORT: THE FUNCTIONAL ASSESSMENT OF CANCER THERAPY-THROMBOCYTOPENIA (FACT-Th) QUESTIONNAIRE Timothy Hunt, Global Outcomes Research, Pfizer Corporation, Peapack, NJ; Jennifer Beaumont, Kimberly Webster & David Cella, Center on Outcomes, Research and Education, Evanston Northwestern Healthcare, Evanston, IL; Linda Elting, Department of Health Services Research, University of Texas MD Anderson Cancer Center, Houston, TX, USA The purpose of this study was to determine if the concerns and symptoms of cancer patients with chemotherapy induced thrombocytopenia (low blood platelet counts) can be captured by patient selfreport and combined with the FACT-G, a general 27-item cancer questionnaire. The development and testing of the 18-item Thrombocytopenia Subscale (ThS) followed a standardized sequence, including item generation, item reduction, scale construction, and reliability/ validity testing. Patient feedback was also obtained to determine if the content domain of the instrument is appropriate relative to its intended use. Patients were recruited from three urban academic medical centers in two cities; assessments occurred at enrolment and about 2 weeks later. At baseline there were 40 severely thrombocytopenic (<50,000/ll) patients and 43 patients with higher platelet counts (>75,000/ll). All subscales and aggregated scores of the FACT-Th showed high internal consistency with an alpha range of 0.71 to 0.96. The new 18-item ThS also demonstrated good internal consistency with an alpha of 0.90 at both assessments. Strong convergent validity was demonstrated with the POMS total score and the SF-36 summary scales with correlation coefficients exceeding ±0.5. Divergent validity was confirmed with the Marlowe–Crowne social desirability scale, where no statistically significant relationships were found to exist with any of the subscales of the FACT-Th. Notably, the new ThS differentiated patients with differing performance status ratings (p < 0.001), and, more importantly, those with thrombocytopenia vs. those with higher platelet counts (p < 0.001). The responsiveness to changes in platelet count of the ThS (p < 0.001) exceeded that found in the SF-36 summary scales (p > 0.2) and the FACT-G subscales (range: p ¼ 0.03 for PWB to p > 0.2 for SFWB) and persisted after controlling for clinically correlated differences in hemoglobin and WBC. The ThS may be a useful targeted endpoint in trials where an important difference in platelet count is anticipated between groups.

736 # 1495/AN ABBREVIATED VERSION OF THE EORTC QLQ-C30 FOR USE IN PALLIATIVE CARE Mogens Groenvold & Morten A. Petersen, Palliative Medicine, Bispebjerg Hospital, Copenhagen, Denmark; Neil K. Aaronson, Division of Psychosocial Research & Epidemiology, Netherlands Cancer Institute, Amsterdam, Netherlands, Peter M. Fayers, Public Health, Aberdeen University Medical School, Aberdeen, UK; Mirjam Sprangers, Medical Psychology, Academic Medical Center, Amsterdam, Netherlands, Jakob B. Bjorner, Copenhagen, Denmark The EORTC QLQ-C30 is being used in thousands of cancer studies. However, some items may not be optimal for palliative care. The aim of this study was to develop a shortened version of the EORTC QLQC30 suitable for palliative care. Results of this shortened version should be compatible (directly comparable) with scores on the corresponding scales of the QLQ-C30. This involved: (1) Use of item response theory methods to abbreviate certain scales of the EORTC QLQ-C30 and to develop algorithms for scoring the shortened scales on the original response scale. The three longest scales were shortened with little or no loss of precision or measurement power. (2) To shorten the EORTC QLQ-C30 by deleting selected scales that are not highly relevant to palliative care. The relevance, importance, and appropriateness of the QLQ-C30 scales/items were assessed in interviews with 41 patients and 66 professionals from six European countries. Most scales and single items were rated as relevant, important, and appropriate. However, as expected, the two first Physical Function items were judged inappropriate; these items were already removed in the shortening of scales. Respondents rated Social Function and Role Function as highly relevant domains but suggested to revise the scales. Items on financial difficulties and diarrhea were judged least important. Based on these results, in May 2003, the EORTC Quality of Life Group will decide on the contents of the new questionnaire, which will be presented. We anticipate that the new questionnaire will be a useful outcome measure for palliative care. For some studies we recommend to combine this instrument with questionnaires assessing other aspects of palliative care, e.g. existential issues. # 1257/THE DEVELOPMENT AND VALIDITY OF THE DUTCH HANDICAP SCALE FOR CHILDREN Symone B. Detmar, Esther J. Hosli, Jeanet Bruil, Ton Vogels & Erik Verrips, Youth, Quality of Life, TNO Prevention and Health, Leiden, The Netherlands Although there is increasing need for the evaluation of social participation in children with a chronic illness, no instruments exists to measure this. This study describes the development and initial psychometric evaluation of the Dutch Handicap Scale for Children (DHSC). This questionnaire is based on the London Handicap Scale (LHS), a valid and reliable utility instrument to measure social participation in adults. A multi-disciplinary research group (pediatrician, researcher in Ql regarding children, child psychologist) was involved in developing the DHSC. The questionnaire was tested in 114 children with a chronic disease and 239 healthy children with an age range from 8 to 18. The assumption that a negative health status would lead to participation problems was tested by relating HUI 3 attributes to corresponding DHSC scores. Questionnaire development resulted in a five-dimension questionnaire: Mobility, Physical independence, Daily activities, Social integration and Orientation. Each dimension contains one item with a six-point response scale. A lower score indicates greater handicap. Feasibility testing with 10 children showed that none of the children experienced difficulties with filling in the questionnaire. Conceptual validity, measured by correlations between the dimensions of the DHSC and HUI Mark 3 was satisfactory. As expected, moderate correlation coefficients between predefined pairs of HUI and DHSC attributes were found; other correlation coefficients were low. Discriminant validity was also satisfactory, shown by large differences between the healthy and the chronically ill group and by several criteria within the chronically ill group. Based on this initial evaluation, the questionnaire seems feasible and valid to be used for children from 8 to 18 years. The following step is to determine children-specific weights.

# 1743/RELIABILITY AND CONSTRUCT VALIDITY OF THE SWEDISH EXTENDED VERSION OF SEIQOL Ann E. Langius-Eklo¨f & Lena Wettergren, Department of Nursing, Magnus Bjo¨rkholm; Department of Medicine Karolinska Hospital, Karolinska Institute, Stockholm, Sweden The purpose of the study was to examine the reliability and construct validity of the Swedish extended version of the Schedule for the Evaluation of the Individual Quality of Life – Direct Weighting (SEIQoL – DW). The sample consisted of 121 long-term survivors of Hodgkin’s lymphoma (HL) and a randomly sample of the general population in Stockholm. The SEIQoL-DW is based on semi-structured interviews where the participants freely nominate areas, both negative and positive, important for their quality of life. Satisfaction with these areas is rated on a category scale and importance is weighted using a colored disk. The rating and the weight for each domain are multiplied and the products aggregated to produce a Generic Quality of Life index score. The Swedish extended version included a disease-specific part asking the HL survivors how the disease influences their life today. The same procedure with the ratings and weightings is used to produce a Disease-specific index. The participants also filled out the SF-12, HAD and SOC scales. The correlations between the Generic QoL index and the Disease-specific index and the other scales are moderate. Ninetyfour of the HL survivors stated that the disease today influenced them. Eighty of these mentioned this as a change in view of life and these statements were judged positively in 70% of the cases, and as both positive and negative in approximately 20% of the cases. Fifty-eight patients nominated health aspects and the statements were generally judged negatively. Data procession correlating negative and positive statements with the scales is in progress. A multiple regression analysis showed that satisfaction of the chosen areas explained nearly 90% of the variance in the Generic QoL index (standardized b coefficient 0.94, R 2 ¼ 0.89, p < 0.001). Preliminary data support the construct validity of the extended Swedish version of SEIQoL-DW. The benefit of using the weighting procedure for producing the Generic QoL and Disease-specific index scores should be discussed. # 1523/ITEM RESPONSE THEORY TO ASSESS THE CROSSCULTURAL VALIDITY OF THE VSP-Am FOR UK AND US ADOLESCENTS RECEIVING GROWTH-HORMONE TREATMENT M.C. Simeoni, S. Robitail, A. Beresniak & P. Auquier, Perceived Health Research Unit, Public Health Department School of Medicine, Marseille, France; J. Cramer, Department of Psychiatry, Yale University, New Haven The VSP-Am is a French generic, multidimensional, self-administered Health-Related Quality of Life (HRQoL) tool, based on the point of view of non-healthy adolescents. It comprises 39 items, exploring 8 dimensions: Relationships with Friends (FR), Sef-Esteem (SE), Vitality (VI), Relationships with Medical Staff (MS), Relationships with Parents (FA), Relationships with Teachers (TE), Psychological Well-Being (PWB), Physical Well-Being (PH). VSP-Am has been cross-cultural validated in UK and US English. This study used item response theory analyses for assessing cross-cultural comparability of measurements obtained with the VSP-Am in UK and US adolescents receiving growth hormone treatments. 185 patients have been involved (UK n ¼ 81 and US n ¼ 104). Confirmatory factor analysis (LISREL) was used to assess the factorial structure. Extended Logistic Model of Rasch (RUMM2010) was used to compute item calibrations, item goodnessto-fit (v2), item-trait test-of-fit (v2) and DIF analyses (ANOVA of standardised residuals). Confirmatory factor analyses indicates that data fits to the model (RMSEA ¼ 0.075). All items present satisfactory goodness-to-fit except one item of SE and one of PWB. Item-trait testof-fit is adequate for every dimension (p > 0.01). Cronbach a (0.78– 0.89) and Person Separation Index (0.75–0.88) shows good reliability for every dimension. DIF analysis are satisfactory for all dimensions except PH. Two items of PH present significant DIF. ICC of item calibration by country was greater than 0.80 for every dimension except PH and PWB. These results confirm the construct validity of the English and American versions of the VSP-Am with adolescents receiving growth hormone treatments. Topics for future research are suggested.

737 # 1301/POPULATION ESTIMATES OF UTILITIES: THE HEALTH AND ACTIVITY LIMITATION INDEX (HALEX) Elena M. Andresen & Angela Recktenwald, Community Health, Kathleen Gillespie, Health Policy and Management, Saint Louis University, St. Louis, MO, USA The American public health plan, Healthy People 2010, sets overall goals, in part from utility-linked information based on the Health and Activity Limitation Index (HALex). The HALex classifies ‘utilities’ for people based on their global health report and six levels of activity limitation (a total of 30 levels). We compared the HALex with a gold standard measure, the quality of well being (QWB) and also with other population summary measures of health-related quality of life (HRQoL). Subjects were enrolled and interviewed using random-digit dialing from a large metropolitan area of the mid west of the USA. Randomly ordered measures included the HALex, QWB, two versions of a time trade off question (TTO 1 & 5 year time frames), and Healthy Days measures from the Behavioral Risk Factor Surveillance System (BRFSS). 200 people aged 18 and older completed interviews. Their mean age was 45.3 years; 70.5% were women, and 67.5% were Caucasian. HALex scale scores were skewed with 33% of subjects classified in the two best health categories; the mean utility score was 0.82 (standard deviation [SD] 0.20). There was some evidence of bias in the TTO scores by time framing; the mean number of days for subjects asked about 1-year was 120 days in a year, and the calculated 1-year score was 102 days per year for those who answered the 5-year question. The mean QWB score was 0.71 (SD ¼ 0.13). Correlations between the QWB and HALex were very weak (r ¼ 0.027). The HALex was also very weakly correlated with TTO (r ¼ )0.225), and modestly correlated with healthy days (r ¼ 0.438), primarily from correlation with physical health days (r ¼ )0.518). The HALex demonstrates a fairly skewed population distribution, and apparently measures utilities quite differently than the QWB. Further study is needed to determine how the HALex measures utilities by comparing it to direct utilities solicitation method. # 1532/A RANDOMIZED MULTI-CENTER TRIAL TO COMPARE UTILITY ELICITATION TECHNIQUES WITH AND WITHOUT HYPOTHETICAL HEALTH STATES Holger J. Schunemann, Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, Ontario, Canada for the GERD Utilities Writing Group Background: Investigators have questioned the validity and responsiveness of utility measures. In particular, the issue of whether patients’ rating of hypothetical health states (marker states) before rating their own health may improve validity and responsiveness is unresolved. Therefore, we performed a large multi-center randomized trial to evaluate the standard gamble (SG) and the feeling thermometer (FT) with and without use of marker states in patients receiving a proton pump inhibitor (PPI) for treatment of gastro-esophageal reflux disease. Methods: We randomized 217 Canadian patients in 18 centers to the FT (self-administered) and SG with marker states (FT/SG+, n ¼ 112) or without marker states (FT/SG), n ¼ 105) before and after 4 weeks of PPI therapy. Patients also completed other HRQL instruments. Results: Marker state status did not impact on baseline scores on either FT or SG (FT+ 0.66, FT) 0.68, SG+ 0.77, SG) 0.78, on a scale from 0 [dead] to 1.0 [full health]). Improvement after therapy was

0.21 in FT+ (p < 0.001) and 0.15 in FT) (p < 0.001) (difference between FT+ and FT) 0.06; 95%CI: 0.01–0.11, p ¼ 0.02). Improvement after therapy in SG+ was 0.07 (p < 0.001) and 0.06 in SG) (p < 0.001) (difference between SG+ and SG) 0.01; 95%CI: )0.04 to 0.07, p ¼ 0.63). In general, correlations of baseline and change scores of other HRQL measures with FT+ and FT) were slightly larger for the FT+ group. Correlations of baseline and change scores of other HRQL instruments were weaker in patients randomized to SG+ compared to SG). Conclusion: Administration of the FT with marker states increased responsiveness of the FT but not the standard gamble, and reduced validity of the SG but not the FT. Compared to the SG, the FT was more responsive to change on the 0–1 scale. Decisions about use of marker states should depend on whether the SG and FT is of primary interest, and the importance of optimal validity and responsiveness in relation to competing objectives such as optimal efficiency. # 1729/A COMPARISON OF FOUR INDIRECT UTILITY ASSESSMENT METHODS IN PATIENTS WITH RHEUMATOID ARTHRITIS (RA) Carlo A. Marra, John C. Woolcott, Aslam H. Anis, Daphne Guh, Jacek Kopec & John M. Esdaile, Health Care and Epidemiology, University of British Columbia, Vancouver, British Columbia, Canada; John E. Brazier, Health Economics Group, ScHARR, University of Sheffield, Sheffield, South Yorkshire, UK Purpose: Utilities can be assessed indirectly using multiattribute theory (MAUT) instruments and used as weightings for quality adjusted life years (QALYs) in economic evaluation. We have directly compared four MAUT instruments (the Health Utilities Index Mark 2 and 3 (HUI-2 and HUI-3), the EuroQoL (EQ-5D), and the Short Form 6D (SF-6D)) in a cohort of RA patients. Methods: A self-completed questionnaire that included the HUI-2, HUI-3, EQ-5D and the SF-6D, and clinical outcomes (pain, functional status (Health Assessment Questionnaire, HAQ), swollen and tender joints, global assessment, RA severity and control). Intraclass correlation coefficients (ICC) were used to assess agreement among the MAUT utilities. The global utility scores were compared across the instruments and within individuals. Exploratory factor analysis was used to identify the dimensions of health represented by the MAUT global utility scores. OLS regression using the Pratt Index was used to assess the relative importance of the identified factors. Results: 313 RA patients (78% female, 67% RF positive, mean (SD) age 61.3 (20.0) years, median [IQR] HAQ 1.13 [1.38], mean (SD) RA duration 13.9 (11.4) years) who were recruited from private rheumatologists completed the questionnaires. Mean (SD) global utility scores were 0.63 (0.24) for the SF-6D, 0.66 (0.13) for the EQ-5D, 0.71 (0.19) for the HUI-2, and 0.53 (0.29) for the HUI-3 (p < 0.0001 for between group and within person comparisons). The ICC across all the indices was 0.67 (95% CI: 0.62–0.71). There was evidence of ceiling effects for the EQ-5D and floor effects for the SF-6D. In RA, the EQ-5D, HUI-2 and HUI-3 global utilities mostly measure mobility, dexterity and pain whereas the SF-6D utilities mostly measure emotional/mental health, dexterity and mobility. Conclusions: There are significant differences in utilities across the different indirect methods between groups and within individuals. Agreement among the instruments is moderate. It is unlikely that these utility values, if used as the weightings for QALYs, would result in comparable estimates.

738 # 1786/COMPARISON OF UTILITY MEASURES WITH ROC Denise R. Globe, Sheng Shi, Tara Knight & Michael B. Nichol, Pharmaceutical Economics and Policy, University of Southern California, Los Angeles, CA, USA The purpose of this study was to provide a method for comparing seven health utility measures (candidates) derived from estimation to two commonly accepted utility measures (standards), HUI2 and HUI3. A good candidate should have higher discriminative power to separate persons of lower standard from persons of higher standard. The area (AUC) under the receiver operating characteristic (ROC) curve of a candidate can be used to quantify such discriminative power. Based on survey data collected from 6923 members of a large managed care organization, AUC of each candidate was computed and compared. Utility score estimates were calculated from the SF-36 and the SF-12 using previously developed mapping algorithms including the SF-6D. When predicting excellent health (HUI3 P 0.97), the Nichol–Sengupta–Globe estimate (NSG-OLS) based on an algorithm which maps the SF-36 to the HUI2, achieved 0.88 of AUC. This is 0.16 points higher than the Visual Analogue Scale (VAS) and 0.04–0.07 points higher than the other five candidates. When predicting the poor health condition (HUI3 < 0.35), the AUC of the NSG-OLS was 0.90, which was 0.14 points higher than VAS and about 0.01–0.03 points higher than the other candidates. Almost all differences were statistically significant at the 5% level. When grouping into poor/fair health (HUI3 < 0.65) and very good/excellent health (HUI3 > 0.90), the NSG-OLS continued to display the highest AUC compared to the other candidates. When health conditions were defined by the HUI2, the NSG-OLS again yielded statistically, significantly higher AUCs compared to the other candidates. This relationship is not surprising, since the NSG-OLS was derived from the analytic data set. Utility measures may vary in terms of scale and focus. The ROC method offers a unified way for cross comparisons. Since the survey data used for this analysis was also used to derive a number of the candidates, further investigation in novel datasets is warranted to determine the utility of the ROC for comparing various utility estimation methods. # 1861/ESTIMATING UTILITIES IN PATIENTS WITH ASTHMA AND COPD: DO THE EQ-5D AND SF-6D YIELD COMPARABLE VALUES? Nancy Kline Leidy, Center For Health Outcomes Research, Medtap International, Bethesda, MD, USA; Elisabeth Sta˚hl, Clinical Science, AstraZeneca R&D Lund, Lund, Maryland, Sweden; Agota Szende, Center For Health Outcomes Research, Medtap International, London, United Kingdom; Klas Svensson, Clinical ScienCE, AstraZeneca R&D Lund, Lund, Sweden Objectives: To compare utility values derived from the EQ-5D and the SF-6D in two respiratory populations with heterogeneous disease severity. Methods: The EQ-5D and SF-6D were administered during two cross-sectional surveys of patients with asthma (n ¼ 206; Hungary) and COPD (n ¼ 174; Sweden). Scores on both measures range from 0 (worst possible health state) to 1 (best possible state). Disease severity was defined according to GINA and GOLD recommendations for asthma and COPD, respectively. Results: Within each patient population, mean EQ-5D and SF-6D scores were similar for the overall group and for those with moderate disease (Table 1). Scores varied for patients with mild and severe disease. Variability was also noted in ceiling and floor effects. One patient in each group scored 1.0 on the SF-6D, while 57 patients in each group (28% asthma; 34% COPD) scored 1.0 on the EQ-5D. Differences were also found in the lower end of the scale. Twenty patients with asthma (10%) and five with COPD (3%) scored <0.5 on the SF-6D, while 47 with asthma (23%) and 19 with COPD (11%) scored <0.5 on the EQ-5D. Conclusions: Results suggest the EQ-5D and SF-6D do not yield consistent utility values in patients with asthma and COPD. Further study is warranted. # 1223/PROSTATE CANCER PATIENTS’ INTERPRETATION OF SF-36 AND EQ-5D: A QUALITATIVE STUDY Ida J. Korfage & Marie-Louise Essink-Bot, Public Health, Erasmus University Medical Center; Tony Hak, Business Administration, Erasmus University, Rotterdam, The Netherlands

First results from the on-going Rotterdam study on the health-related quality of life (HRQoL) effects of primary therapy in localized prostate cancer (PC) showed that urological and sexual dysfunction was generally not associated with a decrease in SF-36, EQ-5D VAS or utility scores. This discrepancy was also found in similar published studies. We undertook a qualitative study to find out if respondents’ interpretation of SF-36 and EQ-5D items can be helpful in explaining this discrepancy. We qualitatively interviewed 15 men who had been curatively treated for localized PC and an age-matched control group of 15 men without PC about their interpretation of EQ-5D, EQ-VAS and four SF-36 scales (Role-Physical, Bodily Pain, General Health and Social Functioning). We applied the Three-Step Test-Interviews methodology (K. vd Veer, T. Hak, H. Jansen): (1) completion of the questionnaire ‘thinking aloud’ while the interviewer makes notes; (2) focused interview aiming at clarifying and completing observational data; (3) semi-structured interview aiming at eliciting experiences and opinions. Interviews were conducted at respondents’ homes or university by one interviewer (IK). Results include the following: (1) when asked about ‘current state of health’ men did not take erectile dysfunction in consideration since they did not regard sex to be part of health; (2) when asked about ‘current state of health’ respondents tended to take expectations about future health in account as well; (3) when asked about ‘future state of health’ men answered not to expect their health to deteriorate, but added that regarding their age of course their health would deteriorate. Additionally we found flaws in items, like ambiguous answer categories. For example, both respondents who consider their health worse than others and those who consider it better than others disagreed with the statement ‘I am as healthy as anybody I know’. We consider results of the qualitative study helpful in explaining the lack of association between disease-specific and generic HRQoL scores in men with localized PC. # 1643/PATIENT-BASED REFERENCE VALUES FOR MLHF QUESTIONNAIRE Montse Ferrer & Olatz Garin, Health Services Research Unit, Municipal Institute for Medical Research, Barcelona, Spain; Ingela Wiklund, Behavioural Medicine, Astra Ha¨ssle Research Laboratoris, Mo´elndal, Sweden; Eric Van Ganse & Guilhem Pietri, Unite´ Pharmacoe´pide´miologie, Centre Hospitalier Lyon-Sud; Martine Mouly, Research & Educational Department, MAPI Research Institute, Lyon, France; Jordi Alonso, Health Services Research Unit, Municipal Institute for Medical Research, Barcelona, Spain Interpretation of Quality of Life scores has been identified as one of the main barriers to generalising the use of these outcomes for physicians. Among heart failure patients, reference values at each New York Heart Association (NYHA) class may facilitate clinicians to interpret individual or patient group scores. References have been mainly developed as country specific for generic questionnaires. However, it is expected that disease-specific quality of life would be more affected by severity than for country differences. The aim of this study was to assess the differences on Minnesota Living with Heart Failure Questionnaire (MLHFQ) scores by country, in order to know if MLHFQ patient-based reference values should or not be country specific. Data was pooled from 8 studies in 13 countries, with a total sample of 2780. General linear model, adjusted by age and gender, was constructed. To test if relationship between MLHFQ scores and NYHA class is consistent across countries, the interaction between them was included in the model (p ¼ 0.363). There were significantly differences by country (USA as reference) only for Australia and Italy (p ¼ 0.003). These two countries also differed from all the others when Bonferroni pair wise multiple contrast was applied. Scores differed significantly by age, gender and NYHA class (p < 0.001). Adjusted means of the total MLHFQ score were: 31 for class II (95%CI: 28.6–33.2), 44 for class III (95%CI: 42.4–45.8), and 58 for class IV (95%CI: 51.8–64.1). Although further research is needed to explore the differences found for Australia and Italy, our results suggest that no country specific patientbased reference values for MLHF questionnaire are necessary. International reference values facilitate validity and interpretation across countries. Supported by the EU Commission (QLRI-CT-2000-00551 (IQOD)) and the CIRIT (2001SGR 00405).

739 # 1243/USING MINIMALLY IMPORTANT DIFFERENCES TO ASSESS INTERACTION AND CONFOUNDING IN MULTIPLE LINEAR REGRESSION OF CHANGE IN QOL Kathleen J. Yost & Elizabeth Hahn, Center, on Outcomes, Research, and Education, Evanston Northwestern Healthcare, Evanston, IL; Alan Zaslavsky & John Ayanian, Health Care Policy, Harvard Medical School, Boston, MA; Dee West, Northern California Cancer Center, Union City, CA, USA Interaction in multiple linear regression can be assessed with significance tests and consideration of the meaningfulness of the interaction effect. Assessing confounding does not involve significance tests and relies solely on an evaluation of the meaningfulness of the effect of the confounding variable. Our objective was to use minimally important differences (MIDs) to assess interaction and confounding in multiple linear regression describing predictors of change in quality of life (QOL). We evaluated data for 569 colorectal cancer patients identified from a population-based cancer registry and surveyed 9 and 18 months post-diagnosis. Baseline QOL, 10 sociodemographic, and 16 healthcare-related variables were assessed as predictors of change from 9 to 18 months in Functional Assessment of Cancer Therapy-Colorectal (FACT-C), FACT-General (FACT-G), and Trial Outcome Index (TOI) scores. General linear models were used to assess predictors of change, interaction terms, and confounders. MIDs for the FACT-C, FACT-G, and TOI were based on published data. An interaction term was added to a regression model if it was statistically significant and resulted in a meaningfully different interpretation of the effect on QOL change scores for at least one of the variables in the interaction. A confounding variable was added if it resulted in a meaningfully different interpretation of the QOL change score for another covariate in the model. Predictors of change in FACT-G included baseline QOL, race, gender, income, general health, quality of care, and an interaction between baseline QOL and quality of care. Predictors of change in FACT-C were the same as for FACT-G plus bowel function and a bowel function by race interaction. Predictors of change in TOI were baseline QOL, age, gender, race, income, comorbidities, general health, bowel function, quality of care and interactions between baseline QOL and bowel function, baseline QOL and age, and bowel function and race. This study illustrates how MIDs can be used to facilitate variable selection in multiple linear regression of change in QOL. # 1814/MEASURING DISPARITIES IN HEALTHY LIFE EXPECTANCIES: RESULTS FROM VARIOUS MEASURES OF THE SAME MODEL Michael T. Molla, Office of Analysis, Epidemiology and Health Promotion, National Center for Health Statistics, Hyattsville, MD, USA The concept of health expectancy was introduced in the 1960s and further developed in the 1970s. Ever since then, it is one of the most frequently used composite measure of health disparities. Healthy life expectancies are calculated using models that incorporate measures of mortality and morbidity. These two components are combined using a mathematical function that transforms the two sets of partial measures into a single composite measure using a life table methodology. In this study, the author will discuss the fundamentals and the application of this measure. To illustrate the practical application of the composite measure, various non-fatal health outcome measures are combined with life table values to estimate a variety of healthy life

measures such as expected life in good or better health, free of activity limitation, without needing help in activities of daily living (ADL) or instrumental activities of daily living (IADL). Healthy life expectancies are estimated by 5-year age group and sex and race. Data from various sources are used for the estimates. Life expectancies by 5-year age group, sex and race are estimated using data from the National Center for Health Statistics (NCHS) and the US Bureau of the Census. Prevalence rates of the various health states are calculated from the person and condition files of the National Health Interview Survey (NHIS) for the 1994 year1996. Preliminary results show that health disparities between males and females varied by measure. The implied variations are smaller at the younger ages and much larger at the older ages. The health disparity pattern between the white and black population is different from the health disparity pattern between males and females. Health disparities by measure varied more when measuring racial health disparities than when measuring health disparities between males and females. The study implies that using different health measures results in different patterns and magnitudes of health disparities between population groups. # 1562/THE PRESENTATION OF QUALITY OF LIFE INFORMATION TO THE GENERAL POPULATION AS COMPARED TO A COHORT OF CANCER PATIENTS Lisa M. Zetes-Zanatta, Michael D. Brundage & Deb Feldman-Stewart, Division of Cancer Care and Epidemiology, Queen’s University Cancer Research Institute, Kingston, Ontario, Canada The purpose was to systematically determine how accurately the general population interprets each of six visual and written formats for communicating HRQL scores, and to compare subjects’ preferences and accuracy to previously studied cancer patients. We report preliminary (n ¼ 32 participants) findings. Subjects were recruited using a ‘snow-ball’ referral process thereby generating a convenience sample with demographics similar to the cancer patients. Each participant interpreted 18 different pairs of (hypothetical) treatments and judged the HRQL scores (three magnitudes of QOL difference between treatments displayed in each of the six formats). Outcome measures were: accuracy of interpretation (correct/incorrect); preference ratings; and helpfulness ratings on each presentation format. A Latin-square design was used to control for presentation order of format and HRQL magnitudes. The two cohorts (31 patients without cancer and 198 cancer patients) had a mean age of 59.94 (SD 8.63) and 65.1 (SD 10.77), respectively; 21 (67.7%) and 90 (45.7%) had post-secondary school education respectively. The average proportion of correct responses differed across formats, ranging from 0.98 to 0.87 (F ¼ 14.44, p < 0.0001). Accuracy rates were associated with education (F ¼ 13.88, p < 0.0001) and age (F ¼ 4.0, p ¼ 0.046), but not with cohort (cancer vs. general population). More importantly, an interaction between format accuracy and cohort was not detected (F ¼ 0.249, p ¼ 0.940). Format preference and helpfulness ratings differed across formats (Kendall’s W ¼ 0.305 and 0.205 respectively, both p < 0.0001), but not between cohorts. Accrual in this study is ongoing to assure adequate power. Our preliminary findings suggest that line graphs displaying mean HRQL scores from clinical trials are the most accurately interpreted and the most preferred by both the general population and by cancer patients. The findings have implications for optimal communication to patients regardless of disease classification, as well as family members, policy makers and other potential consumers of quality of life information.

740 # 1755/COMPARISON OF METHODS FOR GENERATING ITEMS DURING SCALE DEVELOPMENT Francis Guillemin, Michele Baumann & Anne-Christine Rat, EA 3444, University of Nancy, vandoeuvre-les-nancy; Elisabeth Spitz, Health Psychology, University of Metz, Metz; Nathalie Retel-Rude, Besanc¸on & Jacques Pouchot, Internal medicine, University of Paris 6, Garennes-collombes; Joel Coste, Biostatistics, University of Paris 5, Paris, France Background: No methodology is standard for item generation process during scale development. Purpose: To compare methods of item generation for development of a lower limbs osteo-arthritis (OA) quality of life (QoL) scale. Methods: All methods compared relied on extraction of verbatims collected during tape recorded interviews, transcripted and analysed by two teams of health psychologists and sociologists independently. The five methods compared were: health care professionals and caregivers interviewed with standard interview guide with (1) individuals, (2) focus group, and OA patients in (3) focus group, (4) individual standard interview and (5) individual interview using a cognitive interview technique to enhance memories. Results: 83 individual and focus interviews were conducted with 32 caregivers and 96 OA patients. After exclusion of speeches not related to QoL (i.e. satisfaction, coping), 80 verbatims were identified relevant. With caregivers, focus group produced only 35% of verbatims, while individual interviews contributed 80% of verbatims. With patients, focus group produced 74% of verbatims, while both individual methods contributed 99% of resulting items. Caregivers provided a narrower picture of consequences on QoL. Focus group methods contributed some theme more in-depth while individuals gave more broader insights. Guided interview contributed known QoL themes while cognitive interview provided more fine tuned verbatims on exchanges with and perception by others about their condition, as well as perspective in life and fears from future dependency. Selected items were 43 in the final scale. Using this scale as a gold standard, the sensitivity and specificity, i.e. the contribution of each method to the scale, was estimated (table). Comments: The resulting self-administered questionnaire has benefited from the contribution of the various sources. One can recommend to use structured guided interview of individual patients in sufficient number, conducted by interviewers preferably nonspecialist of the condition explor # 1854/DEVELOPMENT AND VALIDATION OF THE WOMEN’S QUALITY OF LIFE QUESTIONNAIRE (WOMQOL) Sarah Gehlert, School of Social Services Administration, University of Chicago; Chih-Hung Chang, Feinberg School of Medicine, Northwestern University; R. Darrell Bock, Department of Psychology, University of Chicago; Ann S. Hartlage, Department of Psychiatry, RushPresbyterian-St. Luke’s Medical Center, Chicago, IL, USA Background and objective: The Women’s Quality of Life Questionnaire (WOMQOL) was developed to measure aspects of the health of women of reproductive age with no known pathology, because no comparable instrument was available. Research design: Women, who were in the study for approximately 60 days, completed the WOMQOL during the follicular and late luteal phases of their menstrual cycles. Demographic data were collected at exit from the study. Subjects: The first 1200 women enrolled in the Women’s Wellness Study, a community-based study of women’s health through time. Measures: Five experts in women’s health and mental health were asked to rate a pool of 140 items as very significant, somewhat significant, or insignificant to women’s health. Sixty-seven items on which raters agreed most were retained and formatted as a self-reported questionnaire with dichotomous responses to each statement (‘true’ and ‘false’) sent to participants. Global Assessment of Functioning scale scores were collected for comparison. Analysis: Dichotomous response data were factor analysed by the item-response theoretic method described by Bock, Gibbons, and Muraki in 1998. Results: Three primary factors were found: Physical, Mental, and Spiritual Health. The pattern held when the sample was divided into age cohorts and into rural vs. urban. Lower scores were found during the late luteal compared to the follicular phase of the cycle. Conclusions: The final 30-item instrument, made up of the highest loading items in each factor, is a psychometrically sound measure that can be used to evaluate the quality of life of women of reproductive age in the general population.

# 1721/DEVELOPMENT OF A GERMAN COMPUTER ADAPTIVE TEST FOR DEPRESSION (DEPRESSION-CAT) BY APPLYING ITEM RESPONSE THEORY Herbert Fliege, Otto B. Walter & Janine Becker, Psychosomatic Medicine, University Hospital Charite´, Berlin, Germany; Jakob B. Bjorner, Copenhagen, Denmark; Burghard F. Klapp & Matthias Rose, Psychosomatic Medicine, University Hospital Charite´, Berlin, Germany Objectives: Depression is one of the most common and relevant mental health problems in medical patients. Cost-effective instruments to assess depression routinely are needed. On the basis of the item response theory (IRT) we developed a computer-adaptive test (CAT) to assess depression along the definition of the APA. Subjects: 3.270 patients (two samples of 1.581 and 1.689) of a university hospital answered sets of questionnaires in diagnostic routine. Research design: Items come from 11 validated mental health questionnaires, five are internationally established and six are German. Three independent reviewers rated 144 out of the available 320 items as representative of the depression construct. Items underwent different analyses to check for the prerequisites of IRT, like unidimensionality, good item response curves or prevention of differential item function. We used the programs Mplus, TestGraf, SAS, and Parscale. Results: Sixty-four items fitted a Generalized Partial Credit Model. We calculated CAT scores using an ‘expected a posteriori’ algorithm and a predefined standard error of O0.32 (reliability P 0.90). A run of the algorithm on simulated data sets yielded that the latent trait could be computed with 7.15 ± 1.39 items, in a range of ±2 standard deviations around the population mean. Real patients’ CAT-scores correlate with scores of all available items (r ¼ 0.95) and with the classic CES-D depression score (r ¼ 0.85). The CAT has a better discrimination at low and high ends of the latent trait continuum. Conclusions: Computer-adaptive testing of mental health as a prominent domain of quality of life can be done economically and with satisfactory precision. # 1818/DEVELOPING A QUALITY OF LIFE MEASURE FOR COMPLEMENTARY & ALTERNATIVE MEDICINE David T. Eton, CORE, Karen Koffler, Integrative Medicine; David Cella, CORE, Evanston Northwestern Healthcare & Northwestern University; Amy Eisenstein, CORE, Evanston Northwestern Healthcare, Evanston, IL; John A. Astin, California Pacific Medical Center, San Francisco, CA, USA Complementary and alternative medicine (CAM) represents a host of ‘unconventional’ therapies not traditionally taught in Western medical schools or generally available in Western Hospitals. We conducted a 3-phase study to develop a quality of life (QoL) outcome measure specific to CAM. In Phase I, we interviewed 30 patients (mean age ¼ 54.5 years; 83% female) receiving CAM therapy (i.e., acupuncture, massage, herbal medicine) and collected self-reports from 12 licensed CAM practitioners (mean cumulative CAM experience ¼ 14.9 years). Participants completed a battery of standardized measures including the Functional Assessment of Chronic Illness Therapy – Spirituality (FACIT – SP), Brief Profile of Mood States (BPOMS), Brief COPE, Patient Reactions Assessment (PRA), and items assessing personal control. Participants were asked to, (1) indicate which of the standardized items were relevant to CAM QoL and (2) suggest additional items relevant to CAM QoL. Thirty-eight items (30 standardized and 8 additional) were generated in this phase. In Phase II, we used a Q-sort methodology with 20 previously untested CAM-receiving patients (mean age ¼ 59.2 years; 55% female) to determine which of the 38 items were most relevant to CAM QoL. Twenty-nine of the 38 items from phase I were deemed at least ‘moderately relevant’ to CAM QoL in Phase II, (10 FACIT-SP, 6 PRA, 5 Brief COPE, 5 additional items, 3 personal control). In Phase III, 4 CAM experts (practitioner/researchers) and 1 QoL measurement expert reviewed the 29 items resulting from Phase II for content validity. Efforts at finalizing the measure are ongoing, however, to date seven domains have been identified: pain, fatigue, physical function, life satisfaction, coping/self-efficacy, existential, and practitioner skill. These domains will be represented in a pilot version of the measure with validation to follow.

741 # 1038/DEVELOPMENT AND VALIDATION OF THE CHINESE QUALITY OF LIFE INSTRUMENT FOR CHINESE MEDICINE Zhao Li, Research & Development Division, School of Chinese m, Hong Kong Baptist University; Leung Kwok Fai, Department of Occupational Therapy, Queen Elizabeth Hospital, Hong Kong; Liu Feng Bin, Department of Internal Medicine, The First Affiliate; Lin Li Zhu, Department of oncology, The First Affiliated Hospita; Lang Jian Ying, Department of Medical Statistics, Guangzhou University of Traditional Chinese Medicine; Fang Ji Qian, Department of Medical Statistics, School of Public H, Sun Yat-Sen University, Guangzhou; Kelvin K.C. Chan, Research & Development Division, School of Chinese m, Hong Kong Baptist University, Hong Kong, China

all differences were of a clinically meaningful magnitude. In a pooled analysis of five clinical trials, we confirmed the previously documented relationship between hemoglobin and fatigue. Further, these data suggest that meaningful improvements in fatigue have a positive impact on HRQOL.

In this study, a new Chinese Quality of Life (ChQOL) instrument incorporating the holistic balance model of health in Chinese Medicine (CM) was developed to evaluate the efficacy of CM and OM (Orthodox Medicine). This instrument consists of a basic core module and several specific modules corresponding to different disease groups defined in CM. Here we mainly focus on the model of this new core module of ChQOL and show some preliminary results on the test of this module to evaluate its validity. As the basic ideas of CM, the theory of Yin and Yang, Essence-Qi-Vitality and Seven-Emotion are used as the basis for the construction of the ChQOL-core. The equilibrium state between Yin and Yang is set as the principal framework for the ChQOL-core and the perfect equilibrium between Yin and Yang reflect the best health and quality of life. The degree of the equilibrium between Yin and Yang are characterized by three aspect of unification between vitality and physique, human and nature, human and society. The unification between vitality and physique can be described by the theory of EssenceQi-Vitality. Compared with the QOL instrument in west, these three unifications are analogy to the physical/functional domain, environment domain and social domain respectively. The theory of Seven-Emotion can be used as an analogy of the psychological domain. As a whole, the ChQOL-core developed by us not only incorporates the key theories and concept of health in CM, but also includes both the negative and positive health/QOL. It is also possible to make the ChQOL-core comparable to the contemporary health concepts in OM based on their similarity in both structure and concepts. We have prepared the field test version of the ChQOL-core and the field test will be conducted soon. Preliminary results will be presented in the ISOQOL annual conference in November 2003. With the results of field test, the domain and facet structures of the ChQOL-core will be refined.

We assessed health-related quality of life in patients with liver cancer using Taiwan versions of WHOQOL-BREF and EORTC QLQ-C30. A total of 172 subjects were consecutively enrolled from patients with hepatocellular carcinoma (HCC) who followed up regularly at the outpatient clinics or during hospitalization at a university hospital from February to April 2002. Internal consistency reliability revealed by the Cronbach’s a coefficient ranged from 0.67 to 0.91 across four domains for WHOQOL-BREF. Exploratory factor analysis identified four factors, which corresponded to the original construct. Each domain score of WHOQOL-BREF correlated significantly with utility measured by visual-analogue scale and standard gamble (Pearson’s correlation range, 0.40–0.59, and 0.17–0.38, respectively), and also positively associated with self-evaluated health status and happiness (Spearman correlation range, 0.26–0.61 and 0.34–0.66, respectively). Significantly moderate to high correlations were found between WHOQOL-BREF and EORTC QLQ-C30 in the aspects of physical, psychological, and general health. These QOL scores were compared with data from 213 healthy persons using general linear model. These patients with liver cancer were found to have significantly lower scores on physical, psychological, social domains. In conclusion, the WHOQOL-BREF can be a valid HRQOL instrument, which is potentially suitable for health outcome evaluation and to supplement clinical decision-making in HCC patients.

# 1581/REDUCING ANEMIA-RELATED FATIGUE IMPROVES HEALTH-RELATED QUALITY OF LIFE IN ANEMIC CANCER PATIENTS: RESULTS FROM FIVE RANDOMIZED CLINICAL TRIALS Joel Kallich, Health Economics, Amgen, Inc., Thousand Oaks, CA; Xiao Xu & Anne McDermott, Outcomes Services, Covance Health Economics and Outcomes Services Inc., Gaithersburg, MD; M. Haim Erder, Health Economics, Amgen, Inc., Thousand Oaks, CA; David Cella, Center on Outcomes, Research and Education, Evanston Northwestern Healthcare, Evanston, IL, USA Anemia in patients with cancer has been associated with fatigue and other symptoms usually associated with health-related quality of life (HRQOL). The purpose of these analyses was to examine the impact of fatigue on other HRQOL outcomes. Data from three dose-finding and 2 phase III randomized clinical trials were combined (n ¼ 1204). Trial patients received darbepoetin alfa, recombinant human erythropoetin, or placebo over a 12-week period. Patients completed the Functional Assessment of Cancer Therapy-Anemia (FACT-An) questionnaire (containing Fatigue, Physical, Social, Emotional, and Functional scales) and numeric rating scales (NRS) of Energy, Activity, and Overall Health at baseline, during interim study visits, and at the end of treatment. HRQOL change scores were computed as the last observed value less the baseline value. Patients with a hemoglobin improvement of 2 or more g/dl had a mean change in FACT Fatigue subscale scores that was 4.2 points greater than the change for those who did not achieve this level of hemoglobin response (p < 0.0001). Patients experiencing a clinically meaningful (P3 points) improvement on the FACT Fatigue subscale reported significantly (p < 0.05) greater improvements in all of the other HRQOL domains, including overall health (19.1% greater improvement) and ability to perform usual activities (27.8% greater improvement). With the exception of social well-being,

# 1652/ASSESSING HEALTH RELATED QUALITY OF LIFE IN PATIENTS WITH HEPATOCELLULAR CARCINOMA Jung-Der Wang & Lukas Jyuhn-Hsiarn Lee, Institute of Occupational Medicine and Industrial Hyg, National Taiwan University, College of Public Health; Kaiping G. Yao, Department of Psychology, National Taiwan University, Taipei, Taiwan

# 1677/CONTRALATERAL PROPHYLACTIC MASTECTOMY: LONG-TERM SATISFACTION, PSYCHOLOGICAL, AND SOCIAL FUNCTION Marlene H. Frost, Mayo Clinic Women’s Cancer Program; Daniel J. Schaid, Jeffrey M. Slezak & Shannon K. McDonnell, Biostatistics; Thomas A. Sellers, Clinical Epidemiology; Lynn C. Hartmann, Medical Oncology, Mayo Clinic, Rochester, MN, USA Contralateral prophylactic mastectomy (CPM) is one option for reducing the risk of a second breast cancer in women with a personal and family history of breast cancer. Few data are available regarding women’s satisfaction, psychological, and social function following CPM. We conducted a descriptive study to examine these outcomes. A total of 792 women with a personal and family history of cancer elected CPM at Mayo between 1960 and 1993. A questionnaire, mailed to all women known to be alive (n ¼ 621) resulted in completion by 94% (n ¼ 583) of the women. The questionnaire included a series of ordinal scales and open-ended items to identify women’s reasons for CPM, perceived risk of breast cancer, satisfaction with CPM, selfesteem, body image, feelings of femininity, sexual relationships, emotional concern about developing breast cancer, stress, and emotional stability. Medical records provided demographic and CPM procedure data. Data were analysed using descriptive and linear regression statistics. A median of 10.7 years after the procedure, the majority of women were satisfied (83%) with their CPM. A smaller number were neutral (8%) or dissatisfied (9%). Variables associated with satisfaction explained 25% of its variance (p < 0.01). Decreased satisfaction with CPM was associated with: decreased satisfaction with appearance, complications with reconstruction, reconstruction after CPM, decreased satisfaction with sexual relationships, and increased level of stress in life. The majority of women experienced no change or favorable effects in respect to their self-esteem (83%), level of stress in life (83%), and emotional stability (88%). Fifty-nine percent of the women reported decreased concern about developing breast cancer. Satisfaction with body appearance, feelings of femininity, and sexual relationships were the most adversely affected with 33, 26 and 23% of the women respectively reporting diminished satisfaction. While most women are satisfied with CPM, women making these decisions should also consider possible adverse effects.

742 # 1747/DOES ADJUVANT CHEMOTHERAPY IMPAIR QUALITY OF LIFE? A TRIAL OF ADJUVANT NAVELBINE/CISPLATINUM FOR EARLY STAGE LUNG CANCER Andrea Bezjak, Tim Winton, Keyue Ding, Barbara Graham, Lesley Seymour & Joseph Ayoub, NCIC Clinical Trial Group, Cancer Research Institute, Kingston, Ontario, Canada; Robert Livingston & Eric Vallieres, Southwest Oncology Group, Seattle; James Rigas & Todd Demmy, Cancer and Leukemia Group, Lebannon, NH; Russell Devore & Kenneth Kessler, Eastern Cooperative Oncology, Knoxville, TN, USA; Frances Shepherd, NCIC Clinical Trials Group, Cancer Research Institute, Kingston, ON, Canada Background: Adjuvant chemotherapy may prolong survival in patients with early stage non-small cell lung cancer (NSCLC), but its toxicity may impact patients quality of life (QOL). We report preliminary QOL results of a randomized study of adjuvant chemotherapy in NSCLC. Materials/Methods: A Phase III randomized study of adjuvant navelbine and cisplatin in completely resected T2 N0, T1-2N1 NSCLC patients was lead by NCIC CTG with participation of ECOG, SWOG, and CALGB. Overall survival was the primary endpoint. QOL was assessed by EORTC QLQ-C30 and lung cancer module at baseline, weeks 5 and 9 of chemotherapy, and at 3, 6, 9, 12, 18, 24, 30 and 36 months. A change score (i.e. difference between follow-up QOL score and baseline score) of 10 points or greater is considered to represent a clinically noticeable change. Results: A total of 482 patients were randomized. Two hundred and nine Canadian patients (out of 234 randomized) and 152 US patients (out of 248 randomized) completed baseline QOL. Baseline QOL data and other characteristics were balanced in the two arms. Impairments at baseline (post-surgery) were seen in global QOL (mean score 58), role functioning (59) and physical functioning (73). Most pronounced symptoms at baseline were fatigue (36), pain (36), dyspnea (36), sleep (36), cough (33), shortness of breath (31) and worry about health (45). QOL impairments during chemotherapy were relatively mild, with worsening of fatigue, nausea and vomiting but a reduction in pain and no change in global QOL. Patients in the observation arm showed considerable improvements in QOL by 3 months (physical, role, social and global QOL change scores all >10). Global QOL improved to similar extent in patients in the chemotherapy arm by 9 months post-randomization, despite this being the peak time for patients to report symptomatic peripheral neuropathy (change score for numbness of fingers/ toes ¼ 29; pins and needles ¼ 21). Conclusion: Despite toxicity of adjuvant chemotherapy, NSCLC patients treated with adjuvant chemotherapy maintain and improve their QOL within the first year after surgery. # 1659/QUALITY OF LIFE IN CANCER PATIENTS AFTER SURGERY: THE MEANING OF PERSONAL RESOURCES AND COPING STRATEGIES Sonja Boehmer & Ralf Schwarzer, Department of Health Psychology, Freie Universitaet Berlin, Berlin, Germany Critical life events like being diagnosed with cancer demand multifaceted and flexible adaptation. From a stress-theoretical perspective, this adjustment process can be facilitated by personal resources such as self-efficacy. The aim of the present study was to investigate how quality of life changes after tumor-related surgery, whether general self-efficacy has a positive effects on patients’ quality of life (QOL) and whether this effect is direct or indirect, i.e., mediated by coping strategies. In a longitudinal setting, patients were approached 1 month

(T1), 6 months (T2) and 1 year after (T3) tumor-related surgery (n ¼ 97, 63% male, aged between 34 and 86, in the majority of patients suffering from gastrointestinal cancer). The questionnaire comprised the EORTC-QLQ-C30, the General Self-Efficacy Scale as well as subscales of the Brief COPE, the Mental Adjustment to Cancer Scale, and the Coping with Surgical Stress Scale. Structural equation modeling was used for data analyses. Patients’ physical function, role function, social function, fatigue, pain, appetite loss, and global QOL improved over time with effect sizes ranging from 0.20 to 0.72. Selfefficacy and the coping strategies of positive reframing, planning, acceptance, internal control beliefs, fighting spirit, finding meaning, and optimistic social comparison 1 month after surgery were found to be positively associated with global quality of life 5 months later, while external control beliefs, anger, and rumination had a negative effect. Regarding the question on mediation, SEM analysis revealed that the positive effect of self-efficacy at T1 on QOL at T2 was completely mediated by positive reframing, while the effect of self-efficacy at T2 on QOL at T3 was mediated only partially by internal control beliefs. It can be concluded that adjustment to cancer is facilitated by general self-efficacy, but this effect is at least partially mediated by coping strategies. # 1120/DISTRIBUTION-BASED CRITERIA FOR CHANGE IN HEALTH-RELATED QUALITY OF LIFE IN PARKINSON’S DISEASE Josephine M. Norquist, Ray Fitzpatrick & Crispin Jenkinson, Department of Public Health, University of Oxford, Oxford, UK To increase the interpretability of results from health-related quality of life (HRQoL) measures, instruments need to detect intra-individual change beyond chance or measurement error. Distribution- and anchor-based methods are often used to determine what constitutes a minimal meaningful change in an instrument’s score. This study uses the standard error of measurement (SEM) and the standard error of the difference (Sdiff) in relation to longitudinal data on individuals with Parkinson’s disease (PD) to calculate the minimum change scores required to be statistically meaningful for each of the dimensions of the PDQ-39. Of the range of disease-specific instruments that have been developed to assess the dimensions of outcome for individuals with PD, the PDQ-39 is the most widely used instrument with the most extensive supportive evidence of measurement properties. Data were examined from two separate studies. A community survey of individuals with PD was carried out in which members (n ¼ 851) of 13 branches of the Parkinson’s Disease Society across England participated in the study. A second study of individuals with PD was clinicbased. Consecutive patients (n ¼ 146) attending neurology outpatient clinics of six hospitals in Anglia and Oxford Regional Health Authority participated. The patterns of SEMs and Sdiffs were similar both across time periods and between samples, for all dimensions except Social Support. The results suggest that, for example, 6 points change on a 0–100 transformed scoring of the Mobility dimension may be considered on distributional grounds a minimum meaningful change. The demonstrated consistency across occasions and types of sample, community and clinic based, and level of HRQoL of SEMs and Sdiff for the majority of the dimensions of the PDQ-39, is empirical evidence of the theoretically claimed advantage of sample independence. The distributional parameters examined here should be explored among other measures of health status and health related quality of life. This approach may offer a general solution to distributional criteria for change.

743 # 1540/A COMPARISON OF INDIRECT AND DIRECT MEASURES OF CHANGE IN SYMPTOMS AFTER ELECTIVE SURGERY Eva M. Bitzer, Christoph Lorenz & Hans Doerning, ISEG, Institute for Epidemiologie, Social Medicine, Health, Hannover, Germany Change measurement is central to quality of life assessment in clinical trials, health services research and quality management. Gold-Standard is the prospective assessment of pre-post intervention changes. This is called indirect measurement, because changes are calculated indirectly by subtracting pre- and post-scores. There are situations were a pre-post-Design is not feasible (i.e. Emergencies) or simply too expensive. That’s where direct assessment of change might be interesting. We compared pre-post measurement with two variants of direct change measurement. Variant A: perceived change compared to the time before surgery (one question per item), Variant B: recalled symptom level at the time before surgery and actual symptom level (two questions per item). In a prospective observational study (185 patients after hernia repair and 130 patients after cholecystectomy; T0: pre-operative; T1: 14 d. post-OP, T2: 6 m. post-OP) changes in disease specific symptoms (Hernia: 9 Items, Cholecystectomy: 8 Items) were assessed prospectively and compared to the two variants of direct change measurement. Comparisons were made according to missing values, relations between indirect and direct measurements (Spearman, Kappa, Kendel’s tau b), the degree to which direct measures over-/underestimate changes. Both Variants are higher (co-)related with indirect change measurement at T1 than at T2 and overestimate changes (in the direction of improvement) with similar frequency. The (co-)relation between indirect measurement and Variant B is higher than that with Variant A in both patient groups, in all items and at T1 and T2. The degree to which overestimation occurs is lower in Variant B than in A. Variant B produces less missing values than Variant A. The retrospective ‘direct’ assessment of change is more optimistic: perceived changes in the direction of improvements are larger than observed in indirect measurement (at least in the context of elective surgery). The data confirm, that Variant B, where pre-operative symptoms are recalled retrospectively, is conceptually closer to the indirect pre-post measurement. # 1139/PATIENT REPORTED OUTCOMES (PROS) IN THE EVALUATION OF NEW DIABETES TREATMENT: RELIABILITY, VALIDITY, AND MINIMAL CLINICALLY IMPORTANT DIFFERENCE (MCID) Clarice P. Hayes & Lee Bowman, Global Health Outcomes Research, Eli Lilly & Company, Indianapolis, Indiana, USA As part of an effort to establish a set of standard patient-reported outcome measures (PROs) for the evaluation of diabetes pharmacologic treatment, we administered the Short Form-36 Vitality Scale and Diabetes Symptoms Checklist-Revised (DSCR) hyperglycemia, hypoglycemia, and fatigue subscales to 161 patients with type 2 diabetes participating in an international, randomized, dosing study of an injectable diabetes treatment. We used baseline data to calculate reliability coefficients and standard errors of measurement (SEM). We used data collected at baseline and end of study to: (1) investigate the responsiveness of the PROs and (2) explore the correspondence between the minimal clinically important differences (MCID) in PRO endpoints and in the clinical endpoint [i.e., glycosolated hemoglobin (A1c)]. Reliability coefficients across countries were acceptable (P0.70). Although the study was not statistically powered for PRO endpoints, we observed significant improvements (p < 0.05) in mean

vitality and hyperglycemic symptom scores. Using SEM for PRO measures and 0.7% for A1c to indicate MCID, we found that 32–38% of patients experienced minimal clinically important differences in vitality or symptoms as compared to 29% who experienced clinically significant improvments in A1c. Of patients who experienced a 0.7% or greater reduction in A1c, 48, 37, 48, and 26% also experienced clinically important improvements in vitality, fatigue, hyperglycemic symptoms, and/or hypoglycemic symptoms, respectively. We found evidence for the reliability and validity of all PROs for the evaluation of diabetes pharmacologic therapy. In addition, using SEM as the definition of MCID for PRO measures, we showed that patients experienced clinically important change in PROs more frequently than they experienced it in glycemic control. The PRO endpoints were as sensitive as A1c, perhaps more so, in detecting improvements in diabetic patient care. # 1722/HEAD AND NECK CANCER PATIENTS DEMONSTRATE IMPROVED MENTAL HEALTH RELATED QOL FOLLOWING SURGERY DESPITE LOSS OF PHYSICAL HEALTH RELATED QOL Assuntina G. Sacco, Otolaryngology – Head and Neck Surgery, University of Michigan Health System, Ann Arbor, MI; Rodney J. Taylor, Otolaryngology – Head and Neck Surgery, University of Maryland Medical System, Baltimore, MD; Theodoros N. Teknos, Carol R. Bradford, Jeffrey E. Terrell, Norman D. Hogikyan, Gregory T. Wolf & Debra Graetz, Otolaryngology – Head and Neck Surgery; Julia Lee, Cancer Center; Douglas B. Chepeha, Otolaryngology – Head and Neck Surgery, University of Michigan Health System, Ann Arbor, MI, USA Objective: To evaluate both physical and mental health related quality of life (QOL) issues for patients treated with surgery for head and neck cancer. Design: A prospective, longitudinal study to evaluate the impact on QOL and shoulder function following surgical management with neck dissection for head and neck cancer by comparing preoperative with 12-month post-operative QOL. Setting: Tertiary care referral multi-disciplinary cancer center. Patients: There were 126 patients completing both pre- and 12-month post-operative evaluations who were enrolled between June 1998 and March 2002. Mean age 56.6 (14–87); M:F, 100:26; Tx: 17/126 (13.5%); T1-T2: 55/126 (44%); T3-T4: 54/126 (43%). Seventy-nine percent of the patients received both surgery and radiation and 83% of the patients had squamous cell carcinoma. Mean duration of follow-up was 34.1 months (12–57). Main outcome measures: After enrollment and prior to treatment, patients received Medical Outcomes Study Short Form 36-Item Health Survey (SF-36), the University of Michigan Head and Neck Quality of Life (HNQOL) and Constant’s Shoulder Function Test. These assessments were repeated 12 months following the completion of surgical treatment. Higher scores denote better QOL and shoulder function. Results: The SF-36 domains which demonstrated higher 12 month post-operative scores were Mental Health (p < 0.0001) and Mental Component Scale (p < 0.0001), whereas lower scores were in Physical Function domain (p ¼ 0.008). Despite the same pre-operative Constant scores, 87.4 vs. 87.5 (p ¼ 0.98), the post-operative Constant scores for modified radical neck dissection (MRND) and selective neck dissection (SND) patients were 72.3 vs. 81.6 (p ¼ 0.01). Conclusions: Despite deterioration in physical health related QOL that occurs after surgical treatment of head and neck cancer, the adaptability of this patient population results in significantly improved mental health related QOL.

744 # 1437/STABILITY OF QUALITY OF LIFE IN LONG-TERM MINORITY PROSTATE CANCER SURVIVORS Miles Y. Muraoka & Carolyn C. Gotay, Clinical Sciences Program, Cancer Research Center of Hawaii, Honolulu, HI, USA

disrupt the factor constructs but they may be kept in the model if discriminating property from IRT was low, thereby avoiding bias. Our results revealed the potential strengths of the application of IRT in the analysis of item reduction.

This report examines stability in quality of life (QOL) in long-term prostate cancer survivors over a three-year period. Participants were identified through a population-based tumor registry, based on prostate cancer diagnosis at 5, 8, or 11 years previously. Participants completed mailed questionnaires at study entry and 3 years later. Data collection is ongoing; this report includes the first 88 respondents out of a total sample of 529. Mean age of participants was 79. Most were married (62%) with about three-quarters non-white, primarily Asians and Pacific Islanders. Measures reported here include the Physical (PHYS) and Emotional (EMO) Functioning subscales of the Rand 36-Item Health Survey and ratings of overall QOL. Predictors of follow-up scores include demographics (age, marital status, education, ethnicity), clinical variables (treatment, years-post diagnosis, and symptoms: urinary, bowel, and sexual functioning), and initial scores. The results of paired t-tests showed that responses were quite stable, with the exception of PHYS, which declined significantly from the first (mean 75.7) to the follow-up (mean 69.3) assessment (p < 0.02). Overall QOL and EMO remained high across assessments. Correlations between first and follow-up measures were all large and positive (r’s of 0.38–0.57, p’s < 0.01). Multiple regression analyses indicated that respondents’ scores on the first questionnaire were the most important predictors of follow up responses (p’s < 0.001). For PHYS and QOL, sexual functioning was also a significant negative predictor (p’s < 0.09). Being 8 years post-diagnosis predicted lower EMO (p < 0.06) (with lower EMO scores longer after diagnosis) and treatment predicted PHYS (p < 0.08), with men treated by radiation therapy only reporting higher scores. While the model explained significant variance (R 2 of 0.34–0.58), no other variables approached significance. This study indicates that QOL in prostate cancer survivors is stable many years after diagnosis. Problems in sexual functioning have broader eects on the QOL of elderly men. Targeted interventions are warranted.

# 1719/THE CERVICAL DYSTONIA IMPACT PROFILE (CDIP-58): 2 – COMPARISON OF TRADITIONAL METHODS AND RASCH TECHNOLOGY FOR DEVELOPING SCALES Stefan J. Cano & Tom T. Warner, Department of Clinical Neurosciences, Royal Free & University College Medical School, London; Jeremy C. Hobart, Department of Clinical Neurosciences, Peninsula Medical School, Plymouth, UK

# 1732/METHODOLOGICAL APPROACH TO REDUCE ITEMS FOR COPD PATIENTS USING WHOQOL-BREF Ting-Hsiang Lin, Department of Statistics, National Taipei University, Taipei, Taiwan; Wen-Miin Liang & Hsin-Wen Chang, Department of Public Health, Li-Jin Huang, Department of Nursing; Chnag-Bi Wang, Department of Public Health, China Medical College, Taichung, Taiwan Quality of life (QOL) of patients is typically measured by lengthy questionnaires, but in clinical practice shorter versions would be more practical and convenient as long as they are equally accurate. There are several shortening methods. Clinicians have used assessments of items that are most important to patients. Psychometricians tend to emphasize items which strengthen factor constructs, such as validity discrimination and internal consistency. Recently there have been more studies on item response theory (IRT). In this study, 262 COPD patients aged greater than 44 years were included. The main outcome measures were quality of life (QOL) scores by using the WHOQOLBREF. Three methods were used to reduce items. First, a confirmatory factory analysis (CFA) was used to obtain an acceptable model by suggesting which items should be removed according to two psychometric criteria: (a) item discriminant validity and (b) item internal validity. Items were removed based on the Lagrange Multiplier Indices from CFA until the goodness of fit indices CFI >0.9 and RMSEA <0.7. Second, the scores evaluating by patients about the importance of each item of the WHOQOL-BREF were used to reduce the items. Third, the item with less degree of discrimination from a graded response model with natural metric of the logistic response function, was considered to be removed. The results showed different item reduction methodologies produced different results. Ten items suggested to be removed from CFA analysis were different from the 10 items suggested by IRT. COPD patients consider taking medicine often to be bothersome and importantly affecting their QOL but this item did not reveal any discriminating properties by IRT analysis and did not show good internal validity by psychometric analysis. In conclusion, no single methodology was ideal. Some important items appeared to

There are a number of different psychometric methods for developing and evaluating rating scales. The literature often terms these as traditional (Classical Test Theory, CTT) and new (Rasch Analysis and Item Response Theory). The relative impact of different methods has received limited attention. This study compared CTT and Rasch technology for reducing an item pool and developing scales during the process of developing the CDIP-58. Responses from 556 people to 150 items concerning the health impact of cervical dystonia were analysed. This pool was reduced to 73 items (7 scales) by CTT, and 58 items (8 scales) by Rasch technology. Rasch identified a separate scale for cervical dystonia symptoms. The other seven scales measured similar constructs but the content of the scales differed notably. Forty items were common to both versions (25% of original pool), although few activities and psychological well-being items were shared. Both versions passed traditional psychometric tests (e.g. Cronbach’s a > 0.88, item–total correlation ¼ 0.57–0.94, test-retest > 0.86), but scales of the CTT version did not fulfil Rasch criteria (e.g. infit > 1.3). Finally, the content of the Rasch version was qualitatively rated as more clinically relevant. The Rasch developed CDIP-58 satisfied criteria for rigorous measurement (CTT and Rasch criteria), but the CTT measure (CDIP-73) did not meet all standards. This suggests that Rasch technology may have advantages over traditional item reduction techniques. # 1858/AN EMPIRICAL DEMONSTRATION OF THE LIMITATIONS OF CLASSICAL TEST THEORY AND THE ADVANTAGES OF NEW PSYCHOMETRIC METHODS Jeremy C. Hobart, Clinical Neurosciences, Peninsula Medical School, Plymouth, Devon, England There is increasing interest in the application of new psychometric methods (Rasch analysis and Item Response Theory). However, traditional psychometric methods (Classical Test Theory) remain the most widely used technology for the development and evaluation of health rating scale, and the analysis of the data they generate. This study compares the CTT with Rasch analysis in order to highlight the limitations of the former, advantages of the later, and potential impact for clinical trials. Multiple Sclerosis Impact scale (MSIS-29) data from 1725 people with MS were analysed using CTT and Rasch item analysis. First, we examined indicators of item difficulty (CCT – item mean score; Rasch – item calibrations) in different subsamples defined by level of person disability. Second, we examined person scores (CTT – summed score; Rasch – person measures) generated by different groups of items defined to measure the same construct. Third, we examined the relationship between person scores (CTT – summed score; Rasch – person measures). Four, we examined the errors associated with person scores generated by each analysis. Using CTT, item difficulty was sample dependent and person raw scores were scale dependent. Using Rasch analysis, item difficulty was sample independent and person measures were scale independent. The plot of raw scores against interval measures indicated that the meaning of raw score differences and changes varies across the scale. It is most concerning at the extremes where a raw score change underestimates linear change 6-fold. CTT produces a single standard error for all person scores whereas Rasch analysis produces a standard error for each person measure. This has implications for the analysis of longitudinal data. New psychometric methods have important advantages over CTT. The limitations of CTT have an impact on the interpretation of clinical trials.

745 # 1829/COMPARISON OF ITEM REDUCTION TECHNIQUES IN THE DEVELOPMENT OF THE QUICKDASH OUTCOME MEASURE Dorcas E. Beaton, Mobility Program Clinical Research Unit, St. Michael’s Hospital; James G. Wright, Orthopaedics, Hospital for Sick Children, Toronto, ON, Canada; Jeffrey N. Katz, Rheumatology, Brigham and Womens Hospital, Boston, MA, USA Questionnaires should be a short as possible while retaining the measurement properties needed for their intended application. The purpose of this study was to reduce the 30-item DASH Outcome Measure (Cronbach’s a ¼ 0.97) down to an 11-item scale and compare item reduction strategies in the process. Methods: Field testing data of the full DASH (n ¼ 407) were used for three parallel approaches to item reduction: concept retention (CR: retaining best item from each domain of the original conceptual framework); equidiscriminative item total correlations (EITC); and one parameter item response theory (IRT). A second cohort (n ¼ 200) of subjects was used to compare the properties of the shortened versions. Criterion for the preferred version included: no items with >40% in one response category, Cronbach’s alpha and test–retest reliability of >0.90, validity and responsiveness (standardized response mean (SRM)). Results: Among three 11 items scales, only two items were in common. CR and EITC had one item with >40% at no difficulty. IRT had three. Alpha/ test–retest reliability were high (CR ¼ 0.92/0.94; EITC ¼ 0.95/0.96; IRT ¼ 0.95/0.97). All correlated well with DASH (>0.98), and with VAS of function (CR ¼ 0.80; EITC ¼ 0.74; IRT ¼ 0.78). SRM for CR (0.79) was closest to full DASH (0.78) (EITC ¼ 0.69; IRT ¼ 0.64). The results, blinded as to approach, were sent to the team of DASH developers who unanimously chose CR version based on ranking of overall impressions of results, and similarity to full DASH. Conclusions: Varying item reduction techniques produced shortened scales with different item content, but surprisingly similar measurement characteristics, possibly reflecting the high correlation of items in the original scale. The measurement properties of the least psychometric version were closest to the full DASH. A concept-based version (clinimetric approach) is recommended for the QuickDASH. # 1096/RESPONSE CONVERSION ON WALKING DISABILITY MEASUREMENTS: A NEW APPLICATION OF ITEM RESPONSE THEORY Marijke Hopman-Rock & Stef van Buuren, Physical Activity and Health, TNO Prevention and Health, 2301 CE Leiden, The Netherlands; Alan Tennant & Sophie Eyres, Psychometric Laboratory for Care Outcomes Measurement, University of Leeds, Leeds, United Kingdom Purpose: The goal of the Health Monitoring Program of the European Commission is to provide relevant and timely information about the health in each Member State. To avoid unnecessary duplication, the new monitoring system will have to be fed by existing data collected through health surveys performed by individual member states. Though the content of these surveys is often quite similar, substantial variations in the actual measurement exist, e.g. in sampling procedures, in the coverage per topic, in the wording of questions and formulation of response categories. Incomparability of information is a major problem in this context. Each Member State has its own tradition in collecting and processing data, and changing established ways of working is not so easy. Methods: Response conversion provides a solution to the problem of divergent formulations of survey questions.

The method transforms incomparable data into a common scale. Where this can be done, comparisons between member states can be made in terms of the common scale. The method consists of two steps. The first step is the construction of a conversion key. This is a relatively complex activity, but needs to be done only once. In the second step, one uses the conversion key to convert prevalence information from individual Member States into a common scale. This step is simple, and can be repeatedly done on a routine basis as new information arrives. The technique works if enough overlapping information in the existing information can be found. A linkage map is a systematic way of arranging overlapping information. The major danger in practical applications is that linkage may not be possible. Statistical models based on item response theory systematically exploit the overlapping information. Properties of such models are well known, but their application in this field brings some fresh methodological problems: how to measure the quality of the conversion key, how to properly account for the uncertainty and translation errors, and so on. The method is illustrated for walking disability differently measured in eight Member States. # 1844/WELLNESS LIFESTYLES Tonya L. Schuster, Sociology, University of California, Irvine, CA; Marnie Dobson, Sociology; Maritza Jauregui, Social Ecology, Robert H. Blanks, Anatomy, University of California, Irvine, USA This research represents an empirical application of a theoretical framework linking use of Network Spinal Analysis (a wellness-oriented form of complementary and alternative medicine), health lifestyle practices, and self-reported health and wellness. We surveyed 2596 patients from 156 offices of members of the Association for Network Care; estimated response rate ¼ 69%. Expanding on the concept of quality of life, we developed a wellness construct consisting of calculated difference scores between two referents, presently and before Network care, for self-reported items representing wellness domains of physical state, mental-emotional state, stress evaluation, and life enjoyment. Positive reported change in nine items assembled into dietary practices, health practices, and health risk dimensions serve as indicators of the construct of changes in health lifestyle practices. The NSA care construct consisted of duration of care in months, awareness of energy and awareness of breathing since beginning Network care. Exogenous variables entered into a structural equation model include gender, age, education, income, marital status, ailments, life change, and trauma. Results demonstrated reported wellness benefits accrue to individuals along a direct path from both self-reported positive lifestyle change (0.22), and from NSA care (0.43). The path (0.65) from NSA care to positive health lifestyle changes indicates that NSA care also has an indirect effect on wellness through changes in health lifestyle practices. Only gender, age, and education remained in the final structural equation model. The Structural Equation model tested in these analyses lends support to a theoretical framework linking wellness, health lifestyles, and CAM. This study shows that our measurements of health and wellness are particularly appropriate for investigating wellness-oriented CAM. NSA care users do tend towards the practice of a positive health lifestyle, which also has a direct effect on reported improvements in wellness. These empirical links are discussed relative to the sociodemographic characteristics of this population and show that use of NSA care is an aspect of a wellness lifestyle.

746 # 1320/THE CULTURAL VITALITY INDEX: INTERSECTING INDICATORS OF ARTS/CULTURE AND PUBLIC HEALTH FOR QUALITY OF LIFE MEASUREMENTS Joaquin Herranz Jr, Department of Urban Studies and Planning, Massachusetts Institute of Technology, Cambridge, MA; Maria-Rosario Jackson & Florence Kabwasa-Green, The Urban Institute, Washington, DC, USA The paper presents indicators of the relationship between art/culture and public health for urban communities, and the development of the Cultural Vitality Index. This indicator research is based on a national study examining the role of community-based arts and cultural activities in contributing to community well-being including public health, social capital, and enriching urban cultural vitality. The paper argues that art/culture conceptual and data limitations have contributed to an under-emphasis of cultural dynamics in quality of life measurements. At the same time, research has found that art and cultural participation contributes to positive health outcomes such as reduced stress, improved healing, and enhanced quality of life. The paper combines two research streams and makes connections between arts/culture and healthy communities. Of particular interest to conference participants is a proposed Cultural Vitality Index (CVI) that is analogous to the infant mortality rate (IMR). Like the IMR, the CVI may be used as an indication of systems of support for creativity. The paper discusses theoretical, methodological, and policy implications of the CVI. Findings are drawn from a 6-year study by The Urban Institute, a national research organization in Washington DC. Methods included interviews and focus groups involving 400 people, and participatory research projects in seven cities (Los Angeles, Boston, Providence, Chicago, Oakland, Denver, Washington, DC). The paper also draws on two other Urban Institute studies: (1) an exploration of support systems for artists in six cities; and (2) a national telephone survey of residents about participation in art/culture activities and in civic activities. Sample references: Ashton J (ed.), Healthy Cities. Keynes, England: Open University Press, 1992, Jackson, Maria-Rosario, Joaquin Herranz Jr. Culture, Creativity, and Communities: A Framework for Measurement. Washington, DC: The Urban Institute Press, 2002. # 1555/QUALITY OF LIFE AND HEALTH PERCEPTION OF OLDER ADULTS (THE WHOQOL-OLD PROJECT) Ramona Lucas Carrasco, Catalan Institute on Ageing; Rosa Monteserı´n Nadal, Equipo de Atencio´n Primaria Sardenya, CAP Sardenya; Salvi Junca Valdor, Subdireccio´n de Planificacio´n y Evaluacio´n, Departamento de Sanidad. Generalitat de Catalunya, Barcelona, Spain We present preliminary results on Quality of Life (QoL) on a sample of older people using the WHOQOL-100 in relation to health perception (healthy/unhealthy) and in relation to the number of health conditions reported. Twenty-three countries are involved worldwide in the measurement of QoL in older adults (The WHOQOL-OLD Project). In Barcelona, Spain, 302 adults 60 and older from PC Centres, Residential Facilities and Community Centres have participated in the pilot study. All participants signed the Informed Consent and completed the following questionnaires: the WHOQOL-100 (a generic QoL questionnaire for adult population which contains six domains: Physical, Psychological, Levels of Independence, Social Relations, Environment and Spirituality), a Module for Older Adults, Importance questions, sociodemographics and a list of co-morbid conditions. The project is funded by the EU in collaboration with the WHO. Nearly 57% were women, 63% were married, 60% were living at home unsupported. 64% reported being healthy, 56% reported one or more health problems affecting their QoL and nearly 93% reported 1 or more health conditions. Overall mean QoL was higher for those reporting themselves as healthy compared to those unhealthy (14.6 and 12.3, respectively). In relation to the WHOQOL-100, people reporting themselves as healthy had higher means in all domains and facets than those reporting themselves as unhealthy. All differences were significant except for one domain (Spirituality) and five facets (Social Support, Home Environment, Health and Social Care, Physical Environment and Spirituality). Older people perceiving themselves as unhealthy and those reporting more health conditions showed lower QoL means. Further research in relation to conditions affecting the QoL of the elderly is needed.

# 1653/HEALTH, INEQUALITY AND QUALITY OF LIFE: AN INTERNATIONAL INVESTIGATION OF EDUCATIONAL STATUS Suzanne M. Skevington & Laura Camfield, Well-being in Developing Countries Research Group (W, University of Bath, Bath, Banes, UK) The World Health Organisation sees the raising of literacy rates to above 70% as fundamental to its strategy of improving health and wellbeing in developing countries. But how far is the receipt of education intrinsic to perceptions of a good quality of life? This study draws on subjective self-report, health-related quality of life data, collected by the WHOQOL Group in 23 countries worldwide. The countries span the range of the Human Development Index which incorporates life expectancy and GDP with literacy rates. During the development of the WHOQOL-Bref (n ¼ 11,830), information about educational status was obtained by inquiring about the level of education completed by participants on a scale from none to tertiary education. Analyses of variance (multiple paired comparisons) confirmed that those with very little or no education reported the poorest quality of life. This was true not only generally, but for each of the 24 WHOQOL facets and domains. Better quality of life was reported by those who had finished primary school, and then a further increment shown for those completing secondary school education on most facets. However for many facets there was no difference in quality of life between those completing secondary and tertiary education. These findings are discussed in relation to the Theory of Human Needs (Doyal and Gough, 1991). The policy implications for health in relation to poverty and inequality, will also be addressed. # 1414/HOW HEALTH RANKS AS A DOMAIN IN QUALITY OF LIFE AMONG THOSE LIVING IN POVERTY Barbara A. Elliott, Department of Family Medicine, School of Medicine Duluth; M. K. Beattie, Duluth Public Schools, Duluth, MN, USA In a north central state in the US, a qualitative study with 731 people living below 200% federal poverty level (FPL) was done to determine their perceived Quality of Life (QoL) and needs. Each 2-h interview was conducted by trained interviewers from the same socio-economic status. Interviews began with structured questions to establish demographics and living circumstances, and continued with open-ended questions: What is positive in your life now for you and your family? and What is not going so well? Probes were specifically used to raise personal, work, community and services areas. Using careful qualitative techniques, the analysis group of researchers and community members coded the written transcripts. The sample was 71% female, 62% Caucasian, and more than half had attended some college. Twothirds had government subsidized incomes, 45% were supporting children, and 12% were homeless. Overall, health was spontaneously named as the third ranking of 20 social indicators that defined quality of life for this group, and it was generally perceived more positively than negatively. Separating the responses into groups distinguishing those from people at 100% or less FPL, 100–150%, and 150–200% revealed varying priorities among the 20 QoL domains, with the importance of health as the most heavily weighted domain by the groups with more resources. Areas within the health domain that were described as critical to QoL also varied with the amount of resources: medicines and supplies were the most difficult aspect for the 150– 200% FPL group, whereas getting access to care was the most difficult for the poorest group. These and other differences documented in this study reflect the huge impact level of poverty has on QoL. The ranking of health among the social indicators that make up perceived QoL is based in the pragmatics of survival, rather than on philosophical or other meaning-based perceptions.

747 # 1505/THE FACIT TREATMENT SATISFACTION MEASUREMENT SYSTEM David Cella, Elizabeth Hahn, Kimberly Webster, Sonya Eremenco, Lauren Lent, Stacie Hudgens, Jennifer Beaumont & Carrie Shonk, CORE, Evanston Northwestern Healthcare, Evanston, IL, USA Patient satisfaction is internationally recognized as an important outcome, especially in chronic illness treatment. The purpose of this project was to develop brief measures of satisfaction with chronic illness care, as part of the Functional Assessment of Chronic Illness Therapy (FACIT) measurement system. Multicultural input was obtained in developing item content, to ensure international applicability in treatment settings or cross-cultural research. During the item generation/scale development phase, satisfaction dimensions were rated by 30 US patients and 31 international clinicians/researchers for relevance. Respondents also suggested specific items for each dimension and provided definitions of, and barriers to, treatment satisfaction. During the instrument validation phase, 28 cancer and 30 HIV patients completed the new treatment satisfaction instrument, a FACIT QOL instrument, the Patient Satisfaction Questionnaire (PSQ-III; Hays et al., 1987), the Eastern Cooperative Oncology Group Performance Status Rating scale, and the Marlowe-Crowne Social Desirability Scale (Strahan and Gerbase, 1972). Item response theory- and classical test theory-based methods were used to evaluate psychometric properties, including reliability, unidimensionality and validity. The internal consistency reliability coefficient was 0.82 for the complete 29-item instrument, and >0.80 for most subscales. Divergent validity correlations were low (r, 0.01–0.33) and convergent validity correlations were high (r, 0.42–0.71). Based on item coverage and clinical applicability, the instrument was split into two measures: the FACITTS-G (general) and the FACIT-TS-PS (patient satisfaction with eight aspects of care). An additional instrument was later developed to evaluate chemotherapy convenience and satisfaction (FACIT-TSCCSQ). The FACIT-TS measurement system provides assessment options that are comprehensive enough to be included in clinical trials or outcomes assessment programs, yet concise enough to be combined with any quality of life assessment in an outpatient or inpatient setting. # 1366/THE DEVELOPMENT OF UTILITY WEIGHTS FOR FUNCTIONAL ASSESSMENT OF CANCER THERAPY-LUNG SCALE (FACT-L) HEALTH STATES Leida M. Lamers, Ivon Buijt & Carin A. Uyl-de Groot, Institute for Medical Technology Assessment, Erasmus University Medical Center, Rotterdam, The Netherlands The Functional Assessment of Cancer Therapy-Lung (FACT-L) Scale is a validated, sensitive, and reliable patient questionnaire that evaluates and quantifies quality of life across several dimensions, including lung cancer related symptoms. The FACT-L was not designed for use in economic evaluation of health care technologies and does not incorporate preferences into its scoring algorithm. The objective of this study is to derive a preference-based measure of health from FACT-L to be used as utilities in economic evaluation of lung cancer treatment. Factor analysis was applied to baseline FACT-L data from a clinical study with non-small lung cancer patients to determine the underlying structure and the principal items within each dimension. Six FACT-L items representing physical, social, emotional and functional well-being, general and disease-specific symptoms were selected to com-

pose an orthogonal set of eight health states. A random sample of 1600 persons from a panel representative for the Dutch population (societal perspective) were asked to value FACT-L health states by completing a questionnaire on the internet. One thousand and seventy-six respondents completed the questionnaire. Each respondent valued egiht Fact-L health states, his own health, the best and worst possible state and dead on a Visual Analogue Scale (VAS) ranging from 0 (worst imaginable health state) to 100 (best imaginable health state). Regression analysis was used to estimate utility weights for the six FACT-L items from the VAS valuations. This set of utility weights was applied on clinical data. For patients with progressive disease the utility reduced from 0.60 at baseline to 0.56 after 4 weeks. For patients with stable disease respectively partial or complete response to treatment the utility increased from 0.62 to 0.64 and 0.65. We conclude the developed utilities were sensitive for clinical changes in health state. This study showed that it is possible to derive value sets for disease specific health related quality of life instruments to use these instruments in economic evaluation # 1328/CROSS-CULTRUAL ADAPTATION AND VALIDATION OF KOREAN VERSION OF EQ-5D IN PATIENTS WITH RHEUMATIC DISEASES Myoung-Hee Kim, Preventive Medicine, Eulji University College of Medicine, Daejon; Sang-Cheol Bae, Internal Medicine, Division of Rheumatology, Hanyang University College of Medicine, Seoul, Republic of Korea Objectives: The EQ-5D is one of the most widely used instruments for health-related quality of life, however, its Korean version is not available. This study aims at translating and adapting the EQ-5D crossculturally into Korean (KEQ-5D), and evaluating its reliability and validity among patients with rheumatic diseases. Methods: Translation, back-translation, and lay assessment of the EQ-5D was done according to the EuroQol Group’s guidelines. Based on the repeated measure data of 65 patients with rheumatoid arthritis (RA), we examined test–retest reliability by intra-class correlation, and responsiveness by effect size and t-statistic. To evaluate validity, we recruited 100 patients with RA, 103 with osteoarthritis (OA), 111 with systemic lupus erythemtous (SLE), 104 with fibromyalgia syndrome (FMS), and 90 with ankylosing spondylitis (AS). For concurrent validity, we explored correlation between the KEQ-5D and KEQ-VAS, KSF-36 global, utility measures such as time-trade off (TTO) and standard gamble (SGM), and disease-specific measures. Concerning construct validity, we compared correlations among each domain of KEQ-5D, KSF-36 physical (KSF-36P), and mental (KSF-36M). Results: Intraclass correlation was 0.635, the effect size 0.683 and t-statistic 5.11 (p < 0.001). Correlations with KEQ-VAS and SF-36 global were significant (0.504 and 0.641, both p < 0.001), however those with TTO and SGM were not (0.082, p ¼ 0.155; 0026, p ¼ 0.660). These findings were replicated in the stratified analysis by diseases. Correlations with disease-specific measures were all significant except in SLE. Correlations between physical domains of KEQ-5D and KSF-36P were higher than those with KSF-36M, on the contrary, correlation between anxiety/depression and KSF-36M was higher than that with KSF-36P. Conclusion: These findings indicated KEQ-5D had stability and responsiveness, and criterion and construct validity were satisfactory. We concluded KEQ-5D could be applied to Korean patients with various rheumatic diseases.

748 # 1545/AGE AND GENDER ADJUSTMENT IN THE CALCULATION OF THE SUMMARY MEASURES OF THE SF-36 Gemma Vilagut, Health services research unit, Institut Municipal d’Investigacio´ Me`dica, Barcelona; Pablo Rebollo, Health Outcomes Research Unit, Hospital central de Asturias, Oviedo; Montse Ferrer, Angels Pont & Jordi Alonso, Health services Research Unit, Institut Municipal d’Investigacio´ Me`dica, Barcelona, Spain Physical (PCS) and mental (MCS) summary measures of the SF-36 facilitate interpretation of scores and allow for a direct comparison with the overall general population. They are calculated based on the standardized scores of the eight SF-36 dimensions using the mean and SD of the general population (conventional standardization). Given that clinical samples tend to differ in age/gender from the distribution of the general population, we assessed the impact of an alternative standardization procedure (namely, by each gender and age, 1 year) to calculate the summary scores. Data of five studies were analysed: 130 carers who cared for a deceased elderly, 710 coronary patients, 170 patients in haemodialysis treatment, 210 kidney transplanted patients, and 321 male patients with COPD. The SF-36 Spanish general population norms (n ¼ 9.151) were used to calculate the SF-36 summary scores of both methods. Compared to conventional standardization, scores obtained with the age and gender standardization in both PCS and MCS were better (higher) among women, differences ranging from 0.5 to 7.5 in PCS and from 1 to 3.5 in MCS. Conversely, among males, scores were worse (lower) for the MCS only (range from )1.2 to )2.7). PCS and MCS summary measures vary depending on whether they are calculated conventionally or using age and gender standardization. These differences may be clinically relevant among some patient populations. The conventional method allows for comparisons between different studies (as reference values are fixed). However, it could lead to misleading interpretation by overestimating the impact on physical and mental health among women and underestimating the impact on mental health among male. Age and gender standardization may provide a more realistic interpretation of the scores when evaluating specific patient samples. Funded by: FIS (G03/202) and CIRIT (2001SGR00405) # 1302/DETERMINING PREDICTORS OF MEN’S REPORT OF SEXUAL PROBLEMS: ADVANTAGES OF A TWO-PART MODEL FOR ZERO SCORES Carol M. Moinpour, Amy K. Darke & Donna K. Pauler, Division of Public Health Science/Southwest Oncology Grp Stat Ct, Fred Hutchinson Cancer Research Center/MP-557, Seattle, WA; Gary W. Donaldson, Department of Anesthesiology/Pain Research Prog, University of Utah, Salt Lake City, UT; Donald L. Patrick, Department of Health Services Res/Social & Behavioral Science, University of Washington, Seattle, WA; John E. Ware, Jr, CEO and Chief Science Officer, QualityMetric Incorporated, Lincoln, RI; Sally A. Shumaker, Department Public Health Science/Social Science & Health Policy, Wake Forest University, Winston-Salem, NC; Patricia A. Ganz, Division of Cancer Prevention & Research, Jonsson Comprehensive Cancer Center/UCLA, Los Angeles, CA; Ian M. Thompson, Jr, Division of Urology, University of Texas Health Sciences Center, San Antonio, San Antonio, TX, USA The Prostate Cancer Prevention Trial (PCPT) is a randomized, double-blind placebo-controlled study of the efficacy of finasteride in preventing prostate cancer. 18,882 men 55 years of age or older were

randomized and are to be followed for 7 years; the study is expected to end spring, 2004. Quality of life was measured at enrollment, randomization, month 6, and years 1–7. This abstract examines predictors of sexual function (as measured by the Medical Outcomes Study Sexual Problems Scale [SPS]) at randomization and at 2 years from enrollment for 8947/9459 men receiving placebo. The SPS has four items combined for a total score ranging from ‘0’ (no sexual problems) to ‘100’ (substantial problems). Covariates were baseline measures of: age; race; SF-36 Mental Health (MH), Physical Function (PF), and Vitality (VT) scales; body mass index (BMI); pack years of cigarette smoking (PCKYR); alcohol drinks per day (DPD); metabolic hours of walking (METHRS); high density lipoprotein cholesterol (HDLC); and the comorbidities diabetes (DIA) and hypertension (HYPER). For the 2-year models, we included the log of the baseline SPS score + 1 (LOG SPS + 1). A common problem with reports of sexual function is underreporting of dysfunction. At baseline, 3840 men (43%) reported zero problems; at year 2, 2674 of 7365 men (36%) reported zero sexual problems. To address this challenge, we used logistic regression (LR) to model predictors of a zero response and ordinary least squares (OLS) to model non-zero responses (Chang and Pocock, 2000). Predictors with p O 0.0004 are reported. Mean report of sexual problems at baseline was 20.01 (sd ¼ 27.05) with a median of 8.00. At 2 years, the mean score was 24.93 (sd ¼ 29.23) with a median of 17.00. Baseline SPS predictors for a zero score were age, VT, PCKYR, DIA, PF, and HYPER; for a non-zero score, age, MH, VT, PCKYR, and DIA. Two-year SPS predictors for a zero score were: LOG SPS + 1 and age; for a non-zero score, LOG SPS+1, age, PF, and VT. Sexual problems at 2 years were primarily determined by baseline sexual function, age, and physical status. # 1279/PROXY REPORTING OF HEALTH STATUS IN ELDERLY PATIENTS WITH HIP FRACTURES C. Allyson Jones & David H. Feeny, Institute of Health Economics, Edmonton, AB, Canada Background: Many patients with hip fractures have cognitive deficits and are unable to respond on their own behalf. Relying on proxy respondents is one approach to avoid excluding these patients from studies and limiting the external validity. Purpose: The purpose is to examine the agreement between patients and family caregivers at different points in time during recovery after hip fracture. Methods: Observational prospective cohort study followed 161 pairs of patients who had MMSE scores >17 and their family caregivers. Patient and proxy respondents were interviewed within 3–5 days after surgery and at 1, 3, and 6 months after surgery. Agreement was assessed for (1) an 11-point pain rating scale, (2) a performance-based disability measure, Functional Independence Measure (FIM), and (3) the Health Utilities Index Mark 2 and 3 (HUI2/3). Agreement was evaluated using intraclass correlation (ICC). Results. The mean age of patients was 80 years (sd 7). The cohort was comprised of mostly women (patient 76%; proxies 77%). ICC values were lower for less observable domains such as pain (0.56) and cognition (0.55); stronger agreement was seen for more observable domains (ambulation ¼ 0.71). Higher agreement was seen at 6 months than baseline. Factors associated with lower agreement were older age of the patient and male proxy respondents. Conclusions: Overall, there was reasonable agreement between self and proxy assessments with the primary outcomes of hip fracture: pain, physical function and overall health status. Agreement improved over time.

749 # 1642/SELF- OR PROXY REPORT IN EVALUATING HRQL IN DEMENTIA? DEVELOPMENT AND VALIDATION OF DEMQOL Sarah C. Smith & Donna L. Lamping, Health Services Research Unit, London School of Hygiene & Tropical Medicine; Sube Banerjee, Health Services Research Department, Institute of Psychiatry, King’s College London, London; Rowan H. Harwood, Department of Health Care of the Elderly, Queen’s Medical Centre, Nottingham; Beth Foley, Paul Smith, Joanna C. Cook & Joanna Murray, Health Services Research Department; Martin Prince, Section of Epidemiology, Institute of Psychiatry, King’s College London, London; Enid Levin, Social Care Institute for Excellence; Anthony Mann, Health Services Research Department, Institute of Psychiatry, King’s College London; Martin Knapp, PSSRU, London School of Economics, London, UK A particular challenge in evaluating HRQL is determining the extent to which self-report can be used with people with cognitive impairment. Although proxy reports are commonly used, these cannot substitute for a person’s own views about the inherently subjective experience of well-being. We validated and compared two new interviewer-administered measures of HRQL in dementia: DEMQOL (self-report) and DEMQOL-Proxy (carer report). Our aim during all stages of questionnaire development and validation was to keep the views of the person with dementia central and to develop a measure suitable for use across the range of severity. First, we used qualitative methods to develop a conceptual framework and preliminary questionnaires. This identified differences between patients and carers in the structure and content of their assessments of HRQL. Then, in field testing with 130 people with dementia and 126 carers, we used a systematic strategy to produce item-reduced versions of both measures (DEMQOL 73 items reduced to 28 items, DEMQOL-Proxy 73 items reduced to 31 items). Finally, in a second independent field test with 79 people with dementia and 99 carers, we evaluated acceptability, reliability and validity. Both measures showed sound and acceptable psychometric properties for people with mild/moderate dementia (MMSE P 10). In people with severe dementia (MMSE < 10), DEMQOL (self-report) is not suitable because of unacceptable levels of missing data, but DEMQOL-Proxy shows promise across the range of dementia severity. Results of both qualitative and psychometric stages indicate that the two measures give complementary but equally valid perspectives on HRQL in dementia, suggesting that the combined use of self- and proxy-reports provides the most comprehensive evaluation of outcome. We critically evaluate the wider implications of these results in terms of: (1) the potential usefulness of self-report with other cognitively impaired respondents; (2) the future role of proxy report in evaluating HRQL; and (3) challenges in the dual use of self- and proxy report. # 1806/PERCEPTION OF QUALITY OF LIFE OF CHRONICALLY ILL CHILDREN: PERSPECTIVE OF PATIENTS, PARENTS AND PHYSICIANS Arieke J. Janse, Wilhelmina Children’s Hospital, University Medical Centre Utrecht, Utrecht; Reinoud J. Gemke, Department of Pediatrics, VU University Medical Centre, Amsterdam; Cuno S. Uiterwaal, Julius Center for Health Sciences and Primary Care; Jan L. Kimpen, Wilhelmina Children’s Hospital; Gerben Sinnema, Department of Pediatric Psychology, Wilhelmina Child, University Medical Centre Utrecht, Utrecht, The Netherlands The care for children with a chronic disease is complex, multifaceted, and extends over a long period of time. The child’s well-being often changes as a function of the child’s condition and the treatment of the disease. Differences in perception of quality of life between patients, parents, and physicians are clinically important as these may have implications for compliance and treatment. A longitudinal study, conducted between July 1999 and January 2002 in four tertiary pediatric care centers in the Netherlands, was performed to investigate the differences in perception of quality of life between children, parents and physicians at diagnosis and after a period of follow-up. Twentythree physicians and 100 parents and patients (aged 10–17 years) diagnosed with acute lymphatic leukemia, asthma, juvenile idiopathic arthritis, cystic fibrosis admitted in day-care or cystic fibrosis admitted for a pneumonia participated in the study. The main outcome measure was health related quality of life assessed by children, parents and physicians using the Health Utilities Index mark 3. As to health and well-being perceptions, differences were evident between physicians,

parents and patients, not only at diagnosis but also after a period of follow-up. Perceptive differences were particularly clear in the subjective attributes emotion and pain. The agreement for the attribute pain was 8%. Physicians assessed the patients to have less pain than the patients and parents did. The differences at follow-up did not change essentially compared to the differences at baseline. At the onset of a serious chronic disease and after a period of follow-up pediatric patients may be misunderstood by health care professionals especially in the subjective but for the patient clinically relevant attributes. # 1260/DISCREPANCIES IN LUNG CANCER PATIENTS AND NURSES ASSESSMENT OF SYMPTOM DISTRESS AND INTENSITY Eva Broberger Wiberg & Carol Tishelman, Department of Nursing, Karolinska Institutet, Huddinge; Louise von Essen, Department of Public Health and Caring Sciences, Uppsala University, Uppsala, Sweden People with lung cancer experience high levels of symptom distress compared to other cancer patients, making symptom management particularly important. Previous research concerning patient-staff perceptions in cancer care report that staff perceive patients to have more distress, especially from psychological symptoms, than what is reported by patients themselves. Such results are generally based on assessments of symptom intensity. The aim of this study was to explore the hypothesis that there may be greater discrepancies between how patients and nurses rate intensity of symptoms than how they rate distress from symptoms. Thirty-three patient-nurse dyads rated patient’s symptom distress using two structured questionnaires; Symptom Distress Scale (SDS): VAS-version measuring intensity of 13 symptoms and a Thurstone scale presenting nine symptoms in every possible subset of two, from which the patient and nurse select the symptom they view as most distressing, irrespective of intensity, for the patient at the time of the interview. Preliminary results on group level show that patients and nurses rated breathing, pain, fatigue and cough as most distressing. Both patients and nurses perceived fatigue and outlook as the most intense symptoms although nurses rated their intensity significantly higher than patients themselves. This support the hypothesis that there are greater discrepancies in how patients and nurses rank intensity of symptoms, compared with the rankings of distress from symptoms. In addition patients and nurses were found to be able to differentiate between the distress from symptoms and their intensity shown by different rankings of distress (Thurstone scale) and intensity (SDS) of symptoms. # 1851/A NOVEL APPROACH TO ASSESSING AGREEMENT BETWEEN RATERS OF MULTI-DIMENSIONAL MEASURES AT THE INDIVIDUAL LEVEL USING FUZZY SET THEORY A. Simon Pickard, College of Pharmacy, University of Illinois, Chicago, IL, USA Background: Analysis and interpretation of agreement between patient and proxy assessments of health-related quality of life is accomplished at the group level using statistics of agreement such as the ANOVAbased intraclass correlation co-efficient (ICC) and Kappa. Objective: To introduce an approach to assessing inter-rater reliability at the individual level. Methods: A measure of causality sensitive to initial condition in the individual patient proposed by Helgason and Jobe was modified and applied to patient-proxy agreement. Stroke patient and proxy assessments of the Health Utilities Index Mark 3 (HUI3) were represented by health state vectors consisting of the single attribute utility scores for eight attributes of the HUI3. For each respondent dyad A and B, the vector size (norm) was computed by summing the scores of the attributes. Based on the geometry of fuzzy logic developed by Kosko, the vector was mathematically interpreted as a fuzzy set and the norm as equivalent to a fuzzy cardinality measure. Results: A fuzzy interconnectivity co-efficient (FIC) was calculated for each individual patient-proxy dyad representing the extent of connectiveness (agreement) between the raters across the domains of HRQL. Discussion: The application of fuzzy set theory to evaluate inter-rater reliability represents a novel departure from present methods of assessing reliability between raters. Interpretation of individual-level FICs can be facilitated using summary statistics. Future research should address limitations of the FIC and exploration of interpretative benchmarks analogous to those for other statistics of agreement.

750 # 1630/COMPUTER ADAPTIVE TESTS AND HANDHELD COMPUTERS FOR PSYCHOLOGIAL ASSESSMENTS Matthias Rose, Psychosomatics, Charite – Humboldt University; Otto Walter, Herbert Fliege, Janine Becker & Rueha Kocalevent, Psychosomatics, Charite, Berlin, Germany; Jakob B. Bjorner, Occupational Health, University of Copenhagen, Copenhagen, Denmark; Burghard F. Klapp, Psychosomatics, Charite, Berlin, Germany

vided QL information in a format compatible with administrative health databases.

Introduction: Computer Adaptive Tests (CAT) offer several psychometric advantages, but the (1) development of the underlying item pool is expensive and (2) applications of CAT’s using a PC platform have some practical limitations. We investigated if the use of personal digital assistants (PDA) can help to overcome both problems. Methods: Since 1990 all patients of the Department of Psychosomatic Medicine at the Charite´ answer a set of standardized paper-pencil questionnaires (e.g. STAI, BDI) within the diagnostic routines. In 1995 we developed a computer assisted assessment on a PDA platform and 2003 CAT’s for depression and anxiety were integrated into this routine. These three different approaches to measure aspects of ‘mood’ were analysed for a total of 9659 patients in regard to economic and psychometric factors using standard techniques and structural equation models with multi group designs (AMOS). Results: The paperpencil and the PDA versions showed almost the same internal structure and similar psychometric characteristics. Using the PDA saved 2/ 3 of the total cost for the assessment. It showed no missing data and enabled us to develop the item pool for two CAT’s. The integration of the CAT’s on the PDA platform was possible but had no further economic advantages. However, it reduced the patient burden as both constructs can be measured with high precision (SE < 0.32) using only approx. 7 items, which is 1/3 of conventional instruments. Conclusion: The use of PDA’s for psychometric assessment provides (1) an economic way to create an item pool for the development of CAT’s, and (2) offers the chance to use CAT’s in research designs too, which demand mobile and inexpensive data collection.

The implementation of quality of life measures in the clinical realm has been hampered by the organisational difficulties in distributing and retrieving forms. While paper and pencil tests have been widly used, alternative forms of assessment such as computerised versions have just begun to be included. For children and adolescents, computer based testing lends itself to application because of the many advantages such an assessment has within this population. The presentation describes the development and implementation of computer based assessment system examine to quality of life in children with or without chronic health conditions within the Disabkids- and Kidscreen group. It is a measurement system which includes a generic part to be applied to children independent of their actual health condition as well as a chronic-generic part and a condition-specific part focussing on the children’s chronic health condition. Different age versions are available. In the presentation, the compositon of these versions as well as the computerised scoring for individual applications and patient groups will be presented. Further steps, such as computer adaptive testing systems will be considered, once the data base available from the Disabkids and Kidscreen projects is large enough to begin such development. Management, maintenance and copyright issues surrounding the computer versions will be discussed. Note: Co-authors are all Kidscreen and Disabkids members.

# 1536/WHAT WILL HAPPEN TO QL MEASURES WITH THE INTRODUCTION OF ELECTRONIC HEALTH RECORDS? ICF TO THE RESCUE Nancy E. Mayo, Clinical Epidemiology; Lise Poissant, Clinical and Health Informatics, McGill University Health Center, Montreal, Quebec, Canada Purpose: E-records provide an accessible medium for capturing patient relevant outcomes at point of contact with the health care system facilitating clinical, population and health services research. QL measures are likely to be included in an e-record as physical, emotional and psychological health are captured, parsimoniously. However, there are many QL measures and the summary scores not readily interpretable. The International Classification of Functioning (ICF) has the potential to be used to provide standard coding for items on QL measures making QL information available to administrative databases. This study determined feasibility of coding the SF-12 using the ICF and assessed the value of information gathered from this process. Methods: The data for this study came from interviews with community-dwelling persons with stroke (n ¼ 604) and age- and gendermatched individuals without stroke (n ¼ 488). Seven rehabilitation professionals independently ICF coded the 12 items. Results: The general health item was not codeable, otherwise, one code per item predominated. Persons with stroke scored 10 points lower on Physical Health than controls (41 vs. 51) and women scored 3 points lower than men. There was a linear trend for increasing Physical and Mental Health with increasing income among controls. Analyses after ICF recoding revealed that persons with stroke had more severe, not more prevalent, impairments of pain, emotional stability, and energy. Activity limitations (stair climbing and housework) and restrictions on community life were both more common and more severe among persons with stroke and restrictions on work were three times more common (60 vs. 20%). Across income, there was a similarly strong gradient for controls and for persons with stroke for decreasing disability with increasing income for pain, stair climbing, housework and community life. Lower income persons with stroke were less likely to report work restrictions than wealthier persons. Conclusion: The ICF coding pro-

# 3001/MAKING A QUALITY OF LIFE ASSESSMENT EASILY USABLE: THE COMPUTERISED VERSION APPROACH Henrik M. Koopman, Department of Paediatrics, Leiden University Medical Centre, NL, The Netherlands

# 1233/ADVANCING THE METHODS OF CHILD SELF-REPORT IN QUALITY OF LIFE RESEARCH Catharine S. Bradley, Graduate Department of Rehabilitation Science, University of Toronto, Toronto, Ontario, Canada; Maru Barrera, Department of Psychology; Brian Feldman, Division of Rheumatology; Victor Blanchette, Division of Hematology/Oncology; Nancy L. Young, Population Health Sciences, The Hospital for Sick Children, Toronto, Ontario, Canada The purpose of this study was to examine and evaluate the state of the art in assessing children’s Quality of Life (QoL) and health status (HS) by self-report (SR). In adults, SR is considered to be one of the best indicators of these constructs. However, one concern with children is that the way questions are perceived may vary with age and compromise validity. While several measures of QoL and HS had been developed to enable SR by children across ages, the strategies used in generating these measures had not yet been synthesized and their effectiveness only minimally assessed. An extensive literature review identified that there are eight commonly used child-focused strategies to adapt measures so that they can be completed by children (e.g., specific response options, avoiding attribution and negative wording). This study examined the prevalence of their application in existing English language child SR measures of QoL and HS. Of the 33 measures reviewed, 8 (4 disease-specific and 4 generic) reported the use of 5 or more of these child-focused strategies. The Activities Scale for Kids (ASK) is among these eight measures, having used seven of the eight strategies. ASK summary scores from 705 children, ages 5– 15 years, were then analysed in relation to respondent age to determine whether the child-focused methods had achieved the goal of accommodating to children’s abilities. This testing demonstrated a weak correlation of r ¼ 0.28 (p < 0.001) between age and summary scores in children with disabilities and a moderate correlation r ¼ 0.50 (p < 0.001) among healthy children. In summary, this study identified eight strategies that may enhance children’s abilities to SR. These strategies have been used to varying degrees in existing measures of QoL and HS. However, despite the ASK having several child appropriate considerations, age was found to be related to ASK scores. This finding suggests that SR in children remains on the cutting edge of child QoL and HS measurement. Future research will investigate if these ‘age effects’ are unique to ASK data and to what extent they are a product of cognition, adaptation or other factors.

751 # 1021/EFFECT OF ORDER OF ADMINISTRATION OF HEALTHRELATED QUALITY OF LIFE INTERVIEW INSTRUMENTS ON RESPONSES Ashley L. Childs & Marta J. Marsh, Department of Ophthalmology, Johns Hopkins University, Baltimore, MD, USA The objective is to determine the effect on patient responses from the order in which general health-related and vision-targeted instruments were administered in a set of randomized trials of intraocular surgery. Patients enrolled in the Submacular Surgery Trials (SST) completed quality of life interviews prior to random assignment to surgery or observation. Interviews were conducted via telephone using a computer-assisted telephone interview system that randomized the order of instrument administration. Either the general health-related instruments were administered first (SF-36 Health Survey (SF-36) and Hospital Anxiety and Depression Scale (HADS)), followed by the vision-targeted instruments (National Eye Institute-Vision Function Questionnaire (NEI-VFQ) and SST-Vision Preference Value Scale), or the vision-targeted questions were asked first, followed by the general health instruments. There were 992 of 1015 patients in the SST who had all four instruments administered in random order: 483 (49%) patients responded to the SF-36 first and 509 (51%) patients responded to the NEI-VFQ first. When the SF-36 was administered first, median scores were higher (indicating better health status) for 8 of the 25 subscales across the 4 instruments, and lower (indicating poorer health status) for 3 subscales. Order of administration produced significantly different scores for three health status subscales: SF-36 mental health (MHI-5), HADS depression and HADS anxiety (p O 0.05, Wilcoxon rank sum test). Overall, the order of administration did not have an effect on responses to the vast majority of subscales in baseline interviews; further the order effect was small when compared to clinically relevant differences in the subscales. However, when the general health instruments followed the vision-targeted instrument the mental health subscales were affected the most, therefore it is important to hold order consistent throughout the trial to minimize bias. # 1530/SENSE OF COHERENCE AND CHRONIC ILLNESS: CAUSES AND CONSEQUENCES Marijke Veenstra, Foundation for Health Services Research, Nordbyhagen; Torbjørn Moum, Department of Medical Behaviour, University of Oslo, Oslo, Norway The concept of sense of coherence (SOC) has attracted a lot of research efforts in health and well-being research. Yet, the search for empirical evidence for the health preserving influence of a high SOC is characterised by the use of a variety of health outcomes, referring to different levels of functioning and used interchangeably and inconsistently. In addition, most designs are cross-sectional and little attention is paid to the possibility of spurious results. The aim of this study is to gain insight in the direction of the causal association between SOC, health and well-being for people with a chronic illness using frameworks from quality of life research and classifications of functioning, disability and health. Our study uses a panel design including 2567 patients with COPD or asthma, diabetes or chronic heart diseases discharged from 17 general hospitals in Norway. Scores on SOC-13 and SF-36 were gathered for these patients in 1998 and 2000. Data were analysed using hierarchical linear regression analysis and structural equations modelling (SEM) using cross-lagged models. The results showed that the strength of the association between SOC and health depends to a large extent on which measure of health is used as the outcome variable. Regression analyses indicated little support for the mediating role of SOC in the association between functional limitations and disability and between disability and wellbeing. However, as the SEM indicated the presence of synchronous effects, causal assumptions related to treating SOC as an outcome were not supported by the data. Future research should be more critical towards the salutogenic influence of SOC. The results in this study indicated that the SOC construct contains elements that are similar to those in life satisfaction and subjective well-being. The presence of feedback loops and synchronous effects imply that a high SOC could be considered a desired outcome involving careful main-

tenance and substitution of resources, as well as a fairly stable disposition with salutogenic qualities. # 1086/THE USE OF GENERAL (SF-36), AND DISEASE SPECIFIC HRQOL INSTRUMENTS IN THE OUTCOMES ASSESSMENT OF NEUROSURGICAL TREATMENTS OF HERNIATED LUMBAR DISC. Michael A. Hagan, Peter D. Angevine & Paul C. McCormick, Neurosurgery, Columbia University, New York, NY, USA HRQoL of patients with a clinical diagnosis of Herniated Lumbar Disc (HLD) was assessed using the SF-36 as well as Neurological Deficit (ND) and Disability Due to Pain (DDP) measures developed specifically for this study. A cohort of 628 patients who received surgical treatment for HLD was followed prospectively for 1 year; with assessments made at baseline, months 3 and 12 post-op. Changes from baseline to month 3 are reported here. Pattern mixture modeling and Heckman 2 stage regression seem to indicate data is Missing At Random. On all scales, patients reported very large increases in HRQoL. Changes reported on SF-36 from baseline: Physical Functioning 36.88 (27.22), Role-Physical 31.52 (37.43), Bodily Pain 38.71 (24.44), Vitality 21.19 (23.40), Social Functioning 32.52 (29.06), RoleEmotional 10.09 (28.23), Mental Health Index 13.90 (19.75), all having p-values of less than 0.0001. ND and DDP showed changes of 7.04 (5.03) and 16.05 (10.80), respectively with p-values of less than 0.0001. Data from this study supports surgery as an excellent treatment for improving the HRQoL of HLD patients as demonstrated by effect sizes of 1.4–0.7 for the SF-36 scales and 1.4–1.5 for the ND and DDP scales, respectively. Regression analyses show a high coefficient of determination (R2 approx ¼ 0.4–0.6) indicating that SF-36, ND and DDP track similar improvements in HRQoL # 1615/COMPARISON OF THE SF-36 AND WHOQOL-BREF IN SUBJECTS WITH CHRONIC LUNG DISEASES Wen-Miin Liang, Department of Public Health, China Medical College, Taichung; Haw-Yaw Shy, Department of Mathematics, National Chunghua University of Education, Chunghua; Li-Jin Huang, Department of Nursing, China Medical College, Taichung, Taiwan The Short Form 36-item (SF-36) and the WHOQOL-BREF questionnaires are commonly used measurements of generic health-related quality of life. The purpose of this study was to compare the psychometric properties of the two questionnaires and to assess their ability to discriminate among different levels of severity groups. A total of 477 (274 males and 203 females) samples aged from 15.2 to 74.7 years were selected by the question asking subjects if they had diagnosed with a chronic lung disease by a physician from a representative population survey (20,855 samples in total) by the National Health Promotion Bureau of Taiwan in 2001. All sample subjects were classified into two groups based on the status of whether or not samples had recently taken medication. The WHOQOL-BREF questionnaire included four domains; physical, psychological, social, and environmental domains. The SF-36 has eight scales; PF, RF, BP, GH, VT, SF, RE and MH. Score distribution, reliability estimates, discriminating ability and item scaling tests, including Rasch analysis were compared. The WHOQOL-BREF domain scores were less skewed and more homogeneously distributed than SF-36. The range of Cronbach’s a in SF-36 was 0.78–0.93, and in WHOQOL-BREF 0.74–0.83. Convergent and discriminant validity of SF-36 and WHOQOL-BREF were examined by the Multitrait–Multimethod (MTMM) matrix, ex. correlation between PH in SF-36 with physical domain in WHOQOLBREF were substantial in magnitude (r ¼ 0.51), while the correlation of PH with social domain was trivial (r ¼ 0.16). Rasch analysis results, such as difficulty calibrations in logits, standard error of each calibration and goodness-of-fit statistics were compared for comparable scales between SF-36 and WHOQOL-BREF. Model fit for IRT model was assessed. The receiving operating characteristic (ROC) curves are presented for comparable scales to discriminate between two levels of severity groups. Our findings suggest that the SF-36 had better internal consistency and scale constructs, while WHOQOLBREF had more homogeneous distribution and provided more power in discriminating among different levels of chronic lung diseases.

752 # 1350/QUALITY OF LIFE FOLLOWING PREOPERATIVE RADIOTHERAPY COMBINED WITH TOTAL MESORECTAL EXCISION FOR RESECTABLE RECTAL CANCER Anne M. Stiggelbout, Medical Decision Making; Corrie A. Marijnen, Department of Clinical Oncology; H Putter, Department of Medical Statistics; Mandy van den Brink, Department of Medical Decision Making; Elma Meershoek-Klein Kranenbarg & Cock J. van de Velde, Department of Surgery, Leiden University Medical Center, Leiden, The Netherlands Short-term preoperative radiotherapy (pRT) and total mesorectal excision (TME) have each been shown to improve local control of disease in patients with resectable rectal cancer. Our multicenter randomized trial showed that the addition of pRT (5 · 5 Gy) reduces the local recurrence rate at 2 years, from 8.2% in the TME-only group to 2.4% in the TME + pRT group (Kapiteijn, NEJM 2001). Results on the quality-adjusted survival have been presented elsewhere. In this paper we present results on the descriptive quality of life data of patients in the trial. Health-related quality of life was assessed using the Rotterdam Symptom Check List (RSCL) at baseline, and at 3, 6, 12, 18 and 24 months after surgery. In the format of the RSCL, three items were added regarding voiding (a 0.76) and seven regarding defecation (a 0.87). Profiles of missing data were graphed; pattern mixture models were fitted to assess the impact of pRT in case of data not missing at random (NMAR). Of the 1530 Dutch patients randomized, 1302 turned out to be eligible (85%). Over the 24 months, few differences were seen between the two arms. Only at 3 months after surgery some differences were seen in favor of the TME-only arm. These patients reported less problems with defecation and with usual activities. It seems that the benefit of pRT comes at few costs regarding quality of life. Results will be presented regarding the patterns of missing data and the robustness of the findings. # 1170/INFERRING QUALITY OF LIFE FROM PATTERNS OF CARE AT THE END OF LIFE FOR WOMEN DYING OF BREAST CANCER Bruno Gagnon & Nancy Mayo, Division of Clinical Epidemiology, McGill University Health Centre, Montreal, Quebec, Canada Purpose: Research on selected terminally ill persons with cancer has shown that Palliative Care has a positive influence on QL. However, it is impossible to measure QL at the level of the population of terminally ill people. But if the care received at end of life is reflective of a palliative care philosophy, a better QL could be inferred. Under such philosophy of care, services should be delivered at home, diminish the need for crisis intervention, and demonstrate a fading away from a curative intent. The objective of this study was to make the inference about QL from the pattern of care received by a population of women with breast cancer at end of life. Methods: The subjects for this study were 2291 women dying of breast cancer in the province of Quebec during the years 1992–1998. Using administrative databases, indices reflective of a palliative care oriented approach during the last 6 months of life were created. Results: Only 7% of women died at home while 70% died in an acute care hospital bed. More than 1/4 of women spent at least 1 month in hospital prior to death and more than 1/2 had multiple admissions. Pattern of care was also characterized by multiple emergency room visits, multiple oncology outpatient visits,

and ongoing chemotherapy but few physician home visits. Few women (25%) had indices indicating any provision of palliative care oriented services during the last 6 months of life. In particular, younger women (<50 years) were less likely (odds ratio: 0.70 [95% CI: 0.54, 0.90]) to access such services compared to their more senior peers. This study suggests that the health services received by these women at the end of life were not conducive to good QL. Concrete health care policies need to be developed to allow a greater number of the terminally ill to have access to care keeping with a palliative care philosophy so that QL at end of life may be improved. Research programs focusing on the ingredients of care and QL are needed. # 1602/RESPONSIVENESS OF GENERIC AND DISEASE-SPECIFIC QUESTIONNAIRES – DISTRIBUTION AND ANCHO-BASED METHODS A.R. Maille´ & C.N.M. Brouwer, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht; R.H. Veenhoven, Department of Pediatrics, Spaarne Hospital Haarlem, Haarlem; E.A.M. Sanders, Department of Pediatric Immunology; D.E. Grobbee, M.M. Rovers, H.F. v. Stel, Julius Center for Health Sciences and Primary Care; A.G.M. Schilder, Department of Otorhinolaryngology, University Medical Center Utrecht, Utrecht, Netherlands Meaningful evaluation of treatment effects on health-related quality of life (HRQoL) and functional health status (FHS) requires instruments that are responsive. Recently developed guidelines recommend use of multiple strategies as well as head to head comparison between generic and disease-specific instruments in the assessment of responsiveness. Our purpose was to assess responsiveness as part of the validation of instruments assessing HRQoL and FHS in children with recurrent acute otitis media (rAOM). Generic FHS, disease-specific HRQoL, and disease-specific FHS were assessed in 383 children aged 1–7 years with rAOM participating in a trial on the effectiveness of pneumococcal conjugate vaccination. Assessment of responsiveness involved two steps: (1) Assessment of sensitivity to change by a paired t-test and calculation of effect sizes; (2) Assigning meaning to change by applying distribution-based (MCID based on a change of 1SEM or 0.3 ES) and anchor-based (using AOM frequency and AOM severity as clinical anchors) methods to estimate minimally clinical important change (MCID). All methods were implemented over two intervals: 0–7 months and 7–14 months follow-up. Results were: (1) Sensitivity to change: change-scores were significant (p < 0.003) for generic and disease-specific instruments. Effect sizes were somewhat higher for disease-specific compared to generic instruments (0.55– 0.95 vs. 0.37–0.60). (2) Interpretation of change: distribution-based methods yielded similar estimates of MCID for generic and diseasespecific instruments: 5–9 points on a 0–100 scale. Anchor-based methods resulted in a larger range of estimates of MCID: 2–17 points on a 0–100 scale for AOM frequency and 3–29 points for AOM severity. Combining distribution-based and anchor-based methods resulted in similar ranges for the MCID for generic (2–12 points on a 0– 100 scale) and disease specific instruments (3–15 points, excluding the NRS Child). In conclusion, both generic and disease specific instruments used in this study show adequate responsiveness to be used in clinical studies on children with rAOM.

753 # 1487/RESPONSIVENESS OF THE SCHIZOPHRENIA QUALITY OF LIFE SCALE (SQLS) AND THE SHORT FORM 36 HEALTH SURVEY (SF-36) IN PATIENTS WITH SCHIZOPHENIA: A COMPARISON STUDY Nan Luo & Shu-Chuen Li, Department of Pharmacy, National University of Singapore; Boon-Kheng Seng, Department of Medical Social Work, Institute of Mental Health; Julian Thumboo, Department of Rheumatology and Immunology, Singapore General Hospital, Singapore We compared the responsiveness of the disease-specific Schizophrenia Quality of Life Scale (SQLS) and the generic Short Form 36 Health Survey (SF-36) in patients with schizophrenia. Outpatients with schizophrenia attending a tertiary-level mental health hospital in Singapore completed an identical English or Chinese questionnaire containing the SQLS and SF-36 twice during two separate routine visits. Subjects also rated changes in their global health status between the two visits. Responsiveness was assessed using Cohen’s effect size, standardized response mean and Guyatt’s responsiveness index. Small, moderate and large magnitudes of responsiveness were defined using Cohen’s criteria (i.e. 0.20, 0.50 and 0.80, respectively). Comparative responsiveness was assessed using relative efficiency. Questionnaires were completed by 104 subjects (English-speaking: 70). English versions demonstrated responsiveness to improved health (n ¼ 31) for 10 of 13 scales (values of indices: 0.25–0.59) except for the energy/motivation (EM), symptoms/side effects (SS) and bodily pain (BP) scales (values of indices: 0.10–0.24). Responsiveness to worsened health (n ¼ 8) was demonstrated for 6 of 13 scales (i.e. the SS, BP, vitality (VT), social functioning (SF), mental health (MH) and MH summary scales, values of indices: 0.29–0.88). Chinese versions demonstrated responsiveness to improved health (n ¼ 22) for 7 of 13 scales (i.e. the psychosocial, SS, role-physical, general health (GH), SF, role-emotional (RE) and physical health summary scales, values of indices: 0.23–0.68) and responsiveness to worsened health (n ¼ 5) for the SF and RE scales (values of indices: 0.38–2.68). The most responsive scales to improved and worsened health were the SF-36 VT and SF scales, respectively for English versions and the SF36 GH and SF scales, respectively for Chinese versions. Both the SQLS and SF-36 are reasonably responsive measures for patients with schizophrenia. Surprisingly, we found that a generic measure appeared to be more responsive than a disease-specific measure. # 1649/VALIDITY AND SENSITIVITY TO CHANGE OF THE RUSSIAN VERSION OF AIMS2-SF IN RHEUMATOID ARTHRITIS (RA) PATIENTS Anna A. Tsepkova, Andrei A. Novik, Tatyana I. Ionova, Tatyana G. Shemerovskaya & Anton V. Kishtovich, Multinational Center of Quality of Life Research, St.-Petersburg, Russia Previously we reported that Russian version of AIMS2-SF with emphasis on cultural relevance for RA patients in Russia was developed; satisfactory results of reliability and factor structure testing were shown. The aim of present study was to assess validity and sensitivity. One hundred and thirty three RA patients (mean age 56.4) were enrolled in the study at acute disease phase followed by the treatment. Known-groups validity was estimated based on assumptions that QoL scales depend on age, gender, clinical and laboratory tests. We assessed scale correlations with external criteria: SF-36 scales, pain, fatigue, laboratory tests and morning stiffness. Sensitivity was studied at entry and 2 months followup. We used verbal rating scale (VRS) for

self-reported disease activity as independent changes indicator. Results showed that physical function, symptom, affect and work scales were worse in female group (affect and symptoms: p < 0.05) as compared with male. QoL scores in 10-years interval groups (range 20–80 years) decreased with aging on physical function (p < 0.01) and symptom scales. The same scales were worse in seropositive patients, in higher functional class and disease activity (physical function: p < 0.01). Correlations were revealed between: (1) physical functions scale and pain, fatigue, morning stiffness, ESR, C-reactive protein (0.33–0.47, p < 0.01); (2) the same criteria and symptom scale (0.22–0.37, p < 0.01), (3) affect scale and pain (0.52, p < 0.01), fatigue (0.44, p < 0.01), morning stiffness (0.17, p < .05). Those scales of AIMS2-SF and SF-36 which measure the same concept correlated well, e.g., social interactions and social functioning ()0.35, p < 0.01). Results on sensitivity were obtained from 53 patients who formed three groups on VRS changes: 37 patients improved, 11 patients with no changes, 5 patients deteriorated. Four of five AIMS2-SF scales improved in cohort of 37 patients: physical function, symptom, work, affect (p < 0.01 except for work). Patients who did not report improvement showed no significant score differences. In conclusion, Russian AIMS2-SF is reliable, valid and sensitive instrument for QoL assessment in RA patients. # 1662/RESPONSIVENESS OF THE SEIQOL-DW IN PATIENTS UNDERGOING STEM CELL TRANSPLANT Lena Wettergren, Department of Nursing, Karolinska Institutet, Huddinge; Magnus Bjo¨rkholm, Department of Medicine, Karolinska Institutet and Karolinska Hospital, Stockholm; Ann Langius-Eklo¨f, Department of Nursing, Karolinska Institutet, Huddinge, Sweden The purpose of the study was to investigate the responsiveness of the Schedule for the Evaluation of the Individual Quality of Life-Direct Weighting (SEIQoL-DW) when measuring quality of life (QoL) over time in recipients undergoing stem cell transplantation (SCT). The study also sought to investigate the convergent validity of the SEIQoLDW by examining the relation between the index scores of the SEIQoL-DW to the SIP (total score) and EORTC QLQ-C30 overall QoL subscale. During October 1998 to December 2001 was a prospective study conducted in patients with malignant blood disorders undergoing high-dose chemotherapy followed by SCT at Karolinska and Huddinge Hospital in Sweden. Patients were evaluated the week before start of high-dose treatment (T1, n ¼ 35), at 6 months (T2, n ¼ 23) and 12 months (T3, n ¼ 24) after SCT. The participants were interviewed with the Swedish extended version of the SEIQoL-DW including a generic and a disease-specific part and filled out SIP and EORTC QLQ-C30 (version 2.0). The SEIQoL-DW is based on semi-structured interviews where the participants are requested to freely nominate areas, both negative and positive, important for their quality of life. Satisfaction with the nominated areas is rated on a category scale and importance is weighted using a colored disk. The rating and the weight for each domain are multiplied and the products aggregated to produce an generic QoL index score and a disease-specific index score, respectively. Preliminary results indicate that the generic and diseasespecific SEIQoL-DW reflect different aspects of life. The former is interpreted as overall QoL and the latter as health-related QoL. A not negligible amount of new domains were nominated at T3 which were not mentioned at T1. The reasons for change of domains and their relation to the scores of SIP and the EORTC QLQ-C30 will be presented and discussed.

754 # 1745/EVALUATION OF THE RESPONSIVENESS OF THE DERMATOLOGY LIFE QUALITY INDEX, AS COMPARED TO THE SF-36, IN PATIENTS WITH MODERATE TO SEVERE PSORIASIS J. Michael Woolley & Deepa Lalla, Global Health Economics, Amgen Inc., Thousand Oaks, CA, USA The objective of this study is to evaluate the validity of the Dermatology Life Quality Index (DLQI), and to compare its responsiveness to that of the SF-36 in patients with moderate to severe psoriasis. In a double-blind, multi-center clinical trial, 118 subjects with moderate to severe psoriasis were randomized to receive therapy with either etanercept or placebo. HRQOL was assessed at baseline and after 12 weeks of therapy using the Dermatology Life Quality Index (DLQI) and the SF-36. The DLQI consists of 10 questions measuring six subscales: symptoms and feelings, daily activities, leisure, work and school, personal relationships, and treatment. All patients who received at least one dose of study drug (n ¼ 112) were included in the analysis and missing data were imputed using the last observation carried forward. Responsiveness was assessed using the effect size statistic (ES), the paired t-statistic (PTS), the standardized response mean (SRM), and the responsiveness statistic (RS). Clinical status was measured using the Psoriasis Area and Severity Index (PASI) score. Results from the analyses show that Cronbach’s a for the DLQI was 0.89 at baseline, and the correlation between the DLQI and the PASI was 0.430. Over 12 weeks, the correlation between the percent change in the DLQI and the percent change in the PASI score was 0.504. As shown in the table, the effect size and other responsiveness measures indicate that the DLQI was very sensitive to changes in clinical status. Furthermore, by every measure in the table, the DLQI was much more responsive than the SF-36. These results support the hypothesis that the Dermatology Life Quality Index is a valid measure of health-related quality of life in patients with moderate to severe psoriasis. In addition, it shows excellent responsiveness to change in clinical status. The Dermatology Life Quality Index is likely to be a useful measure for assessing patient-reported outcomes in clinical trials in psoriasis. # 1069/THE SHORT FORM HEALTH SURVEY (SF-36): TRANSLATION AND VALIDATION STUDY OF THE IRANIAN VERSION Ali Montazeri, Mariam Vahdaninia & Azita Goshtasebi, Iranian Institute for Health Sciences Research, Tehran, Iran This was a large population-based study to develop the Iranian version of the SF-36 for use in health related quality of life assessments in Iran. After successful translation of the English version of the SF-36 into Persian (the Iranian language), a culturally comparable questionnaire was developed and pilot tested. Then, the Iranian version of the SF-36 was administered to a random sample of 4163 healthy individuals aged 15 years and over in Tehran. The mean age of the respondents was 35.1 (sd ¼ 16.0) years, 52% were female, mostly married (58%), and the mean years of their fromal education was 10.0 (sd ¼ 4.5). Reliability was estimated using the internal consistency (Cronbach’s a) and validity was assessed using convergent and known groups comparison analysis. The analysis showed that all eight SF-36 subscales met the minimum reliability standard, Cronbach’s alpha coefficients ranging from 0.77 to 0.90 with the exception of the vitality (a ¼ 0.65). Convergent validity was assessed using the correlation of each item with its hypothesized scale and all showed satisfactory results (all correlation above 0.40, ranging from 0.57 to 0.95). Known groups comparison showed that in all dimensions the SF-36 discriminated between men and women as expected (women had lower score than men and all p values less than 0.05). In general, the Iranian version of the SF-36 performed well and the findings suggest that it is a reliable and valid measure of health related quality of life among general population. In deed the results from this cross-cultural study contributes to the existing knowledge on the international development of the SF-36 worldwide.

# 1286/EVALUATION OF SPIRITUAL WELL-BEING IN RELATION TO PSYCHOLOGICAL DISTRESS, SOCIODEMOGRAPHIC AND CLINICAL VARIABLES IN CANCER PATIENTS IN URUGUAY Juan J. Dapueto & Carla Francolino, Medical Psychology; Liliana Servente, Biophysics and Statistics, Facultad de Medicina, Universidad de la Repu´blica, Montevideo, Uruguay The association between spiritual well-being (SWB) and health status has elicited considerable attention from researchers. Most of the studies have been carried out on samples of patients reporting high rates of religious affiliation and religious practice. This study aimed to evaluate the FACIT Spiritual Well-Being Subscale (FACIT-Sp) in a traditionally lay South-American culture where agnosticism is deeply ingrained in the social scheme. Data were collected using the FACITSp from 361 cancer patients with various tumor sites at different stages of disease undergoing different treatments. Factor analysis was performed on the 12 items of the scale confirming the existence of two factors labeled Meaning and Peace (M/P) and Faith (F). Reliability of the FACIT-Sp was evaluated using Cronbach’s a which showed high internal consistency for the total scale as well as the M/P and F subscales. Construct validity showed moderate correlations with measures of quality of life (QOL) and psychological distress. The relation between SWB and clinical and sociodemographic characteristics was evaluated by means of ANOVA or t-test. Associations were observed between FACIT-Sp M/P and performance status, in vs. outpatient and educational level. The FACIT-Sp F was significantly related to religious affiliation, health coverage and living arrangements. In a multivariate regression analysis, SWB was a major determinant of QOL assessments when controlling for clinical and psychosocial variables. SWB was associated with better contentment with quality of life even in the presence of severe pain and other symptoms. The results showed that, even when studied in a predominantly agnostic population, (1) the FACIT-Sp subscale had good reliability, (2) SWB was independently associated with QOL when controlling for clinical and psychological variables, (3) in turn, assessments of SWB were influenced by clinical and demographic features and (4) SWB influenced patients’ appraisal of symptoms. # 1140/ARE THERE COUNTRY-SPECIFIC DIFFERENCES IN OBESITY-RELATED QUALITY OF LIFE? Lee Bowman & Clarice P. Hayes, Global Health Outcomes Research, Eli Lilly & Company, Indianapolis, IN To determine whether there may be country-specific differences in the perception of the impact of weight on quality of life (QOL), the Impact of Weight on Quality of Life Scale (IWQOL) was administered to a sample of obese individuals [Mean body mass index (BMI) ¼ 31.3, 63% P1 self-reported comorbidity, mean age ¼ 41 years, 58% female] in Germany (n ¼ 557), Italy (n ¼ 510), United Kingdom (UK) (n ¼ 547), and United States (n ¼ 570). The IWQOL consists of 65 items forming seven subscales: Health, Social/Interpersonal, Sexual Life, Work, Self-Esteem, Mobility, and Activities of Daily Living. Country-specific samples differed significantly (p < 0.001) in demographics. For example, the UK respondents on average were significantly younger with fewer reported comorbidites than in the other three countries, but had significantly greater average BMI than Italy or Germany. To detect differences across countries, analysis of variance was performed with country as the independent variable and IWQOL total and subscale scores adjusted for BMI, total number of self-reported comorbidities, gender, and age as the dependent variables. The results indicated significant differences (p < 0.001) in IWQOL Total Scale scores and all subscales except Sexual Life. For example, as compared to the other three countries, the UK had significantly better mean Total and Mobility scores; Italy had significantly worse mean Health scores; and Germany had significantly worse mean Social scores. Linear regression analysis confirmed country as an independent predictor of obesity-related QOL. QOL is an important outcome in the evaluation of weight loss interventions (e.g., drug therapy). Differences in the perceived impact of weight on QOL across countries found in this study suggest the need for cross-cultural adaptation of the IWQOL or other QOL measures before they are used in international evaluations of obesity interventions.

755 # 1239/PSYCHOMETRIC PERFORMANCE OF 10 LANGUAGE VERSIONS OF THE BENIGN PROSTATIC HYPERPLASIA PATIENT IMPACT MEASURE (BPH-PIM) Mona L. Martin, Mountlake Terrace, WA; Donald M. Bushnell, Health Research Associates, Inc., Mountlake Terrace, WA, USA; Asha V. Hareendran, Outcomes Research, Pfizer, Sandwich, Kent, UK Objective: The BPH-PIM is a unique measure benefiting from the strengths of both standardized and individualized items. Good psychometric performance has been reported for the original measure. This abstract presents the psychometric properties of the Canadian, Czech, Estonian, Finnish, Lithuanian, Norwegian, Polish, Slovakian, Australian and UK versions of this newly developed measure. Methods: Males 40 years or older, with documented evidence of benign prostatic enlargement and an International Prostate Symptom Score (IPSS) score of >12 were recruited from clinical trials. Internal consistency was measured using Cronbach’s a test–retest reproducibility by intraclass correlation coefficients (ICC) between screening and baseline (2-weeks), convergent validity using the IPSS total score and the American Urological Association (AUA) QoL item, and discriminant validity by severity groupings based on tertiles of IPSS scores. Responsiveness was assessed 4 weeks after baseline using effect sizes. Results: Results were analysed on a per country basis using populations of n ¼ 47 (Australia) to n ¼ 180 (Lithuania). Mean ages ranged between 59 and 67 across the 10 countries. The measurement model (total score plus 3 subscales) was confirmed using factor analyses for all versions. a Coefficients ranged between 0.89 and 0.93, and ICC values were between 0.72 and 0.93. The BPH-PIM total score showed moderate to strong associations across countries with the IPSS score (0.31–0.50; p < 0.05) and with the AUA QoL item (0.24–0.61; p < 0.05). BPH-PIM scores were able to discriminate between levels of severity in all countries (F ¼ 5.34 to 16.93, p < 0.01). Patients reporting small amounts of change in their condition had improvements in BPH-PIM scores ranging from 3.46 (Slovakia) to 14.02 points (Australia) with effect sizes ranging between 0.26 and 1.12. Conclusion: The BPH-PIM performs well across these 10 language versions and is capable of demonstrating statistically meaningful improvement in BPH patient impact in multiple cultures. # 1622/METRIC PROPERTIES OF THE ‘MINIMUM CORE SET OF SIX QUESTIONS’ PROPOSED FOR PATIENTS WITH LOW BACK PAIN Angels Pont, Health Services Research Unit, Institut Municipal d’Investigacio´ Me`dica; Ferran Pellise´, Unidad Cirugia Raquis, Hospital Universitari de la Vall d’Hebron, Barcelona, Spain; Luis Alvarez, Madrid, Spain; O. Escudero, Unidad Cirugia Raquis, Hospital Universitari de la Vall d’Hebron; Jordi Alonso, Montse Ferrer, Health Services Research Unit, Institut Municipal d’Investigacio´ Me`dica, Barcelona, Spain A Minimum Core Set of Six Questions (6Q-Core) has been proposed for outcome measurement of patients with spinal disorders (Deyo et al., Spine 1998; 23: 2003–2013), but its metric properties have not been studied yet. The aim of this study is to evaluate them. Patients consecutively recruited in a University Hospital in Barcelona suffering degenerative low back pain (n ¼ 81) and patients from a hospital in Madrid with subacute osteoporotic fracture-vertebroplasty (n ¼ 50) completed the 6Q-Core. For test–retest analysis, the questionnaire was readministered after 15 days in 28 patients with degenerative low back pain. Responsiveness was assessed among all the patients with subacute osteoporotic fracture-vertebroplasty after 3 months of surgery. For the 6Q-Core, a score for each of five domains (Back Function, Generic Health Status, Pain, Work Disability and Patient Satisfaction) and a total score have been calculated. Eighty-two women and 49 men with a mean of 58.7 years were evaluated. Missing items were low for all questions except for Work Disability and Back Function (52 and 22% of patients, respectively). The total score of the 6Q-Core achieved the internal consistency standard for individual comparisons (Cronbach’s a ¼ 0.91). Floor effect was high for Back Function, Generic Health Status and Work Disability (37, 64 and 43%). Moreover, Work Disability presented a ceiling effect of 41%. Test–retest showed stability with Intra-class Correlation Coefficient equal or higher than 0.8 for all domains and 0.96 for total score. Responsiveness was high with an effect size of 4.1 for total score. These

results suggest that the 6Q-Core could be a reliable and responsive instrument for outcome assessment in spinal disorders allowing for a routine evaluation in the clinical practice. However, Work Disability score presented some problems (high floor, ceiling and missing data) that merit further research. Funded by: CIRIT 2001SGR 00405. # 1751/IS QUALITY OF LIFE RESEARCH THEORY BASED? Astrid K. Wahl & Tone Rustoen, Faculty of Nursing, Oslo University College, 0167 Oslo; Berit R. Hanestad, Department of Public Health and Primary Health care, University of Bergen, 5009 Bergen, Norway; Geraldine Padilla, School of Nursing, University of California San Francisco, CA, USA There has been a rapid development of the concept QOL. However quality of life research has been critiqued by the lack of theory based research. Many of the instruments used in quality of life research are empirically developed without a theoretical framework which makes hypothesis generation and interpretation of results difficult. The aim of this presentation is to give an overview of theory based quality of life research. This review analysed papers published in the journal, Quality of Life Research, from 1995 until the end of 2002. The analysis addressed six research questions: (1) how is quality of life defined, (2) how is quality of life distinguished from other related concepts such as happiness or well-being, (3) what are the targeted domains of studies (4) what theoretical framework(s) or models are used, (5) do studies include theory testing hypotheses, and (6) what reasons guide the choice of measures? Preliminary results showed that quality of life research is generally atheoretical. Few papers address the definition of the concept quality of life, nor do they use theoretical frameworks or theoretical models to guide the research. There is a need to explicate the implicit conceptual context and theory-development contributions in quality of life research. The field should move from psychometrics and descriptive epidemiological studies to theory generating and testing research. Future quality of life studies should attempt to provide a definition of the concept, describe the theory from which the research questions develope and explicate the conceptual model illustrating the relationship among the concepts and measures. # 3002/QUALITY OF LIFE IN SCHIZOPHRENIA: THE NEED FOR CONCEPTUAL MODELS George Awad, University of Toronto, Canada Schizophrenia is a chronic and disabling psychiatric disorder, accompanied by acute psychotic relapses, which often require hospitalization. The illness significantly impacts on various behavioural, mental and psychological domains resulting in long-term disability. Schizophrenia presents as one of the most appropriate disease models, whereby QOL should be a significant component in its management. The last three decades have witnessed an increase in interest and publications about QOL in schizophrenia. While the majority of published studies report data about QOL measurement in schizophrenia, it is unclear about what is measured or what it means, in terms of linking quality of life to important outcomes such as better adherence to medications or improved psychosocial functioning. The lack of interest in developing conceptual models for QOL in schizophrenia has limited the meaningful development of such an important construct. Additionally, the ongoing unresolved debate about the subjective/objective dichotomy has uncovered significant inconsistencies in the application of the construct of QOL in schizophrenia. This presentation will review the available conceptual models, including our conceptual model, which continues to be the only major model for quality of life in medicated persons with schizophrenia. It is our contention that, without coherent and reliable conceptual models that can be experimentally validated; the field of quality of life in schizophrenia will be limited in its impact on clinical management. It may also be that one conceptual model in schizophrenia could prove inadequate in dealing with various stages of the illness and its varied outcomes. Possibly, different conceptual models are required, which can be appropriate for the specific population, the stage of the illness and its treatment, as well as societal expectations at a particular point of time. Similarly, research interests have to look beyond QOL, not only as an outcome by itself but also as a mediator for other important outcomes.

756 # 1248/TESTING A MODEL OF QUALITY OF LIFE (QoL) FOR FAMILY CAREGIVERS OF STROKE SURVIVORS Carole L. White, Sharon Wood-Dauphinee & Nancy Mayo, Epidemiology & Biostatistics, McGill University, Montreal, Quebec; Sylvie Lauzon, School of Nursing, University of Ottawa, Ottawa, Ontario; Mark Yaffe, Susan Scott & James A. Hanley, Epidemiology & Biostatistics, McGill University, Montreal, Quebec, Canada Although previous studies have examined selected factors influencing QoL, the complex interplay of these variables in family caregivers of stroke survivors has not previously been investigated. The purpose of this study was to test a model of the caregiver’s QoL that incorporated proposed relationships among factors in the caregiving situation, the caregiver’s environment, caregiver characteristics, and overall QoL. A sample of 202 caregivers completed a battery of instruments measuring the stroke survivor’s function and behavior, the caregiver’s social support and positive affect, quality of the relationship between the caregiver and stroke survivor, and the caregiver’s overall QoL (represented as a latent factor). Structural equation modeling, using maximum likelihood estimation, was used to assess the model. The model represented a good fit to the data (robust CFI ¼ 0.97; RMSEA ¼ 0.07) and explained 73% of the common variance among the indicators of QoL. Factors making up the caregiving situation had both direct and indirect pathways to QoL. The total effect of behavioral disturbances on the caregiver’s QoL was )0.61, interpreted as an increase in stroke survivor behavioral disturbances of 1 SD associated with a 0.61 SD decrease in the caregiver’s QoL. Caregiver provision of assistance also demonstrated both direct and indirect effects on the caregiver’s QoL with a total effect of )0.24. Social support was associated both directly and indirectly with the caregiver’s QoL (total effect: 0.34). The quality of the relationship between the caregiver and the care-recipient (0.10) and the caregiver’s positive affect (0.20) had small direct effects on QoL. This model supports the hypothesis that the caregiver’s QoL is the outcome of a complex interplay among factors associated with the caregiving situation, factors in the caregiver’s environment, and characteristics of the caregiver. The strength of direct and indirect paths suggests the importance of interventions aimed at addressing the stroke survivor’s behavior, enhancing social support, and maintaining the relationship between the caregiver and the care-recipient. # 1658/SYMPTOM DISTRESS IN PATIENTS WITH LUNG CANCER: ARE WE MEASURING THE RIGHT DIMENSIONS? Carol Tishelman, R&D Unit, Stockholms Sjukhem Foundation, Department of Nursing, Karolinska Institutet, Stockholm, Sweden; Lesley F. Degner, Faculty of Nursing, University of Manitoba, Winnipeg, Manitoba, Canada; Ann Rudman, R&D Unit Stockholms sjukhem Foundation, Department of Nursing, Karolinska Institutet, Stockholm; Eva Broberger Wiberg, Department of Nursing, Karolinska Institutet, Huddinge; Sara Runesdotter, R&D unit, Stockholms sjukhem Foundation, Department of Nursing, Karolinska Institutet, Stockholm; Petter Gustavsson, Department of Nursing, Karolinska Institutet, Huddinge, Sweden Despite the particularly vulnerable situation for patients with lung cancer, published findings on symptom experience remain relatively sparse. The distinctions between the dimensions of intensity and distress have generally not been clarified in existing research, nor has the question of how distressing patients perceive different symptoms to be, irrespective of their current intensity levels, been specifically addressed. The primary aim of this project was therefore to explore

patterns of intensity and distress from symptoms in newly diagnosed lung cancer patients, in order to provide a framework for nursing and palliative interventions for these patients. Four hundred patients with newly diagnosed inoperable lung cancer have been interviewed up to six times during the first year post-diagnosis. The main dependant variable was the ranking of distress from nine common symptoms, using a Thurstone scale of pairwise comparisons. Instruments assessing perceived symptom intensity and quality of life were also completed at each time point. This data has also been completed with inductive inteview data, both structured and unstructured, to heighten the validity and understanding of generated results. Data analysis indicates that the ranking of perceived intensity differs from the ranking of perceived distress. The mean ranking of distress in the total group remains constant over time, with breathing, pain and fatigue rated as most distressing, while the mean ranking of intensity fluctuates and varies by subgroup and over time. The implications of this data for introducing a paradigm of preventive palliation will be discussed. The usefulness of this data for considering both group and individual needs will also be addressed. # 1767/LINKING RULES FOR MAPPING OUTCOME MEASURES TO THE INTERNATIONAL CLASSIFICATION OF FUNCTIONING, DISABILITY AND HEALTH Alarcos Cieza, Szilvia Geyh, Martin Weigl, Edda Amann & Barbara Kollerits, Physical Medicine and Rehabilitation, University of Munich, ¨ stu¨n, CAS-Group, World Health Munich, Germany; T. Berdihan U Organization, Geneva 27, Switzerland; Gerold Stucki, Physical Medicine and Rehabilitation, University of Munich, Munich, Germany The International Classification of Functioning Disability and Health (ICF) provides a language and framework for the description of health and health-related states. We have previously developed 10 rules for linking the concepts contained in health-status measures to the ICF in a specific and precise manner. The aim of this study was to test these rules and, when necessary, to refine then. Different outcome measures used to assess the effects of interventions in patients with chronic health conditions in RCTs documented in MEDLINE were linked to the ICF. The following difficulties were identified during the linking process: (1) Items concerning health in general and quality of life could not be specifically linked to the ICF. (2) If an item is not contained in the ICF, this item is assigned ‘nc’ (not covered by the ICF) according to linking rule 10. Therefore, nc is used when personal factors (pf) are to be linked, since they have not yet been classified. Much information is thereby lost. (3) Clinical outcomes (e.g. count of tablets) can be used for different purposes (e.g. compliance or pain severity), depending on the clinical setting. (4) The wording of questions in clinical outcomes used to diagnose diseases very often masks the aim of the outcome (e.g. measures to assess the cognitive mental status of patients). To deal with these challenges, the following amendments to the linking rules are proposed: ad (1) nd-gh (not definable-general health) should be used for items concerning health in general and nd-qol (nd-quality of life), for items concerning quality of life in general. (2) It will be differentiated among pf-s (pf-satisfaction), pf-sp (pf-self perception), pf-po (pf-perception of others, pf-kn (pfknowledge). This list of abbreviations can be progressively amended to gather information about relevant personal factors. (3) and (4) Three different pieces of information are to be extracted from clinical outcomes: what is measure, how it is measured, and with which aim it is measured. The latter piece of information is to be linked to the ICF using the original rules.

757 # 1147/WHAT DO WE KNOW ABOUT ASSESSING HEALTH-RELATED QUALITY OF LIFE (HRQOL) IN CANCER? FINDINGS FROM REPORTS OF THE CANCER OUTCOMES MEASUREMENT WORKING GROUP (COMWG) Carolyn C. Gotay, Cancer Research Center of Hawaii, University of Hawaii, Honolulu, HI; Joseph Lipscomb & Claire F. Snyder, Division of Cancer Control and Population Sciences, National Cancer Institute, Bethesda, MD, USA Purpose: The COMWG is a National Cancer Institute initiative to improve the science of outcomes measurement in cancer. Thirty five researchers from the US, Canada, and the Netherlands prepared comprehensive reports on the state-of-the- science of outcomes measurement in the four most frequent cancers (breast, colorectal, lung, prostate) across the continuum of care. Methods: This paper focuses on a compilation and synthesis of findings related to HRQOL. Findings were based on published and unpublished literature. Results: The following points summarize researchers’ conclusions and caveats: (1) HRQOL is multidimensional BUT there is not agreement of what those dimensions are; (2) Validated questionnaires to measure HRQOL in cancer patients and survivors are available BUT there are no questionnaires that should be currently regarded as ‘gold standards’; (3) Using disease-specific or dimension- specific measures maximizes chances to detect specific treatment-related effects BUT minimizes opportunities to compare findings across studies; (4) HRQOL has received much more attention than other non-clinical outcomes (e.g., satisfaction, economics) in clinical trials BUT the opposite is true in analyses of patterns of care and policy; (5) HRQOL data have provided useful information in clinical trials BUT the relationships between HRQOL and clinical outcomes (e.g., survival) are not consistent. Further progress in this field will be facilitated by the following: (1) Research techniques (e.g., item response theory, copmuter adaptive testing) promoting comparability of findings across HRQOL questionnaires; (2) Theoretical models to define HRQOL and distinguish antecedent from outcome indicators; (3) Integration of HRQOL and other outcomes; and (4) Consideration of how HRQOL data can be made useful for a range of purposes: clinical trials, observational studies, cost-effectiveness analysis, population surveillance, policy development, and patient-provider decision making at the bedside. # 1525/NORMATIVE AND EMPIRICAL ANALYSES OF THE USE OF QOL IN THE NCI-SUPPORTED COMMUNITY CLINICAL ONCOLOGY PROGRAM David R. Buchanan, Lori Minasian, Joseph Kelaghan & Ann O’Mara, Division of Cancer Prevention, National Cancer Institute, Bethesda, MD, USA The purpose of the research is to examine QOL objectives and measures in 246 symptom-management clinical trials supported by the Community Clinical Oncology Program of the US National Cancer Institute over the past 20 years. Methods included empirical and normative analyses of QOL measures used in these trials. Results show that 72% of the CCOP studies have QOL objectives or measures, most often as a secondary endpoint. Thirty-six percent (n ¼ 90) are currently active. Studies include treatment trials of symptoms in patients with the following cancers: breast, prostate, lung, colorectal, oral cavity, pancreatic, and brain. Symptoms addressed in the CCOP include fatigue, pain, nausea and vomiting, hot flashes, sleep disorders,

anemia, osteoporosis, cognitive functioning, sexual dysfunction, anorexia and cachexia, anxiety, and depression. In the CCOP trials that have QOL measures, 59% investigate the efficacy of new pharmacological interventions, 27% examine the effectiveness of complementary and alternative medicines (e.g., valerian, ginger, St. John’s Wort, acupressure, etc.), and 9% behavioral interventions (stress management strategies, diet and exercise, etc.). The most frequently used general quality of life measures used in these trials are the FACT, SF-36, EORTC QLQ-C30, QLQ-LC-13, LASA, Spitzer Quality of Life Index, and Uniscale. Empirically, the analysis identified two major issues. Conceptually, the definition of ‘quality of life’ varies enormously across these studies. Theoretically, hypothesized relationships between various symptoms treated and general QOL or dimensions of QOL (e.g., physical, emotional, social, functional) are under-specified. Normatively, based on philosophical work by Parfit, Griffin, and Brock, the QOL measures used in the CCOP trials were found to fall almost exclusively into the category of ‘hedonist’ theories, to the neglect of ‘preference satisfaction’ and ‘objective’ theories of human well-being. The paper concludes with recommendations for improving the conceptual, theoretical, and philosophical underpinnings of QOL measures in cancer control trials # 1127/QUALITY OF LIFE AND COPING IN YOUNG ADULT SURVIVORS OF CHILDHOOD CANCER Heleen Stam, Martha A. Grootenhuis & Bob F. Last, Psychosocial Department, Emma Children’s Hospital, Academic Medical Centre, Amsterdam, The Netherlands As a result of advances in childhood cancer treatment the number of survivors reaching adulthood has increased enormously in the last decades. The examination of QoL of childhood cancer survivors becomes more and more important. This study compares the QoL of 353 survivors of childhood cancer in The Netherlands (49.6% male, 50.4% female, mean age 24.3 years) with the QoL of 508 Dutch young adults (47.0% male, 53.0% female, mean age 24.2 years) who did not suffer from cancer. Coping strategies of the survivors were also investigated. HRQoL was measured with the RAND-36. The Cognitive Control Strategies Scale (CCSS) was used, which measures four coping styles: predictive control (having positive expectations about the future), vicarious control (attributing power to medical caregivers and treatment), interpretative control (searching for meaning and information), and illusory control (wishful thinking). MANOVA was conducted to test group differences on the RAND-36, correcting for age and sex. Multiple regression analyses were performed to test the predictive value of the coping strategies on HRQoL, while correcting for age, gender, medical variables (diagnosis and treatment variables) and health status. The results of MANOVA showed small differences between survivors and controls on three RAND-scales: physical functioning, social functioning, role limitations due to physical problems. Coping seemed to be an important predictor of HRQoL, especially for mental functioning and general health perceptions. More reliance on predictive control was correlated with better HRQoL on all scales of the RAND-36. Higher scores on interpretative control were associated with more role limitations due to physical problems. Health status (current health complaints/disease and chronic health complaints/disease) also attributed strongly to the model. Worse health status predicted lower HRQoL, especially worse physical and social functioning, and pain. The attributions of gender, current age, diagnosis and treatment variables to HRQoL were much lower.

758 # 1284/TRAJECTORIES OF RECOVERY AFTER HEMATOPOIETIC STEM CELL TRANSPLANT (HSCT): MATERNAL PERSPECTIVE Susan K. Parsons, Pediatric Oncology, Dana-Farber Cancer Institute; Mei-Chiung Shih, Clinical Research Program, Children’s Hospital, Boston, MA; Richard Martini, Child and Adolescent Psychiatry, Children’s Memorial Hospital, Chicago, IL; Jaime Ostroff, Psychiatry and Behavioral Sciences, Memorial Sloan-Kettering Cancer Center; William H. Redd & Derald H. Ruttenberg Cancer Center, Mount Sinai School of Medicine, New York, NY; Sharon L. Manne, Psychooncology Program, Fox Chase Cancer Center, Philadelphia, PA, USA Although HSCT is a treatment option for many cancers and hematological disorders of childhood, the impact of HSCT on children’s health-related quality of life (HRQL) has not been well described. We performed a longitudinal analysis using the Child Health Ratings Inventories-Parent Report of 179 mothers of children (5–18 years) at four time periods over the first year of the HSCT. Three major factors derived from principal components analysis were the focus of the analysis: physical (PF), social (SF), and emotional functioning (EF).The primary goal of the analysis was to quantify the trajectory of the child’s HRQL, linking HRQL with clinical status. The analysis was performed using a pattern mixture model to compare the three principal transplant types (autologous; allogeneic, related; allogeneic, unrelated), adjusted for the timing and severity of graft vs. host disease (GVHD) and relapse as time varying covariates. Trajectories were estimated separately for subjects who survived greater than 12 months and those who did not. Overall, 30% of children died; this did not vary by transplant type. One-third of all allo recipients developed GVHD (82% was acute; 25% was severe). In addition, 13% of patients relapsed within 12 months of transplant. Table 1 shows the estimated mean scores (s.e.) of PF for patients who survived more than 12 months and had no severe GVHD or relapse within 12 months. The mean PF varied significantly by transplant type; the unrelated group had the lowest functioning at 3 months. Nevertheless, recovery by 12 months was similar across all transplant types. Further, subjects with severe acute GVHD had significantly worse PF, while the impact of severe chronic GVHD on PF was smaller. The impact of relapse within 12 month was minimal. The trajectory among non-survivors varied by time of death. These results and the trajectories for SF and EF will also be presented. In conclusion, parents perceive the impact of HSCT on their children’s HRQL to be considerable. While compelling, an understanding of the children’s perceptions is still needed. # 1533/QUALITY OF LIFE (QOL) OF WOMEN AT INCREASED HEREDITARY RISK OF OVARIAN CANCER WHO OPT FOR PROPHYLACTIC SURGERY vs. PERIODIC SCREENING Joanna Madalinska & Eveline Bleiker, Psychosocial Research & Epidemiology; Marc van Beurden, Gynecology, The Netherlands Cancer Institute, Amsterdam, The Netherlands; Heiddis Valdimarsdottir, Psychiatry, Memorial Sloan-Kettering Cancer Center, New York, NY; Henk Boonstra, Gynecology, University Medical Center, Nijmegen; Neil K. Aaronson, Psychosocial Research & Epidemiology, The Netherlands Cancer Institute, Amsterdam, The Netherlands Little is known of the QOL effects of preventive health strategies in women at increased genetic risk of ovarian cancer. High-risk women may opt either for prophylactic oophorectomy (PO) or periodic gynecologic screening (GS). The aim of the current multicenter, longitudinal study is to determine the psychosocial and physical consequences of both preventive strategies. To date, 146 out of 168 eligible women at

increased ovarian cancer risk who sought gynecologic advice on preventive measures have agreed to participate in the study (response 87%). Assessments by questionnaires took place shortly after the first gynecological consultation (baseline) and at 3- and 12 months postsurgery. Baseline data were available on 146 women of whom 41% intended to undergo PO, 44% opted for GS and 15% had not yet decided on the preventive option. All women reported high levels of generic QOL (SF-36, QLQ-C30) and sexual functioning (SAQ). As expected, menopausal women (23% of the study sample) had significantly more endocrine symptoms (FACT-ES) than did pre-menopausal women (p < 0.05). Women who intended to undergo PO reported significantly more cancer worries and anxiety, and perceived their risk of developing ovarian cancer as significantly higher than did women who opted for GS (p < 0.01). The choice of PO was associated with mutation status (BrCa1/2 carrier), parity (having at least one child), education (low) and an active coping style. Sample accrual and data collection are still on-going. New patients will be included in the study until a PO group of 130 women is obtained. Thus far, 60 women have undergone PO and are scheduled for the follow-up assessments. Preliminary longitudinal analyses will examine whether PO affects QOL of these women. # 1733/IDENTIFICATION AND EMPIRICAL EVALUATION OF SCALES FOR ASSESSING THE PATIENT EXPERIENCE OF COGNITIVE IMPAIRMENT Lori Frank, MEDTAP International, Inc, Bethesda, MD; Jennifer A. Flynn, Eli Lilly and Co, Indianapolis, IN; Louis Matza, Mary Kay Margolis & Leah Kleinman, MEDTAP International, Inc, Bethesda, MD; Lee Bowman, Eli Lilly and Co, Indianapolis, IN, USA Mild cognitive impairment (MCI) and mild dementia of the Alzheimer’s type (DAT) impact individuals profoundly yet standard neuropsychological instruments do not always detect impairment, and the patient perspective has not been rigorously measured. The PROCOG was developed to address the need for patient-based assessment of the impact of cognitive impairment. Items were created based on clinical input, patient and informant focus groups in the US and Europe, and the literature. Seven hypothesized subscales were identified to address relevant domains of the cognitive impairment experience. Empirically identified subscales from initial psychometric validation data were compared to hypothesized domains and psychometric properties evaluated. Subjects were 60 MCI, 72 DAT, and 34 cognitively intact control subjects age 64+ recruited through outpatient neurology and memory clinics in the US. Items were submitted to exploratory factor analysis (EFA) using principal components. Factor solutions were reviewed and compared with the seven hypothesized subscales. The 4 and 5 factor solutions did not provide adequate separation of conceptually distinct dimensions. A 6 factor solution most closely approached simple structure following oblique rotation, explained 74% of variance and improved interpretability relative to the 7 factor solution. Factors were affect/self-esteem, skill loss, semantic memory, memory for recent events, cognitive functioning, and social impact, and a long term memory (LTM) item. Scaling success was 90% or above. Internal consistency reliability was excellent: a > 0.82. Two-week test–retest reliability and reproducibility was supported by high ICCs (0.78-0.96 for all but LTM). Skill loss and memory for recent events subscales had highest discriminant validity between MCI and DAT subjects; controls were distinct from MCI and DAT subjects (post hoc t-tests, p < 0.05). EFA results improved on conceptually identified domains and resulted in distinct, valid, internally consistent subscales to enhance measurement of the patient perspective on cognitive impairment.

759 # 1625/DOES CAREGIVER WELL-BEING INFLUENCE THE RATING OF THEIR CHILDS HEALTH-RELATED QUALITY OF LIFE? – A STUDY IN CHILDREN WIHT RECURRENT ACUTE OTITIS MEDIA (RAOM) C.N.M. Brouwer, M.M. Rovers & A.R. Maille´, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht; R.H. Veenhoven, Department of Pediatrics, Spaarne Hospital Haarlem, Haarlem; D.E. Grobbee, Julius Center for Health Sciences and Primary Care; E.A.M. Sanders, Department of Pediatric Immunology; A.G.M. Schilder, Department of Otorhinolaryngology, University Medical Center Utrecht, Utrecht, Netherlands Chronic conditions in a child may not only have a considerable impact on the child’s well-being, but also on family-life and caregiver wellbeing. This may influence caregiver perceptions of their child’s functioning as well as their judgement of the child’s health-related quality of life (HRQoL) and functional health status (FHS). Our objective was to explore how caregiver HRQoL influences the rating of their child’s HRQoL. Caregivers of 383 children with rAOM aged 1–7 years completed questionnaires on the impact of the child’s rAOM on their own HRQoL (NRS Caregiver), and on family functioning (FFQ), as well as questionnaires on FHS (OM-6), and HRQoL (NRS Child) of their child. The influence of caregiver HRQoL on the relation between AOM frequency and their rating of child HRQoL was examined by multiple regression modeling with inclusion of caregiver HRQoL as an independent variable. Interaction of AOM frequency and caregiver HRQoL was included to examine modification of the relationship between AOM frequency and child HRQoL by caregiver HRQoL. Family functioning and caregiver HRQoL were judged poorer in children with 4 or more AOM episodes per year compared to children with 2–3 AOM episodes. In children with equal AOM incidence, caregivers with poorer HRQoL rated their child’s HRQoL lower than caregivers with better HRQoL. Furthermore, caregiver HRQoL appeared to modify the relationship between AOM frequency and child HRQoL. We conclude that recurrent AOM in a child negatively affects caregiver functioning and family life, which appears to influence the association between AOM frequency and child HRQoL and FHS. Studies on HRQoL in children should therefore include assessment of caregiver psychological adjustment and HRQoL. # 1163/CAN HEALTH-RELATED QUALITY OF LIFE BE MEASURED IN INFANTS AND TODDLERS? EXAMPLE OF THE CHILD HEALTH QUESTIONNAIRE INFANT/TODDLER (CHQ-IT) Hein Raat, Public Health, Erasmus MC – University Medical Center Rotterdam, Rotterdam, ZH, The Netherlands; Jeanne M. Landgraf, HealthAct Inc., Boston, MA; Eveline Bunge, Public Health, Erasmus MC, Rotterdam, ZH; Margot Buskop, OKZ, Thuiszorg Salland, Ommen, OV; Lisette W. A. van Suijlekom-Smit & Henriette A. Moll, Pediatrics; Marie-Louise Essink-Bot, Public Health, Erasmus MC, Rotterdam, ZH, The Netherlands Can generic health-related quality of life be measured in a reliable and valid way in infants and toddlers? Young children have different functions than older children and show rapid growth and development. In this study we evaluated separately in 2–12 month (infants) and 1– 3 year olds (toddlers), the 103-item Child Health Questionnaire Infant/ Toddler (CHQ-IT) questionnaire for parents with 11 scales (9 for infants) covering physical and psychosocial domains, developed by Landgraf in the US. A random general population sample of 500 parents of 0–3 year olds was mailed the CHQ-IT. Test–retest reliability was measured in a random subgroup (n ¼ 118). Clinical CHQ-IT data were collected for children with asthma (n ¼ 84) and rheumatic dis-

ease (n ¼ 19). Mailed questionnaire response was 82% (2–12 months n ¼ 92; 1–3 years n ¼ 318). In all samples all scales showed Cronbach’s as >0.70; in the subgroup infants 1 a (‘Parental-Time’) was <0.70. Test–retest correlations of all scales were statistically significant (ICC 0.42–0.82; p <0.01). CHQ-IT mean scores were different between children with few/many self-reported chronic conditions and few/many visits to the doctor (p < 0.01). E.g., the highest Cohen’s effect size (1.30) was found for ‘General health perceptions’ that had a lower mean score in children with P2 self-reported conditions (64, SD ¼ 15; n ¼ 57) than in those with no self-reported conditions (84, SD ¼ 11; n ¼ 240); in the subgroup infants this effect size was equally high (1.34). Furthermore, the two clinical samples showed distinct score profiles in the expected directions in comparison with the general population score profile (e.g. a very low mean ‘General health perceptions’ score in the asthma sample and a low mean ‘Bodily pain’ score in the rheumatic disease sample). The findings support the conclusion that it is feasible to measure health-related quality of life in very young children, even in infants 2–12 months old. Internal and test–retest reliabilities of the CHQ-IT scales were satisfactory. The data support the construct and discriminative validity of the CHQ-IT. # 1874/ANAMNESTIC COMPARATIVE SELF ASSESSMENT (ACSA) I. SENSITIVITY, RELIABILITY AND PRONENESS TO RETROSPECTIVITY BIAS AND RESPONSE-SHIFT Peter Theuns, Psychology, Vrije Universiteit Brussel, Brussel; J L. Bernheim, Human Ecology, Vrije Universiteit Brussel, Jette; Ghislain Ledure & Michel Souris, Medical Psychology, Universitair Medisch Centrum St Pieter, Brussels; Darius Razavi, Medical Psychology, Instituut Jules Bordet, Brussels, Belgium Background: To measure global subjective QOL, ACSA uses the respondent’s memories of his best (+5) and worst (-5) periods in life as the extreme ends of the rating scale. Respondents rate their average QOL during the last e.g. 2 weeks (short-term retrospective), or during long intervals (long-term retrospective) relative to these anchor points. ACSA aims to reduce the trivialization of responses, trait and mood-ofthe-day effects, proximity and cultural biases and response shifts that one may elicit when asking the conventional ‘How-have-you-been?’ question. Research questions: (1) Sensitivity, (2) Test–retest and inter-rater reliability, and (3) Proneness to response shifts of ACSAratings. Methodology: Over a period of >1 year an oncologist routinely obtained 6–25 ACSA ratings relative to the interval since the last visit to the clinic (short-term retrospective) from 40 cancer patients in treatment or follow up in a university clinic. For each patient, integral mean ACSA-ratings were computed. At the end of the observation period of >1 year, non-clinic-affiliated psychiatrists obtained two additional ACSA-ratings relative to: (1) the recent inter-visit interval (short-term retrospective), and (2) the whole observation period (longterm retrospective). Results: Inter-patient ratings varied by up to the whole range of 11 scale points. Intra-patient variation over the whole observation period was >5 points for each patient. Inter-rater reliability for short-term retrospective ACSA was r ¼ 0.92. The correlation between integral mean ACSA-ratings and the long-term retrospective ACSA equaled r ¼ 0.61. Conclusions: (1) ACSA ratings are highly sensitive to changing situations. (2) Short-term retrospective ACSAratings were very reliable. (3)The correlation of retrospective ACSAratings and integral mean ACSA is strong in 2/3 of patients, but weak or absent in 1/3, probably due to response shifts. These results, together with other demonstrating validity and discriminative power between patients with different diseases, suggest ACSA should be considered for QOL studies.

760 # 1569/TO WHAT DEGREE ARE SINGLE ITEM ASSESSMENTS OF QOL STABLE IN A PRIMARY CARE SETUP Yasmin Maor, Internal Ward A, Sheba Medical Center, Ramat Gan, Israel; Hedi Feivel, Zeev Aharonson & Joseph Rosenblum, Medical Wing, Kupat Holim Meuhedet, Tel Aviv, Israel One of the obstacles for using QOL tools in routine clinical practice is the reluctance of health-care professionals to perform structured QOL evaluations due to time constraints. Thus, physicians often use a global question assessing patients QOL. We sought to quantify the amount of change patients’ exhibit in their repeated response to a QOL question and to investigate factors affecting patients differing responses. We present preliminary data on 62 patients participating in a study assessing hypertension (HTN) in a primary care practice. Patients responded to questionnaires in the following order: sociodemographic details, a question assessing QOL (QOL1), symptom severity (TMI), depressive symptoms (CESD), and then the QOL item was repeated (QOL2). Patients were interviewed again after a mean of 15.9 days (SD 11.5) and the QOL question was repeated for the third time (QOL3). We assessed how many patients changed their response to the QOL item, and compared patients responses (paired t test). Variables affecting patients QOL scores were investigated using multivariate linear regressions. Dependent variables were differences between QOL assessments. Candidate variables were age, sex, education, body mass index (BMI), duration of HTN, TMI, and CESD. Mean age was 57 years (SD 8.7), mean BMI was 29.6 (SD 4), mean QOL1, QOL2 and QOL3 scores were 63.2 (SD 18.8), 66.3 (SD 19.1) and 69.3 (SD 19.0), respectively. 10 (16%) patients had a difference greater than 10 points (10%) between QOL1 and QOL2 items, 16 patients (30%) between QOL1 and QOL3, and 15 (29%), patients between QOL2 and QOL3. The paired differences between QOL1 and 2 and QOL1 and 3 were significant (p < 0.05). The main variable identified in the regressions as affecting the differences between the various QOL assessments was CESD. To conclude, a significant portion of patients change their response to a single item assessing QOL even during a short interview. Depressive symptoms seem to play a role in this process. This data emphasizes that despite their appeal and common use, single item QOL measures are problematic. We plan to further investigate this issue. # 1110/PSYCHOMETRIC VALIDATION OF THE TREATMENT SATISFACTION SCALE FOR MEN WITH BENIGN PROSTATIC HYPERPLASIA (TSS-BPH) Lucy Abraham & Asha Hareendran, Outcomes Research, Pfizer Ltd, Sandwich, Kent; Helen Doll & Darren Clayson, Oxford Outcomes, Oxford, UK The study was designed to demonstrate the psychometric validity of the newly adapted 13-item treatment satisfaction scale for men with benign prostatic hyperplasia (TSS-BPH). The TSS-BPH was completed by 556 men with lower urinary tract symptoms due to benign prostatic obstruction at the end of (or at withdrawal from) a Phase II study. Other measures used included flow rate (Qmax), the International Prostate Symptom score (IPSS), the BPH Patient Impact Measure (BPH-PIM), and the Global Rating of Change (GRC). The psychometric properties of the TSS-BPH were assessed using standard methods, including the use of Rasch models, and the estimation of Minimally Important Difference (MID). Factor analysis confirmed that the TSS-BPH can be considered as a three-scale measure (Satisfaction with Efficacy, Satisfaction with Dosing, and Satisfaction

related to Side Effects). A second order factor analysis demonstrated that an overall score can be calculated. The three subscales explained 66% of the variance, with the Efficacy and Overall Satisfaction scales having Cronbach’s a statistics of 0.80–0.90, demonstrating good internal reliability. Statistically significant correlations between all TSSBPH scales (except the Dosing scale) and the BPH-PIM and IPSS QoL scores demonstrated convergent reliability. Discriminant validity was shown by statistically significant associations in the expected directions between all TSS-BPH scales (but the Side Effects scale) and IPSS, Qmax, and BPH-PIM scores. The TSS-BPH Side Effects scale was strongly associated with the number of adverse events and their severity. Best estimates of the MCID for the three scales and the overall scale were calculated in terms of GRC responses and adverse events. The results of the Rasch analysis suggested that four of the TSS-BPH items could be administered alone. The TSS-BPH has been found to have good psychometric properties although it would be possible to administer 4 of the 13 items with little loss of information. # 1548/THE CONCEPTUAL FOUNDATIONS FOR A GENERALIZED PATIENT TREATMENT EXPECTATIONS QUESTIONNAIRE Anusha Sinha, Health Outcomes, Quintiles Late Phase, San Francisco, CA; Mark J. Atkinson & Clayton R. Rowland, Outcomes Research, Pfizer, Kalamazoo, MI; Amy E. Keating, Health Outcomes, Quintiles Late Phase, San Francisco, CA; Steven L. Hass, Outcomes Research, Pfizer, Kalamazoo, MI; Shoshana S. Colman, Health Outcomes, Quintiles Late Phase, San Francisco, CA, USA Patient Expectations of Medications (PEM) has been identified as the strongest predictor of Treatment Satisfaction with Medication (TSM), yet the causal effects of prior expectations and satisfaction is small. Moreover, the majority of expectations measures are disease-specific, thus limiting the comparability of measurement constructs across different treatment types. This study builds on a Generalized Conceptual Model (GCM) for TSM. The objective was to enhance the theoretical foundation of PEM and TSM through the content development of a conceptually grounded generalized patient expectations questionnaire (TSQM-Baseline) designed to evaluate treatment satisfaction at baseline. Key dimensions of PEM, as related to TS, were identified through a literature review and subsequently evaluated using two parallel qualitative approaches: (1) internet-based bulletin board focus groups (BBFGs) and (2) in-depth interviews. The review citations were indexed between 1990 and 2003 in major health-related literature. Findings from the literature review and the GCM for TSM (used to develop the TSQM) guided the development of a common discussion guide used both for the BBFGs and in-depth interviews. Two online BBFGs were convened with 35 patients representing the therapeutic areas of asthma, arthritis, cancer, cardiovascular disease, depression, diabetes, infectious disease, migraine, and psoriasis. In parallel, 15 indepth interviews were conducted representing similar therapeutic areas. The guide was iteratively revised in both approaches. Findings from each approach were independently analysed and over 95% agreement on main thematic content was found. Moderators of the relationship between PEM and TSM were defined, and six core measurement domains of PEM were identified: (1) Effectiveness, (2) Side Effects, (3) Convenience, (4) Impact on Daily Life, (5) General Expectations, and (6) Cost. Based on these construct dimensions, a draft questionnaire was created and pretested on 15 patients. The resulting TSQM-Baseline item bank consisted of 51 items, inclusive of mediators/moderators and core PEM items.

761 # 1417/PATIENT SATISFACTION ASSESSMENT WITH PHARMACOLOGICAL PAIN TREATMENT: DEVELOPMENT AND VALIDATION OF A NEW MEASURE Eva Baro´ & Alfonso Casado, Outcomes Research, 3D Health Research, Barcelona; Carles Garcı´a, Medical Department, Lacer Laboratorios, Barcelona, Catalunya; Salvador Ribas, Outcomes Research, 3D Health Research, Barcelona, Spain Patient satisfaction assessment and QoL are increasingly important in capturing patient-perspective on health care, but few validated measures exist to assess satisfaction with pharmacological pain treatment. This study aim to develop a reliable, valid and sensible tool for assessing satisfaction with pharmacological pain treatment. Sources to define content of the measure were bibliographic review, focus groups with patients, and expert opinion. An initial self-administered version of 14 items was obtained and tested in a prospective study. Item–total statistics and exploratory factor analysis (FA) were performed for item reduction. The final version was psychometrically validated by assessing feasibility, internal consistency (Cronbach’s a), convergent validity (visual analogic scale (VAS) for pain intensity) and discriminant validity for patients presenting pain relief (ROC curves). Additionally, test–retest reliability (intraclass correlation coefficient, ICC) and responsiveness (effect size, ES) were assessed in patients maintaining and changing treatment, respectively. The initial version was administered to 362 patients (women 58%, 51 years old, 48.1% suffering from osteoarthritis). Four items were deleted from the initial version owing to a low item-scale homogeneity, leading-weight in FA and/or contribution to alpha values. The final FA explained 70.2% of the variance. Four dimensions were identified concerning adverse events (AE), speed-length of effect, functional benefit and overall satisfaction. Final version was answered completely for 97.2% of sample. Alpha for the global score (GS) was 0.88 and over 0.80 for dimensions. ICC for GS was 0.73 and ranged from 0.59 (functional benefit) to 0.80 (AE). Correlations with VAS ranged from 0.48 to 0.55 (except AE, 0.20). AUC for GS was 0.78 and over 0.65 for dimensions (except AE, 0.57). ES were large for GS and dimensions (0.8–2.5). This new measure provides a short, reliable, valid and sensitive means of measuring satisfaction with treatment in ambulatory patients with acute or chronic pain. # 1247/DEVELOPMENT AND VALIDATION OF A DISEASE-SPECIFIC TREATMENT SATISFACTION QUESTIONNAIRE FOR GERD (TSQ-G) Karin S. Coyne, Center for Health Outcomes Research, MEDTAP International, Bethesda, MD, USA; Ingela Wiklund, Research and Development, AstraZeneca, Molndal, Sweden; Jordana Schmier, Exponent, Alexandria, VA, USA; Katarina Halling, Alessio degl’ Innocenti, Research and Development, AstraZeneca LLP, Molndal, Sweden; Dennis Revicki, Center for Health Outcomes Research, MEDTAP International, Bethesda, MD, USA Currently, no disease-specific, patient-based treatment satisfaction (TxS) instruments related to gastroesophageal reflux disease (GERD) exist. The purpose of this study was to develop and validate a TxS questionnaire in GERD. Six focus groups of GERD patients (n ¼ 36) were conducted to ascertain patient perceptions of TxS. Complete and immediate symptom relief was the most important aspect of TxS with side effects, provider relationships, and cost being secondary. The resulting 64-item questionnaire was piloted in 64 patients and reduced to 41 items. The 41-item questionnaire was then validated in treated GERD patients; 25% were randomized to a 2-week test–retest. Ancillary validation measures were: SF-36, QOLRAD, GSRS, SDRS, PSQ-18, and physician and patient global measures of symptoms and satisfaction. Exploratory factor analysis and Rasch analysis were used

to examine the item and subscale structure. Internal consistency, test– retest reliability, and construct validity were assessed using Cronbach a, intra-class correlations, ANOVA, and t-tests. A total of 198 GERD patients participated with a mean age of 50.7 years; 68% female, and 84% Caucasian. Clinician-rated severity of GERD was: mild (32%); moderate (50%); and severe (18%); 83% were on a PPI. The final TSQ-G consisted of 28 items with seven subscales (Symptoms, Satisfaction, MD relationship, Cost, PRN, Expectations, and Bother). Cronbach as ranged from 0.58 (MD) to 0.94 (Symptoms). Correlations were moderate to strong within the expected subscales of the SF-36, QOLRAD, GSRS, and PSQ-18. No significant correlations were present with the SDRS. Age and gender did not affect TSQ-G responses. TSQ-G subscales, particularly Symptoms and Satisfaction, significantly discriminated among levels of clinician-rated disease severity, number of symptom days, and patient and clinician ratings of satisfaction. Test–retest correlations ranged from 0.65 to 0.92. The TSQ-G has excellent reliability and concurrent and discriminative validity. It appears to be a useful tool for evaluating treatment satisfaction among GERD patients. # 1708/SHORTENING THE QUESTIONNAIRE OF SATISFACTION FOR HOSPITALIZED PATIENT (QSH): PRELIMINARY DEVELOPMENT OF THE QSH-19, A 19-ITEM SHORT FORM S. Antoniotti & C. Sapin, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille; J. Labare`re, Medical Evaluation Department, School of Medicine, Grenoble; M.C. Simeoni, L. Boyer, D. Debensason & M. El Khammar, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille; L. Gerbaud, Epidemiology, Health Economics & Public Health Department, School of Medicine, Clermont Ferrand; C. Colin, Statistics & Medical Information Department, School of Medicine, Lyon; P. Auquier, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille, France The ‘Satisfaction Questionnaire for Hospitalized patients’ (QSH) is a French self-administered questionnaire based on patients’ viewpoints, designed for inpatients whatever the medical, surgical or obstetrical ward. The QSH comprises 45 items, allowing the calculation of nine dimensions as well as two composite scores (staffs and structure indexes). The purpose of this study was to determine whether a derived short-form would yield similar psychometric findings as the full-length version QSH. Data used in this study came from two sources: data from patients’ sample involving in the reduction step of the QSH development (n ¼ 1274) were used to select the subset of items from the 45-item QSH; validation step sample data (n ¼ 2736) were used to cross-validate population-based scoring algorithms for QSH-19 indexes and to perform preliminary tests of validity, reliability. Regression methods were used to select 19 items from the QSH to reproduce its two composite scores. The resulting 19-item short-form achieved an adjusted R2 of 0.98 in prediction of the stas and structure indexes. Several tests of empirical validity were replicated for the QSH-19, including comparisons between groups known to dier in terms of gender or age. All the significant results shown by the QSH indexes were also encountered by the QSH-19. The ability of QSH-19 to discriminate between medical, surgical and obstetrical wards or hospital structure specificities (university-teaching hospital vs. general hospital) was also proven. These first results of validation from this study are encouraging about the feasibility of shortening questionnaire in order to improve their use in large sample survey with a minimal loss in measurement precision. The QSH-19 indexes reached the same statistical conclusions about hypothesized group dierences as did the two indexes provided by the QSH. Questionnaire length was reduced by more than two-thirds. The choice of this short version would be more justified in studies with large sample size.

762 # 1853/DEVELOPMENT AND VALIDATION OF A FAMILY CAREGIVER BURDEN SCREENING AND ASSESSMENT INSTRUMENT (BURDEN) Chih-Hung Chang & Linda L. Emanuel, Feinberg School of Medicine, Northwestern University, Chicago, IL, USA Background: End-of-life care has benefited from recognition that quality patient care entails inclusion of the family in the circle of care. However, the recognition has not yet been paralleled by rigorous development of measurement instruments to screen and assess the extent of family caregiver’s burden. Methods: As part of a larger study, we conducted a series of focus groups and interviews with family caregivers and professionals followed by a survey of a nationally representative sample of 799 family caregivers for patients with a terminal diagnosis. From the focus groups we derived a framework of areas that are important to caregivers. These areas were then empirically tested using item response theory-based analysis of the caregivers’ survey responses. Results: Six areas were identified from the focus groups, with several themes in each area. Five of these areas were evident in the empirical analysis. The remaining one was underrepresented in the survey questions but had sufficient merit to retain. A mnemonic was constructed to correspond to the instrument’s name and acronym, BURDEN, in order to aid in its routine use. That is: Burden of the caregiver role; Understanding illness information; Role related care skills; Depression and other mental health needs; Economic matters; and the Nexus of community. Conclusions: BURDEN is the first empirically derived, comprehensive tool designed from a validated framework and tested survey questions for clinical use in family caregiver screening and assessment in end-of-life care. Evaluation of its performance in another population is needed to complete BURDEN’s fuller evaluation. # 1715/LONGITUDINAL COMPARISON OF PATIENT’S AND CAREGIVER’S PSYCHOLOGICAL MORBIDITY DURING END-OFLIFE CARE FOR BREAST CANCER Eva Grunfeld, Medical Oncology, Ottawa Regional Cancer Centre; Jennifer Clinch, Clinical Epidemiology Program, Ottawa Health Research Institute; Teresa Janz, Ottawa; Timothy Whelan, Medical Oncology, Hamilton Regional Cancer Centre, Hamilton, Ontario; Len Reyno, Dalhousie University, Halifax, Nova Scotia, Canada; Craig Earle, Dana-Farber Cancer Institute, Boston, MA, USA; Robert Glossop, Vanier Institute of the Family, Nepean; Andrew Willan, Centre for the Evaluation of Medicine, St Joseph’s Hospital, Hamilton, Ontario, Canada There is growing awareness of the impact of caregiving on the psychological well-being of the family caregiver. We conducted a longitudinal cohort study in order to better understand the end-of-life experiences of patients and their family caregivers. Eighty-four women with advanced breast cancer and their principal family caregiver were enrolled in the study and followed up until patient death or completion of the study at 2 years, whichever came first. QL data were collected at three monthly intervals during palliative illness and at biweekly intervals during terminal illness (the last 4 weeks of life). This longitudinal study allows us to make paired comparisons of the patientcaregiver dyad over the course of the illness. Patients and caregivers both completed general health status measures (EORTC QLQ-C30

and SF36, respectively) and domain specific measures (HADS). During palliative illness, caregivers were more anxious (mean score 8.37; SD 4.34) than patients (mean score 7.25; SD 4.16) (difference 1.12; 95% CI 0.09–2.15; p ¼ 0.03). Neither caregivers nor patients became more anxious over time. During palliative illness there was no difference in the proportion of caregivers and patients with depression (10.7 vs. 11.9%, respectively). However, between palliative and terminal illness there were both more depressed caregivers (8.7 vs. 30.4%; p < 0.03) and more depressed patients (20.0 vs 40.0%; p ¼ 0.04). These results point to the importance of ongoing assessment of both caregivers and patients for psychological morbidity. Interventions aimed at caregivers to manage both anxiety and depression may prevent psychological morbidity during patient illness and bereavement. # 1219/THE QUALITY OF LIFE OF PARTNERS OF PATIENTS WITH HIGH-GRADE GLIOMA: RESULTS OF A PROSPECTIVE, LONGITUDINAL STUDY Martin J. Muller, Division of Psychosocial Research & Epidemiology, The Netherlands Cancer Institute; Martin Klein, Department of Medical Psychology, Vrije Universiteit Medical Center; Neil K. Aaronson, Division of Psychosocial Research & Epidemiology, The Netherlands Cancer Institute, Amsterdam, The Netherlands Little attention has been paid in the research literature to the healthrelated quality of life (HRQL) of patients with cancer, in general, and brain cancer, in particular. Given the nature of the disease and its treatment, one might expect that the partners of these patients would experience significant problems in various HRQL domains. As part of a larger, prospective study, we examined the HRQL of 52 partners of patients with high-grade glioma. The partners’ HRQL was assessed with the SF-36 Health Survey: shortly after the patient had received the diagnosis and the tumor had been surgically resected, but before start of radiotherapy; after completion of radiotherapy (approximately 4 months following diagnosis), and at 4-monthly intervals thereafter. The partners’ SF-36 profiles were compared to those of an age- and gender-matched normative sample drawn from the Dutch general population. At baseline, the partners exhibited significantly lower mean scores than the normative comparison group on the role-physical, vitality, social functioning, role-emotional and mental health scales, and the mental component score of the SF-36. Effect sizes ranged from 0.66 to 1.5. During the follow-up period, the observed differences in SF-36 mean scores between the partners and the comparison group were smaller, but remained statistically significant, with effect sizes of approximately one-half standard deviation. The partners’ self-reported HRQL was not associated significantly with the patients’ clinical or neurocognitive status, but was with the patients’ self-reported HRQL, and particularly his/her mental component scores. In conclusion, the HRQL of the partners’ of patients with high-grade glioma is significantly impaired around the time of diagnosis and initial treatment. With the passage of time, the partners’ HRQL profiles improve, but do not return to the levels typical of the general population. These results suggest the need to offer psychosocial support services to the partners of patients with brain tumors throughout the course of treatment and follow-up.

763 # 1222/HEALTH RELATED QUALITY OF LIFE (HRQOL) AND BURDEN OF FAMILY CAREGIVERS OF DIALYSIS PATIENTS Fernando Alvarez-Ude, Medicina Interna, Hospital General De Segovia, Segovia; Covadonga Valdes, Nefrology I, Hospital Central Asturias, Oviedo; Carmen Estebanez, Medicina Interna, Hospital General Segovia, Segovia; Pablo Rebollo, Nefrology I, Hospital Central Asturias, Oviedo, Spain Some studies have shown a decrease of the HRQoL of caregivers of chronic patients. The aim of this study was to evaluate the HRQoL and burden of family caregivers of dialysis patients, and to analyze which variables were associated to it. A sample of 221 pairs of patient-carer, stratified by age and gender, was randomly selected out of 14 dialysis units: 152 patients were on hemodialysis and 69 on peritoneal dialysis. Patients and carers answered the SF-36, obtaining a Physical (PCS) and a Mental (MCS) Component Summary scores standardized by age and gender, and the Duke-UNK Functional Social Support (FSS). Carers also answered the scale Caregiver Burden Interview of Zarit (ZS). Mean PCS and MCS scores of carers were 48.4 ± 13.8 and 48.0 ± 11.3, respectively. Multiple regression analysis showed that the variables associated to lower PCS of the carer were: higher ZS and older patient age R2 ¼ 0.15; p < 0.001).Variables associated to lower MCS were: higher ZS and lower FSS of the carer, and lower MCS of the patient (R2 ¼ 0.29; p < 0.001). Variables associated to a higher ZS of carers were: lower FSS and lower PCS and MCS scores of the career and higher age and lower PCS and MCS scores of the patient (R2 ¼ 0.49; p < 0.001). Carers with a MCS O42 points (cut off point associated with depression) were 28.3% (95%CI ¼ 22.4–34.8). Logistic regression analysis showed that variables associated to having a MCS O 42 points were: higher ZS and lower FSS of carer. So, (1) HRQoL of caregivers of dialysis patients is slightly worse than that of the general population of the same age and gender. (2) Physical health status is more damaged in those caregivers suffering higher burden and caring of older patients. Mental health status is more damaged in those suffering higher burden, feeling lower social support and caring of patients with worse mental health status. (3) The burden experienced by family carers depends on perceived social support, age of patient and physical and mental health status of carer and patient. (4) A significant percent of carers of dialysis patients have depression which is associated to higher burden and lower social support perceived. # 1645/HEALTH RELATED QUALITY OF LIFE DURING THE BEREAVEMENT PERIOD OF CARERS OF A DECEASED ELDERLY Pablo Rebollo, Health Outcomes Research Unit, Hospital central de Asturias, Oviedo; Jordi Alonso, Health services Research Unit, Institut Municipal d’Investigacio´ Me`dica (IMIM); Isabel Ramon, Vic, Spain, Gemma Vilagut, Rosalı´a Santed, Health services Research Unit, Institut Municipal d’Investigacio´ Me`dica (IMIM); Ramon Pujol, Internal Medicine Service, Hospital General Universitario de Bellvitge, Barcelona, Spain Few empirical studies have addressed the health-related quality of life (HRQL) impact of bereavement. The objective of this study was to evaluate the HRQL of the main carer after the death of a close senior person. The final 142 carers participating in the study ‘Dying elderly in Catalonia’ were invited to participate in a phone interview with the SF36. General population SF-36 norms were used to estimate gender and age standardised PCS and MCS scores. A MCS below 43 was

used as an indicator of possible depression. Bivariate and multivariate (logistic regression) analyses were performed. Additionally, databases of several patient studies were used to facilitate interpretation of results. A total of 130 carers (91.5%) participated, the majority being women (80%), middle age (56%), and the spouse (22%) or son/ daughter (56%) of the deceased person. The interview was carried out a median of 6 months after the death. Carers’ SF-36 dimension scores were lower than the expected for Mental health, Role emotional and Social functioning. Their MCS was below the expected in 0.6 SD units, whereas the PCS was above the expected by 0.2 SD The proportion of individuals with a possible depression was 33.8%. A first-degree family relationship with the deceased (OR ¼ 4.5), and a lower age of the deceased (OR ¼ 3.8) were significantly associated (p < 0.02) to a possible depression, while a lower level of education showed a marginal higher risk of depression (OR ¼ 2.8, p ¼ 0.05). The recent death of a senior is associated with a poor emotional health of the carer. Although previous emotional was unknown, results indicate the need to research whether the identified factors could flag the search of an effective intervention. Funded by: FIS (99/1109) and CIRIT (2001SGR00405). # 1202/EMOTION, PAIN AND OVERALL HRQL DEFICITS ASSOCIATED WITH COMORBIDITIES IN TYPE 2 DIABETES Sheri L. Maddigan & Jeffrey A. Johnson, ACHORD, Institute of Health Economics, Edmonton, AB, Canada The purpose of this analysis was to assess the burden of comorbities (heart disease, depression, and osteoarthritis) on emotion, pain and overall health-related quality of life (HRQL) in type 2 diabetes. The study sample included 4462 respondents identified as having type 2 diabetes from the 2001 cycle of the Canadian Community Health Survey, a cross-sectional survey of noninstitutionalized Canadians aged 12 years and older. The Health Utilities Index Mark 3 (HUI3) was used to assess pain and discomfort, emotion and overall HRQL deficits. Weighted regression analyses were used to control for the effects of sociodemographic characteristics on HRQL. Utility scores were compared between individuals with no comorbidities, single comorbities and all possible combinations of the three identified comorbidities. The average overall HUI3 score for individuals with type 2 diabetes without comorbid heart disease, depression or osteoarthritis (OA) was 0.82 (95% CI: 0.80–0.84), while average utility scores on the emotion and pain attributes were 0.96 (95% CI: 0.95–0.97) and 0.92 (95% CI: 0.89–0.94). The presence of any single comorbidity or combination of comorbiditities was associated with statistically signficant and clinically important deficits in overall HUI3 scores relative to individuals without any of the three comorbid conditions. The largest overall burden was seen in individuals with the combinations of OA and depression (mean HUI3 score ¼ 0.37, 95% CI: 0.31–0.44) and heart disease and depression (mean HUI3 score ¼ 0.43, 95% CI: 0.36–0.49). Relative to individuals without comorbidities, the largest pain burden was seen in individuals with OA and depression (mean difference ¼ 0.52, p < 0.001), heart disease and OA (mean difference ¼ 0.31, p < 0.001) and heart disease and depression (mean difference ¼ 0.29, p < 0.001). Heart disease or OA combined with depression was associated with the largest emotional burden. These analyses highlight the contribution of comorbidities to the disease burden of type 2 diabetes and illustrate the importance of managing such comorbities in patients with the disease.

764 # 1446/EXPLORING THE PREDICTIVE FACTORS OF HRQOL & RELATED CHANGES FOR THE ELDERLY IN VETERAN HOMES Kuan-Lang Lai, Institute of Life Science; Yun-Chang Wang & MingHao Kuo, School of Public Health, National Defense Medical Center; Herng-Ching Lin, School of Health Care Administration, Taipei Medical University; Senyeong Kao, School of Public Health, National Defense Medical Center, Taipei, Taiwan Objective: The purpose of this study was to explore the predictive factors of health-related quality of life (HRQOL) and related changes for the aging elderly. Method: The Taiwan’s abbreviated version of the World Health Organization Quality of Life (WHOQOL-BREF) was administered at baseline and longitudinally followed-up 1 year later on 692 male elders who lived in five veteran homes in Taiwan. The data on demographics, socio-economics, disease status, health resources access, and health behaviors were collected. A stepwise multiple regression was performed to explore the possible predictive factors. Results: Mean age of the 692 participants was 77.8 years. Stepwise multiple regression indicated that regular exercise, self-rating health status, number of outpatient visits, and happiness were predictive factors of the physical domain (R2 ¼ 0.46). Happiness, self-rating health status, leisure activity, and regular exercise were predictive factors of the psychological domain (R2 ¼ 0.42). Happiness, primary caregiver, and hospitalization days were predictors of the social relationship domain (R2 ¼ 0.28). Dierent veteran homes, income, disability, leisure activity, and self-rating health status were predictive factors of the environment domain (R2 ¼ 0.22). A total of 472 eligible elders (68.2%) were successfully followed for 1 year after eliminating those who were dead, hospitalized, emigrated, or lost contacts. There was dramatic change in the score on physical domain (p < 0.0001), with a decrease of 4.7% in average. In addition, there was also a significant decrease in the score on environment domain (p < 0.005), but the scores on psychological and social relationships domains did not show significant changes. The R2 of predictive factors for score change over 1 year across domains is between 0.17 and 0.06. Conclusion: The results suggest that several factors can be used to predict the HRQOL on dierent domains for aging elderly in veteran homes. But longer follow-up period is needed to further explore the predictive factors of the related HRQOL changes. # 1741/SELF-EFFICACY AS A PREDICTOR OF HEALTH-RELATED QUALITY OF LIFE IN MULTIPLE SCLEROSIS Afsane Riazi & Alan J. Thompson, Neurological Outcome Measures Unit, Institute of Neurology, London; Jeremy C. Hobart, Peninsula Medical School, Derriford Hospital, Plymouth, UK Certain psychosocial variables are relatively unexplored as possible predictors of health-related quality of life (HRQoL) in multiple sclerosis (MS). Self-efficacy is one such variable. If self-efficacy is a predictor of HRQoL, it may be an important area to target in clinical practice. This study examined whether changes in multiple sclerosis self-efficacy predict changes in HRQoL. Eighty-nine people with multiple sclerosis completed the Multiple Sclerosis Self-efficacy Scale (MSSE function and control scales), the Multiple Sclerosis Impact Scale (MSIS-29), the Multiple Sclerosis Walking Scale (MSWS-12) on admission to an inpatient rehabilitation unit (n ¼ 43) or steroid treatment for relapses (n ¼ 46). The measures were again completed on discharge (rehabilitation group) or 6 weeks later (steroid group). Multiple regression analyses were used to determine whether changes in self-efficacy predict changes in HRQoL. Changes in MSSE function scores were independently associated with changes in MSIS-29 physical (Beta ¼ )0.42; t ¼ 4.1; p < 0.001), MSIS-29 psychological (Beta ¼ )0.28; t ¼ )2.9; p < 0.01), and MSWS-12 scores (Beta ¼ )0.50; t ¼ )4.3; p < 0.001). Changes in MSSE control scores were independently associated with changes in MSIS-29 physical (Beta ¼ )0.49; t ¼ )5.1; p < 0.001), MSIS-29 psychological (Beta ¼ )0.50; t ¼ )5.9; p < 0.001), and MSWS-12 scores (Beta ¼ )0.40; t ¼ )3.4; p < 0.01). That is, increases in self-efficacy scores from baseline to follow-up, were significantly associated with decreases (improvement) in physical, psychological and walking scale scores. This relatively unexplored patient-centered variable in multiple sclerosis is potentially modifiable and may offer new ways to intervene to improve HRQoL in MS.

# 1336/SELF-REPORTED SYMPTOM TREND IN HIV-INFECTED PEOPLE TAKING HAART Rita Murri & Massimo Fantoni, Infectious Diseases; Cosmo Del Borgo, Infectious Diseases, INMI L. Spallanzani; Gilda Deiana, Infectious Diseases, Catholic University of Rome, Rome, Italy Objective: To assess predictors of change over time in the symptom score in HIV+ pts taking highly active antiretroviral therapy (HAART). Methods: Prospective, multicenter, cohort study on Italian HIV+ pts at any stage of disease. Consecutive patients were enrolled in seven Italian Centers. The MOS-HIV was used as tool of the study. Two summary scores for physical (PHS) and mental health (MHS) were obtained. A list of 15 symptoms more often experienced by HIV-infected people taking therapies was added and a symptom score calculated. A single item on satisfaction with sexual life was also included into the questionnaire. Outcome of the study was considered the change in symptom score from the baseline to the fourth follow-up visit. A last observation carried forward approach was used; people with only the baseline evaluation were excluded from the analysis. Results: 585 patients were included into the study. Median of followup: 798 days. >35 years: 44%, females: 32%, intravenous drug users: 48%, CDC’s group C: 32%. At enrollment, median CD4+ cells count was 259 mm)3 and rate of people with undetectable HIV RNA was 32%. At fourth follow-up visit, 90% were taking HAART (61% with a protease inhibitor [PI]-containing regimen, 25% with a nonnucleoside reverse transcriptase inhibitor [NNRTI] and 14% with both). Symptom score increased from 7.6 (SD 6.1) of baseline to 8.3 (6.8) at fourth follow-up visit (p at paired t-test 0.02). Among the single symptoms, diarrea significantly increased over time (p ¼ 0.02). On multivariate linear regression analysis, factors associated with change in symptom score were baseline CD4 (Stand. beta coeff. 0.10; p 0.01), time on nelfinavir (Stand. beta coeff. 0.09; p 0.03), change in PHS (Stand. beta coeff. )0.24; p 0.001), change in MHS (Stand. beta coeff. )0.48; p < 0.0001), and symptom score at baseline (Stand. beta coeff. )0.63; p < 0.0001). Conclusions: Symptom score was influenced by HAART, especially protease inhibitors, and was strongly correlated to both physical and mental health. # 3000/GENDER DIFFERENCES IN HEALTH RELATED QUALITY OF LIFE IN CHILDREN AND ADOLESCENTS. U. von-Rueden, J. Bruil, S.B. Detmar & U. Ravens-Sieberer, Child and Adolescent Health, Robert Koch-Institute, Berlin, Germany Gender plays a major role in subjective health assessment. Differences in health related quality of life (HrQoL) in adults have been documented in various studies assessed with different measurements (e.g. WHOQOL, SF36). In general, women report lower HrQoL scores in physical well-being, emotional well-being and self-esteem than men. The present study examines HrQoL data from the pilot test of the European KIDSCREEN project (Screening for and Promotion of Health related Quality of Life in Children and Adolescents: A European Public Health Perspective). In total 1001 children (50% girls and 50% boys) between 8 and 11 years of age and 1441 adolescents (53.1% female and 46.1% male adolescents) between 12 and 18 years of age from seven countries (A, CH, D, E, F, GB, NL) responded to the generic self-administered KIDSCREEN HrQoL questionnaire. Comparisons between girls and boys showed that younger boys scored significantly higher (p < 0.001) on the HrQoL subscale Self Perception (Body Image and Self-esteem), whereas girls displayed higher scores on Cognition and School Functioning (p < 0.001). In the adolescents group similar and additional effects were found. Male adolescents showed higher HrQoL scores in Physical Well-being (p < 0.001) and Psychological Well-being (Positive (p ¼ 0.005) and Negative emotions (p < 0.001)), whereby female adolescents reported a better HrQoL concerning Social Support and Peer Relations (p ¼ 0.008). Gender differences are increasing with age. The findings indicate that gender is an important determinant for children’s and adolescent’s HrQoL even at young age. Possible reasons and underlying mechanisms need to be further addressed to understand gender specific health perception according to age.

765 # 1119/YOU KNOW IT’S THERE BUT HOW DO YOU CAPTURE IT? CHALLENGES FOR THE NEXT PHASE OF RESPONSE SHIFT RESEARCH Carolyn E. Schwartz, Behavioral Science Research, QualityMetric Incorporated, Concord, MA, USA; Mirjam Sprangers, Medical Psychology, Academic Medical Center, Amsterdam Zuidoost, The Netherlands There is an increasing recognition that when people experience changes in health, they may change their internal standards, their values, and/or their conceptualization of health-related quality of life (HRQOL). These ‘response shifts’ make comparisons over time difficult to interpret, but could be used to provide substantial clues about adaptation and the tolerability of treatments. The purpose of this presentation is to discuss the primary theoretical, definitional, and methodological challenges for the next phase of response shift research. Using empirical examples where possible, we will discuss how implicit theories may confound response shifts in prospective and retrospective assessments, by discussing expectations, denial, impression management, effort justification, cognitive dissonance responding, and recall bias. We will describe six dimensions of definitional confusion that may lead to miscommunication about response shift, including bias vs. meaningful change, measurement vs. subject characteristics, explanation vs. phenomenon, temporary vs. long-lasting, event vs. passage of time, unrelated to health vs. related to health. Finally, we will highlight theoretical (e.g., characterizing a catalyst) and methodological (e.g., psychometrics of the then-test, IRT/SEM) topics in need of attention for the continued elucidation of response shift effects in HRQOL research. # 1186/ASSESSING RESPONSE SHIFT IN QL: THE THENTEST AND A STRUCTURAL EQUATION MODELING APPROACH COMPARED Mechteld R. Visser, Frans J. Oort & Mirjam Sprangers, Medical Psychology, Academic Medical Centre, Amsterdam, The Netherlands Response shift is a challenge for measurement and interpretation of QL-changes. It is defined as a change in the meaning of one’s selfevaluation of QL as a result of a change in internal standards (recalibration), values (reprioritization) or reconceptualization (Sprangers and Schwartz, 1999). Recalibration is commonly assessed with a baseline follow-up design extended with a retrospective baseline assessment (thentest) at follow-up. The difference between baseline and thentest scores is indicative of recalibration. On the basis of baseline and follow-up data only, Oort’s structural equation modeling approach (SEM, 2003) assesses recalibration, reprioritization and reconceptualization, operationalized by a change in the means of specific factors, values and patterns of common factor loadings, respectively. To compare the thentest and the SEM approach in detecting response shift, we assessed the QL (SF-36) of 170 cancer patients prior to (baseline) and 3 months after surgery (follow-up). At follow-up a thentest version of the SF-36 was also administered. Baseline followup differences and baseline thentest differences (recalibration) were analysed with t-tests. Baseline follow-up differences showed significantly (p < 0.01) improved Mental Health (MH), and Role limitations due to Emotional problems (RE), no changes in General Health (GH) and Social Functioning (SF), and deterioration in Physical Functioning (PF), Role limitations due to Physical problems (RP), Vitality (VT) and Bodily Pain (BP). Both the thentest and SEM found recalibration on BP and RP, although the RP recalibration effect on the thentest did not reach conventional levels of significance (p ¼ 0.08). With the thentest an additional recalibration effect on GH was found, whereas SEM indicated reconceptualization of GH. Notably, these effects were in opposite direction. This first comparison of the thentest and SEM showed agreement about the (non)occurrence of recalibration response shift on seven of the eight SF-36 scale. # 1345/USING SEIQOL-DW FOR MEASURING RESPONSE SHIFT IN QUALITY OF LIFE IN PALLIATIVE TREATMENT OF SMALL CELL LUNG CANCER PATIENTS Marjan J. Westerman, Social Medicine, Institute for Research in Extramural Medicine, VU Un, Amsterdam, The Netherlands

M J. Westerman, T. Hak, B.A.M. The, H. Groen, G. van der Wal, VU University Medical Center, Institute for Research in Extramural Medicine, Department of Social Medicine, Amsterdam, The Netherlands. Quality of life (QoL) is considered a necessary end point in clinical trials in cancer patients. Expected deterioration in QOL often does not occur even in cases of serious illness such as small cell lung cancer. These contradictory findings highlight the importance of studying response shift in QOL. In our study ‘Response shift in Quality of Life in palliative treatment of small cell lung cancer patients’, a qualitative exploratory longitudinal multiple case-study with 30 patients (cases), each patient is interviewed at equivalent ‘assessment points’ in their illness trajectory. Part of this study is an exploration of SEIQoL-DW as instrument for measuring response shift. The SEIQOL-DW is a measure of QOL that seek to incorporate the unique views of the individual. SEIQoL-DW allows respondents to ‘define’ QoL by nominating the five most important aspects of their lives at that moment (‘elicited cues’) and by indicating the relative contribution (‘weight’) of each cue to their overall QoL. SEIQoL methodology entails discussing the nominated cues with the respondents in order to have a good understanding of the life domain and/or if needed to make the answers more, less detailed/abstract. We reported (ISOQOL 2001, Amsterdam) that this process is vulnerable for interviewer bias, which is a complication that arise in the measurement of response shift with SEIQOL-DW. This presentation will demonstrate SEIQOL-DW data of patients which are interviewed three to four times during first line chemotherapy. Findings from our data suggest that reconceptualization as indicated by cue changes and changes in values as indicated by different weighting do occur as a result of changes in health status. Respondents, aware of there short life-expectancy, are more likely to name and value social and emotional issues rather than physical or functional. # 1406/INDIVIDUALIZED HEALTH-RELATED QUALITY OF LIFE (HRQL) POST STROKE: EVALUATION RESPONSE SHIFT Sara Ahmed, Epidemiology and Biostatistics; Nancy Mayo, School of Physical and Occupational Therapy; James Hanley, Epidemiology and Biostatistics; Sharon Wood-Dauphinee, School of Physical and Occupational Therapy; Robin Cohen, Oncology and Medicine, McGill University, Montreal, Canada Individualized measures of HRQL allow persons to define the areas that are most relevant for their own quality of life and may be useful for evaluating response shift by examining changes in the domains (reconceptualization) selected, and in their corresponding weights (change in values). Individualized measures may be most valuable for nonhealthy individuals who are likely to experience a response shift; but the use of this type of instrument may be compromised by the practical difficulties respondents face when completing such a measure. The objective of this study was to assess reconceptualization of HRQL, and change of individual values among individuals with stroke during the first 6 months of recovery. Persons were recruited through a randomized trial of acute post-stroke care. Individualized HRQL was evaluated at 6 and 24 weeks post-stroke using the Patient Generated Index (PGI). At 24 weeks a semi-structured interview was administered to assess whether verbalizations given by subjects indicated that they had experienced a response shift. From a possible 167 subjects, 92 had complete PGI information at the 6- and 24-week evaluations, and of these, 47 completed the semi-structured interview. Between the 6- and 24-week evaluations, the domains selected were: exactly the same for 10 (11%) subjects, reduced for 27 (29%), expanded for 11 (12%), and completely different for 44 (48%) subjects. For domains that remained important by the 24-week evaluation, there were significant changes in weights. The average change in domains and weights was not significantly associated with change in physical function. The semi-structured interviews indicated a response shift for 14 (30%) subjects. Fifteen(31%) subjects who completed the interviews had changes in domains and weights, but also reported improved ability to perform the domains initially selected. The feasibility of using an individualized measure among individuals with stroke is limited by the added complexity of completing and interpreting such a measure.

766 # 1408/I FEEL GOOD ABOUT THINGS: POSITIVE QUALITY OF LIFE OUTCOMES FOLLOWING A STROKE Deborah Buck & Ann Jacoby, Primary Care, University of Liverpool, Liverpool, Merseyside, UK Society perceives people with disabilities as enjoying a less fruitful existence than the population as a whole. The general public and health professionals often possess negative attitudes and expectations concerning people with disabilities. Studies show that health professionals underestimate perceived health status, psychological well-being and quality of life (QOL) of patients. However, some who experience disability or illness first-hand perceive a good QOL against all odds. This apparent discrepancy has been termed the disability paradox which highlights the importance of the personal experience of disability in shaping one’s self-identity and view of the world. This presentation will draw on qualitative data from a study to develop a stroke-specific QOL measure. This enabled scrutiny of the apparent anomaly between presence of disability and/or chronic illness and perceptions of a good QOL. In-depth interviews were conducted with 14 male and 14 female stroke survivors. Age range: 27–86 years. Time since discharge from hospital: range 1–15 months. Level of disability at discharge: range none to severe (measured by the Barthel Index, range 3–20). Analysis was based on the Framework method. The qualitative data presented here focus on favourable outcomes following stroke as revealed by participants. Examples of positive underlying attitudes are provided in an attempt to gain greater understanding of the lived experience of illness. Positive aspects included: perceived transformations in self-identity; a more positive outlook on life; greater appreciation of life; enjoying new activities or roles; improved personal relationships. Themes within underlying attitudes included: coming to terms; motivation; determination; comparison with other people; positive thinking. Survivors used a variety of strategies to find meaning and recognise something positive that had come from their illness experience. Underlying psychological disposition is likely to shape the way and degree to which people come to terms with illness or disability. Response shift may also account for the apparant paradox of good QOL in illness. # 1507/A COMPARISON OF FOUR METHODS OF SAMPLE SIZE ESTIMATION, USING THE SF-36 Michael J. Campbell & Stephen J. Walters, School of Health and Related Research, University of Sheffield, Sheffield, S. Yorkshire, United Kindgom We describe and compare four different methods for estimating sample size and power, when the primary outcome of the study is a health related quality of life (HRQoL) measure. These methods are: 1. assuming a Normal distribution and comparing two means; 2. using a non-parametric method; 3. Whitehead’s method based on the proportional odds model (Whitehead 1993); 4. the bootstrap. We illustrate the various methods, using data from the SF-36. The results show that if the HRQoL outcome has a limited number of discrete values, say less than 7, such as the ‘role limitation’ domains or if the proportion of cases at the boundaries is high (scoring 0 or 100), then we would recommend using Whitehead’s method. Alternatively, if the HRQoL outcome has a large number of distinct values and the proportion at the boundaries is low, then we would recommend using method 1. If a pilot or historical dataset is readily available (to estimate the shape of the distribution) then bootstrap simulation (Method 4) based on this data may provide a more accurate and reliable sample size estimate than method 1 or 2. In the absence of a reliable pilot set, bootstrapping is not appropriate and conventional methods (1 or 2) of sample size estimation or simulation will need to be used. Fortunately with the increasing use of HRQoL outcomes in research, historical datasets are becoming more readily available. Strictly speaking our results and conclusions only apply to the SF-36 outcome measure. Further empirical work is required to see whether or not these results hold true for other HRQoL outcomes. However, the SF-36 has many features in common with other HRQoL outcomes, such as the NHP and QLQC30, for example, multi-dimensional, ordinal or discrete response categories with upper and lower bounds, and skewed distributions. Therefore we believe these results and conclusions using the SF-36 may be appropriate for other HRQoL measures. Whitehead J. Sample

size calculations for ordered categorical data. Statist Med 1993; 12: 2257–2271. # 1325/RESAMPLING PROCEDURES TO ADJUST FOR MULTIPLE ENDPOINTS IN QUALITY OF LIFE STUDIES Diane L. Fairclough, Colorado Health Outcomes Center, University of Colorado Health Sciences Center; Pamela Wolfe, Wolfe Statistical Consulting, LLC, Denver, CO, USA One of the salient features of quality of life (QOL) data is the multiplicity of domains, or outcomes, produced by any but the most simplistic instruments. While it may be evident a priori which restricted subset is most important in given clinical trial, this is not always the case. Regardless, there will eventually be an interest in assessing all the QOL domains. Because the domains in the QOL instruments are correlated to some extent, multiple comparison procedures that assume independent hypotheses will give overly conservative results. The problem is compounded in studies with three or more treatment arms. Resampling methods are generally more robust, since they incorporate distributional characteristics of the data, and since the dependence structure between the tests is incorporated in the resampling-based estimates, there is a demonstrable gain in power. Although these methods are well-researched, their application to QOL data is rare. We apply a well-known step-down procedure to a series of clinical trials with both multiple QOL domains and multiple comparisons within each domain, and contrast the results to a step-down bootstrap procedure with 10,000 resampled datasets. In particular, the power for each approach is estimated for both a restricted subset of hypotheses and a series of broader hypotheses. # 1234/WHAT IS THE ADDED CLINICAL VALUE OF HEALTH RELATED QUALITY OF LIFE (HRQOL) STUDIES IN RANDOMISED CONTROLLED TRIALS (RCTS)? RESULTS FROM MORE THAN 100 TRIALS Andrew Bottomley, Fabio Efficace & Veerle Vanvoorden, Quality of Life Unit, EORTC Data Center, Brussels, Belgium; Galina Velikova, Cancer Medical Research Unit, Cancer Research UK Clinical Centre, Leeds, United Kingdom; Eva Greimel & Karl Franzens Universitaet, Obstetrics and Gyneacology, Graz, Austria; George Van Andel & Onze Lieve Vrouwe Gasthuis, Department of Urology, Amsterdam, The Netherlands In the last decade HRQOL has gained more acceptance as an outcome in oncology. However, critics have questioned the added value of HRQOL data from oncology clinical trials. We therefore conducted a series of systematic reviews in selected disease sites and report here an overview of the added clinical insights from over two decades of HRQOL RCTs. Systematic reviews were performed for six disease sites. Articles were selected from 1980 to 2002 according to predefined eligibility criteria; e.g. only patient assessed HRQOL, adult patients undergoing any medical treatment. Articles were selected from MEDLINE, Cochrane library and Cancerlit. Eligible RCTs were independently assessed by three reviewers on 30 predefined criteria. Provisional pooled findings identify 124 RCTs with 36,705 patients. Five trials examined brain or ovarian cancer, 26 were identified for advanced breast cancer, 29 for NSCLC, 25 for prostate and 34 for colorectal cancer. HRQOL was a secondary endpoint in most studies. Some 107 (86%) trials tested for survival differences, which were statistically significant in 30 (28%) studies, and significant differences in HRQOL were seen in 17 (57%) of these. In RCTs where no significant differences in survival were shown (n ¼ 77, 72%), HRQOL data provided additional significant results in 40 (52%) trials. In total about 20% of the trials (mostly the more recent ones) reported clinical significance. A few trials failed to test for HRQOL differences (n ¼ 13) because of problems of poor compliance or no a priori analysis intentions. In advanced breast cancer HRQOL studies contributed to clinical decision-making in RCTs comparing different doses of hormonal or chemotherapy or different chemotherapy schedules. Overall, HRQOL data appears to add valuable information even when survival endpoints offer no additional insights into the clinical care of the patients. In summary, in carefully conducted RCTs HRQOL data can provide complementary information to understand treatment outcomes. This can supplement traditional clinical data such as survival.

767 # 1384/THE IMPACT ON CLINICAL DECISION MAKING OF RANDOMIZED CONTROLLED TRIALS INCLUDING HEALTH-RELATED QUALITY OF LIFE IN COLORECTAL CANCER. Fabio Efficace & Andrew Bottomley, Quality of Life Unit, EORTC Data Center, Brussels, Belgium; Jane M. Blazeby, Division of Surgery, Bristol University, Bristol, United Kingdom; Veerle Vanvoorden, Quality of Life Unit, EORTC Data Center, Brussels, Belgium Whilst the number of randomized controlled trials (RCTs) including health-related quality of life (HRQOL) in colorectal cancer is increasing, no systematic reviews exist of the evidence relating to the clinical impact of these studies. Hence the aim of this systematic review was to evaluate whether the inclusion of HRQOL, as a part of the trial design, has supported decision-making for planning future medical treatment in colorectal cancer. A Literature search from 1980 to March 2003 was performed to identify RCTs of colorectal cancer patients who had undergone a HRQOL assessment. Articles were identified mainly by MedLine, the Cochrane Library and the online National Cancer Institute database. All studies enrolling at least 50 patients and using a HRQOL patient self-reported measure were included. Both HRQOL and clinical reported outcomes were systematically analysed to evaluate their consistency and their relevance for supporting decisionmaking to aid in the planning of future medical treatment. The ‘minimum standard checklist for evaluating HRQOL outcomes in cancer clinical trials’ was used to evaluate the robustness of the reported outcomes. A total of 34 RCTs enrolling 10,180 patients were identified. The majority of studies (74%) examined metastatic patients. 26% of the RCTs examined HRQOL as a primary endpoint. 35% of the studies (testing for survival difference) showed difference between treatment arms whereas 55% showed a difference in terms of HRQOL outcomes. Overall 26% of the RCTs reported high quality HRQOL outcomes providing a comprehensive picture of the cancer treatment. These studies were also a valuable source of information supporting clinical decision-making and helping in planning future phase III RCTs. Yet recent studies were better designed and provided robust outcomes. In summary, bridging the gap between research and clinical practice is the next challenge that HRQOL science has to face, and this is likely to be achieved by robust outcomes from RCTs, such have been seen in the last 5 years. # 1736/INFLUENCE OF MEASURING SYSTEMATICALLY QUALITY OF LIFE ON PATIENTS OUTCOMES Sophie Moser, Epide´miologie et e´valuation cliniques (EA 3444), CHU, Nancy; Fabienne Empereur, DIM, CHR Metz-Thionville, Metz; Emmanuel Desandes, DIM, CAV, Vandoeuvre-les-Nancy; Phi Linh Nguyen Thi, Sylvie Klein, Francis Guillemin & Serge Briancon, Epide´miologie et e´valuation cliniques (EA 3444), CHU, Nancy, France Purpose: Despite potential benefits to the patient, measures of health related to quality of life are not been routinely used in the daily care of individuals. The objective of this study was to examine the effect of measuring systematically quality of life (QoL) in clinical decisionmaking and in patient QoL after discharge. Methods: The study sample includes 2704 patients admitted in the Nancy University Hospital, in short stay services. A first randomisation assigned one patient out of five in a control group having not QoL measures at admission. The intervention group received QoL measures, ‘four generic questionnaires (SF36, Duke, Euroquol and GHQ-12) and one or two specific questionnaires’. After the assessment, a second ran-

domisation, with a 2 * 2 factorial design, affected patients to four groups: returning results to physician, returning results to nurses, returning results to both or not returning results. Main outcomes was health related QoL as measured by SF36, 15 days after discharge. All practices of care management during the stay were collected. Analysis of variance for repeated measures, adjusting for age and sex was used. Results: 2035 males and 1368 females participated. Mean age was 55.5 ± 14.5 years. The response rate was 74.8%. There was no differences of QoL between intervention group with not returning results to the care staff and control group. In intervention groups, mental functioning score increased, pain score was stable, and physical functioning scores decreased. These differences were not statistically significant. General health perception scores decreased (p < 0.05). This decrease was lowest in the physician group, but not in the physician/nurses group (interaction term: p < 0.05). Nevertheless, there was no difference in care management between groups. Measuring QoL at the beginning of hospitalisation did not lead to show an important improvement of quality of life 15 days after discharge, whatever the way of returning the results to the professional staff. # 1145/DO THE EFFECTS OF QUALITY IMPROVEMENT FOR DEPRESSION DIFFER FOR MEN AND WOMEN? Cathy D. Sherbourne, Health, RAND, Santa Monica; Robert Weiss, Department of Biostatistics, Naihua Duan, Center for Community Health, UCLA, Los Angeles; Chloe Bird & Kenneth Wells, Health, RAND, Santa Monica, CA, USA Purpose: Landmark studies demonstrate that quality improvement (QI) interventions designed to increase treatment to depressed primary care patients improve short-term outcomes and some longer-term HRQOL outcomes. However, before committing substantial resources to disseminating QI interventions, policymakers need to know whether the intervention prototypes successfully reduce existing gender disparities or inadvertently exacerbate them. This analysis examined whether QI programs for depression in primary care equally benefitted men and women. Sample and Methods: 1299 primary care patients sampled from clinics in managed care organizations screened positive for depression and enrolled in a 2-year group-level randomized controlled trial. Clinics were randomly assigned to usual care (UC) or one of two QI programs providing nurse assessment and patient education, and resources to support medication management (QI-Meds) or psychotherapy (QI-Therapy). We conducted intent-to-treat analysis for longitudinal models of repeated outcomes. We examined over 30 different covariance structures within patients to determine the one that best fit the data. Results and Conclusions: We found gender differences in the effect of QI for some outcomes (receipt of appropriate care, mental HRQOL, and employment status) but not for others (receipt of any depression care, probable clinical depression). For some outcomes, the intervention that provided extra resources to encourage psychotherapy was particularly effective for men. For example, men in QI-Therapy showed an early 6-month improvement in mental HRQOL, relative to UC, which was sustained through 18 months. This was not true for men in QI-Meds. For women, both QI-Therapy and QI-Meds improved mental HRQOL relative to UC, but the beneficial effect of QITherapy was delayed until 12 months. These and other results suggest that the QI interventions did address initial gender disparities that were found for men and women under usual care conditions. Results provide some guidance for how future QI interventions for depression should be refined.

768 # 1691/CAN FAMILY PSYCHOEDUCATION IMPROVE QUALITY OF LIFE OF RELATIVES OF SCHIZOPHRENIC PATIENTS? Lucie Motlova, School of Public Health, UC Berkeley, Berkeley, CA, USA; Eva Dragomirecka, Department of Social Psychiatry, Prague Psychiatric Center, Praha 8 – Bohnice, Czech Republic; Dagmar Dzurova, School of Public Health, UC Berkeley, Berkeley, CA, USA; Filip Spaniel, Eva Goppoldova & Pavla Selepova, Praha 8 – Bohnice, Czech Republic Family psychoeducation in schizophrenia is an evidence-based psychosocial intervention. It is designed to reduce family burden and distress by delivering relevant information and improving patient functioning, family coping and social networking. The goal of the present study was to explore the influence of family psychoeducation on quality of life of the relatives of schizophrenic patients and to compare their quality of life with a control population. The group psychoeducation program included 51 patient relatives (age mean 49.4). The Subjective Quality of Life Analysis (SQUALA) questionnaire – a self-administered questionnaire that includes 23 areas of life covering health status, everyday activities, social interactions and inner reality – was administered before the program and 3 months later. The relatives’ QOL scores before and after the program were compared with the control group (case–control design). Comparing recipients with non-recipients: There were no significant differences between relatives receiving psychoeducation and the control group in the overall QOL score, but the QOL profiles were different. The relatives attach more importance to Hobbies and Beauty and arts domains, while the controls value Money more. The relatives are less satisfied than controls with Mental wellbeing, Family and Children domains. Comparing recipients before and after the program: The QOL score of the relatives who entered the program with low quality of life improved significantly after the program (p ¼ 0.000). Male relatives (n ¼ 23) have significantly worse quality of life before the program than women (n ¼ 28) (p ¼ 0.008). The QOL improved in men significantly after the program in several domains: Health, Mental well-being and Safety, while in female relatives QOL improved only in Resting domain. The results showed that family psychoeducation proved to be beneficial to most participants in our study, particularly fathers and other male participants. Sources of funding: The research is supported by CNS LN00B122MSMT CR and Fogarthy Program D43 TW05810-01. # 1695/DOES A CUSTOMISED SELF-MANAGEMENT PROGRAMME HAVE ANY IMPACT ON THE QUALITY OF LIFE OF PEOPLE WITH ULCERATIVE COLITIS? Jan Lecouturier, On behalf of the CESP Research Team, University of Newcastle Upon Tyne, Newcastle, UK The aim of this research is to determine whether a self-management programme (SMP) improves quality of life, especially emotional wellbeing and social functioning, in people with ulcerative colitis (UC). Other studies have demonstrated the effectiveness of SMPs in improving psychosocial well-being and physical functioning (heart disease) and reducing hospital use (asthma). Further research was warranted to refine the methodology and role of SMPs and test whether the findings apply to other conditions. Other SMPs have been designed from a professional perspective but this programme was designed for the purposes of the study using a patient-centred approach to inform the content. This randomised controlled trial com-

pared the effects of a SMP (intervention) to the effects of a factual information package (control). Data were collected by postal questionnaires at four time-points, screening, baseline, 6 and 12 months. Outcome measures include: IBDQ, HADS and scales of self-management and self-efficacy. A sample of people with UC were sent a screening questionnaire containing the Inflammatory Bowel Disease Questionnaire (IBDQ). Those with a lower score on the social and emotional domains (indicating a poorer quality of life) of the IBDQ were invited to participate in the trial. Comparison of baseline and 6 month IBDQ scores revealed a significant improvement for intervention patients on 6 of the 32 items (frequency of frustration, depression, irritability, loose bowel movements, feeling bloated, feeling relaxed and problems with gas/wind). In contrast control patients had improved on one item (frequency of urgency). For intervention patients only there was an improvement in self-efficacy related to confidence in obtaining information from community resources and emotional support from others outside friends and family. While the majority of intervention patients still used self-management techniques introduced in the programme (e.g. guided imagery) at 6 months, we will report on the 12-month data and whether these techniques are still in use. # 1780/PROVIDING AUDIOTAPES OF PRIMARY ADJUVANT TREATMENT CONSULTATIONS TO MEN WITH PROSTATE CANCER: IMPACT ON QUALITY OF LIFE, MOOD STATE, RECALL, AND SATISFACTION Tom F. Hack, Faculty of Nursing, University of Manitoba, Winnipeg, Manitoba; Tom Pickles, Radiation Oncology, British Columbia Cancer Agency, Vancouver, BC; Barry D. Bultz, Psychosocial Resources, Dean Ruether, Medical Oncology, Tom Baker Cancer Centre, Calgary, AB; Lesley F. Degner, Faculty of Nursing, University of Manitoba, Winnipeg, Manitoba, CA, USA The purpose of this multi-site, randomized controlled trial was to examine the efficacy of providing men with prostate cancer with an audiotape of their primary adjuvant treatment consultation. A total of 425 men and 15 oncologists from cancer centres in three Canadian cities participated. The men were block randomized to 1 of 4 consultation groups: (1) Standard Care Control Group – Consultation not audiotaped; (2) Audiotaped – Patient not given audiotape; (3) Audiotaped – Patient given audiotape; and (4) Audiotaped – Patient offered choice of receiving the audiotape. Patients received the audiotape immediately after completing the post-consultation measures. Patients completed the Decisional Role Preferences Scale (Degner) before and after the consultation, a measure of communication satisfaction postconsultation, and measures of information recall, mood disturbance (Profile of Mood State), and breast-specific quality of life (FACT-B) at 3 months post-consultation. A feedback questionnaire was mailed to all participating oncologists. Results showed that patients who received the consultation audiotape reported having been provided with significantly more treatment and disease information than patients who did not receive the audiotape. The regression analysis showed that, among those who received the audiotape, patients who assumed an active role in decision making during the consultation had significantly higher prostate cancer-specific quality of life at 3-months post-consultation than passive patients. Patients rated the audiotape intervention positively, with an average score of 83.0 out of 100. Fifty-six percent of oncologists stated that either most or all patients should be provided with an audio-taped copy of their adjuvant consultations.

769 # 1813/A STRUCTURED MULTIDISCIPLINARY PSYCHOSOCIAL INTERVENTION IMPROVES THE QUALITY OF LIFE OF PATIENTS WITH ADVANCED CANCER Jeff A. Sloan, Health Sciences Research; Teresa Rummans & Matthew Clark, Psychiatry; Marlene Frost, Nursing; Pamela Atherton, Health Sciences Research; Michael Bostwick, Psychiatry; Gail Gamble, Rehabilitation; Mary Johnson, Chaplaincy; Jim Martenson, Oncology; Jarrett Richardson, Psychiatry; Mashele Huschka & Jean Hanson, Health Sciences Research, Mayo Clinic, Rochester, MN, USA Background: Patients with advanced cancer face multiple challenges to their quality of life (QOL). The present study tested a multidisciplinary intervention to improve QOL of patients with advanced cancer. Design: Eligible patients were within 12 months of diagnosis, were receiving radiotherapy, had no cognitive impairment nor an untreated psychiatric disorder. Patients were randomly assigned to either a structured eight-session multidisciplinary intervention over 4 weeks, or standard medical care. Each 90-min session was led by either a psychologist or psychiatrist, or co-led with a nurse, physical therapist, chaplain or social worker. Topics included physical therapy, information on cancer treatment, group support, spiritual reflection, relaxation training, and cognitive-behavioral therapy. QOL was assessed at baseline, 4, 8, and 27 weeks. The primary endpoint was overall QOL (Uniscale) at week 4. Results: 115 patients were enrolled from 10/02/ 2000 to 10/28/2002. Overall QOL at week 4 averaged 10 points higher in the intervention group than in the control group (80 vs. 70, p ¼ 0.047), an increase of 3% from baseline in the intervention group vs. a decrease of 9% in the control group (p ¼ 0.009). The number of patients reporting a clinically meaningful improvement of 10 points was three times higher in the intervention group (30 vs. 9%, p ¼ 0.004). Spiritual QOL at week 4 showed the largest improvement of all assessments (15% average difference between groups, p ¼ 0.003, 25 vs. 14% showing a 10% improvement, p ¼ 0.14). Changes in physical and emotional QOL single-item scores were superior in the intervention group (0 vs. )10%, p ¼ 0.02 and 3 vs. )5%, p ¼ 0.046). Differences were observed in POMS scores (84 vs. 81, p ¼ 0.01) and changes from baseline in Symptom Distress Scale scores ()4 vs. )9%, p ¼ 0.027). QOL differences did not sustain after 8 and 27 weeks from randomization. Conclusions: A structured multidisciplinary psychosocial intervention results in clinically meaningful advantages in QOL for patients with advanced cancer undergoing active medical treatment. # 1005/SYSTEMATIC METHOD FOR DETECTING FLOOR & CEILING EFFECTS IN NEUROLOGICAL HRQoL ASSESSMENT Steven Albert, Sergievsky Center; Michael A. Hagan, The Spine Center at the New York Neurological Instit, Columbia University, New York, NY, USA Analyses of an essential tremor case–control study showed the relationship between health professional assessments and HRQoL to be ‘S’-Shaped. This can be a phenomenon intrinsic to the data or a ceiling/flooring effect. The algorithm below P differentiatesPbetween the two. EðSUMCjSUMHÞ ¼ A þ Beaþbð HÞ =ð1 þ ea þ bð HÞÞ. Four parameters were evaluated simultaneously; and as the usual logistic

coefficients; ‘A’; ‘A + B’ as additional parameters used to determine presence of an artifact. The solution is the minimizer of the equation: P P P f Ci  ½A þ BeaþbðRHÞ =ð1 þ ea þ bð HÞÞg2 . Objective of this neurological case–control study was to measure impact of tremor on ADL. Ninety seven cases were matched (age, gender) with 99 normalvolunteer-controls. Patient Self-Reported Functional Rating of Tremor (SUMC) aggregates 33 items reflecting patients’ performance of ADL [0 ¼ no disability, 19 ¼ maximum disability]. Independent Rating of Performance Tests (SUMH) reflects a health care professional’s assessment. Data from semi-structured interviews (subjects performed various tests) aggregates patient’s motor skills, neurological assessment, and severity of the clinical condition [100 ¼ maximum, 0 ¼ no disability]. Iterative logistic regressions, converged to A ¼ 0 and B ¼ 78 (maximizing R2). In the extreme impairment range, the curve converges to ‘A + B’ ¼ 78 or 79 (controls exert little influence). Since the value is other than 100 (anchor) this eect is real. When graphing these curves it becomes evident that restricting data to subjects with tremor, produces a ‘jump’ in ‘No Tremor Disability’. Comparing results from the whole dataset to the results from the sub-set of cases, we can infer that since the values for ‘A + B’ are close a real eect exists at ‘Severe Tremor’ end of curve. Since ‘A’s are dierent, the discriminative power of the instrument is working hence we can distinguish between eects in a general population and the tremor population. Both results are valid depending on the perspective used. # 1448/DOES ITEM RESPONSE THEORY BASED SCORING IMPROVE THE MEASUREMENT ABILITIES OF THE EORTC QLQ-C30 SCALES? Morten A. Petersen & Mogens Groenvold, Palliative Department, Bispebjerg Hospital; Jakob B. Bjorner, National Institute of Occupational Health, Copenhagen, Denmark; Neil K. Aaronson, Division of Psychosocial Research & Epidemiology, The Netherlands Cancer Institute, Amsterdam, The Netherlands; Peter Fayers, Department of Public Health, Aberdeen University Medical School, Aberdeen, UK; Mirjam Sprangers, Department of Medical Psychology, Academic Medical Center, Amsterdam, The Netherlands Most health-related quality of life questionnaires include multi-item scales. The scale scores are usually estimated as a sum (or mean) of the item scores. Such sum scores are simple to construct and interpret. However, scoring procedures utilizing more of the information in the items might improve the measurement abilities and thereby reduce the number of patients needed in studies. We investigated whether item response theory (IRT) based scoring improved the measurement abilities of the EORTC QLQ-C30 physical functioning (PF), emotional functioning (EF), and fatigue (FA) scales. Using a database of 13,010 subjects we estimated the relative validity (RV) of IRT scoring compared to sum scoring using known-groups comparisons. An RV > 1 indicates that IRT scoring results in better measurement abilities. The mean RVs were 1.04 (PF), 1.03 (EF), and 0.97 (FA). None of the RVs were significantly larger than 1. Thus, no gain in measurement abilities using IRT scoring was found. Possible explanations for this include that the scales are relatively short and that the items in the scales are not constructed or selected for IRT scoring. We conclude that IRT scoring of the three longest QLQ-C30 scales does not improve measurement abilities compared to the traditional sum scoring.

770 # 1788/BIAS REDUCTION OF WARM’S WEIGHTED LIKELIHOOD ESTIMATION IN COMPUTERIZED ADAPTIVE TESTS WITH POLYTOMOUS ITEMS Otto B. Walter, Matthias Rose, Herbert Fliege, Janine Becker & Burghard F. Klapp, Clinic for Psychosomatics, Medical School of Humboldt-University, Berlin, Germany

understand the variance in scores not explained by obvious variables, and aim at developing more precise valuation methods # 1774/DIFFERENTIAL RESPONSE OF JAPANESE AND NORTH AMERICAN PATIENTS WITH OSTEOARTHRITIS: WHAT ARE THE IMPLICATIONS FOR POOLING DATA Daniel S. Himmelberger & David U. Himmelberger, Health Outcomes Group, Palo Alto, California, Shinichi Uemura, Outcomes Research, Pharmacia YY, Tokyo, Japan

An essential step in computerized adaptive tests (CATs) is the computation of an ability estimation of the latent trait. Warm’s weighted likelihood estimation (WLE) deserves particular attention because this approach was designed to reduce the first-order bias term of maximum likelihood estimation (MLE). We investigated the properties of bias reduction of WLE in comparison to expected a posteriori (EAP) estimation in a simulation study with our CAT-engine for two CATs for depression and anxiety. These tests were developed from responses from 3270 (Depression-CAT) and 2348 (Anxiety-CAT) psychosomatic patients. Item response curves were analysed according to Muraki’s generalized partial credit model. Item banks consisted of 64 and 50 polytomous items, respectively. For each 0.25-interval of the latent trait between )3.5 and 3.5 we generated responses of n ¼ 100 fictitious patients (simulees) with known ability levels. CAT scores based on WLE and EAP estimation were compared with the ability levels of the simulees. Simulated runs of the Depression-CAT for both estimation methods were generated from real responses of n ¼ 1010 patients. CAT scores were compared to n(0,1) standardized scores achieved in the State-Trait-Anxiety-Inventory (STAI). In our analysis of CAT scores based on simulated responses, WLE yielded markedly lower bias than the EAP estimator for ability levels < 1.0 and >1.0. Simulated runs of the Depression-CAT based on real responses showed very high correlations ranging from 0.88 to 0.94 between n(0,1) standardized STAI values and CAT scores for both estimation methods. However, deviation from standardized STAI values for ability levels < 1.0 and >1.0 was significantly lower for WLE scores. Our results show that WLE reduces bias for ability estimation in CATs with polytomous items. Thus, the application of WLE is recommended in conditions where unbiased estimates are needed. This is particularly relevant when cut-off scores are set on the ability scale, or when a comparison of CAT scores and paper- and-pencil tests is intended.

Most researchers rely on a careful translation to provide comfort that the translated questionnaires completed by patients speaking another language in another culture behave in a manner that is similar to the source version of the questionnaire; and therefore it is acceptable to pool data from both populations for a unified statistical analysis. This study examined the performance characteristic of both the source and translated versions of a well known questionnaire used in osteoarthritis and the implications of these findings for a pooled analysis. A total of 1006 patients in North America and Japan with OA of the knee or hip were analysed to evaluate whether the performance characteristics of the Japanese and Canadian English versions of the WOMAC LK are equivalent. Using a 1-1 matching scheme, patients in North America were matched with Japanese patients on the basis of age, gender and extent of disease at baseline. This resulted in 318 patients in both treatment groups and 185 in each placebo group. Patients differed only with respect to their cultural background and language. The performance characteristics of the Japanese and Canadian English version of the Likert form of the WOMAC were similar for all three scales of the global score for internal consistency (a ¼ 0.82–0.95); test–retest reliability (ICC ¼ 0.37–0.51); and construct validity. However, for responsiveness, the scores for the Japanese patients averaged about 20% lower than similar patients in North America and the effect sizes for the Japanese patients were only half as large as those for North American patients. In spite of the apparent large differential in effect sizes, the 95% confidence interval overlapped for a domains and the global score. Explanations for these differences will be discussed and results will be presented to illustrate the impact of such differences when pooling data for statistical analysis.

# 1094/UNDERSTANDING TIME TRADE-OFF VALUATIONS: FROM AGE TO RESPONSE TENDENCIES Marja C. Stuifbergen, Marie-Louise Essink-Bot & Willem Jan Meerding, Public Health, Erasmus Medical Center, Rotterdam, The Netherlands

# 1559/STRATIFICATION OF PATIENTS USING QOL PARAMETERS BY THE METHOD OF INTEGRAL PROFILES Andrei A. Novik, T.I. Ionova, A.V. Kishtovich, A.A. Tzepkova & S.V. Mylnikov, Multinational Center of Quality of Life Research, St. Petersburg, Russia

Time trade-offs (TTOs) measure valuations of health states, using time as a (hypothetical) commodity to trade. The measurement outcome contains both the valuation of the health state and the value the respondent attaches to time. We hypothesized that individuals from different ages may hold different values for time. This may explain the association of respondent age and the valuation of health states sometimes found in the literature. We investigated the role of age and other respondent characteristics in time trade-offs. A computerised questionnaire was designed to measure valuations for six health states with a Visual Analogue Scale, a time trade-off using 10 years to trade, and a time trade-off using the respondents’ remaining life expectancy. Respondents were recruited from three age groups: 20–25, 45–50 and 60–65. Analysis was done with G-theory, allowing to estimate the relative contribution of three sources of variance simultaneously: health state, valuation method, and age group. Two hundred and twelve respondents participated. Valuation scores differed significantly but not systematically between valuation methods. In a G-study, the percentage of variance explained by the health state was 37%, method 2%, and age group 0%. Further analysis revealed that a relatively large proportion of variance was explained by respondent related variation in scores. Cluster analysis showed that systematic differences in scale use were the main source of remaining variance. This variance was not related to respondent characteristics measured, including respondent age, health and experience with illness. We found no systematic age effect in VAS nor two TTO variants, nor an effect of other respondent characteristics. We did find large differences in scale use by respondents. We conclude that age is not an important variable in the valuation of health states using these methods. Research in this field should focus on differences in scale use by respondents, to try and

It is quite clear for physicians that population of new patients is heterogeneous in terms of different parameters, including QoL. Selection of treatment regimen requires an assessment of prognosis. Poor prognosis is suggested by lower QoL parameters. Therefore stratification of patients is necessary. QoL questionnaires used in clinical setting are multidimentional. Integral QoL index which might be used for stratification is worthwhile. The aim of the study was to identify the method to stratify patients using QoL parameters of Russian version of AIMS2SF. One hundred and thirty three patients with rheumatoid arthritis (RA), mean age 56.4, SD 11.8, male/female 22/111, in acute disease phase were enrolled in the study. Patients filled in AIMS2-SF at baseline and in 2 months of treatment. The method of integral profiles (MIP) was used to calculate integral QoL index. Patients were stratified basing on integral QoL index by standardization. The groups number was calculated as a ratio of the range and SD. Power and sample size analysis was used to compare QoL index in MIP with the mean of the values of QoL scales and with the sum of QoL scales. Six groups of patients with RA were formed after stratification of patients by MIP. The first group had the highest integral QoL index, the sixth – the worst one. The differences between means of AIMS2-SF scores in groups were statistically significant by ANOVA (p < 0.05). Power and sample size analysis for two groups comparison (p < 0.05) showed that in order to obtain the same power (0.80) of the dependent samples t-test MIP needed less sample size (n ¼ 22) as compared to means of QoL scales (n ¼ 27) and sums of QoL scales (n ¼ 27). MIP allows to interpolate the missing QoL scales. In addition, it gives ground to assess the weight of each scale in QoL index. The method of integral profiles to calculate integral QoL index is proposed. The method allows to provide more exact stratification of patients using QoL parameters. Further studies to test the method using different profile questionnaires are necessary.

771 # 1641/WHAT SAMPLES SIZES FOR RELIABILITY AND VALIDITY STUDIES II: A PROSPECTIVE STUDY Stefan J. Cano & Tom T. Warner, Clinical Neurosciences, Royal Free & University College Medical School, London; Jeremy C. Hobart, Clinical Neurosciences, Peninsula Medical School, Plymouth, UK Last year we presented a retrospective analysis of data from 713 people with multiple sclerosis (MS) suggesting that low sample sizes (n > 20 for reliability, and n > 40 for validity analyses) can provide useful indicators of psychometric estimates. This study is a prospective analysis in a different disease group. A prospective analysis was carried out on data from 391 people with cervical dystonia (CD) who had completed the CD Impact Profile (CDIP-58) and a selection of other health status questionnaires (SF-36, Euroqol, General Health Questionnaire, Hospital Anxiety and Depression Scale). First, the total sample was reduced by decrements of 50% according to when they were recruited into the study to generate prospective samples of n ¼ 320, 160, 80, 40, 20. In addition, the total sample was reduced randomly and non-randomly to generate samples of the same size, in order to test findings in our previous study. Indicators of reliability (Cronbach’s a, item-total correlation, test–retest reproducibility) and validity (within scale correlations, convergent and discriminate construct validity) were undertaken in all sub-samples. Reliability estimates remained constant and fulfilled criteria in all subsamples. Validity analyses were less stable in smaller samples (n < 40). However, in samples exceeding n ¼ 40 recommended criteria were reached and hypothesis testing reflected the original sample scores. This study provides further evidence that low sample sizes (n > 20 for reliability, and n > 40 for validity analyses) can provide useful estimates of psychometric properties. # 1357/EVIDENCE-BASED INTERPRETATION GUIDELINES FOR HEALTH-RELATED QUALITY OF LIFE MEASURES: A NOVEL APPLICATION OF META-ANALYSIS ILLUSTRATED WITH THE FACT-G Madeleine King, Centre for Health Economics Research (CHERE), University of Technology (UTS); Martin Stockler, Sydney Cancer Centre, Royal Prince Alfred & Concord Hospitals, Sydney, NSW, Australia; David Cella, Center on Outcomes, Research and Education (CORE), Evanston Northwestern Healthcare (ENH), Evanston, IL, USA; David Osoba, QOL Consulting, Vancouver, BC, Canada; David Eton, Amy Eisenstein, CORE, ENH, Evanston, IL, USA; Joanna Thompson, CHERE, UTS, Sydney, Australia Interpreting the clinical significance of effects observed on health-related quality of life (HRQOL) scales is problematic because their units of measurement are unfamiliar. Interpretability is not readily established; it is something that develops as a body of evidence accumulates. The aim of this research was to use meta-analytic methods to develop evidence-based interpretation guidelines for the Functional Assessment of Cancer Therapy – General (FACT-G), a valid and widely used cancer-specific instrument. A comprehensive search of on-line bibliographic databases yielded 123 papers with FACT-G scores; 81 of these provided informative contrasts (i.e., scores of two or more groups at one time or one group at two or more times). These were reviewed independently by three clinical experts (DC, DO, MS), who were blinded to the HRQOL scores. They assessed each contrast in terms of the clinical categories and circumstances, and judged whether it would produce a trivial, small, moderate or large difference in HRQOL. The experts’ judgments were then linked with the observed differences in HRQOL, and weighted average differences and effect sizes were calculated for each size class (trivial, small, moderate and large). These evidence-based interpretation guidelines may inform both the design and interpretation of clinical trials that use FACT-G. # 1276/ACQ – IS FIVE OUT OF SEVEN ITEMS ACCEPTABLE IN LARGE CLINICAL STUDIES? Klas Svensson & Ann-Christin Mo¨rk, Clinical Science, AstraZeneca R&D, Lund, Sweden; Elizabeth F. Juniper, Qoltech Ltd, Bosham, West Sussex, UK The Asthma Control Questionnaire (ACQ) is a validated questionnaire for assessing if patients’ asthma is adequately treated. It consists of seven items (5 · symptoms, rescue bronchodilator use and FEV1% of

predicted normal). A validation study of the ACQ has suggested that in large clinical studies measurement of FEV1 and rescue bronchodilator use may not be needed to obtain an accurate estimate of asthma control (1). This is possibly because the heterogeneity in patients with inadequate asthma control is lost in large group analyses. However, this has never formally been tested in clinical studies. With this in mind we wanted to investigate the loss of information from ACQ when removing the items concerning FEV1% of predicted normal and rescue bronchodilator use. We compared three variants of a reduced ACQ: ACQ5 (5 symptoms alone), ACQ6a (symptoms and FEV1) and ACQ6b (symptoms and rescue use) with the original ACQ (ACQ7) using data from a large clinical study with respect to systematic difference and ability to detect treatment differences. Agreements between the full ACQ index (ACQ7) and the reduced versions were high: ICC ¼ 0.9517 for ACQ5, 0.9893 for ACQ6a and 0.9671 for ACQ6b. The difference D5 ¼ ACQ5-ACQ7 was ranging from )0.857 to 1.057 with a mean value of )0.013 and SD ¼ 0.280; the quartiles are )0.200 and 0.171. For D6a ¼ ACQ6a-AQCQ7 the range was from )0.643 to 0.500 with a mean value of 0.028 and SD ¼ 0.123; quartiles ¼ )0.048 and 0.095; for D6b ¼ ACQ6b-ACQ7 the range was from )0.8571 to 0.6190 with mean ¼ )0.039 and SD ¼ 0.221; quartiles being )0.190 and 0.119. No significant difference between the two treatment groups was found, neither for D5 (p ¼ 0.87), D6a (p ¼ 0.76) nor D6b (p ¼ 0.99). We conclude that all three variants of the ACQ index should be possible to use in clinical studies of the type described above. (1) Juniper EF, O’Byrne PM, Roberts JN. Measuring asthma control in group studies: do we need airway calibre and rescue b2agonist use? Respiratory Medicine 2001; 95: 319–323. # 1639/INTEGRATING HEALTH PROFILE WITH SURVIVAL FOR QUALITY OF LIFE ASSESSMENT Jung-Der Wang, Institute of occupational medicine and industrial hyg, National Taiwan University, College of Public Health; Chiun Hsu, Department of Oncology, National Taiwan University Hospital; KaiPing G. Yao, Department of Psychology, National Taiwan University, Taipei, Taiwan In cohort studies or clinical trials, measurements of quality of life (QoL) were averaged across available individuals for each group at given points in time to produce single measures for comparisons. However, estimates of these single measures may be severely biased if substantial mortality occurs over time. We have developed a method that integrates QoL measurement and survival for long-term evaluation of health services. We defined a mean QoL score function over time for an index population as the average QoL score of all individuals both alive and dead at each time point in the population. While a living subject’s QoL can be assessed by asking one’s subjective preference, the score of a decedent can be assigned a fixed value depending on the specific facet on health profile. The mean QoL score function over time is reduced to a single measure of expected cumulative QoL score, which is the area under the curve of mean QoL score function over a given time interval and can be estimated by taking a random sample from a cross sectional survey. For the QoL score function to be extrapolated to life-long, it requires the assumption that the disease causes premature death or a long-term moderate impairment of QoL. We provided methods and computer programs for estimating mean QoL score functions and the reduced single measures for use in comparisons. A cohort of 779 breast cancer patients from Chiangmai, Thailand were followed for 12 years to demonstrate the proposed methods. The data included the 12-year complete survival records and QoL scores on 233 patients collected from a cross-sectional survey using WHOQOL questionnaire and standard gamble method. The expected cumulative QoL scores using utility and psychometric scales were compared among patients in four groups of clinical stages in this cohort for time from onset up to 12 years and life-long. We conclude that such an integration of QoL measurement with survival can be useful for the evaluation of health service and clinical decision. (prerequisties: for the attendees: concept or experience of survival or life table analysis)

772 # 1470/DIMENSIONAL REDUCTION OF THE 3 SCALE INSTRUMENT TO EVALUATE COPING SKILLS FOR USE IN THE CLINICAL SETTING. VALIDATION IN A POST MYOCARDIAL INFARCTION POPULATION A. Ribera, G. Permanyer-Miralda, P. Cascant, I. Ferreira, Cardiology Department, Hosp. Vall d’Hebron, Barcelona, Spain Relationship of the coping process with disease and health related quality of life (HRQL) has only recently been addressed. The Brief COPE evaluates 13 coping strategies and has been designed to be used in the clinical setting, but they are still impractical for routine use. The main objectives of this study were: To find an interpretable subscale solution that facilitates to study the relationship between general coping patterns and HRQL, and to describe these patterns in patients after a first myocardial infarction (AMI). The Brief COPE was administered to patients participating in a wider longitudinal study of HRQL after AMI during recovery in the Cardiology Department of a tertiary hospital. In adition, sociodemographic and clinical variables were collected. Of the 144 patients asked to participate during a one year period, 105 answered the study questions. We hypothesised that at least three groupings would arise from factor analysis: one with high loadings on strategies considered to be adaptive, a second with those considered to be maladaptive and a third with the strategies that imply seeking support. We conducted factor analysis and the three interpretable hypothesised factors were confirmed: (1) acceptance, planning, active coping and positive reappraisal; (2) denial, behavioural disengagement, and self blame; and (3) seeking social support for instrumental and emotional reasons, religion and venting. Scores of the three factors were compared by age, gender, social class and educational level. Women, manual workers and less educated people had significantly lower levels of adaptive coping, whereas less educated people had significantly higher levels of the other two factors. There was a decreasing tendency of adaptive coping and an increasing tendency of maladaptive coping and seeking support with age. In conclusion: The Brief COPE can be summarized in three interpretable scores which can be used to evaluate their relationship with other health related variables and to understand subjective determinants of HRQL # 1162/REDUCING EXACERBATIONS LEADS TO A BETTER HEALTH-RELATED QUALITY OF LIFE IN PATIENTS WITH COPD Elisabeth Stahl, Clinical Science, AstraZeneca R&D Lund, Lund, Sweden; Paul W. Jones, Medical School, University of London, St. George’s Hospital, London, United Kingdom Budesonide (B)/formoterol (F) in a single inhaler (SymbicortR) improves health-related quality of life (HRQL) and reduces exacerbation frequency in patients with COPD (GOLD stages IIB-III). The aim was to assess the relationship between the number of exacerbations requiring medical intervention (hospitalisation and/or use of oral steroids/ antibiotics) and HRQL. COPD patients (mean age 64 years, FEV1 0.99 L, 36% predicted normal) from the two double blind, 12-month studies mentioned above (study 1, n ¼ 812; study 2, n ¼ 1022) were included in the analysis. Exacerbation data were collected prospectively. HRQL was assessed using the St. George’s Respiratory Questionnaire (SGRQ); low value means better HRQL and a 4-unit reduction in Total score is a clinically meaningful difference. Mean SGRQ Total scores were lower in patients experiencing fewer exacerbations in both studies (Table). The significant difference in score was also clinically relevant comparing patients with 0 vs. >1 exacerbation. Lower exacerbation frequency correlates consistently with better HRQL. It can be concluded that a reduction in exacerbations is an important goal in the treatment of COPD.

# 1583/THE EFFECT OF CO-MORBIDITY ON THE ASSOCIATION OF CARDIOCEREBROVASCULAR DISEASES AND HEALTH-RELATED QUALITY OF LIFE IN A CHINESE ELDERLY POPULATION Cheng-Chieh Lin, Department of Family Medicine, China Medical College Hospital; Tsai-Chung Li, Institute of Chinese Medical Science, China Medical College; Chia-Ing Li, Department of Medical Research, China Medical College Hospital, Taichung, Taiwan, Republic of China Introduction: Co-morbidity is an important dimension of health status in older adults because the concurrent presence of co-morbidity is prevalent. Co-morbid illness may modify the primary factor of interest, and co-morbidity itself may also independently be a risk factor for adverse outcomes. The objectives of the present study are to examine the association of cardiocerebrovascular diseases and health-related quality of life (HRQoL) in a Chinese, community-based elderly population and to examine whether the existence of the co-morbidity will modify these associations. Methods: All elders aged 65 and over who resided in Chung-Shing-Shin-Tseun, Taiwan were recruited with an overall response rate of 89.13% (n ¼ 921) in 1998. Face-to-fact interviews were used to collect information. An elder had any cardiocerebrovascular diseases if he/she had been diagnosed as having heart disease, hypertension, stroke, or diabetes. HRQoL was measured by SF-36. The co-morbidity consisted of eye, pulmonary, and musculo-skeletal disorders, ulcer and liver disease and other disease, and anemia. Results: After controlling for the effects of age, gender, and co-morbidity, the estimated effect of cardiocerebrovascular diseases on all subscales was negative except mental health and role limitations due to emotional problems, often as much as a 5–10 point reduction on this 100-point scale. Co-morbidity exerts independent effect on physical functioning, role limitation due to physical and emotional problems, and mental health. On the other hand, the existence of musculo-skeletal disorder enhance the effect of cardiocerebrovascular diseases on pain; eye disorder on general perception of health and vitality; and other diseases on social function. # 1573/WHAT IS ASSOCIATED WITH POOR HEALTH-RELATED QUALITY OF LIFE AMONG WOMEN WITH CORONARY DISEASE? Bernice Ruo, Internal Medicine, San Francisco VA; Medical Center, University of Calif, San Francisco, CA; Naji Younes, Biostatistics and Epidemiology; Paula Friedenberg, Biostatistics, George Washington University Biostatistics Center, Rockville, MD; Ivan Barofsky, The Quality of Life Institute, East Sandwich, MA, USA As more people with coronary disease are living longer, it is increasingly important to manage this disease to minimize not only its progression but also its impact on patients’ health-related quality of life (HRQL). Surprisingly, studies of the association between measures of disease severity and HRQL have shown little or no association. It is unclear what other factors might influence HRQL among patients with coronary disease. To address this issue, we used data from the Women’s Angiographic Vitamin and Estrogen study, a randomized, double-blind trial of 423 post-menopausal women with coronary disease. Participants were followed for an average of 2.8 years, had entry and exit coronary angiograms to assess disease progression, and completed the Seattle Angina Questionnaire (SAQ) and the Medical Outcomes Survey Short Form 36 (SF-36) to assess HRQL. Poor HRQL was defined as the lowest quartile of scores in this population for each HRQL instrument. Each participant also completed the Center for Epidemiological Studies Depression scale, the Medical Outcomes Survey sleep questionnaire, and the Duke Activity Scale Index. Using logistic regression, we examined the association between poor SAQ angina frequency score or SF-36 scores (mental component scale, MCS and physical component scale, PCS) and age, co-morbidities, hormone treatment, disease progression, depression score, sleep score, and functional status. In adjusted analyses, only higher depression scores were associated with poor SF-36 MCS scores. Worse functional status scores were associated with poor SF-36 PCS scores and poor SAQ scores. Worse sleep scores were associated with poor SAQ scores. Angiographic coronary disease progression was not associated with poor SAQ, SF-36 MCS, or SF-36 PCS scores. Depressive symptoms, poor sleep, and poor functional status are more strongly associated with worse HRQL than angiographic disease progression. Efforts to improve HRQL among women with coronary disease should include assessment of depressive symptoms, sleep, and functional status.

773 # 1553/EMPLOYING EUROQOL-5D IN HEALTH NEEDS ASSESSMENT IN CARDIAN PATIENTS: A COMPARISON WITH OTHER QUALITY OF LIFE TOOLS Mohsen Asadi-Lari & David Gray, Division of Cardiovascular Medicine, University of Nottingham, University Hospital (QMC), Nottingham, UK Background: Improving the quality of health services depends on how facilities are tailored to fulfil patients’ needs, therefore satisfying healthcare needs of patients – in particular cardiac – is one of the main objectives of the National Health Service. The Euroqol-5D (EQ-5D) questionnaire is a valid and reliable quality of life (QOL) tool; no published study has evaluated the EQ-5D as a health needs assessment (HNA) tool in suspected and confirmed myocardial infarction. Method: To investigate the relationship of QOL measurement and HNA two generic (SF-12 and EQ-5D) and specific (Seattle Angina Questionnaire: SAQ) QOL instruments were administered in a crosssectional study on randomly selected patients (mean age ¼ 69.7, SD ¼ 11.7) admitted to the acute cardiac unit, Nottingham (n ¼ 242). The needs assessment included health care, social (employment, accommodation, finance and transport) and information needs. Main results: 58% experienced no major cardiac interventions. QOL domains were related to almost all of HNA fields; specifically, healthcare needs were mostly detected by the disease-specific and social and general needs by generic QOL tools. EQ-5D showed reasonable ability to detect various needs in these patients. EQ-5D was the only tool to detect the significant difference of QOL perception in various social classes (p < 0.01); the lower social class the poorer scores in the visual analogue scale (VAS). EQ-5D was also related to satisfaction domain, access to health facilities, and patients’ concerns in HNA survey. (q ¼ 0.24–0.56, p < 0.001) Correlation between EQ domains and SF-12 were higher than those with SAQ, although EQmobility was similar to SAQ physical domain and physical component score (PCS) in SF-12. (R ¼ 0.69, p < 0.001) Conclusion: EQ-5D has considerable merit as a measure of health-related quality of life in patients with suspected or confirmed infarction, providing health service researchers with a useful shortcut and alternative to relatively time-consuming formal health needs assessment. # 1471/COMPARISON OF THE SPANISH VERSION OF FOUR HEALTH-RELATED QUALITY OF LIFE (HRQL) INSTRUMENTS FOR ADOLESCENTS IN A HEALTHY SAMPLE Jose-Maria Elorza-Ricart & Luis Rajmil, Research Area, Catalan Agency for Health Technology Assessment; Jordi Alonso, Municipal Institute of Medical Research; Silvina Berra & Marta Aymerich, Research Area, Catalan Agency for Health Technology Assessment, Barcelona, Catalonia, Spain; Ulrike Ravens-Sieberer, Child and Adolescent Health, Robert Koch Institute, Berlin, Germany; MarieClaude Simeoni, Department of Public Health, University Hospital of Marseille, Marseille, France; Donald Patrick, Social and Behavioral Sciences Program, School of Public Health and Community Medicine, Seattle, WA, USA The Kidscreen is a seven European country project for developing a HRQL questionnaire for children and adolescents. The Spanish pilot study was carried out in April 2002, including the Spanish versions of other instruments: VSP-A (France), KINDL (Germany) and Youth QoL (YQOL, USA). The study purposes were to compare the psychometric

properties of these instruments and their ability for discriminating by age, gender, socio-economic status and self-perceived health. One class per grade (12–18 years) was selected in three schools of Catalonia, Spain. HRQL measures were collected from adolescents, and information on general health and use of health services was asked to parents. Both single index and dimension scores of each instrument were computed following authors’ algorithms. Classical psychometric properties were assessed and expected HRQL score differences were compared using ANOVA. Responses were obtained from 291 (82%) adolescents and 195 (67%) parents. Missing data were lower than 5% for all instruments. Internal consistency was adequate (>0.7) for most of the scores (21 out of 25), being the ‘school’ dimension of KINDL the lowest (a ¼ 0.41). The VSP-A single index was more centred distributed (mean 63.9; [SD ± 13.5]) than all the others, the YQOL index being the most skewed towards better HRQL (76.6; [SD ± 14.7]). Ceiling effect (>15%) was found only for four dimensions, with no floor effect. Albeit all the single index discriminated among the factors studied (p < 0.05) the Kidscreen discriminated the most among genders in adjusted models. Our results suggest that all the analysed questionnaires could be useful and suitable for evaluating HRQL in Spanish healthy adolescents. Additional analyses are necessary. # 1263/PRELIMINARY EVALUATION OF A SHORT PSYCHO-EDUCATIVE GROUP INTERVENTION FOR ADOLESCENTS WITH IBD Marieke de Boer, Martha A. Grootenhuis, Bert H. Derkx & Bob F. Last, Psychosocial Department, Emma Children’s Hospital, Amsterdam, Netherlands Inflammatory Bowel Disease (IBD), comprising Crohn’s disease, ulcerative colitis and indeterminate colitis, has its onset in adolescence. In a recent study we demonstrated that disease activity and the use of a predictive coping style (having positive expectations) is strongly associated with HRQoL in children. The aim of this study is to evaluate whether a short psycho-educative group intervention (aiming to enhance efforts to appraise the disease positively, strengthening reliance upon the doctor, information seeking and relaxation) can influence the coping style of adolescents with IBD and have a positive effect on their QoL. Two hundred and three Children (aged 12–18 years) registered at the Crohn and Colitis Ulcerosa Association were invited to participate in the group at the Emma Children’s Hospital in Amsterdam. Thirty patients responded positively. However, 18 lived too far away to attend and served as a control group. Twelve children were enrolled and attended in groups of six children in six group sessions, supervised by two psychologists. Using reliable and valid instruments the child’s coping styles, feelings of competence and health related QoL was assessed before and after the intervention. The parents were asked to fill in the Child Behaviour Check List (CBCL). Multiple regression analyses were performed to test whether group-participation was predictive of the outcome measures while correcting for the first measurement and gender. First results show more use of a predictive coping style and an increased quality of life after participation. These were especially for the domains self-esteem, self-image and emotional functioning. The parents reported decreased externalising behaviour after participation. The preliminary analyses show promising results to continue this intervention.

774 # 1510/VALIDATION OF THE RUSSIAN TRANSLATION VERSION OF THE PEDSQL4.0 GENERIC CORE SCALES FOR QUALITY OF LIFE RESEARCH IN 8–12 YEARS OLD RUSSIAN CHILDREN Tatiana P. Nikitina, St. Petersburg, Russia; Anton V. Kishtovich & Tatyana I. Ionova, Multinational Center of Quality of Life Research, St. Petersburg, Russia; James W. Varni, College of Architecture and Medicine, Texas A&M University, College Station, TX, USA The purpose of this study was to validate the Russian translation version of the PedsQL4.0 Generic Core Scales, a generic pediatric QoL questionnaire, for QoL research in 8–12 years old Russian children. The results of translation and cultural adaptation were reported earlier. Target sample within validation included 43 healthy Russian children of 8–12 years old, 43 Russian children of the same age with acute health conditions (asthma before treatment, cancer before-, on-, o-treatment) and parents of above children. Data quality was assessed and acceptable results were founded (low percentage of missing values, sucient item-response distribution). Cronbach-a’s coecient was calculated, it exceeded 0.70 on all scales of child- and parent reports excluding of School Functioning scale of parent report (0.68). Cronbach’s-a on Total Scale of child- and parent reports was 0.90. Construct validity was assessed: (1) QoL Scores in healthy children were higher than in children with acute health conditions on all scales in child- and parent reports (p < 0.01); (2) positive significant Spearmen correlations between the same scales of child- and parent reports were founded (r ¼ 0.43–0.66, p < 0.01); (3) QoL Means in child report of children with acute health conditions were significantly higher than in parent report on all scales (p < 0.01), except of Emotional Functioning and Social Functioning; (4) QoL Scores for Physical Functioning and Total Scale were significantly higher for cancer patients with higher Karnofsky Index (p < 0.01) in child- and parent reports. Sensitivity to clinical differences was calculated by comparison of QoL Means between children with cancer on- (n ¼ 13) and off-treatment (n ¼ 20). QoL Means in children off-treatment were significantly higher than in children on-treatment on all scales of child report and on Total Scale of parent report (p < 0.01). Thus, the Russian translation version of the PedsQL4.0 Generic Core Scales (ages 8–12 years) demonstrates acceptability, reliability, validity and sensitivity for pediatric QoL research in Russia. # 1337/NEW QUALITY OF LIFE(QOL) QUESTIONNAIRE FOR CHILDCARE Rika Hayashida, Kouko Hama, Michiko Kobayashi, Tomoko Shimada, Yoshiko Naka, Kazumi Uchiyama, Michiko Kinoshita, Akemi Nakamura, Sumie Nagano, Emiko Mizomichi, Suzuko Watanabe, Miyako Sato, Yumiko Yamaguchi, Kiyomi Katayama, Yuko Kamatsu & Miyoko Heima, Siebold University of Nagasaki, Nishisonogi-gun, Nagasaki; Takashi Mandai, Japanese Society of Quality of Life Research, Kobe, Hyogo, Japan The purpose of this study was to develop the desirable QOL questionnaire for childcare. Two hundred and thirty-one mothers who take care of children under 12 months participated in this study. Our new original self-administrated questionnaire consisted of 43 questions divided into 12 categories. Cronbach’s a coefficients of our questionnaire were high enough to accept for clinical use: 0.92 in environmental problems, 0.89 in economy, 0.82 in dietary problem, 0.78 in mental function, 0.77 in well-being, 0.75 in total infant–maternal interaction including 0.92 in touching subcategory and 0.86 cradle song

subcategory, 0.75 in sleep, and 0.74 in social participation, etc. respectively. Our questionnaire contained 12 mainfactors and cumulative contribution was 0.71. There were significant positive correlation between mental function and infant–maternal interaction (p < 0.01), between well-being and social participation (p < 0.01), between mental function and well-being (p < 0.01), and between social participation and infant–maternal interaction (p < 0.01) etc. For the high QOL childcare, infant–maternal interaction especially touching and support for childcare are most important for childcaring mother from the viewpoint of mental control. We had believed that QOL in childcaring mother had been too difficult to evaluate, but these findings indicated that our new QOL questionnaire has high enough reliability and potency of validity to use for childcaring mother. We would like to use this questionnaire to evaluate the QOLs of childcaring mothers and apply new findings to establishment for the more excellent childcare in future. Although childcare is most important social problems in isolated and civilized urban life in the world recently, scientific QOL evaluation had not been done yet due to the absence of available QOL questionnaire for childcaring mother. We must pay more attention to many kinds of childcaring mother’s problems on various developmental stage of children from the standing point of QOL in all over the world. # 1393/HEALTH-RELATED QUALITY OF LIFE IN 8-YEAR OLD CHILDREN WITH DOWN’S SYNDROM Jeanet Bruil, Minne Fekkes & Erik Verrips, Child Health, TNO Prevention and Health, Leiden, The Netherlands Introduction: Aim of the study was to evaluate the health, health-related quality of life, school participation, parenting stress and their determinants in a cohort of 8-year old children with Downs syndrome. Methods: All children known by the Dutch Down Syndrome Foundation (SDS) born in the years 1992–1994 were included in the study. Parents were asked to fill out a questionnaire on health, health-related quality of life (TNO AZL Children Quality of Life Questionnaire (TACQOL)), school participation and the short form of the Parenting Stress Index (PSI). Results: Parents of 207 children (111 boys and 96 girls) were included in the study (response rate 65%). Mean age of the children was 8.1 years (7.01–9.4). For 90% of the children chronic health problems were reported. Most prevalent chronic health problems were eye abnormalities (51%), chronic airway infections (44%) and heart disease (42%). Comparison of means of children with Downs syndrome and healthy children revealed significant differences on gross motor functioning, autonomy, cognitive functioning and social functioning (p < 0.01). Depending on the scale, 37% (autonomy) to 84% (pain and symptoms) of the health status problems reported were associated with negative emotional reactions of the child. 75% of the children had participated in regular day care centers or schools lifetime. At the age of eight, 50% participated in regular primary schools, most in the first or second level. Mothers reported significantly more parenting stress than the reference group from an open population, but far less than a clinical group of mothers (p < 0.01). Conclusions: As a group, parents of children with Downs syndrome reported more chronic health problems and a lower quality of life of their children. In the current study, half of the group participated in regular schools at the age of eight, whereas in earlier days most children were educated in special schools in the Netherlands. Parenting stress is higher in the mothers of children with Downs syndrome, but did not reach the clinical range.

775 # 1390/FIELDTESTING OF THE HAEMO-QOL QUESTIONNAIRE FOR CHILDREN AND ADOLESCENTS WITH HAEMOPHILIA IN EUROPE Sylvia von. Mackensen & Monika Bullinger, Institute and Policlinics for Medical Psychology UKE, University of Hamburg, Hamburg, Germany Introduction: In spite of an increased interest in the assessment of quality of life in children, an instrument for children with haemophilia until now was not available. Epidemiological studies, clinical trials and quality of life assurance programs in the paediatric haemophilia area are in need of such an instrument. Because of the low prevalence of the condition, such an instrument should also be cross-culturally applicable. Methods: In the study presented, a quality of life assessment instrument for children with haemophilia (originally cross-culturally developed and pilot-tested in 58 children) was filed-tested in six countries (France, Germany, Italy, The Netherlands, Spain and the United Kingdom). The psychometric properties of the Haemo-QoL questionnaire were examined in 339 children with haemophilia as well as their parents. The Haemo-QoL is a self-reported questionnaire for children in the age ranges 4–7 (I), 8–12 (II), 13–16 years (III) as well as for parent rating. It assesses 9–11 dimensions (depending on age group versions) of quality of life of special relevance for the condition of haemophilia. Psychometric testing involved examination of reliability (internal consistency, retest reliability) and validity (convergent and discriminant). Results: The results showed, that the three age group versions of the Haemo-QoL had acceptable Cronbach a reliability values ranging from 0.80 to 0.92, as well as possessing sufficient discriminant validity (e.g. number of bleeds) and convergent validity (e.g. correlation with KINDL). However, in young children (4–7 years) when compared to older children, these indicators were less satisfactory. Conclusion: The Haemo-QoL full version is now available for children of age group I (21 items), age group II (64 items) and age group III (77 items) as well as their parents. The Haemo-QoL is currently included in an EU-funded clinical study (ESCHQoL). Acknowledgement: The Haemo-QoL Study-Group includes clinicians from six countries. The project was funded by BAYER research grant. # 1661/STRUCTURAL EQUATION MODELING TO ASSESS THE VALIDITY OF THE PRELIMINARY VERSION OF THE EUROPEAN CROSS CULTURAL INSTRUMENT DISABKIDS PROXY MEASURE D. Debensason, M.C. Simeoni & M. El Khammar, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille, France; C. Petersen, S. Schmidt & M. Bullinger, Department of Medical Psychology, University Hospital, Hamburg, Germany The aim of European DISABKIDS project is to improve the QoL of children with chronic diseases by assessing QoL from the perspective of children and their parents using a process of multinational codevelopment involving seven countries (A, Fr, G, Gr, Nl, Sw, UK). It emphasizes the problem of the appropriateness of a structural model between children and parents measures in order to allow the comparison of children and parents reports. The first step of development led to the generation of a questionnaire comprising two parts: one chronic generic (119 items), and one specific to various conditions (232 items in total). For the chronic generic structure issued from children’s answers, the pilot study identified a six dimension-structure including 56 items: Treatment, Physical, Emotion, Independence,

Social exclusion, Social inclusion.The purpose of this study is to know if children’s model fits to parents’ data using confirmatory factor analysis and IRT. Similarities and dissimilarities between children and parents reports are studied (paired T-test, effect size). The pilot study included 380 children with chronic conditions: asthma (146), diabetes (64), cerebral palsy (21), epilepsy (37), rheumatoid arthritis (54), cystic fibrosis (29) and skin diseases (29), in three age-groups (4–16 years). The dimensions of the children’s and parents’ chronic generic modules have satisfactory internal consistency (Cronbach’s a > 0.7). A confirmatory factor analysis and item response theory-based approaches were used to assess the psychometric properties and statistical equivalence of the two different measures. The correlations between children’s and parents’ dimension scores ranged from 0.41 to 0.56. Significant differences are found between parents’ and children’s scores for all the dimensions (p > 0.001) except for Independence. The parents underestimated the HRQL of their children. The effect sizes are higher than 0.35 except for Physical (0.21) and Social inclusion (0.16). These first results stress the interest in conducting further researches. # 1419/QUALITY OF LIFE IN PEDIATRIC INFLAMMATORY BOWEL DISEASE James M. Perrin, Pediatrics; Karen Kuhlthau, Aziz Chughtai, Center for Child and Adolescent Health Policy; Christine King & Harland Winter, Pediatrics, MassGeneral Hospital for Children, Boston, MA, USA Measurement of quality of life (QoL) in children and adolescents with inflammatory bowel diseases (IBD) has had little attention, despite the importance of understanding key factors affecting QoL, especially for measuring the effects of treatment for IBD, a typically exacerbating but invisible condition. The main purpose of this pilot study is to examine the impact of clinical severity, treatment, and social factors on QoL of a diverse population of children and adolescents with IBD. The study aims to (1) determine the associations of clinical characteristics (condition type, activity/severity, and treatment) with specific components of general health-related quality of life and IBD-specific QoL; (2) describe the effects of sociodemographic characteristics (SES, age, and gender) on these measures; and (3) compare the views of different observers (parent and child with IBD) of the child’s QoL. The study applies both general and condition-specific QoL measures among a random sample of 250 children and adolescents with IBD, ages 8–18 years, in six US clinical sites. Data collection began in April 2003. We obtain measures of QoL from both the child and a parent in each case. The study gathers additional data regarding the subjects’ clinical condition (condition type, severity/activity, treatment [including surgery], age of onset) and socioeconomic status (household structure and income). Main analyses will confirm factors on an IBD-specific QoL measure, next compare general and specific measures of QoL, and then examine the influence of clinical and sociodemographic variables on QoL, through multivariate regression techniques. We will also examine the differences in child and parent assessments of QoL. The information from this study will provide a stronger base for future studies of treatment and natural history of IBD, as well as health services research comparing different patterns of IBD care. It will help to clarify the life domains that are affected by IBD and will inform interventions to improve QoL for children with IBD.

776 # 1820/ASSESSING RELEVANCE: CHILDREN’S ROLE IN DEVELOPING A MEASURE OF QUALITY OF LIFE Nancy L. Young, Population Health Sciences, The Hospital for Sick Children; Catharine S. Bradley, Graduate Department of Rehabilitation Science, University of Toronto; Cindy D. Wakefield, Division of Hematology/Oncology, The Hospital for Sick Children, Toronto, Ontario; Dorothy Barnard, Division of Hematology/Oncology, I.W.K. Grace Health Centre, Halifax, Nova Scotia; Victor Blanchette, Division of Hematology/Oncology; Patricia J. McCusker, Department of Pediatrics and Child Health, University of Manitoba, Winnipeg, Manitoba, Canada Several measures of quality of life (QoL) are available for children. Many have been adapted from adult forms or are based on the opinions of parents or health care experts. However, there is little evidence to suggest that children will choose the same items to describe their QoL as would adults. The purpose of this study was to examine the differences between child and adult perspectives in developing a disease-specific measure of QoL. An initial list of items was generated from the literature and five discussion groups of children with Hemophilia, parents and nurses. These items were presented to three focus groups with children and three with parents who generated new items and eliminated irrelevant items. Each person then identified the 10 most relevant items. The perspectives of 24 Canadian Hematologists were gathered by survey. Item reduction was based on strict rules to value children’s input most heavily. The items were compiled into a questionnaire and pilot tested with 10 children to obtain feedback on formatting and wording. After pilot testing, item reduction was repeated with the rules altered to value parents’ input most heavily. Results were examined for differences in the measure depending on whose perspective was valued. The process resulted in a 79-item version of the Canadian Hemophilia Outcomes – Kids’ Life Assessment Tool (CHO-KLAT). Examination of the origin of these items found that 50% of items from children were retained. When item reduction was rerun emphasising parents’ perspectives, children and parents agreed on the inclusion of only 37 items. In comparison to the physicians, 50% of items they discarded were kept by children and 20% of items they valued were discarded by children. It is widely recognised that QoL is a construct that should be based on the perceptions of the individual. In this study, children provided a unique perspective on their QoL that was not represented by others. This research suggests that including children in the development of measures may enable a more valid and comprehensive assessment of children’s QoL. # 1759/RESULTS OF THE COMPARISON OF THE FRENCH VERSION OF THREE HEALTH-RELATED QUALITY OF LIFE (HRQoL) INSTRUMENTS FOR ADOLESCENTS IN A HEALTHY SAMPLE M.C. Simeoni, S. Robitail & D. Debensason, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille, France; Ulrike Ravens-Sieberer, Robert Koch Institute, Kidcreen Group, Berlin, Germany; Donald Patrick, Washington University, Seattle, USA; P. Auquier, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille, France The KIDSCREEN is a European project, involving seven countries, developing a HRQoL questionnaire for children and adolescents. The French pilot study was carried out in May 2002 in adolescence attending school, including other instruments: VSP-A (39 items-France), and YQOL-S (eight items-USA). We compare their correlation and their ability to discriminate by gender, age, socio-economic status, family status, health status and use of health services. Each grade of secondary school (11–16 years) was represented. Both single index and dimension scores of each instrument were computed following

authors’ algorithms. Classical psychometric properties were assessed and expected HRQL score differences were compared using ANOVA. Responses were obtained from 437 randomised adolescents and their parents. Acceptability was satisfactory for YQOL-S , VSP-A and preliminary KIDSCREEN research instrument. Missing data were lower than 5% for the three instruments. Item internal consistency and item discriminant validity are satisfactory for the VSP-A and the preliminary KIDSCREEN research instrument. Internal consistency was good (Cronbach’s a above 0.70) for all of the 21 scores except three ranged between 0.65 and 0.70. Divergent and convergent validity was supported by relevant correlations between YQOL-S, VSP-A and preliminary KIDSCREEN research instrument. All dimensions present satisfactory distributions. Ceiling effect (>15%) was found only for two dimensions, without any floor effect. The three instruments discriminated among demographic and socio-economic status. Among the three instrument, the VSP-A and the preliminary KIDSCREEN research instrument found the most significant differences according to health status. Adolescents who used health services showed lower HRQoL scores for the three instruments. This study provides promising results of the reliability and the validity of the three instruments in the French adolescent population. Better understanding of similarities and discrepancies between HRQoL instruments is a topic of interest for future research. # 1467/DIFFERENCES IN THE PROCESS OF DEVELOPMENT AND THE CONTENT OF FOUR HEALTH-RELATED QUALITY OF LIFE (HRQL) QUESTIONNAIRES FOR CHILDREN AND ADOLESCENTS Luis Rajmil, Research Area, Catalan Agency for Health Technology Assessment; Jordi Alonso, Municipal Institute of Medical Research; Jose-Maria Elorza-Ricart, Silvina Berra & Marta Aymerich, Research Area, Catalan Agency for Health Technology Assessment, Barcelona, Catalonia, Spain; Ulrike Ravens-Sieberer, Child and Adolescent Health, Robert Koch Institute, Berlin, Germany; Marie-Claude Simeoni, Department of Public Health, University Hospital of Marseille, Marseille, France; Donald Patrick, Social and Behavioral Sciences Program, School of Public Health and Community Medicine, Washington, Seattle, USA Although there is agreement on the multi-dimensional nature of HRQL, the content of measuring instruments has been insufficiently reviewed, especially for the younger ages. The objective of this study was to analyse and compare the content of four instruments included in the pilot test of the Kidscreen project, a new HRQL instrument developed in Europe. Instruments included were: the VSP-A (France); the Kindl (Germany); the Youth QoL (the US); and the Kidscreen (seven European countries). Information reviewed included: initial item pool generation, item reduction methods, the final items and dimensions, and the scoring methods as well as the populations on which they were carried out. A modified MOT Scientific Committee Criteria to evaluate HRQL instruments was used for assessing their relative quality. Items had been generated through focus group with adolescents in three cases, and lay and professionals for the remaining (Kindl). Item reduction was based on Classical Testing in all cases, but for the Kidscreen Item Response Theory was also used. The number of dimensions ranged from four dimensions (Kindl and YQOL), to 10 (Kidscreen). The total number of items ranged from 24 (Kindl) to 53 (Kidscreen), with more items addressing social than physical functioning. All questionnaires allow to built a single index. The Kidscreen and VSP-A included only healthy adolescents in its development, while Kindl and YQOL included both healthy and disabled children or with chronic conditions. MOT scale scores of the four instruments were similar. HRQL instruments for adolescents and children evaluated are multidimensional and have followed a satisfactory development.

777 # 1214/IMPACT OF HYDROTHERAPY CARE ON THE QUALITY OF LIFE OF CHILDREN Therese Nocera & Francois Verriere, Medical Direction, Laboratoires Dermatologiques Ave`ne; Eric Myon & Charles Taieb, Pharmaco Economics & Quality Of Life Department, IRPF, Boulogne Billancourt, France

# 1748/VALIDATING THE PSORIATIC ARTHRITIS QUALITY OF LIFE (PSAQOL) INSTRUMENT Stephen P. McKenna, Lynda C. Doward & Diane Whalley, Galen Research, Manchester; Alan Tennant, Paul Emery, Douglas J. Veale, Rheumatology & Rehabilitation Research Unit, University of Leeds, Leeds, UK

Introduction: Quality of life enables patients to assess the different consequences of the disease and/or treatment. Skin disorders have a strong impact on the patient’s physical and psychological well-being and the numerous quality of life studies carried out among adults are here to confirm this. However, very few studies have been performed to assess the quality of life in children. This is mainly attributable to the fact that there is a lack of questionnaires to do this. Method: A specific scale developed by Professor A. FINLAY (C-DLQI – Children-Dermatology Life Quality Index) is completed by each child before starting the course of treatment at Ave`ne Hydrotherapy Center, after 3 weeks of treatment, after 3 months and after 6 months. Results: Every year, Ave`ne Hydrotherapy Center welcomes a growing number of children suffering from chronic dermatosis. It thus seemed relevant to assess the quality of life of these children (4–15 years old) The following initial analysis has been obtained from the 30 first questionnaires completed upon inclusion and after 3 months. The DLQI score upon inclusion was 23.58 and 11.60% after 3 months. The results show an improvement in the patient’s quality of life 3 months after the hydrotherapy course of treatment. This difference is statistically significant (p ¼ 0.010). After 3 weeks of treatment, the C-DLQI score was 20.3% (p ¼ 20.32). Conclusion: These results clearly show that even some time after the end of the hydrotherapy course of treatment, quality of life in children suffering from chronic dermatosis continues to be improved.

Purpose: Psoriatic Arthritis (PsA) is an inflammatory form of arthritis occurring in association with psoriasis affecting almost 1% of the population. The multi-faceted nature of PSA results in complex manifestations of impairment and disability, which cause difficulties for impact assessment. The PsAQoL, a needs-based quality of life (QoL) instrument, has recently been developed to help evaluate the effectiveness of alternative treatments for the condition. The content of the instrument was derived from in-depth qualitative interviews with 50 patients and tested by interviews with a new sample of PsA patients. A 46-item measure went forward for psychometric assessment. This paper reports the results of the psychometric testing of the PsAQoL. Methods: Item reduction was achieved by administering a postal survey to 94 PsA patients. Rasch analyses were applied to identify misfitting items and those that worked differently by age or disease duration. A second, test–retest postal survey was conducted with a new sample of 286 patients to assess psychometric and scaling properties. Results: The first survey identified a preliminary 35-item measure with good scaling properties and minimal differential item functioning. Following the second postal survey a 20-item version of the PsAQoL was derived with good item fit and item stability. It was shown to have excellent test–retest reliability (0.89) and internal consistency (0.91). Scores on the measure were related to important clinical variables (including pain and fatigue) and to current perceived health and whether respondents were having an arthritis flare-up. It is concluded that the PsAQoL should prove to be a valuable instrument for assessing patient-based outcome in clinical trials and practice.

# 1815/QUALITY-OF-LIFE AND ECONOMIC IMPACT OF PRIMARY AXILLARY HYPERHIDROSIS: RESULTS FORM A NORTH AMERICAN CLINICAL STUDY PROGRAM Jonathan W. Kowalski, Global Health Outcomes Strategy & Research, Allergan, Irvine, CA; D. Glaser, St. Louis University School of Medicine, St. Louis, MO, USA; N.J. Lowe, 3 Cranley Clinic for Dermatology, London, UK; A. Ravelo, Allergan, Inc., Irvine, CA, USA Purpose: To describe the effect of primary axillary hyperhidrosis (HH) on patients’ daily lives and medical resource utilization. Methods: Patients enrolled into a prospective study conducted at 18 North American centers completed at baseline the validated 41-item baseline module of the Hyperhidrosis Impact Questionnaire (HHIQ). Patient inclusion criteria included: (1) gravimetric measurement of at least 50 mg of spontaneous sweat production in each axilla over 5 min at rest and; (2) score of 3 or 4 on the Hyperhidrosis Disease Severity Scale (i.e., patient report of underarm seating as barely tolerable or intolerable, and frequently or always interfering with daily activities). Descriptive and summary statistical analyses were performed. Results: 322 patients enrolled and completed the HHIQ. 81% indicated being somewhat or very dissatisfied with their ability to perform current work activities due to their HH. The percentage of patients at least moderately limited in various daily activities ranged from 77% (meeting people for the first time) to 43% (sexual activities). HH caused 85% to feel less confident than they would like, 48% to become frustrated with many daily activities, and 36% to be unhappy or depressed. 75% changed clothing 2 or more times daily and 26% spent 15 min or more daily treating their HH. 50% reported being at least moderately emotionally affected by HH. Previous treatment for HH included: topical antiperspirants (93%), prescription medications (18%), herbal remedies (9%), or iontophoresis (6%); however, 89% reported dissatisfaction with their current treatment. 19% reported at least 1 visit to a dermatologist and 30% reported at least 1 visit to a general practitioner in the past 12 months due to HH. Conclusion: Primary axillary hyperhidrosis results in occupational, psychological, and physical impairment. Both patients and society may benefit from increased recognition of this debilitating condition, and from the development of, and access to, new safe and effective therapies for primary axillary hyperhidrosis.

# 1217/IMPACT OF HYDROTHERAPY CARES ON THE QUALITY OF LIFE OF PATIENTS’ SUFFERING FROM SKIN DISEASES Chantal Segard, Francois Verriere & Therese Nocera, Medical Direction, Laboratoires Dermatologiques Ave`ne; Eric Myon & Charles Taieb, Pharmaco Economics & Quality of Life Department, IRPF, Boulogne Billancourt, France Introduction: Skin diseases have a strong impact on the physical and mental well-being of the patient. This is confirmed by the large number of quality of life studies that exist. It is clear that dermatological diseases affect not only the life of the patient but also that of his/her partner and family. The Ave`ne Dermatological Hydrotherapy Center, welcomes over 2500 patients a year suffering from skin diseases. The objective of the study is to demonstrate the relevance of the long term effects of hydrotherapy treatments on patients Quality of Life. Method: A specific scale (DLQI – Dermatology Life Quality Index) are completed by each patient at their arrival at the Ave`ne Dermatological Hydrotherapy Center (inclusion), at the end of hydrotherapy cares (3 weeks), and 6 months after the hydrotherapy cares. The completed questionnaires were returned by post. Results: In this first analysis, the first 200 patients suffering from the three following conditions: psoriasis, atopic dermatitis were taken into account and analysed at inclusion and at the end of the hydrotherapy cares. The DLQI score at inclusion was 34.55% At the end of hydrotherapy treatments the DLQI score was 19.17% (p < 0.00001). The first 30 patients were taken into account and analysed at inclusion and at 6 months after the hydrotherapy cares. For these 30 patients, the DLQI score at inclusion was 32.1%. At 6 months the DLQI score was 24.2% (p < 0.02). Conclusion: These first results show evidence of an improvement of patients’ quality of life after 3 weeks of hydrotherapy treatments (p < 0.00001). These results also demonstrate the hydrotherapy cares persistency effect as shown by the improvement of patients’ quality of life 6 months after hydrotherapy treatments (p < 0.02).

778 # 1198/PSORIASIS & QUALITY OF LIFE: SPOT STUDY, RESULTS AT 3 MONTHS Eric Myon & Charles Taieb, Pharmaco Economics & Quality of Life Department, IRPF, Boulogne Billancourt, France Context: Psoriasis is a chronic disease that affects 4.7% of the European population. Even though it is not life-threatening, it is a very disabling condition in daily life because of the damage that it does to the body image. Therefore, the improvement of the patients’ quality of life is one of the medical treatment’s priorities. Methods: SPOT is a longitudinal, prospective observational program that follows 800 patients with psoriasis in France. One group was treated with topical retinoid, the other was treated with a vitamin D3 derivative, and the specific quality of life was evaluated through the PDI questionnaire. The patients are questioned at the time of inclusion and at 3 months. Objective: To assess the impact of a treatment on the patients’ quality of life. Results: In this first analysis, the first 167 patients were included and were analysed at inclusion and 3 months after initiation of the treatment. Retinoid group n ¼ 66, vitamin D3 derivative group n ¼ 101. The PDI score at inclusion was: Retinoid group: 14.7% (SD ¼ 14.6), vitamin D3 derivative group: 13.6% (SD ¼ 13.0). The PDI score after 3 months was: Retinoid group: 11.8% (SD ¼ 13.9), vitamin D3 derivative group: 11.1% (SD ¼ 12.2). These first results show evidence of an improvement of the patients’ quality of life 3 months after treatment (p < 0.003 for the retinoid group, p < 0.011 for the vitamin D3 derivative group). PDI provides a finer analysis based on five dimensions, which are: daily activities, work, personal relationships, leisure and treatment. The vitamin D3 group was improved significantly with regard to the leisure dimension (p ¼ 0.0087) while the improvement was borderline significant for the work dimension (p ¼ 0.058). The retinoid group was improved significantly in terms of the daily activities (p ¼ 0.0009) and the treatment (p ¼ 0.0001) while the improvement was borderline significant for the work (p ¼ 0.056) and leisure dimensions (p ¼ 0.059). Conclusion: As shown by the PDI, an improvement in the patients’ quality of life was observed following their management. # 1197/PSORIASIS AND QUALITY OF LIFE SPANISH RESULTS Juana M. Delmolino, Accion Psoriasis; Eric Myon & Charles Taieb, Pharmaco Economics & Quality of Life Department, IRPF, Boulogne Billancourt, France Introdution: Skin diseases have a strong impact on the physical and mental well-being of the patient. This is confirmed by the large number of quality of life studies that exist. Objectives: To evaluate the effect of psoriasis on quality of life of patients in spain. METHOD: 4500 anonymous questionnaires (comprised of two scales : the Psoriasis Disability Index (PDI) plus 10 questions concerning treatment and evolution of psoriasis) were sent, via a Psoriasis Patient Support Group (AccionPsoriasis). Results: 1900 questionnaires were returned (june 2002): response rate 42%. An analysis of the first 810 questionnaires was realised. The sex ratio Men (M)/Women (W) was: 49/ 51. Mean age: 42 years. Mean age of diagnosis: 21.8 years. The average to the total score was 8.47 (SD 7.2 rank 0–39) i.e. 18.82 (SD: 17.2) when reported to a scale of 0–100. Significative difference was observed between M and W for the global handicap score 20.73 vs. 16.95 p < 0.002. Two groups was identified: pouse´e de psoriasis (84%), pas de pousse´e de psorisis (12%) – no answers (4%) – Psoriasis had a greater adverse affect on patients with crise: PDI score ¼ 20.24 (SD 17.52) than the patient without actual crise PDI score ¼ 8.98 (SD 11.34). This difference was significative p < 0.00001. Conclusion: These results highlight the value of appropriate and relevant psychological and medical environment for patient suffering from psoriasis. # 1196/PSORIASIS AND DEPRESSIVE SYMPTOMATOLOGY: FIRST RESULTS FROM A EUROPEAN STUDY Eric Myon & Charles Taieb, Pharmaco Economics & Quality of Life Department, IRPF, Boulogne Billancourt, France Objective: Chronic skin diseases have a heavy impact on the physical and mental well-being of the patient. This is confirmed by the large

number of quality of life studies that exist. Paradoxically, although numerous evaluations of depressive symptomatology have been performed in the USA and in Canada, no study has evaluated depressive symptomatology in European patients. This was why it was decided to perform, in association with the three largest patient associations in Europe – the APLCP in France, Accion Psoriasis in Spain, and ADIPSO in Italy – an evaluation of depressive symptomatology in patients suffering from psoriasis. Tools: The CES-D scale (Center for Epidemiologic Studies-Depression scale) was developed in the USA at the National Institute Of Mental Health to perform epidemiological studies of depressive symptomatology in the general population. Method: In total the CES-D scale was sent or remitted to 7800 patients with psoriasis in 2002 and 2003 throughout Europe (1500 for France, 5000 for Italy and 1300 for Spain). The completed questionnaires were returned through prepaid envelopes. Results: The response rates in France, Spain and Italy were respectively: 44, 23 and 60%. The male/female gender ratio was as follows: 46/54 for France, 54/46 for Spain and 61/39 for Italy. In France depressive symptomatology was observed in 47.3% of men (average score 17.7) and 49.1% of women (average score 23.4), in Spain depressive symptomatology was observed in 44% of men (average score 17.8) and 55% of women (average score 21.0) whereas in Italy depressive symptomatology was observed in 34% of men (average score 17.2) and 62% of women (average score 21.0). Conclusion: These results show that despite the differences within these three countries, a highly satisfactory level of participation of results is observed which converge towards an important number of patients with depressive symptomatology; the ration being higher in patients suffering psoriasis than in the general population. # 1230/WHAT DO THE DERMATOLOGY LIFE QUALITY INDEX (DLQI) SCORES REALLY MEAN IN TERMS OF PATIENTS’ OVERALL QUALITY OF LIFE Y. Hongbo, Dermatology, UWCM; M.A. Harrison & M.S. Salek, WSP Centre for Socioeconomic Research, Cardiff University; A.Y. Finlay, Dermatology, UWCM, Cardiff, UK The measurement of patients health related quality of life (HRQoL) in dermatology is becoming increasing important to monitor change and aid decision making. The DLQI is a dermatology-specific measure of patients HRQoL which has been adopted extensively. Although in general high scores equate to high impairment, basic information about interpretation of detailed meaning of scores is still missing both for DLQI and for other HRQoL measures. The aim was to determine the relationship between DLQI scores and the patients’ view of their overall impairment of HRQoL. Nine hundred and forty two general dermatology outpatients referred to University Hospital of Wales, who met the inclusion and exclusion criteria, were posted a DLQI questionnaire and an Overall Question (OQ) asking for their view of their overall HRQoL. 675 (71.7%) responded in the first 6 weeks. Of these 567 (60.2%) were evaluable: male 237, female 330; mean age 56.3, SD 20.8, range 16–98 years. The mean DLQI score (max 30) was 5.07, SD 6.09, range 0–30; and the mean of the single HRQoL OQ score (max 4) was 1.23, SD 1.22, range 0–4. The relationship between the DLQI scores and OQ scores was examined. For patients with DLQI scores 0–1 (n ¼ 224), the corresponding OQ score ¼ 0 (No effect at all on patient’s overall HRQoL); DLQI scores 2–5 (n ¼ 153), OQ score ¼ 1 (A small effect on patient’s overall HRQoL); DLQI scores 6– 12 (n ¼ 119), OQ score ¼ 2 (Moderate effect on patient’s overall HRQoL); DLQI scores 13–22 (n ¼ 55), OQ score ¼ 3 (Very large effect on patient’s overall HRQoL); DLQI scores 23–30 (n ¼ 16), OQ score ¼ 4 (extremely large effect on patient’s overall HRQoL). In order to examine the correlation between aggregated DLQI scores and individual patient scores, 30 patients were randomly chosen to correlate their scores with the remaining 537 patients. There was a very high correlation (r ¼ 0.95, p < 0.01); this demonstrates the representativeness of the DLQI scores. There was no gender difference in the comparisons. These findings will contribute to the meaningfulness of the DLQI scores, which may lead to wider acceptability in a clinical setting.

779 # 1808/HOPE AND QUALITY OF LIFE IN THE NORWEGIAN GENERAL POPULATION Tone Rustoen & Astrid K. Wahl, Faculty of Nursing, Oslo University College, Oslo; Berit R. Hanestad, Department of Public health and Primary health care, University of Bergen, Bergen, Norway The purpose of the study was to analyze the association of the perception of hope and quality of life in the Norwegian general population. Hope is often being referred to as a coping strategy important for quality of life experience. Four thousand Norwegian citizens, aged 19– 81 years randomly drawn by Statistics Norway from the National Register were sent a questionnaire in the year 2000. The sample consists of 1893 persons (52% female, mean age 45.2, SD 15.8 years). Respondents filled in the WHOQOL-BREF (26 items). Scores range from 4 to 20 in the physical, psychological, social relationship and environmental domains respectively, with higher scores denoting better quality of life. A second questionnaire, Herth Hope Index (HHI), 12 items, had scores ranging from 12 to 48, with higher scores indicating higher hopefulness. The mean score for the HHI was 34.9 (SD 3.7) and for the WHOQOL-BREF, 15.8 (SD 2.8), 15.2 (SD 2.4), 14.9 (SD 2.7) and 15.3 (SD 2.4) for the physical, psychological, social relationship and environmental domains respectively. Regression analysis show that the explained variance of HHI in the WHOQOL-BREF domains was 8% (stand b 0.28) in the physical domain, 22% (stand b 0.47) in the psychological domain, 11% (stand b 0.33) in the social relationship domain and 9% (stand b 0.30) in the environmental domain (for all p < 0.0001). These results indicate that hope is associated with the quality of life domains measured by the WHOQOLBREF. The strongest relationships occurred between hope and psychological and social aspects of quality of life. The importance of hope for improving quality of life is an issue for further research. # 1680/QUALITY OF LIFE AND SOCIAL CHANGE IN THE CZECH REPUBLIC AFTER YEAR 1989 Hnilicova Helena & Janeckova Hana, Institute of Medical Humanities, 1st Medical School, Prague, Czech Republic, Czech Republic In frame of the research project of the ‘Goals of Medicine and Quality of Life’ we were interested in impact of the political and social changes during 90’s on the real life. Project is based on the empirical study investigating the quality of life of the ‘healthy’ population in a comparison with patients suffering from chronic illness. The main objective was to understand how people themselves experienced it in the various areas of their life. We tried: (1) the main attributes of the QoL of the Czech people in the 90’s. In frame of the research project of the ‘Goals of Medicine and Quality of Life’ we were interested in impact of the political and social changes during 90’s on the real life. Project is based on the empirical study investigating the quality of life of the ‘healthy’ population in a comparison with patients suffering from chronic illness. The main objective was to understand how people themselves experienced it in the various areas of their life. We tried: (1) the main attributes of the QoL of the Czech people in the 90’s (2) To clarify the impact of the social changes on the subjectively perceived QoL during transition period 1990–2000. We found that 10 years after, the Czech population was in principal quite satisfied with overall QoL. There were significant differences according to age and gender. Younger generation and men were more satisfied that older people and women. The most valued were changes related to the basic human rights and freedom. Changes in economic situation and in job opportunity were valued less. The worst impact was identified for the interpersonal relations.

# 1432/EVAULATION OF PSYCHOLOGICAL, SOMATOVEGETATIVE, AND SEXUAL SYMPTOMS OF NORMAL KOREAN MALE POPULATION USING THE KOREAN-VALIDATED AGING MALES’ SYMPTOMS’(AMS) RATING SCALE Sehyun Kim, Preventive Medicine, College of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do; Ji Y. Yi, Public Health, Korea University, Seoul; Hye J. Park, Nursing; Young G. Kang, Family Medicine, College Of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do, Korea A standardized instrument to measure the severity of symptoms for aging males is not yet available in Korea. The objective of the study was twofold: to validate Korean translation aging males’ symptoms’(AMS) rating scale, and to evaluate symptoms of all three dimensions of AMS in normal male population in Korea. A total of 350 males aged over 40 were randomly selected and invited to participate in the study. We have been following rigorous translation steps for Korean translation of AMS. We used the same reference cutoff values of the original German AMS when evaluating psychological, somatovegetative, and sexual factors of AMS. One hundred fifty four returned completed Korean AMS (response rate 44%). Mean age of the subjects is 51.6 (SD ¼ 4.47, range 41–71). Based on self-reported background information, most subjects were healthy normal. The Korean AMS had high reliability overall and in all three domains (Cronbach’s a 0.83–0.90). Those who are not currently working or who had experienced stressful situations (chest pain under stress or experienced life-threatening events) scored worse for psychological and somatovegetative symptoms than those who are not (p < 0.05). Most had mild to severe symptoms for psychological factor, mild to moderate symptoms for somatovegetative factor, and moderate to severe symptoms for sexual factor. The Korean translated AMS has been shown to have excellent scaling properties, and this scale can be applied easily in medical practice for the assessment of severity of aging males’ symptoms in males in Korea. The Korean male population has shown higher degree of aging males’ symptoms in psychological and sexual factors than somatovegetative factor. # 1664/THE MEASUREMENT OF PREFERENCES TOWARD HEALTH STATES USING THE CLASSIFICATION AND MEASUREMENT SYSTEM OF FUNCTIONAL HEALTH (CLAMES) Julie Bernier & Sarah Gorber, Health Analysis and Measurement Group, Statistics Canada, Ottawa, Ontario, Canada While constructing a Summary Measure of Population Health (SMPH), one needs to associate a morbidity score with each of the health states that are to be included in the calculation of that measure. Elicitation of preferences is an interesting way of achieving that goal since it produces a measure that is on a continuous interval scale. Such an exercise was conducted in Canada using the CLAMES to describe the health states under study. The CLAMES is a multi-attribute system made of 11 attributes each having 4 or 5 possible levels. The measurements have been done through a series of focus groups across Canada. Each group was made of 10 or 11 participants. A session is a day long and consists of preference measurement exercises. In order to do these exercises, the health states have been split into two groups: the anchor states and the regular states. The anchor states are 12 selected states that have been measured collectively in every group using a standard gamble procedure. It is then possible to use the anchor states to measure inter-group variability or to do comparisons within specific socio-demographic groups. The preference score for these anchors and the corresponding analysis of the variance will be presented. In addition to the anchor states, 193 regular states were evaluated which are a combination of states generated by the definition process and hypothetical states created to meet some design requirements. These states are used to construct an estimation function that allows for the estimation of the score associated with any combination of the attribute levels. One possible approach when trying to construct such a function is the statistical approach proposed by Brazier et al. (2002). This method has been applied to the data and results will be presented.

780 # 1327/COMPARISON OF EURO-QOL EQ-5D MEASURED AMOUNG RURAL RESIDENTS WITH URBAN FACTORY WORKERS IN KYOTO, JAPAN Yoshiyuki Watanabe, Mari Fujita, Junko Nagura, Kyohei Hayashi & Kotaro Ozasa, Department of Epidemiology for Community Health and Medic, Graduate School of Medical Science, Kyoto Prefectura, Kyoto; Chisato Hamashima, Cancer Etiology Section, Cancer Information and Epid, National Cancer Research Institute, Tokyo, Japan We conducted EQ-5D surveys to investigate the health status in a rural Japanese population and among urban factory workers. All the residents aged 30 years old or older in a rural area of Kyoto were asked to answer the Japanese clinical version of EQ-5D using a selfadministered questionnaire from June 1999 through February 2000. All the urban factory workers (age range: 18–65) were also asked to answer the same questionnaire in May 2001. A total of 4900 (70.0%) out of 7000 rural residents and 1368(65.1%) out of 2130 factory workers responded. The proportion of ‘no problems’ decreased with age in both groups for all of the five questions of EQ-5D except for – Anxiety/Depression among factory workers. Factory workers in their thirties showed the lowest proportion of – no problems for – Anxiety/ Depression compared to other age groups. The proportion of a high tariff-QOL score (=1) calculated using a basic tariff for Japanese was higher among urban factory workers than rural residents in each age group. The proportion also decreased with age among rural residents though there was no such relationship among urban workers. These data suggest that there is a healthy worker’s effect among urban factory workers. # 1582/VETERANS’ WITH POSSTRAUMATIC STRESS: HOW BAD IS THEIR QUALITY OF LIFE WHEN COMPARED TO OTHER VETERANS AND WITH THE CIVILIAN POPULATION? Graeme Hawthorne, Psychiatry, Melbourne University, Heidelberg Heights, Victoria, Australia Posttraumatic stress (PTSD) is a chronic condition with which veterans live over many years. The symptoms are trauma-related hyper-arousal (e.g. vigilance, startle reflex), intrusive memories (e.g. nightmares), emotional numbing and avoidance behaviours. The sequelae include depression, anxiety and substance abuse. This paper explores the HRQoL of veterans attending PTSD treatment. HRQoL instruments were the WHOQOL-Bref and the AQoL. The scores were compared with Australian population norms and with randomly selected older veterans without PTSD. On all four WHOQOL-Bref domains the PTSD-veterans obtained scores significantly lower than both the community veterans and the population. AQoL findings were similar, showing PTSD-veterans had worse HRQoL. When scores were examined by PTSD severity, there were monotonic relationships: those in the worst PTSD quartile obtained values indicative of extremely poor HRQoL. There were also relationships with other measures and with the use of health care resources. WHOQOL-Bref and AQoL items were ordered by their relative efficiency compared with community veterans. Those areas where PTSD-veterans reported particularly poor HRQoL were losses with respect to anxiety, self-worth or negative feelings, poor concentration, relationships (including intimacy, family roles and friendships), and activities of daily living. The results indicate that PTSD is associated with major losses in almost all dimensions of HRQoL, and that veterans seeking treatment have significantly lower HRQoL when compared with the general population or with non-treatment seeking veterans. Comparisons with other health conditions reveals these losses have shocking effects on PTSD-veterans’ HRQoL. Measuring HRQoL among veterans quantifies the extent of their problems, enables comparisons with other conditions, and provides veterans with a personal voice at the system level. Both the WHOQOL-Bref and AQoL appear suitable instruments for this purpose. Based on the findings, treatment should address those areas identified as causing particular distress

# 1628/DOES MOOD AFFECT SEIQOL-DW IN ELDERLY FEMALES? RESULTS OF A MOOD INDUCTION STUDY Hanne Bruhn, Saskia Teunisse & Louise H. Phillips, Department of Psychology, University of Aberdeen, Aberdeen, United Kingdom Individual Quality of Life (IQoL) is an increasingly popular concept, but is IQoL subject to transient influences, such as mood, as are global QoL questions? Global QoL questions have been found to be influenced by mood using mainly student samples (see Kahneman, Diener & Schwarz, 1999, for a review). The current study focuses on the impact of mood on IQoL in the ageing population using a well known measure: the Schedule for the Evaluation of Individual Quality of Life (SEIQoL): a Direct Weighting procedure for Quality of Life Domains (SEIQoL-DW (O Boyle, Browne, Hickey, McGee, & Joyce, 1995). Forty older females (mean age 71.3 years, SD 5.6) took part. All were screened for absence of depression and cognitive impairment. The baseline QoL measure was the Satisfaction With Life Scale (SWLS) (Diener, Emmons, Larson & Griffin, 1985). Film clips, selected on the basis of two pilot studies, were used as sad or neutral mood induction procedure (MIP). This procedure has been found to be the most effective for mood induction (Gerrads-Hesse, Spies, & Hesse, 1994). Participants watched the film clip according to MIP condition and completed the ratings of their IQoL areas immediately after. There were no significant differences between MIP conditions in terms of demographic variables, MMSE, depression scores, and baseline QoL. There was a significant change in mood in the sad MIP condition (p < 0.05). There were no significant differences between MIP conditions in mean SEIQoL-DW index. However, a significant difference between MIP conditions was found for one of the SEIQoL index components. The implications of these findings will be discussed. # 1103/FEASIBILITY OF VERY FREQUENT WEB-BASED QOL HOME SELF-ASSESSMENT AFTER HSCT Nigel E. Bush, Biobehavioral Sciences; Carol M. Moinpour, Southwest Oncology Group Statistical Center, Fred Hutchinson Cancer Research Center, Seattle, WA; Gary W. Donaldson, Department of Anesthesiology, University of Utah, Salt Lake City, UT; Mel R. Haberman, College of Nursing, Washington State University, Spokane, WA, USA Survivors of hematopoietic stem cell transplantation (HSCT) face late complications that may be debilitating or even life threatening. Our own data suggest that sensitive QOL measures after HSCT can function as leading indicators of adverse clinical events such as relapse and death. If we can measure QOL frequently enough, we may enable clinicians to identify early signs of adverse events and intervene prospectively to minimize complications. However, conventional QOL measurements typically are infrequent, and almost certainly miss clinically salient, short-term spikes in QOL. The emergence of the Internet has made frequent data capture potentially convenient and reliable. We have developed and implemented a comprehensive webbased system for collecting HSCT patient’s QOL self-assessments from their homes using tailored online questionnaires administered at daily and monthly intervals, during the critical first year of recovery after transplant. In a field-test of the feasibility of this system, we found excellent patient compliance and high user satisfaction. Eighty-four adult HSCT patients with home web-access were recruited posttransplant immediately prior to returning home. Of those, 72(86%) agreed to participate and were enrolled on study. Subsequent attrition due to death, relapse or loss of internet access reduced this number to 63. The feasibility field-test period covered the first 29 weeks after return home during which patients completed daily and monthly QOL self-assessments online. At the time of analysis, 42 of the 63 patients had returned home and were completing online assessments. Compliance exceeded our requirements of 2–3 completions a week, with patients completing nearly four daily online questionnaires per week. Fewer than 1 of every 200 online questions were missed. All our larger monthly questionnaires were completed. More than 90% of patients were satisfied with online daily reporting and found the website easy and convenient. The new technology and analytical strategies we have introduced for dynamic, frequent, QOL assessment should prove widely applicable to other disease populations.

781 # 1685/FEASIBILITY STUDY OF A COMPUTER-BASED VERSION OF THE SEIQOL-DW IN GASTRO-INTESTINAL CANCER PATIENTS A˚sa K. Lindblad, Department of Pharmacy, Pharmaceutical Outcomes Research, Uppsala, Sweden; Lena Ring, Department of Psychology, Health Services Research Centre, Dublin, Ireland; Rebecka Jansson, Department of Pharmacy, Pharmaceutical Outcomes Research, Uppsala; Preben Bendtsen, Department of Health and Society, Social Medicine and Public Health, Linko¨ping; Bengt Glimelius, Division of Oncology, Radiology and Clinical Immunology, Uppsala, Sweden An approach to QoL assessments that captures individual concerns may be useful in cancer patients where treatment often is palliative. The Schedule of Evaluation of Individual Quality of Life-Direct Weighting (SEIQoL-DW) allows for individualized QoL assessments. The objective of the study was to assess whether a prototype of a computer-administered version of the SEIQoL-DW is feasible in clinical practice. A second objective was to obtain a deeper understanding of gastro-intestinal (GI) cancer patients´ definition of QoL, using a semistructured qualitative interview. A consecutive sample of 20 patients with varying stages of GI-cancer agreed to participate. The computer test and the interview took place in relation to a regular hospital visit. The participants completed the SEIQoL-DW on a touch screen computer assisted by the interviewer if needed. An optional prompt list of areas was included in the computer-administered version. The interviewer then used a protocol to evaluate the feasibility of the program, followed by the semi-structured interview regarding patients’ appraisal of QoL. Most participants thought that the computer-administered SEIQoL-DW was easy to understand (85%) and showed great interest in completing the computer test (90%). A majority (95%) believed that the result of the computer-administered SEIQoL agreed with their own present appraisal of QoL. The participants were able to nominate in median four areas without using the prompt list. Median time required to complete the test was 10 min (range 6–19). Family was a frequent nominated area as were social network (besides family) and leisure activities. Although few participants nominated health in the computeradministered version of SEIQoL-DW, health was frequently brought up in the following interviews. These preliminary data indicate that a computer-based version of SEIQoL-DW is feasible in clinical practice. # 1369/EVALUATING THE IMPORTANCE OF SENDING REMINDERS WHEN USING MAIL QUESTIONNAIRES Jose M. Quintana, Nerea Gonzalez, Felipe Aizpuru, Jose A. San Sebastian, Antonio Escobar & Cristobal Esteban, Unidad de Investigacion, Hospital de Galdakao, Galdakao, Vizcaya, Spain Mail questionnaires have several advantages, but also disadvantages. One of the most important is the amount of non-respondents. To avoid this, researchers recommend having as much as two reminders, after the first mailing, to increase the response rate. The goal of this study was to evaluate the characteristics of patients and if the level of satisfaction of those who respond to a first mailing was different from to those who respond to subsequent reminders. Patients randomly selected from four hospitals were contacted to fill out a hospitalised patient satisfaction questionnaire. The questionnaire included six domains. A total of 2600 patients were selected (650 by each hospital). The questionnaires were sent by mail 2 weeks after the patient had been discharge from the hospital. Patients who did not answer were sent a first reminder with a letter asking for returning the questionnaire and a second reminder with a new copy of the questionnaire if an answer had not been received two weeks afterwards. Multivariate linear regression analysis was performed. Response rate, after the two

reminders, was 73.5% for the whole sample, though only 40% of patients respond to the first mailing. Patients who respond to the first mail were younger, with a higher proportion of university education level, compared with those who respond to subsequent reminders. Also, those who answer to the first mailing had more satisfaction with the health care while hospitalised. Specifically, they showed higher scores, meaning higher satisfaction, on domains such as information, human care, intimacy and comfort, than those who answer to the reminders. Differences remain even after adjustment by socio-demographic variables. Main conclusion of this study is the importance of have as much as two reminders when using mail questionnaires. The socio-demographic characteristics of patients differed depending when they answer, and, also, their satisfaction level. Also, the response rate to a single mailing would be considered low, and the nonresponse rate high. Having only one mailing may importantly bias the conclusions of one study. # 1535/SELF-ADMINISTRATION AND STANDARDIZATION OF THE CHRONIC RESPIATORY QUESTIONNAIRE: A RANDOMIZED TRIAL Holger J. Schunemann, Clinical Epidemiology & Biostatistics, McMaster University, Hamilton, Ontario, Canada; Milo Puhan, Horten Centre, University of Zurich, Zurich, Switzerland; Gordon H. Guyatt, Clinical Epidemiology and Biostatistics, McMaster Universtity, Hamilton, Canada; Michaela Behnke, Pulmo Research, University of Hamburg, Hamburg, Germany; Marco Laschke, Medicine, Klinik Barmelweid, Barmelweid, Switzerland; Alfred Lichtenschopf, Medicine, Rehabilitationszentrum, Weyer/Enns, Austria; Otto Brandli, Medicine, Zurcher Hoehenklinik, Faltiberg/Wald, Switzerland Background: The Chronic Respiratory Questionnaire (CRQ) has demonstrated excellent measurement properties in patients with COPD, but in its original form it is limited by the requirement for interviewer-administration and the individualised dyspnoea questions. There is no validated German translation of an instrument to measure health related quality of life in patients with chronic obstructive pulmonary disease (COPD). The objective of this randomised controlled trial was to examine the measurement properties of the interviewer and self-administered German CRQ as well as of a standardised CRQ dyspnoea domain (all patients received the same questions). Methods: We conducted a multinational trial using a 2 · 2 factorial design. We randomly allocated 71 patients with COPD to complete the interviewer administered CRQ (CRQ-IA) or the self-administered CRQ (CRQ-SA) and other validation measures at the beginning and end of a respiratory rehabilitation program. In addition, we randomised patients to receiving either the standardised or individualised dyspnoea questions first. We assessed and compared responsiveness, cross sectional and longitudinal construct validity of each domain (dyspnoea, fatigue, emotional function and mastery). Results: Seventy-one patients participated in the study. We found good discriminative and longitudinal validity for both the CRQ-IA (n ¼ 38) and CRQ-SA (n ¼ 33). Responsiveness of the fatigue domain was higher for the CRQ-SA compared to CRQ-IA (p ¼ 0.02), but there was no difference in responsiveness on the other domains. Compared to the standardised dyspnoea domain the individualised dyspnoea questions tended to show greater responsiveness (greater ability to detect change in quality of life) for both the CRQ-IA and CRQ-SA (p ¼ 0.07 for the CRQ-IA and p ¼ 0.10 for the CRQ-SA). Conclusion: Individualization of questions may lead to greater responsiveness, but researchers in COPD, in particular those in German-language countries, can utilize any one of four CRQ formats that have proved both valid and responsive.

782 # 1828/A COMPARISON OF OBSERVED AND SELF-REPORTED HRQL IN A HOME CARE POPULATION Colleen J. Maxwell, Community Health Sciences, University of Calgary, Calgary, AB; Walter P. Wodchis & Adriana Venturini, Toronto Rehabilitation Institute, Toronto, ON; Jennifer D. Walker, Jenny X. Zhang & David B. Hogan, Community Health Sciences, University of Calgary, Calgary; David F. Feeny, Institute of Health Economics, Edmonton, AB, Canada

# 1840/QOL OF WOMEN ENGAGED IN TREATMENT FOR TRAUMA AND CO-OCCURRING SUBSTANCE ABUSE AND MENTAL DISORDERS Marion Becker, Sonal Pathak & David Thornton, Department of Mental Health Law and Policy, Univerisity of South Florida, Tampa, FL; Mark J. Atkinson, Outcomes Research, Pfizer, Kalamazoo, MI, USA

While broad multi-dimensional health status assessment is common in home care programs, such assessments are typically based on observer assessments of client functioning and rarely identify self-reported health-related quality of life (HRQL). This study examined the relationship between observed and self-reported health related quality of life (HRQL) in a community-dwelling home care population. Three hundred home care clients were assessed with the Minimum Data Set for Home Care v2.0. Self-reported HRQL was measured using the Health Utilities Index Mark 2 (HUI2) HRQL survey, and observed HRQL was measured by the Minimum Data Set-Health Status Index (MDS-HSI). Both provide summary HRQL scores and single-attribute scores for sensation, cognition, self-care, mobility, pain and emotion. The correspondence between both measures was assessed by examining the intra-class correlation (ICC) and the consistency of the relationship between HRQL and client characteristics. Both self-report and observed HRQL measures provided similar results. The overall HUI2 (self-report) score was 0.49 (SD ¼ 0.18) while the overall MDSHSI (observed) score was 0.54 (SD ¼ 0.17). Both observed and selfreported HRQL declined proportionately with increased age, cognitive impairment and dependency in activities of daily living. The overall ICC between the observed and self-report measures was 0.49. Singleattribute domain scores were most closely correlated for sensation and least for emotion. This study found analogous population-level results between observed and self-reported HRQL scores. Both observed and self-reported measures demonstrated a parallel relationship with cognitive and physical function. However, ICC analysis of variance results suggested that individual differences did exist between individual client and observers’ assessment of HRQL, particularly in the emotion domain.

Introduction: Improving health related quality of life (HRQL) is an overarching goal of treatment for substance abuse and mental disorders, yet little is known about this outcome among dually and triply diagnosed women. Purpose: The aim of this substudy was to examine the factorial dimensions of the Wisconsin Quality of Life Index (WQLI) among women being treated for trauma and co-occurring mental and substance abuse disorders. Sample: Participants included 349 women in community-based mental health treatment who met the criteria for a DSM-IV Axis I mental disorder or an Axis II personality disorder, as well as an Axis I substance-use disorder and a history of physical or sexual abuse. All were participants in the Triad Women’s Project, one of nine projects in the 5 year longitudinal Women, Co-occurring Disorders and Violence Study (WCDV). Methods: Among persons with severe and persistent mental illness the W-QLI has been shown to consist of three factorial dimensions (i.e., Life Satisfaction, Symptoms/ Outlook & Social Relations/Support) composed of items that assess eight apsects of HRQL: life satisfaction, occupational activities, psychological well-being, physical health, social relations, economics, ADL, and mental health symptoms. A confirmatory factor analysis was used to determine the reproducability of the WQLI factor structure in this triply diagnosed population. Findings: The original factor structure and loading pattern of the WQLI was confirmed using data from this sample of women. Also supporting the validity of the WQLI in this population, WQLI scores were shown to be higher among those with more education and social support, and lower among those with greater illness severity and living alone. Moreover, WQLI scales were shown to be responsive to changes in HRQL occurring over the course of 6–12 months of treatment. Conclusions: The three factorial dimensions of the WQLI were replicated in this unique mental health population and the resulting scales are responsive to important HRQL changes in these womens’ lives.

# 1593/QUALITY OF LIFE IN OUTPATIENTS WITH DEPRESSION COMORBID WITH SOCIAL PHOBIA Marcelo P. Fleck, Marcelo T. Berlim, Marco A. Caldieraro & Daniele P. Pavanello, Psychiatry and Forensic Medicine, Federal University of Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brazil

# 1591/QUALITY OF LIFE IN DEPRESSED BRAZILIAN OUTPATIENTS WITH SUICIDAL IDEATION Marcelo P. Fleck, Marcelo T. Berlim, Daniele P. Pavanello & Marco A. Caldieraro, Psychiatry and Forensic Medicine, Federal University of Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brazil

Objective: To compare the scores of quality of life of depressed patients with and without a diagnosis of comorbid social phobia. Methodology: Thirteen depressed individuals with comorbid social phobia and 39 depressed individuals without any psychiatric comorbidity according to the Mini International Neuropsychiatric Interview were investigated. All the patients consulted at the Mood Disorders clinic of the Hospital de Clı´nicas de Porto Alegre. To determine the quality of life we used the World Health Organization Quality of Life Instrument Short Version (WHOQOL BREF) and to quantify the intensity of the depressive episode we used the Beck Depression Inventory (BDI). Results: The investigated groups did not differ in terms of age, gender, school level, main diagnosis, marital status, family history of depressive disorder and presence of physical disease. After the statistical analysis, the depressed patients with comorbid social phobia presented significantly lower scores of quality of life in all the assessed domains (psychological, physical health, environmental, and social relationships) when compared with the depressed patients without a psychiatric comorbidity. There was no difference between the study groups in terms of the intensity of the current depressive episode. Conclusion: The comorbidity of a depressive disorder with social phobia is associated with a deficit of quality of life that is superior to that observed in patients that only present pure depression.

In this study the authors evaluated the quality of life (QOL) of depressed patients with and without suicidal ideation. The main objective was to quantify the magnitude of the impact of suicidality in the subjective well-being and day-to-day functional ability of individuals with depressive disorders. For this purpose the authors investigated 70 outpatients with depression attending a university hospital in southern Brazil. They were examined using the Portuguese versions of the World Health Organization Quality of Life Instrument Short Version (WHOQOL BREF), and of the Beck Depression Inventory (BDI). Results indicated that suicidal ideation was present in 50% of all patients. Additionally, depressed patients with suicidal ideation scored significantly worse (p < 0.05) in all QOL domains (i.e., physical, psychological, social relations, and environmental) when compared with those without suicidality. This finding, in sum, reinforces the notion that suicidal ideation is associated with appreciable morbidity in terms of pervasive QOL deficits.

783 # 1683/QUALITY OF LIFE AFTER SEXUAL SURGICAL REASSIGNMENTS Jitka Potribna, Department of Psychiatry, University Hospital Ostrava, Ostrava 1; Eva Dragomirecka, Department of Social Psychiatry, Prague Psychiatric Center, Praha 8, Bohnice, Czech Republic Sexual surgical reassignments have been offered to persons with severe sexual identity disorders since 1950s. In spite of progress in the surgical procedure the results are still ambiguous. The treatment, however, helps a number of transsexuals to live in accordance with their gender identity. The aim of this study was to compare whether quality of life of transsexuals after the sexual reassignments differs from quality of life of persons without sexual identity disorders living in the same sexual role. 21 persons (17 female-to-males and four maleto-females) that were treated in the Ostrava University Psychiatric Outpatient Department between 1973 and 2002 completed the Subjective QUAlity of Life Assessment (SQUALA). The SQUALA comprised of the Importance and Satisfactions scales in 23 domains of life, allowing assessment of value orientation and individual satisfaction. The partial quality of life score of each domain is the product of the importance and the satisfaction values of the given domain; the global quality of life score represents the sum of all these partial scores. The results were compared with those of persons of the same age, sex, marital status and education. The results showed that the transsexuals attributed higher importance to the domains mental well-being and work and lower importance to the domain SEXUALITY. On the satisfaction scale, the transsexuals were significantly more satisfied with the domain self-care while the controls with the domain children. The control group had higher-quality of life in two domains: children and sexuality. The total quality of life score of the transsexuals did not differ significantly from the control group. The global quality of life of transsexuals after change of their sexual identity seems to be comparable to the matched persons. Encouraging is the fact that no differences were found in the domains of interpersonal relations, love and family. However, they still had problems in their sexual lives and parental roles. Supported by research project CEZ MZ 00000023752. # 1347/THE VALUE OF HEALTH-RELATED QUALITY OF LIFE IN PRIMARY CARE PATIENTS WITH MAJOR DEPRESSIVE DISORDER Christophe Sapin, Altipharm, Paris; Bruno Fantino, Agoras, Lyon, France; Marie-Laure Nowicki, ICR - Psychosis, H. Lundbeck A/S, Paris, France; Paul Kind, Center for Health Economics, University of York, York, United Kingdom Major depressive disorder (MDD) is a prevalent psychiatric disorder associated with impaired patient functioning and reductions in healthrelated quality of life (HRQL). The present study describes the impact of MDD on patient HRQL as measured in terms of utility, and examines differences in variations by patient sociodemographic and clinical characteristics. Ninety-five French primary care practitioners recruited 250 patients with a DSM-IV diagnosis of MDD for inclusion in an 8week follow-up cohort. Patient assessments included the Montgomery Asberg Depression Rating Scale (MADRS), the Clinical Global Impression of Severity (CGI), the EuroQoL (EQ-5D) and the Short Form36 Item scale (SF-36). EQ-5D was converted into a single index summary score using previously published utility weights. The mean EQ-5D utility at baseline was 0.33, and 8% of patients recorded a current health state with a negative value worse than dead. There were no statistically significant differences in utilities by demographic features. Significant differences were found in mean utilities by level of disease severity assessed by CGI. Different clinical response profiles assessed by MADRS were revealed by EQ-5D at endpoint, with a mean of 0.85 for responder-remitters, 0.72 for responder-non-remitters, and 0.58 for non-responders. Even allowing for a moderate correlation between SF-36 and EQ-5D , differences in SF-36 between endpoint and baseline explained only 40% of the variance of the difference in EQ-5D during the study period. Self-reported patient valuations for depression can easily be obtained using EQ-5D, and are important for economic evaluations of new antidepressant compounds.

# 1367/THE SUBJECTIVE PERCEPTION OF COGNITIVE IMPAIRMENT IN PATIENTS WITH PSYCHOSIS: THE GEOPTE SCALE Julio Sanjuan, GEOPTE, Facultat de Medicina, Universitat de Valencia, Valencia; Prieto Luis, Health Outcomes Research Unit, Eli Lilly & Co, Alconbendas, Madrid; Jose´ M. Olivares, GEOPTE, Complejo Hospitalario Xeral-Cies, Vigo; Salvador Ros, GEOPTE, Hospital del Mar, Barcelona; Angel L. Montejo, GEOPTE, Hospital Universitario, Salamanca; Francisco Ferre, GEOPTE, Servicio de Salud Mental, Alcobendas; Fermı´n Mayoral, GEOPTE, Complejo Hospitalario Carlos Haya, Ma´laga; Miguel A. Gonzalez-Torres, GEOPTE, Hospital de Basurto, Bilbao; Manuel Bouson˜o, GEOPTE, Universidad de Oviedo, Oviedo, Spain Aims: Despite there is a growing interest in the assessment of the subjective perception of cognitive impairment in patients with psychosis there is a lack of standardized instruments specifically designed to evaluate such type of construct. The aim of this article is to present the characteristics and psychometric properties of a scale for measuring cognitive impairment as reported by patients with psychosis. Methods: The new GEOPTE scale gathers information from two sources: the patient’s subjective perception of his deficits and that of the informant or caregiver. It consists of 15 items, 7 for basic cognitive functions and 8 for social cognition. The scale was applied to 107 patients with a diagnosis of psychosis (according to DSM-IV), and gathered general clinical data, overall clinical impression, mood and degree of insight. Results: The GEOPTE scale presented an excellent internal consistency (Cronbach’s a: 0.84 for patient and 0.87 for informants). Factorial analysis identified two factors, which explained a total variance of 39%. The first factor was related to – basic cognitive function – items and the second to – social cognition – items. As regards the validity of the construct, the grading on the scale is closely related to Overall Clinical Impression, degree of insight and depressive symptoms. Conclusions: The GEOPTE scale for measuring social cognition in psychosis has an excellent psychometric behavior both in the degree of internal consistency and in the correlation with overall clinical variables, mood and degree of insight. # 1031/RESPONSE SHIFT IN WOMEN WHO HAVE BEEN PREGNANT WITH CANCER Mary Tederous-Williams, Counseling, School & Educational Psychology; James P. Donnelly, Counseling, School and Educational Psychology, University at Buffalo/SUNY, Amherst, NY, USA It is difficult to imagine two more medically and psychologically contradictory events co-occurring in a woman’s life than cancer and pregnancy. Yet cancer is in fact diagnosed in approximately one of every thousand pregnant women. Perhaps because the co-occurrence of these conditions is relatively infrequent, there have been no studies of the quality of life of such women. The present study represents a first attempt to do so via the analysis of 19 cases using a coding scheme derived from response shift theory. The 19 individuals in this study wrote narrative descriptions of their experience of cancer treatment during pregnancy that were published in a national support group newsletter from 1997 to 2002. The method of studying narratives employed the following steps: (1) Development of a coding scheme based on the model of response shift and quality of life described by Schwartz and Sprangers (2000). The coding scheme included the following elements: catalyst, antecedents, mechanisms, response shift and perceived quality of life. (2) Coding each narrative by two raters. Discrepancies in coding were resolved via discussion. (3) Analysis of the coded data via both basic qualitative and quantitative (proportion of sample describing specific aspects of response shift model) methods. The data were summarized in a graphic model illustrating the relative frequency of each specific component of the model (e.g., proportion of sample using ‘reframing expectations’). The analyses suggest that response shift is evident in the narratives of these women and that the components of response shift described in the Schwartz and Sprangers model were well represented in the data. The method and sample were inherently limited in that they are survivors who are able to report positive outcomes. Despite the limitations, the study may have implications for narrative analytic approaches to response shift, and may bring attention to an understudied and important population.

784 # 1187/INCORPORATING THE THENTEST INTO THE STRUCTURAL EQUATION MODELING (SEM) APPROACH TO RESPONSE-SHIFT DETECTION Frans J. Oort, Mechteld R.M. Visser & Mirjam Sprangers, Medical Psychology, AMC/University of Amsterdam, Amsterdam, The Netherlands In the SEM approach to response-shift (RS) detection, true change, recalibration, reprioritization, and reconceptualization are operationalized as changes in common factor means, specific factor means, common factor loadings, and common factor patterns, respectively. However, recalibration can only be detected through SEM if, Assumption 1, it does not affect all variables to the same extent (i.e., no non-differential recalibration). Another way of RS detection is by extending the pretest–posttest design with a thentest. True change and (recalibration) RS can then be assessed by calculating posttest– thentest and pretest–thentest differences. However, the posttest– thentest comparison is only valid if, Assumption 2, both tests are equally affected by RSs, and the pretest–thentest comparison is only valid if, Assumption 3, all occurring RSs are of the recalibration type. These three assumptions can be investigated by incorporating the thentest into the SEM approach, as is illustrated with quality of life (QOL) data from 170 cancer patients. QOL was measured with the SF36, administered before and after invasive surgery. Using a threefactor measurement model, SEM showed two RSs that affected the posttest only: reprioritization of Social Functioning (p ¼ 0.014) and recalibration of Role Physical (p ¼ 0.006); two RSs that affected the thentest only: reconceptualization of Vitality (p ¼ 0.010) and recalibration of General Health (p ¼ 0.007); and three RSs that affected both posttest and thentest: reconceptualization of General Health (p < 0.001) and recalibration of Bodily Pain (p ¼ 0.002) and Vitality (p ¼ 0.002). SEM also showed true change in common factor means: deterioration of physical QOL (p < 0.001) and general fitness (p < 0.001), and improvement of mental QOL (p < 0.001). Non-differential recalibration was tested by comparing the pretest and thentest means of the three common factors, but these differences were not significant (p ¼ 0.084, 0.773, 0.672). Concluding, in this illustrative example Assumption 1 seemed valid, but Assumptions 2 and 3 had to be rejected for four SF36-scales. # 3005/SELF-REPORTED PSYCHOLOGICAL WELL-BEING IN SCHIZOPHRENIA PATIENTS: ASSOCIATION WITH LEVEL OF INSIGHT AND CLINICAL SYMPTOMS Dennis Revicki, Center for Health Outcomes Research, MEDTAP International, Inc., Bethesda, MD, USA Schizophrenia is associated with significant impairment to psychological well-being and HRQL. HRQL assessment requires the self report and subjective evaluation which may be affected by schizophrenia related psychopathology. We evaluated the relationship between clinician and patient rated psychological well-being (PWB) and the association between PWB, level of insight, deficit syndrome, and clinical symptoms. The sample included 687 schizophrenia patients from a clinical trial. Clinical symptoms were measured at baseline and over the 6 week acute treatment period using the PANSS, a deficit syndrome classification, MADRS, and level of insight. Patient function was assessed using the Quality of Life Scale (QLS) and the SF-36 mental health index (MHI) at baseline and 6 weeks. Regression analysis was used to evaluate the relationships between level of insight, deficit syndrome, PANSS, and MADRS on QLS total and MHI scores. Average age was 39 years, with 68% males, 65% Caucasian, and 26% classified as deficit syndrome. Baseline MHI scores were significantly associated with clinician rated MADRS scores in nondeficit (p < 0.05) but not in deficit patients (p > 0.05). For the nondeficit patients, baseline to 6-week changes in MHI scores were associated with changes in depression status (p < 0.001) but not in deficit patients (p > 0.05). For deficit patients, level of insight was associated with inconsistent reports on PWB (p < 0.05). Baseline to 6week changes in QLS total scores were related to changes in positive

symptoms (p < 0.01), negative symptoms (p < 0.01), MADRS (p < 0.01) and EPS (p < 0.05). Baseline to 6-week changes in MHI scores were associated with changes in positive symptoms (p < 0.01) and MADRS scores (p < 0.01). Most (75%) patients with schizophrenia reported their PWB consistently with clinician rated depression. Deficit patients, with lack of insight, provided inconsistent reports. Functional outcomes are associated with changes in clinical symptoms and treatment-related movement disorders. # 1880/SOCIALOGICAL ASPECTS OF QUALITY OF LIFE IN A TRANSITION COUNTRY Eva Krizova, Institute of Medical Ethics, 3rd Faculty of Medicine, Charles University, Praha (Prague), Czech Republic The poster presents findings of an empirical research on health-related quality of life in the period of social transition. A questionnaire was distributed in 930 adult persons. During multiple analyses (Answer Tree, SPSS) the interrelations between different variables from health, micro-social and macro-social domains were examined and their impact was compared. In each domain the most predicting variable was defined. In the final analysis the subjectively perceived existence of a close person as a source of emotional support emerged as the crucial factor influencing the quality of life more likely than pain, health and/or sleep disorders or economic conditions. Gender and education played a significant role. Quality of life was significantly more decreasing with a lower economic status in women while education played a stronger role in achieving high score of quality of life in men. Health disorders and a worse economic situation had the highest impact on quality of life in the middle-aged, most probably due to overall competition and performance tasks in the 90’s. Unlike our original hypotheses the macro-social aspects associated with the period of historical and cultural transition like change of property and power relations, negative transition events as unemployment, violance or non-voluntary change of place of residence etc. were less influencing the overall quality of life than the aspects of personal life. The most profiting population group was the young generation and the right-wing oriented respondents who favored most the changes. # 1250/IS PERSONALITY RELATED TO HEALTH-RELATED QUALITY OF LIFE? A SYSTEMATIC REVIEW I-Chan Huang & Albert W. Wu, Health Policy and Management, Johns Hopkins University, Baltimore, MD, USA Purpose: Personality is one of the variables that may potentially affect Health-Related Quality of Life (HRQOL) measures. The purpose of this study was to determine (1) whether personality variables are associated with specific HRQOL domains and (2) how much of the variance in HRQOL is be explained by personality variables. Research methods: This was a systematic review of the literature using the following search strategies: literature search, study selection based on inclusion and exclusion criteria, data extraction, and data synthesis. We searched MEDLINE and PsycINFO combining the key words ‘quality of life’ and ‘personality’. We excluded studies that focused on personality disorders and that reported relationships of personality and HRQOL strictly for instrument validation. All data were abstracted and independently by each of the two reviewers, with discrepancies resolved by a third expert. Summary of results: The literature search identified studies of which 54 met study criteria. Among the 54 studies, 52% assessed the relationship between neuroticism on HRQOL, and 30% assessed the relationship with extroversion. In general, some but not all personality variables were associated with HRQOL measures. Personality variables were more likely to relate to psychosocial than physical aspects of HRQOL. Personality explained varying proportions of the variances in specific domains of HRQOL. Personality variables explained more variance in psychosocial HRQOL (often >15%, range 3–39%) than physical HRQOL (often <5%, range 0–29%). The systematic collection of personality data concurrent with HRQOL measures may be helpful in research, clinical practice and health policy evaluations.

785 # 1407/SELF-RELATED HEALTH AND QUALITY OF LIFE: ARE THEY DIFFERENT CONSTRUCTS? Deborah Buck & Ann Jacoby, Primary Care, University of Liverpool, Liverpool, Merseyside, UK The terms health status and quality of life (QOL) are often used interchangeably. However, a recent meta-analysis demonstrated that the two appear to be distinct constructs. Mental health domains were found to have a far greater impact than physical functioning on global QOL. Physical functioning domains were more influential than mental health domains on self-rated health (SRH). This presentation reports findings from a new study that further examined the relationship between these concepts. Informants (stroke survivors) rated their overall health and QOL on a 5-point scale. The relationship between these global variables and mean scores on 11 specific QOL domains was examined. Specific domains and age and gender were entered into a stepwise regression analysis with the global items as the dependent variables. Scores on 6 of the 11 specific QOL domains (Mobility, ADL/ Self-care, Cognition, Sleep, Fatigue, Feelings) were significantly related to SRH. Scores on all but 1 domain (Vision) were significantly related to ratings of overall QOL. High scores (i.e. greater perceived impact) on Sleep and Vision domains were the strongest predictors of lower SRH. High scores on Feelings and Sleep domains were the strongest predictors of lower global QOL. These findings shed further light on the much-used concepts of SRH and QOL. Although lending some support to the notion that SRH and global QOL may be distinct constructs, the evidence is far from conclusive. For example, issues around the concept of Sleep seem to reflect both physical and mental well-being. Sleep emerged as a predictor, albeit not the strongest, of global QOL. Moreover, moderately high correlations were found between SRH and global QOL. It is possible that the extent to which people distinguish between overall QOL and overall health may depend: current subjective health status; the objective nature of any illness or condition they may have. This is important since global items such as those described here are often used to determine the validity of more lengthy health outcome measures. # 1538/THE RELATIONSHIP BETWEEN SELF-CLARITY AND QUALITY OF LIFE Chia-huei Wu, Cheng-ta Yang & Kai-ping Yao, Psychology, National Taiwan University, Taipei, Taiwan, R.O.C. Self-concept has been considered as an important factor for individual health. It has revealed that the evaluative component of self-concept, often measured in terms of self-esteem, has positive impact on individual health or well-being. However, the association between individual health and the knowledge component of self-concept, which consists of the beliefs about one’s attribution, was examined rarely. Thus, the purpose of present study is to investigate the relationship between quality of life and self-clarity, which is defined as the extent to which the contents or self-beliefs are clearly and confidently defined. It is hypothesized that the higher self-clarity one has, the healthier he/ she would be. Ninety-three students of National Taiwan University participated in this study. They completed the self-clarity rating scale and the WHOQOL-BREF Taiwan version. Different indices of selfclarity were computed from self-clarity rating scale. Those were the average absolute deviations form the scale midpoint, the standard deviation of responses, the number of extreme responses, and the average confidence rating for items. Results revealed that different indices of self-clarity, generally, correlated positively with quality of life, suggesting that the higher self-clarity one has, the healthier one would be. Besides, results also suggested that clarity on negative traits has more impact on quality of life than the clarity on positive traits. # 1807/LIVING WELL AFTER BREAST CANCER: DEVELOPING A CONCEPTUAL MODEL TO UNDERSTAND HOW WOMEN WITH BREAST CANCER CAN BEST COPE AND IMPROVE THEIR QUALITY OF LIFE Daniel R. Longo, Rebecca J. Swaney, Elise Radina, Kevin W. Craig, Kevin D. Everett, Glen T. Cameron, Armer M. Jane & John E. Hewett, Family and Community Medicine, University of Missouri-Columbia, Columbia, MO, USA

The overall objectives are: (a) to both identify and confirm those factors that foster/inhibit women’s abilities to cope with breast cancer as it relates to their quality of life and (b) to identify variations in these factors among women of diverse cultural backgrounds. We will use focus groups (data will be audio-taped and transcribed) to clarify findings from qualitative pilot studies currently underway and to ensure accurate representations from patients’ perspectives of the lived experiences of women with breast cancer (i.e., – member checks: see Bloor, 1983; Lincoln & Guba, 1985). We conducted six focus groups (three focus groups of non-Hispanic and three focus groups of Hispanic women with breast cancer). This method allows us to utilize empirical literature in light of the real world experiences of women living with breast cancer. Focus group inquiry guides were developed following completion of data analyses from the qualitative pilot studies. These analyses were used to modify the current conceptual model that guides our research program. This modified conceptual model was the focus of inquiry with the focus groups. The initial inquiry guide contains questions aimed at eliciting reactions from participants regarding the modified conceptual model. After each focus group the conceptual model was modified to reflect salient reactions of focus group participants. This model reflects the importance of coping methods to improve quality of life including the role of spirituality which we had not anticipated would have been shown to play such a vital role in the ability of these cancer survivors to cope and improve their overall quality of life. The development of this model is unique in the use of ‘member checks,’ that is the participation of women with breast cancer, the subjects of our study, in the development of the conceptual model. The model will be empirically tested in subsequent studies. # 1794/CONTENT COMPARISON OF HEALTH-RELATED QUALITYOF-LIFE (HRQL) INSTRUMENTS BASED ON THE INTERNATIONAL CLASSIFICATION OF FUNCTIONING, DISABILITY AND HEALTH (ICF) Alarcos Cieza & Gerold Stucki, Physical Medicine and Rehabilitation, University of Munich, Munich, Germany The International Classification of Functioning Disability and Health (ICF) is increasingly being applied in clinical research and clinical practice. ICF-based instruments will be used in the future in clinical and epidemiological trials, as well as in health reports. Since health related quality of life (HRQL)- and ICF-based approaches have both strengths and weaknesses, it is expected that they will often be used concurrently in clinical practice, research, and health reporting. Therefore, it is essential to understand the relationship between HRQL instruments and the ICF. The objective of our study was to examine the relationship between six short HRQL instruments and the ICF. The SF-36, the Nottingham Health Profile, Spitzer’s Quality of Life Index, the WHOQOL-BREF, the WHO Disability Assessment Schedule, and the European Quality of Life instrument were linked to the ICF separately by two health professionals according to 10 linking rules developed specifically for this purpose. The items in the HRQL instruments were linked to 89 different ICF categories, 17 categories of the component body functions, 58 categories of the component activities and participation, and 14 categories of the component environmental factors. No items were linked to the component body structures. Twelve items could not be linked to the ICF at all. In the component body functions, only emotional functions are covered by all examined instruments. In the component activities and participation, all instruments cover aspects of work, but they scarcely cover aspects of mobility. Only four of the six instruments address environmental factors. The ICF proved highly useful for the comparison of HRQL instruments. Based on the linkage, it was possible to study the heterogeneity of health-status measures regarding their representation of body functions, activities, participations, and environmental factors. The comparison of selected HRQL instruments may provide clinicians and researchers with new insights when selecting health-status measures for clinical studies.

786 # 1341/DOES RELIGIOSITY PROMOTE ADULT QUALITY OF LIFE? Chia-huei Wu & Kai-ping Yao, Psychology, National Taiwan University, Taipei, Taiwan, R.O.C. The purpose of current study is to investigate the relationship between religiosity and quality of life by examining the effect of intrinsic orientation, extrinsic-personal orientation, extrinsic-social orientation, and religious involvement on four different domains of quality of life (QOL) according to WHOs definition, including physical health, psychological health, social relationships, and environmental domains. One hundred and two Christians participated the study. They completed the standardized questionnaires, including the Religiosity battery, and the WHOQOL-BREF Taiwan version. Results of regression analysis indicated that intrinsic orientation has positive effects on physical health, psychological health and social relationships. However, the two extrinsic orientations do not have any effect on various domains of quality of life. Religious involvement has positive effect on psychological health, social relationships and marginal significant effect on environment domain. Besides, results also suggested that the effect of intrinsic orientation, generally, is greater than religious involvement. These results not only revealed that religiosity does promote quality of life, but also suggested that religious orientations and religious involvement have different effect on various domains of QOL. # 1776/QUALITY OF LIFE IN CHILDREN AND ADOLESCENTS WITH JUVEILE CHRONIC ARTHRITIS AND THEIR PARENTS Esther Mueller-Godeffroy, Klinik fuer Kinder- und Jugendmedizin, Universitaetsklinikum Schleswig-Holstein, Luebeck; Hartwig Lehmann, Klinik fuer Kinder- und Jugendmedizin, Saechsisches Krankenhaus Hubertusburg, Wermsdorf; Rolf Michael Kuester, Klinik fuer Kinder- und Jugendmedizin, Rheumaklinik Bad Bramstedt, Bad Bramstedt; Ute Thyen, Klinik fuer Kinder- und Jugendmedizin, Universita¨tsklinikum Schleswig-Holstein, Luebeck, Germany The aim of this study was to investigate HRQOL in children and adolescents with Juvenile Chronic Arthritis (JCA) and their parents and to determine the impact of children’s physical functioning, stress level, and behavioural problems. Using a cross sectional design, we included 72 children with JCA, ages 8–16, and both parents. We measured sociodemographics, clinical variables, HRQOL of children (Revised Children Quality of Life Questionnaire/KINDL) and parents (SF-36), children’s impairment of activities of daily living (Child Health Assessment Questionnaire/CHAQ), and children’s behavioural problems (Child Behaviour Checklist/CBCL). Compared to healthy children, children and adolescents with JCA reported significantly reduced HRQOL in some dimensions of QOL. Children reported lower selfesteem, adolescents reported reduced physical wellbeing and reduced total QOL. Older children reported lower QOL than younger children, adolescent girls reported lower QOL than adolescent boys. Children with JCA had more emotional and behaviour problems compared to healthy children. Parents’ HRQOL was impaired in the area of emotional health. Lower HRQOL in children was associated with a higher degree of functional impairment and behaviour problems. Children’s and parents’ overall HRQOL showed only moderate correlations. Selfreported HRQOL may be a relevant outcome measure for children with Juvenile Chronic Arthritis as well as parents. # 1825/HEALTH RELATED QUALITY OF LIFE OF CHILDREN PRIOR TO STEM CELL TRANSPLANTATION: DISEASE TYPE AND AGE EFFECTS Maru Barrera & Alisa Kenny, Psychology, Hematology/Oncology Program; Fred Saunders, Pediatrics, Hematology/Oncology Program, the Hospital for Sick Children, Toronto, Ontario, Canada Children undergoing stem cell transplantation (SCT) are medically fragile and differ in many aspects of their disease history including the

type of diagnosis. Health variables as well as personal and familial factors may influence the success of the SCT and children’s healthrelated quality of life (HRQOL) outcomes. The purpose of this study was twofold: (1) To examine the HRQOL of children prior to undergoing a SCT by diagnosis type and age, (2) To assess the concurrent validity of a generic and two disease-specific measures of HRQOL. The sample consisted of 111 children (mean age ¼ 10.4 years, SD ¼ 5.11 years) diagnosed with Acute Lymphoblastic Leukemia (22%), other leukemias (26%), Neuroblastoma (19%), other solid tumors (18%), and metabolic and blood disorders (15%). The Child Health Questionnaire (CHQ, generic measure), and the Pediatric Oncology Quality of Life Scale (POQOLS) and the Play Performance Scale (PPS), disease specific measures, were completed by the caregiving parent within a week prior to SCT. Disease specific measures of HRQOL discriminated among children with different diagnoses. Children treated for ALL and Neuroblastoma were reported to have lower quality of life, measured by POQOLS and PPS, compared to children with other diagnoses, and compared to norms for children with cancer. Adolescents were perceived as having the lowest HRQOL. The POQOLS scores were significantly correlated with the physical and psychosocial CHQ scores. In conclusion, disease specific measures of HRQOL appear to be better able to differentiate between disease types than generic measures of HRQOL; prior to undergoing SCT, children with ALL or Neuroblastoma may be at greater risk for poor HRQOL compared to children with other diagnoses. The HRQOL of adolescents is likely to be more compromised than that of younger children, probably because of conflictive developmental and treatment demands. The results of this study have clinical implications for the care of children undergoing SCT. # 1756/IMPACT OF DETERMINANTS ON ADOLESCENTS’ HEALTHRELATED QUALITY OF LIFE (HRQL) A. Remacle-Bonnet, M.C. Simeoni, A. Loundou, S. Antoniotti, J.L. San Marco & P. Auquier, Perceived Health Assessment Research Unit, School of Medicine, Marseille, France Adolescents filled in the VSP-A, a self-administered HRQL questionnaire, validated and developed from the adolescents’ point of view, as well as a questionnaire (derived from the Health Behaviour in School aged Children questionnaire) gathering information about five major determinants: Health status, Socio-demographic characteristics, School environment and social support, Risk behaviors and way of life. HRQL scores were compared (two-sample T-tests, nonparametric Mann Whitney tests or ANOVA). Logistic regression models were performed to determine the factors that predict the lowest HRQL levels (VSP-A index). The representative sample included 895 students: age range ¼ 10–17 years, boys: 51%. 66% were living with their two parents; 47% of the parents were employees or workmen, and 35% had high socioeconomic status. Out of the sample, 8% reported a health problem occurred during the past month and 82% reported psychosomatic complaints during the past 6 months. Two-third of the adolescents were smokers and 49% alcohol users. More than 40% were not satisfied with their body image. Two-third of the pupils were not satisfied with their school grades and 16% declared low parental support. About half of the adolescents were violent or suffered from violence. Six factors predicting the lowest HRQL level on the VSP-A index were identified: gender (OR ¼ 2.5; IC95% ¼ [1.4; 4.7]), friends relationships (OR ¼ 4.6; IC95% ¼ [2.5; 8.7]), violence (OR ¼ 2.9; IC95% ¼ [1.2; 6.9]), load of school work (OR ¼ 1.7; IC95% ¼ [1.2; 2.0]), parental support (OR ¼ 1.5; IC95% ¼ [1.1; 2.1]), psychosomatic complaints (OR ¼ 1.4; IC95% ¼ [1.1; 1.7]). Longitudinal studies would be helpful to further analyse the impact of these different factors on the adolescents’ HRQL.

787 # 1577/THE DAILY EXPERIENCE OF FAMILIES LIVING WITH AN ADHD YOUTH Donald L. Patrick, Tari D. Topolski & Todd C. Edwards, School of Public Health and Community Medicine, University of Washington, Seattle, WA, USA Anecdotal data suggest that when there is a child with attention deficit/ hyperactivity disorder (ADHD) in a family, the family experiences a significant impact on daily routines, which may lead to reduced quality of life for family members. Few studies, however, have examined the daily experience of living with this disorder. In this study we examine the daily behaviors associated with ADHD within the context of family impact. Literature and focus groups were used to create a 19-item Daily Diary of Events and Behaviors (DDEB). This instrument provides a measure of the impact on the family by assessing frequency, bothersomeness, and time of day the events/behaviors occur as reported by the primary caregiver validated against other measures of ADHD symptomatology and Family Impact. Caregivers from 26 ADHD families and 16 control families were recruited through ADHD specialty clinics and a general pediatric clinic respectively, and completed the DDEB over 9 consecutive days. All ADHD participants were on medications and most (85%) were on extended release medications with reportedly good adherence. The results of validation of the DDEB showed that it was significantly related to measures of ADHD symptom and functioning (r ¼ 0.50–0.61, p < 0.01), and to family impact (r ¼ 0.26–0.65, p < 0.05). Looking at differences between the groups, more negative behaviors (t ¼ )4.34, p < 0.001), rated as more bothersome (t ¼ )2.94, p < 0.01), were reported among ADHD families relative to control families. Time period variation was observed for ADHD families, while control families generally reported the same number of symptoms across time periods. This study supports the validity and importance of obtaining this type of data. Although adherent to taking long-acting ADHD medication, negative behaviors persisted especially during morning and evening hours, among these youth with a negative impact on the family unit. # 1885/THE RELATIONSHIP BETWEEN SCHOOL ADJUSTMENT AND LIFE SATISFACTION IN ADOLESCENTS IN POLAND Joanna T. Mazur, Department of Epidemiology, National Research Institute of Mother And Child; Barbara Woynarowski, Faculty of Pedagogy, Warsaw University; Agnieszka Malkowska, Department of School Medicine, National Research Institute of Mother And Child; Hanno Kololo, Faculty of Pedagogy, Warsaw University, Warsaw, Poland The purpose of this study is to investigate the relationship between school adjustment and life satisfaction among students in Poland. The study is based on the fourth series of Health Behaviour in School-Aged Children (HBSC). A WHO Cross National Study conducted in Poland in 2002 as a school based survey. A sample was selected as representative for three age groups’ mean age 11.7; 13.7 and 15.7 years (n ¼ 6383). Life satisfaction was measured on the one-item scale (0– 10) in accordance with Cantril approach. As the main outcome measure the percentage of students with low (<6 points) life satisfaction was defined. Series of logistic regression models were estimated to calculate odds ratio (OR) with 95% confidence intervals (CI) and trend in odds ratio estimates. Three aspects of student’s school adjustment

were studied. Each question was analysed separately and then all variables in the final model. All analyses were adjusted for age, gender and socio-economic status. Results showed, that 21.0% of Polish students (17.7% of boys and 22.5% of girls) perceived their life satisfaction as low-mean 7.35, standard deviation 2.09. The rate of low life satisfaction is growing from 13.7% in 11-years olds to 27.7% in 15years olds. Low satisfaction with school, poor academic achievements, and high school-related stress are independent predictors of low life satisfaction with odds ratio 2.02 (1.53–2.68), 3.34 (2.43–4.59) and 1.80 (1.33–2.44), respectively. However, the impact of socioeconomic status is much higher. The percentage of students perceiving their well-being as low is growing from 3.5% for very well off families to 60.5% for not at all well off families, OR ¼ 27.7, CI(OR) O 10.08–76.14. In generic QoL instruments developed for children and adolescents school-related items should be included. Socio-economic status is an important factor modifying subjective well-being and the real impact of it’s components and determinants. This analysis should be repeated for other HBSC countries to test cross-cultural differences. # 1024/USE OF QUALITY OF LIFE AS AN ENDPOINT IN RANDOMIZED CONTROLLED TRIALS IN JAPAN: A SYSTEMATIC REVIEW Mariko Naito, Takeo Nakayama, Department of Medical System Informatics; Shunichi Fukuhara, Department of Epidemiology and HealthCare Research, Kyoto University School of Public Health, Kyoto, Japan Use of Quality of life (QOL) as an endpoint in randomized controlled trials (RCTs) has not been systematically assessed in Japan. In order to investigate trend of utilization of QOL and to systematically assess the quality of such studies in Japan, we searched articles published during 1970–2001 using PubMed. Counting number of articles with the combination of relevant publication type and MeSH, we found that utilization of QOL increased from 0 to 3.7% in RCT reports worldwide, while from 0 to 1.5% of reports published from Japan. Among 48 studies published from Japan, 36 studies that met our criteria were evaluated. Authors used established QOL instruments in seven studies (19%), quoted a reference that described their methodology in 9 (25%), developed original instruments for their studies in 5 (14%), only assessed the symptoms or performance status in 13 studies (36%), and methods were unclear in 2 (6%). The most commonly studied condition was cancer (61%), followed by cardiovascular disease (11%) and urologic disorders (11%). The most common interventions were drugs (67%) followed by modes of care (22%). QOL was the primary endpoint in five studies (14%). No study used generic health-related QOL instruments. Authors conceptually defined QOL in four articles (11%) that had been published since 1991. The response rate for QOL measurement was given in 20 articles (56%), however, authors unclearly stated responders in 10 articles (28%). In conclusion, though utilization of QOL as endpoints in RCTs has increased over time worldwide, in Japan much fewer such studies have been reported and research methodology in designing and utilizing QOL measurement is of limited quality. More rigorous and constructive discussion and/or some guideline, which addresses when, what and how QOL should be utilized in RCTs, may be required for more frequent and more scientifically sound use of QOL as an endpoint in RCTs in Japan.

788 # 1567/ASSESSING QUALITY OF LIFE IN PARENTS OF CHILDREN WITH CANCER Tatiana P. Nikitina, Multinational Center of Quality of Life Research, St. Petersburg; Eugenia I. Moiseenko, Institute of Pediatric Oncology and Hematology, Moscow; Anton V. Kishtovich, Tatyana I. Ionova & Andrey A. Novik, Multinational Center of Quality of Life Research, St. Petersburg, Russia Cancer is detrimental not only for patients but for family members, especially if a patient is a child. Information about parents quality of life (QoL) is to be taken into account while management of children with cancer. The purpose of this work was to study QoL of parents of children with different types of cancer and identify if parents QoL parameters are dependent on child age and type of cancer. Russian version of SF-36 was administered to parents of children of 2– 18 years old with cancer as well, as to parents of healthy children of the same age. Ninety-four parents of children with solid tumors and blood cancer before- (n ¼ 16), on- (n ¼ 36), off-treatment (n ¼ 42) and 16 parents of healthy children were enrolled in the study. ANOVA method was used for statistical analysis. Social Functioning and General Health (p < 0.01) of parents of healthy children were significantly higher that in parents of children with cancer before-treatment. Social Functioning of parents of children with cancer on-treatment (p < 0.01) was worse than of parents of healthy children. Social Functioning of parents of children off-treatment was significantly higher that in parents of children before-treatment (p < 0.01). No differences were found in QoL scores between parents of healthy children and parents of children with cancer off-treatment. QoL of parents did not depend on child age and type of cancer. Thus, SF-36 is appropriate tool to study QoL in families of children with cancer. Social Functioning and General Health of parents of children with cancer is impaired as compared to parents of healthy children. After treatment of a child restoration of QoL parameters of their parents takes place and they do not differ from the ones of the parents of healthy children. There is no dependence between QoL parameters of parents and child age and type of cancer. Our findings confirm the importance of QoL assessment in family members of children with cancer. # 1040/THERE IS NO PLACE LIKE MOTHER’S HEART Ljiljana Z. Trajanovic, Zagorka S. Stevanovic & Olivera Z. Zikic, Day Hospital, Clinic for Mental Health Protection, Clinical Center, Nis; Srbobran Miljkovic, Department of Psychiatry, Medical Faculty, University of Nis, Nis, Serbia, Serbia and Montenegro Family caregivers are at considerable risk of alteration of their health and deterioration of health-related quality of life (HRQOL). This is especially true for family caregivers of persons with serious mental illness. The object of this research was to study family relationship, within the nuclear family group of patients with paranoid schizophrenia, and its impact on quality of life. The research included 100 family members of 60 persons with diagnosis of paranoid schizophrenia. The sample consisted of 29 fathers, 34 mothers, 20 spouses, 8 siblings and 9 children (age 18 and more). Medical Outcomes Study 36-item Short-Form Health Survey (SF-36) was used. In general, the results show that family relationship with paranoid schizophrenia person has a significant influence on HRQOL. Differences were observed in all of eight domains, with greatest impact on general health perception (p < 0.000001), physical functioning (p < 0.00001) and mental health (p < 0.001). Mothers were the most affected group of the relatives and they have the lowest scores (physical functioning ¼ 55.73; role physical ¼ 48.52; bodily pain ¼ 51.47; general health ¼ 44.52; vitality ¼ 40.73; social functioning ¼ 54.04; role emotional ¼ 43.13; mental health ¼ 44.35). Fathers and then spouses followed them, although there were not significant differences between these two groups. Total number of sibling and adult children was too small to permit a definite conclusion, but they looks like having satisfactory HRQOL. These finding are consistent with facts about a parents as primary caregivers of people with persistent and severe mental illness. The mothers are particularly sensitive for child disease, which is, with no doubt, a risk factor for their health and quality of life. Those who’s love is unlimited should have unlimited attention of medical professionals.

# 1503/CROSS-CULTURAL DEVELOPMENT OF A HEALTH RELATED QUALITY OF LIFE MEASURE FOR AIDS NATURAL CAREGIVERS: FIRST RESULTS P. Auquier, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille, France; J. Cramer, Department of Psychiatry, Yale University School of Medicine, New Haven, Boston, MA, USA; S. Robitail, D. Debensason, A. Beresniak & M.C. Simeoni, Perceived Health Research Unit, Public Health Department, School of Medicine, Marseille, France Since the progress of therapeutics, AIDS is a chronic progressive disease with strong social and emotional impact not only on patients but also on natural caregivers. Measuring this impact, in terms of the quality of life of natural caregivers allows to describe consequences of care procedures provided to these patients. The aim of this study was to carry out a French and US cross-cultural instrument to measure the quality of life specific to AIDS natural caregiver. Content analysis of 24 interviews of AIDS natural caregivers were conducted in France. English linguistic validation was made in USA, using forward and backward methodology. After reconciliation a final pool of 110 items was generated. Item reduction of the preliminary item pool involved 143 AIDS natural caregivers in France (n ¼ 62) and USA (n ¼ 81). Sociodemographic information was asked to patients and their caregivers. Clinical data and generic QoL questionnaire (SF36) were collected respectively for patients and caregivers. Exploratory factor analyses relying on parallel analysis identified eight dimensions: Psychological Well-Being, Private life/Privacy, Financial issues, Empathy, Social support, Leisure, Rejection and Sentimental life. Item reduction selected 61 items. Lisrel confirmatory factor analyses of the model is satisfactory (RMSEA ¼ 0.074). Item internal consistency and item discriminant validity are both above 90%. Internal consistency was adequate for every dimension (Cronbach’s a range from 0.73 to 0.94). Criterion validity (SF-36) was satisfactory. This new instrument provides a promising tool for studying and understanding quality of life in AIDS natural caregivers. Topics for future research are suggested. # 1781/GENDER DIFFERENCES IN HRQOL AMONG INFORMAL CAREGIVERS OF MULTIPLE SCLEROSIS PATIENTS Marta Aymerich & Imma Guillamo´n, Catalan Agency Health Technol Assessment & Research; Albert J. Jovell, Fundacio´ Biblioteca Josep Laporte, Barcelona, Spain Chronic disabling diseases have an impact in people indirectly affected by the disease, that is, informal caregivers. However, few studies take caregivers’ HRQOL into account. Multiple sclerosis (MS) is one of these chronic diseases which an important number of patients evolve to partial or total disability. The objective of the study was to show the impact of the disease on caregivers’ HRQOL. A cross-sectional multicentre study was designed to assess HRQOL and psychological wellbeing on informal caregivers of MS patients. The HRQOL was measured by the SF-12 which is a generic instrument with two dimensions: the physical and the mental components. A SF-12 score below or over 50 (SD:10) means, respectively, worse or better HRQOL than general population. Psychological well-being was measured by the GHQ-12 which is a mental health screening instrument where a score of three or more means a probable case of mental disorder. Overall, 551 informal caregivers (280 women) were recruited consecutively in nine hospitals during more than 3 years. MS informal caregiver was mostly the partner (66%), with a mean age of 45 years (range 13–78), with children (80% of the sample), 2% were illiterate, and 15% had a university degree (similar education to general population). They have been living a mean of 11.2 years with the patient and giving care for 3.25 hours per day on average. Their HRQOL, according to SF-12 scores, was worse compared to the general population, not for the physical but for the mental component. Stratifying by gender, the SF12 physical component showed no differences between men and women: 49.6 (95% CI: 48.4–50.8); whereas for the mental component there were statistically significant differences: 48.4 (95% CI: 47.2– 49.6) in men, and 44.4 (95% CI: 43.1–45.8) in women. Moreover, according to GHQ-12 scores, a mental disorder was more probable in women than in men (29.5% vs. 24.3%; p ¼ 0.003). These data showed that the MS has an impact on informal caregivers’ mental HRQOL, and significantly more on women than men. More data is needed to know if these caregivers receive appropriate care.

789 # 1232/VALIDITY, RELIABILITY AND SENSITIVITY TO CHANGE OF THE SMILE: A SPECIFIC MEASURE FOR WOMEN IN MATURE LIFE Hanne Thorsen, Department of General Practice, University of Copenhagen, Copenhagen, Denmark; Sally Shumaker, Wake Forest University School of Medicine, US; Thomas Kohlmann, University of Greifswald, Germany; Mauro Niero, University of Verona, Italy; Anne Marciniak, Pfizer, Sandwich; Linda Abettz, MAPI Values, Manchester, UK The aim of the study was to validate the SMILE, a measure developed specifically to address health issues faced by women after the menopause. A 118-item version of the SMILE was tested in five countries (Italy, Germany, Denmark, Spain and US) on 823 women aged 55–70 representing healthy women and women with osteoporosis or cardio-vascular conditions. A subset of 266 women were retested after 2 weeks. A cohort of 297 healthy women starting postmenopausal hormone therapy (HT) at baseline completed the measure at baseline and after 12 weeks. Aspects of reliability, validity, and responsiveness were assessed using multi-trait analyses, comparisons of known groups, and measures of change. The measure was reduced to 61 items with 10 dimensions (happiness, mood, perspectives about the future, social interaction, appearance, comfort/mobility, memory/concentration, sleep, energy, and genito-urinary health). Analysis of this version showed scaling success for item convergent validity (97% of items correlated P0.4 with their own dimension), item discriminant validity (all items correlated higher with their own dimension), internal consistency reliability (a range: 0.68–0.92). Reproducibility was good (ICC > 0.7). Known-groups validity tests of scale scores consistently yielded significant differences (p ¼ 0.05) between women: healthy vs. with disease (on 7/10 scales, effect size range 0.25–0.58) and healthy vs. healthy with hot flushes (on 6/10 scales, effect size range 0.29–0.72). Comparison between SMILE scores and several concurrent measures indicated correlation as hypothesised. The SMILE was sensitive to HT related change. Effect sizes for the improved health status ranged 0.30–0.61 (>0.4 for 8/10 scales). Results indicate that the SMILE is a reliable and valid instrument and appropriate for cross-cultural use in evaluation of interventions in women after menopause. The SMILE distinguishes between women with different health states and is responsive to treatment effect. The SMILE is a promising instrument filling a gap in existing population specific measures. # 1435/HEALTH-RELATED QUALITY OF LIFE ASSESSMENT IN MENOPAUSAL WOMEN: COMPARISON BETWEEN HRT USERS AND NON-USERS Watcharee Lermankul, Clinical Pharmacy, Faculty of Pharmacy, Silpakorn University; Tippayaporn Somsak, Pharmacy, Nakhon Pathom Regional Hospital, Nakhon Pathom, Thailand The purpose of this study was to determine the effect of hormone replacement therapy on the quality of life in menopausal women comparing between users and non-users. A total of 135 patients aged 40–62 years were recruited from two menopausal clinics. Fifty-two of them were currently receiving hormone therapy (age ¼ 51.0 ± 5 years) and 83 were not (age ¼ 51.4 ± 4.8 years). No major socioeconomic differences were presented. Participants were asked to completed two questionnaires, i.e. MENQOL (Thai version) and a climacteric symptoms form, by themselves. The MENQOL consisted of 29 items in four domains including Vasomotor, Psychosocial, Physical and Sexual. Cronbach’s a coefficients of MENQOL were 0.80–0.87. There were no statistically significant differences in any domain of the MENQOL and overall health between users and nonusers. Postmenopausal women who were on hormone therapy had better scores on all domains of MENQOL than the non-users on Physical (p ¼ 0.015) and Sexual (p ¼ 0.011) domains. Women with high symptom scores receiving hormone therapy had a better quality of life than the non-users in Psychosocial domain (p ¼ 0.016). Among hormone users, perimenopausal women had better quality of life than postmenopausal women in Sexual (p ¼ 0.021) domain, however comparison between groups of high and low symptom scores revealed no significant differences on any domain. In the non-users group,

perimenopausal women were not significantly different from postmenopausal women. Nonetheless, women with a low symptom score had a better quality of life in all domains (p < 0.001). In conclusion, although the quality of life of users and non-users were not different, benefits were evident in a subgroup of postmenopausal women treated with HRT. # 1141/RELATIONSHIP BETWEEN COMORBIDITIES AND BODY MASS INDEX FOR THE IMPACT OF WEIGHT ON QUALITY OF LIFE SCALE Lee Bowman & Clarice P. Hayes, Global Health Outcomes Research, Eli Lilly & Company, Indianapolis, IN, USA The Impact of Weight on Quality of Life Questionnaire (IWQOL) consists of 65 items forming 7 subscales: Health, Social/Interpersonal, Sexual Life, Work, Self-Esteem, Mobility, and Activities of Daily Living (ADL). To determine the relationship between obesity-related quality of life (QOL) and both comorbidities and BMI, we administered the IWQOL to 2184 obese individuals (Mean body mass index, BMI ¼ 31.3, mean age ¼ 41 years, 58% female) in four countries. By self-report, 31% of respondents had been diagnosed with hypertension, 30% high cholesterol, 22% migraines, 14% arthritis, 14% asthma, and 9% diabetes. We used linear multiple regression with IWQOL subscales and total scale as dependent variables and BMI, gender, age, and total number of comorbidities as independent variables. The results indicate that one standard deviation unit change in number of comorbidities produces the greatest change in QOL as measured by the IWQOL Health, Social/Interpersonal, Sexual Life, Self-esteem subscales and Total scale while BMI produces the greatest change in the Mobility and ADL subscales. Significant (p < 0.001) variance accounted for by the independent variables ranged from 34% for the Health Subscale to 9% for the Work Subscale. Using the same regression procedure, separate analyses were conducted across 4 BMI categories (25–29, 30–34, 35–39, and 40 or greater). The results indicate that, for the two highest BMI categories, number of comorbidities produces the greatest change in QOL. For these categories, number of comorbidities was the only statistically significant (p < 0.001) predictor of Total Scale and Health, Mobility, and ADL subscale scores. The correlations between the IWQOL and BMI and number of comorbidities, appears to anchor the immediate impact of obesity on QOL and also the long term impact of obesity (via its complications or comorbidities). Therefore, the instrument discriminates respondents’ weight and their health status separately. # 1430/HEALTH STATUS AND QUALITY OF LIFE: THE DIFFERENCES BETWEEN MEN AND WOMEN Neusa S. Rocha & Marcelo P. Fleck, Psychiatry, Federal University of the State of Rio Grande do Sul, Porto Alegre, RS, Brazil We selected a convenience sample composed by healthy (n ¼ 119) and sick people (n ¼ 122), paired by sex and age. Our objectives were: (1) to verify the relation between health status and quality of life; (2) to verify the differences in quality of life between men and women; (3) to verify if the association between gender and depressive symptoms is associated with lower quality of life, being healthy or not; (4) to verify if the differences persist after adjusting for confounding factors. We used as instruments for measure: quality of life (WHOQOL-100); depressive symptoms (BDI). Sick women (n ¼ 61) had poorer quality of life in physical (mean ± SD ¼ 48 ± 14.9) and psychological (mean ± SD ¼ 59.5 ± 11.8) domains, compared to sick men (mean ± SD ¼ 52.3 ± 14.5; mean ± SD ¼ 65 ± 11.9; p < 0.05; respectively). This differences did not appear between healthy men and women. The occurrence of depressive symptoms is higher in sick women (mean ± SD ¼ 12.9 ± 9.7) than sick men (mean ± SD ¼ 8.6 ± 6.7; p < 0.05). Comparing health status (autoavaliation), there is no difference between sick men and sick women. The difference between sick men and sick women in the psychological domain disappear when we adjusted by confounding factors such as: age, health status, social-economic level and depressive symptoms. Our findings suggest that the poor women’s quality of life can be mediated by depressive symptoms.

790 # 1785/ICF CORE SET FOR PATIENTS WITH STROKE: A PRACTICAL TOOL FOR APPLYING THE INTERNATIONAL CLASSIFICATION OF FUNCTIONING, DISABILITY AND HEALTH IN CLINICAL PRACTICE AND RESEARCH Gerold Stucki & Alarcos Cieza, Physical Medicine and Rehabilitation, University of Munich, Munich, Germany; Jan Schouten, Department of Epidemiology, University Hospital Maastricht, Maastricht AZ, Netherlands; Szilvia Geyh & Thomas Ewert, Physical Medicine and Rehabilitation, University of Munich, Munich, Germany; Nenad Kostanjsek, CAS-Group, World Health Organization, Geneva 27, Switzerland The purpose of this study was to define a large ICF Core Set for clinical practice and a short ICF Core Set for research based on the International Classification of Functioning Disability and Health (ICF) for patients with stroke. An ICF Core Set can generally be defined as a list of ICF categories that are relevant to most patients with a health condition. A pre-selection of ICF categories was derived through empirical data from patients, international Delphi surveys conducted with international experts, and systematic reviews. The results derived from these three independent approaches were presented to a panel of 42 international experts in the field of stroke at an international consensus conferences organized in collaboration with the WHO. The ICF categories to be included in the ICF Core Sets were defined according to a decision-making and consensus process at the conference, The large ICF Core Set for stroke encompasses 130 ICF categories. 18 of the 130 categories constitute the short ICF Core Set: 6 body functions (b110 Consciousness, b114 Orientation, b140 Attention, b144 Memory, b167 Mental functions of language and b730 Muscle power), two body structures (s110 Brain and s730 Upper extremity), seven activities and participations (d310 Communicating with – receiving – spoken messages, d330 Speaking, d450 Walking, d510 Washing oneself, d530 Toileting, d540 Dressing and d550 Eating), and three environmental factors (e310 Immediate family, e355 Health professionals and e580 Health services, systems and policies). The validity, as well as the feasibility, of the ICF Core Set will be tested in different countries and settings within the scope of a collaboration project by the University of Munich and WHO. The development of the ICF Core Set for stroke meets the emerging need in neurological-neuropsychological rehabilitation for multi-level assessment, considering not only isolated sensomotoric and cognitive deficits, but also including everyday activities, participation, and patients’ social and physical environment. # 1629/ANALYSIS OF QUALITY OF LIFE IN POLYTRAUMATIZED AND RESUSCITATED CRITICAL CARE PATIENTS Laura Wolowicka, Iwona Trojanowska, Tomasz Torlinski & Alicja Bartkowska-Sniatkowska, Intensive Care and Pain Clinic, University of Medical Sciences, Poznan, Wielkopolskie, Poland Background: Quality of Life should be included for outcome evaluation of intensive therapy life-threatening states. Survival prognosis methods are more and more precise but quality of life after intensive therapy however is still unknown. Aim of the study was to assess quality of life in correlation with clinical factors Material and methods: In 1998– 2001 124 patients with severe trauma (Group I) and 100 patients after cardiac arrest (Group II) were included into the study. The control group (Group C) consisted of 110 healthy persons with the same demographic data. WHO QOL-Bref questionnaire – our first Polish version from 1998 was used in all groups of patients. This instrument was combined with HAD and SOC assessment. Results: Overall quality of life in Group I was 12.4 ± 3.1 pc and general health 11.5 ± 2.8 pc. In the Group II overall quality of life was 13.5 ± 2.50 pc and general health 13.0 ± 2.6 pc. Control group results were: overall quality of life 14.6 ± 2.9 pc and general health 14.3 ± 3.8 pc. The statistical significance of differences between: Group I and C (p < 0.0001) in both assessments and Group II and C (p < 0.05) according to overall quality of life and (p < 0.001) according to general health. In particular domains: physical, psychological, social and environment there were no difference between Group I and II. There were no significant differences in QOL according to such clinical factors as: place of resuscitation, mechanism of cardiac arrest in Group I.

In Group II was positive correlation between GCS scale and social domain and negative correlation between physical activity domain and length of unconsciousness and ICU stay (p < 0.05). In assessment of HAD in Group II there were no significant correlation between negative emotional state and clinical factors. In the same group correlation (Spearman coeff.) between QOL and coherence was found in context of psychological (p < 0.02) and social (p < 0.005) domain only in meaningfulness. Conclusions: Overall quality of life in the polytrauma and resuscitated patients was on the relatively good level. # 1291/AGE PREFERENCE FOR HEALTH FAVOURING THE YOUNG Marja C. Stuifbergen, Willem Jan Meerding & Marie-Louise EssinkBot, Public Health, Erasmus Medical Center, Rotterdam, The Netherlands An assumption of cost-effectiveness analyses (CEA) is that the value of health is independent of age. We investigated if age preferences for the distribution of health exist in a sample of the Dutch public. Two hundred and twelve respondents (aged 17–72, 53% female) gave their opinion on three hypothetical choice situations regarding a micro- and macro level of decision making. Choices were (1) allocating a donor kidney to one of two kidney patients of different age (pairs 20–45, 45– 70, 20–70), (2) determining the sequence of vaccination against a lethal influenza virus by age group, and (3) gaining life years in different age groups by screening for cancer (ages 20–45, 45–70, 20– 70). Reasons for the preferences could be provided in open questions. With regard to the kidney patients, 91, 92 and 68% of respondents preferred the youngest patient for the pairs 20–70, 45–70 and 20–45 respectively. For the vaccination problem the most frequently given ranking of preferences was from young to old. With respect to the cancer screening most respondents chose the youngest age group when the gain in lives was equal, and about half of the respondents had switched preference when 10 old lives were sacrificed to gain 1 young life. Interpretation of the open questions showed that a form of ‘fair innings’ or ‘productivity ageism’ was the main reason for age preference. At the same time however, 47% of respondents did not endorse age as a decision criterion. In this research we found preferences for health at ages below 70. More research is needed into the generalizability of this result, and the role of age in combination with other possible decision criteria. Ethical reflection is necessary to decide if and if so, how, decisions regarding priority setting should take age preferences into account. # 1049/LEADERSHIPS AND FAMILY Dolors S. Mun˜oz, Organization, Business and Design of Product; Carme C. Saurina, Economics, UdG, Girona, Spain Purpose of research: This poster explores how gender and organizational levels interact to influence ratings of leadership and work satisfaction. The research also considers that transformational leadership allows women to simultaneously carry out leadership and gender roles. Subject sample and statement methods: quantitative research: Self-administered questionnaire by mail over a representative sample of 100 leadership’s women. Qualitative research: In depth interview over a sample of 19 leadership’s women. Summary of results (adequate to support conclusion): This study shows that women with managerial positions devote more than 10 h per week for family matters and household chores, with percentages ranging between 65 and 75%, and even arriving at 90% in cases of women with children. The combination of family and work acts against the educational chances and professional progress in women’s life, as well as against the possibility of personal and leisure development. Finally, one of the main features of this qualitative analysis is the feeling of guilt that women experience for not devoting more time to their families. This fact makes a considerable difference to their male partners and it points out that a balanced level of co-responsability has not been reached yet. Still, women express a high degree of fulfillment in their professional and family lives.

791 # 1339/GENDER DIFFERENCES IN REPORTING OF HRQOL IN INTENSIVE CARE UNIT SURVIVORS Kelsey D. Juzwishin, Deirdre A. Hennessy, Gloria Bowen, Dean Yergens & Christopher J. Doig, Community Health Sciences, University of Calgary, Calgary, AB, Canada This was a prospective, longitudinal cohort study comparing healthrelated quality of life (HRQOL) in male and female patients admitted to any one of three multi-system ICUs in the Calgary Health Region from September 2001 through December 2002. HRQOL was assessed by SF-36 and Euro-qol at baseline, 1 and 6 months post-ICU discharge. The study sample included 257 ICU survivors that met the study criteria (ICU LOS >48 hours, age >18 years). For this sample, mean age was 52.8(±17.8) years, mean ICU LOS was 8.3(±8.6) days and males made up 57% of the sample. The sample was stratified by gender and into five age categories (18–34, 35–49, 50–65, 66–75, and 76–100). With regards to the SF-36 physical functioning component score (PCS), at baseline, females (aged 35–49) reported worse scores than their male counterparts (p < 0.05). In contrast, 1 month after ICU discharge, men (aged 35–49) reported greater worsening (mean change) in their PCS scores than did females of this strata (p < 0.05). Six months after ICU discharge, men of this age group continued to report more significant impairments than did females (p < 0.05). The mean PCS scores did not differ significantly between males and females in any other age group. With regards to the SF-36 mental component score, at all age strata, males and females showed no difference in mean MCS at baseline. At 1 month after ICU discharge, males (aged 76–100) reported a larger mean decreases in MCS than did females (p < 0.05). Six months after ICU discharge males (aged 35–49) reported a larger mean decreases in MCS than did their females counterparts (p < 0.05). Among ICU survivors, females appear to recover more quickly than do males. Stratification on age and gender provides important information into differences in both baseline and follow up HRQOL in ICU survivors. # 1143/COMPARING SF-36 SCORES OF PARTICIPANTS IN THE WOMEN’S HEALTHY EATING AND LIVING STUDY, WOMEN’S HEALTH INITIATIVE, AND MEDICAL OUTCOMES STUDY Kathleen J. Yost, Center on Outcomes, Research, and Education, Evanston Northwestern Healthcare, Evanston, IL; Mary N. Haan, Epidemiology, University of Michigan, Ann Arbor, MI; Richard A. Levine, Mathematics and Statistics, San Diego State University, San Diego, CA; Ellen B. Gold, Epidemiology and Preventive Medicine, University of California, Davis, CA, USA The Medical Outcomes Study (MOS) 36-item Short-Form Health Survey (SF-36) is a widely used measure of general health-related quality of life (HRQL), which facilitates the comparison of HRQL across independent studies. The objectives of our study were (1) to compare scores of eight scales and two summary scales of the SF-36 across three samples of women with different health profiles, and (2) to determine whether statistically significant group differences were also meaningful differences. We compared the SF-36 scores of 358 participants in the UC Davis site of the Women’s Health Eating and Living (WHEL) study, 360 Women’s Health Initiative-Dietary Modification trial (WHI-DM) participants (UC Davis site), and 360 women from the MOS. WHEL participants had a recent history of breast cancer, WHIDM participants were healthy, post-menopausal women, and MOS participants had a history of hypertension, heart disease, diabetes, and/or depression. General linear models adjusting for the covariates age, education, and marital status were used to identify statistically significant differences in scale scores. Distribution-based methods were used to identify minimally important differences (MIDs). Score differences were considered meaningful if they were greater than their respective MIDs. MIDs ranged from 5 to 8 points for the mental health scale to 12–18 points for the role-physical scale. MIDs for the physical and mental component summary scales were 3–5 and 3–4 points, respectively. The HRQL of the WHI-DM and WHEL samples was similar. Both the WHI-DM and WHEL samples had significantly better HRQL than the MOS sample, specifically with respect to scales associated with physical health. However, several of the statistically significant group differences were below their respective MIDs and, therefore, were not meaningful. This study illustrates that relying solely on statistical significance can lead to misleading conclusions about

group differences in HRQL scores. Using MIDs to assess the meaningfulness of group. The Medical Outcomes Study (MOS) 36-item Short-Form Health Survey (SF-36) is a widely used measure of general health-related quality of life (HRQL), which facilitates the comparison of HRQL across independent studies. The objectives of our study were (1) to compare scores of eight scales and two summary scales of the SF-36 across three samples of women with different health profiles, and 2) to determine whether statistically significant group differences were also meaningful differences. We compared the SF-36 scores of 358 participants in the UC Davis site of the Women’s Health Eating and Living (WHEL) study, 360 Women’s Health Initiative-Dietary Modification trial (WHI-DM) participants (UC Davis site), and 360 women from the MOS. WHEL participants had a recent history of breast cancer, WHI-DM participants were healthy, postmenopausal women, and MOS participants had a history of hypertension, heart disease, diabetes, and/or depression. General linear models adjusting for the covariates age, education, and marital status were used to identify statistically significant differences in scale scores. Distribution-based methods were used to identify minimally important differences (MIDs). Score differences were considered meaningful if they were greater than their respective MIDs. MIDs ranged from 5 to 8 points for the mental health scale to 12–18 points for the role-physical scale. MIDs for the physical and mental component summary scales were 3–5 and 3–4 points, respectively. The HRQL of the WHI-DM and WHEL samples was similar. Both the WHIDM and WHEL samples had significantly better HRQL than the MOS sample, specifically with respect to scales associated with physical health. However, several of the statistically significant group differences were below their respective MIDs and, therefore, were not meaningful. This study illustrates that relying solely on statistical significance can lead to misleading conclusions about group differences in HRQL scores. Using MIDs to assess the meaningfulness of group differences should complement statistical comparisons of HRQL. # 1600/MEDICAL STUDENT TIME PREFERENCE FOR HEALTH Kashyap Trivedi, La Jolla, CA; Theodore G. Ganiats, Family and Preventive Medicine, University of California, San Diego, La Jolla, CA, USA Context: Differences in time preference for health between patients and physicians may create barriers in shared medical decision-making. Previous work demonstrated a difference between patient and physician time preference for health. This study gives insights into whether the difference is inherent or learned as a result of medical training. Methods: Headache and chickenpox vignettes were used. The headache vignette had four unique time-to-future-treatment versions; each subject received only one version. The chickenpox vignette had only one version. Random packets of vignettes were distributed to all 519 local medical students. Logistic regression was used to calculate median discount rates for the headache vignette using a population-based approach; simple proportion is reported for the chickenpox vignette. Results: 205 of 519 (39%) medical students surveyed returned the vignettes. Students were grouped into preclinical (first 2 years) and clinical (last 2 years) for analysis. For the headache vignette, time to future health was a significant factor (p < 0.000) and medical specialty choice was nearly significant (p ¼ 0.06) in influencing the choice between choosing treatment now or later. Medical student group (preclinical vs. clinical), gender, and other career choices were not significant. The median discount rate for the migraine vignette was 43% (95% CI 23–74%). For the chickenpox vignette, 60% chose to expose their child to chickenpox now rather than later, without variation by student group. Prior work with patients showed a migraine vignette discount rate of 116% (95% CI 41–191%) and a rate in the chickenpox vignette of 59%. Conclusions: Medical students appear to have different time preferences than patients for the headache vignette, but approximately the same for the chickenpox vignette. More data (currently being collected) will further elucidate the time preferences of physicians. For now, given the lack of variation by student group, it appears at least some of the time preference difference between physicians and patients is inherent and not a result of training.

792 # 1574/OUTCOME ONE YEAR FOLLOWING TREATMENT FOR PRIMARY BONE TUMOUR Aileen M. Davis, Elaine P. Whittingham & Janet A. Parsons, Research, Toronto Rehabilitation Institute; Anthony Griffin, Jay S. Wunder & Robert S. Bell, Orthopaedic Surgery, Mount Sinai Hospital, Toronto, Ontario, Canada We have minimal understanding of the health-related quality of life (QOL) of patients treated for primary bone tumour. This study evaluated different constructs of QOL and determined the predictors of these constructs in order that we might inform improved interventions. Two hundred and two patients (upper extremity (UE), n ¼ 54; lower extremity (LE), n ¼ 164) were evaluated using the Musculoskeletal Tumor Society Scale (MSTS; symptoms and impairments), the Toronto Extremity Salvage Score (TESS; physical disability) and the Reintegration to Normal Living (RNL; participation in work, social and leisure activities, comfort with personal relationships and coping skills). Higher scores indicated higher QOL (MSTS range 0–35, TESS range 0–100; RNL range 0–110). Pre- and 1-year post treatment scores were compared using a paired t-test. Factors predictive of outcome were determined using regression. The relationships among the measures were evaluated using correlation methods. Mean age was 37 years (SD ¼ 16.5); M:F ¼ 1:1. One-third of patients received chemotherapy in addition to surgery of simple excision (15%), joint preservation (35%), joint resection and reconstruction (41%) or fusion (8%). There was no difference in the pre- and post-treatment MSTS score for UE or LE patients (p > 0.63). The TESS improved from 75.1 to 81.9 (p ¼ 0.05) and 67.1 to 75.3 (p < 0.001) in the UE and LE patients respectively. The RNL showed no difference in the UE patients but the LE patients improved from 81.8 to 88.2 (p ¼ 0.04). Having a LE tumour, specifically a pelvic tumour, a surgical procedure requiring joint resection and treatment complications was predictive of a lower MSTS and TESS score. The RNL was predicted by treatment complications. The MSTS was moderately correlated (r ¼ 0.63) with the TESS and weakly correlated with the RNL (r ¼ 0.38). The correlation between the TESS and RNL was moderate (r ¼ 0.52). The QOL concepts represent different constructs. This study has identified patient groups in which different aspects of QOL are affected providing a basis for developing targeted interventions that might improve outcome. # 1674/QUALITY OF LIFE FOR CHILDHOOD BRAIN TUMOR SURVIVORS Jin-shei Lai, Center on Outcomes, Research and Education (CORE), Evanston Northwestern Healthcare & Northwestern University, Evanston; Stewart Goldman, Division of Hematology – Oncology, Children’s Memorial Hospital at Chicago (CMH) & NU, Chicago; David Cella, CORE, ENH & NU, Evanston; Aimee Carlson, CMH, Chicago, IL, USA Brain tumors are the most common form of childhood solid tumor. With the improved treatment technology, presently over half survive longer than 5 years. However, quality of life (QOL) has become a concern due to potentially severe long-term effects of the tumor and/or its treatment. This study intended to advance our knowledge in this area by interviewing survivors, their parents, and experts with the goal of developing a QOL scale for this population. Subjects included 20 survivors (Ss), 20 parents (Ps), and 12 experts (Es) (five clinicians and seven teachers) with experiences in treating/teaching brain tumor survivors. Mean age of Ss was 9.3 (SD ¼ 1.5); mean number of years post-treatment was 4.0 (SD ¼ 2.1). Concerns addressed by subjects

were classified into: physical well-being (PWB), emotional well-being (EWB), social well-being (SWB), family well-being (FWB) and general cognitive functioning (GCF). In PWB, though few Ps (4/20) and Es (3/ 12) mentioned ‘fatigue’, most Ss were concerned about it (13/20). Ps and Es were concerned about ‘balance/falling’ (8/20) and ‘sensory impairments’ (9/12), respectively. Within EWB, Ps were concerned about the survivor being ‘moody’ (13/20) and ‘easily frustrated’ (12/20) while Es focused on survivors’ feelings of ‘being different from other kids’ (7/12) and ‘depression’ (6/12). We did not ask EWB-related questions of the Ss. Ss and Ps identified few problems in SWB, while Es were concerned that survivors ‘did not have many friends’ (6/12). Regarding FWB, Ss and Ps identified ‘arguing with parents’ (6/20) and ‘siblings’ (5/20) as major problems, while Es were concerned about ‘siblings’ reactions to the survivor’ (6/12). For GCF, Ss, Ps and Es all identified ‘poor memory’ and ‘poor attention span’ as major problems. In conclusion, survivors identified different problems in QoL than did parents and experts, particularly in PWB and SWB, highlighting the importance of taking survivors’ perspectives into account when measuring their QOL. Utilizing these results, we are currently developing a new QOL scale for childhood brain tumor survivors. # 1362/CHEMOTHERAPY-INDUCED NAUSEA AFFECTS PATIENTS’ QUALITY OF LIFE (QOL) MORE THAN VOMITING Benny C. Zee, Winnie Yeo, Maria Lai, Jane Koh, Emily Chu, Angel Lee & Tony Mok, Department of Clinical Oncology, Chinese University of Hong Kong, Shatin, Hong Kong Background: Chemotherapy-induced nausea and vomiting (CINV) are common complications of cancer treatment, affecting 70–80% of patients. Previous symptom control studies had used vomiting as primary endpoint and treat nausea as secondary endpoint. Objectives: To determine if nausea affects patients’ QOL more than that of vomiting. European Organization of Research and Treatment of Cancer (EORTC) Quality of Life Questionnaire (QLQ-C30) and the Functional Living Index for Emesis (FLIE) were used. Methods: We entered 46 cancer patients receiving moderately or highly emetogenic chemotherapy in the study. All patients completed diary for 5 days for EuroQoL Visual Analogue Scale (EQVAS), nausea and emetic episodes. The QLQ-C30 and FLIE were completed prior to chemotherapy and at the end of day 5. A change score was used in the analysis. For nausea and EQVAS, 5 days averages were used. Spearman rank correlations were used to assess the association between QOL and CINV. Results: Diary nausea was associated with QLQ-C30 nausea (r ¼ 0.78, p < 0.0001) and global QOL (r ¼ )0.30, p ¼ 0.0399), FLIE nausea subscale (r ¼ 0.80, p < 0.0001), FLIE vomiting subscale (r ¼ 0.73, p < 0.0001), FLIE overall score (r ¼ 0.86, p < 0.0001), EQVAS (r ¼ )0.51, p ¼ 0.0003). The QLQ-C30 global QOL associated with FLIE overall score (r ¼ )0.38, p ¼ 0.0107), FLIE nausea subscale (r ¼ )0.46, p ¼ 0.0014), but not with FLIE vomiting subscale (r ¼ )0.24, p ¼ 0.1064). The EQVAS score was significantly correlated with QLQ-C30 nausea (r ¼ )0.38, p ¼ 0.0103), global QOL (r ¼ 0.41, p ¼ 0.0051), FLIE nausea subscale (r ¼ )0.54, p < 0.0001), FLIE vomiting subscale (r ¼ )0.41, p ¼ 0.0060) and FLIE overall score (r ¼ )0.54, p < 0.0001). Conclusion: FLIE overall score is an aggregate measurement of 18 items and has a highest correlation with diary CINV. However, only FLIE nausea subscale but not FLIE vomiting subscale were significantly related to QLQ-C30 global QOL. The EQVAS and diary nausea score have higher correlations with FLIE nausea subscale than FLIE vomiting subscale indicating that nausea affects patients’ QOL more than vomiting. Study supported by Merck Sharp & Dohme (Asia) Ltd.

793 # 1527/QUALITY OF LIFE (QOL) FOLLOWING BONE MARROW TRANSPLANT (BMT) IN SPANISH-SPEAKING AND ENGLISHSPEAKING SURVIVORS Marcia Grant & Betty Ferrell, Nursing Research and Education, City of Hope National Medical Center, Duarte, CA, USA Purpose: The purpose of this study was to compare QOL assessment in Spanish-speaking BMT survivors with a matched cohort of Englishspeaking BMT survivors. Sample and methods: A group of 53 Mexican-American Spanish-speaking BMT survivors were matched on age, gender, time since transplant, and type of transplant with 74 Englishspeaking BMT survivors. Subjects were selected from the City of Hope survivor database. The questionnaire of 43 items is organized into the City of Hope (COH) four-dimensional framework (physical, psychological, social, and spiritual well-being). Translation procedures included translation into Mexican-American Spanish, back translation, and pilot testing for accuracy. Reliability and validity of both questionnaires were established in previous studies. Analysis/Evaluation: Initial analysis established the accuracy of the matched sampling procedure. Comparative analysis, using SPSS, was conducted for demographic and general health concerns, and across the four QOL dimensions. v2 and t-tests were used in the analysis. Comparisons included each dimension, as well as individual items. Findings and Implications: Significantly more Spanish-speaking patients were unable to identify whether or not graft-vs.-host disease was present. Fewer Spanish-speaking patients returned to their previous jobs and more Spanish-speaking patients reported not having health insurance. Both groups had highest QOL subscale scores in spiritual well being, with Spanish-speaking patients significantly higher. Lowest QOL scores were in social well being, with no differences between groups. Spanish-speaking survivors reported significantly lower scores in coping, and were more fearful of recurrence, second cancers, and metastases. Spanish-speaking patients had significantly more family distress but were more likely to feel they had a purpose in life, and were more hopeful. Results can be used as a basis for tailoring supportive interventions based on cultural differences. # 1836/CRITICAL REVIEW OF QUALITY OF LIFE ASSESSMENT IN PANCREATIC CANCER PATIENTS: FOCUS ON WEIGHT LOSS Lucie Kutikova & Lee Bowman, Global Health Outcomes Research, Eli Lilly and Company, Indianapolis, IN, USA The reported impact of weight loss on health-related quality of life (HRQOL) in pancreatic cancer patients (pts) translates primarily to three types of concerns: (1) low level of energy, fatigue, nutritional status, (2) body image changes, and (3) family/social impact. Weight loss has been recognized as an important prognostic factor and symptom in pancreatic cancer pts and has been incorporated into a clinical benefit algorithm used for the evaluation of treatment in advanced disease. Review of published literature and meeting abstracts from 1990 to 2003 was conducted to compile existing information regarding weight loss/cachexia, anorexia and HRQOL in pancreatic cancer pts. Additionally, content review of the HRQOL questionnaires used in the studies was performed. Among 11 studies where HRQOL of pancreatic cancer pts was evaluated with validated questionnaires, one study examined relationship between objectively measured weight loss and overall HRQOL. Patients with weight loss had significantly lower HRQOL scores. One generic, one gastrointestinal functionspecific and four cancer-specific validated questionnaires were used across the 11 studies. Use of pancreatic cancer-specific modules (EORTC QLQ-PAN-26, FACT-Hep) was not reported. Two cachexiaspecific questionnaires, constructed for use in cancer pts, were identified in the literature search. All but two questionnaires addressed energy/fatigue and family/social impact concerns. However, only two questionnaires and QLQ-PAN-26, FACT-Hep and cachexia-specific modules included items addressing body image issue. In conclusion, HRQOL questionnaires used in pancreatic cancer studies in the past did not fully address weight loss-related issues. As new treatments for pancreatic cancer emerge and their impact on HRQOL in pancreatic cancer pts is evaluated, pancreatic cancer-specific questionnaires should be used. Weight loss-specific questionnaires are available, however their length may be of concern. Further research should be conducted to examine effect of weight loss on HRQOL-related issues of pancreatic cancer pts as well as their caregivers.

# 1754/PREDICTORS OF SURVIVAL IN PROSTATE CANCER Charles Taft, Jan Karlsson & Marianne Sullivan, Health Care Research Unit; Erik Holmberg, Oncological Centre, Sahlgrenska University Hospital, Go¨teborg, Sweden Prostate cancer is one of the most frequently diagnosed malignancies in men and the second leading cause of male cancer deaths. The disease and its treatment are associated with adverse effects on condition-specific aspects of health-related quality of life (HRQL), e.g. urinary, bowel and sexual function. General HRQL domains appear to be only moderately impaired in patients with localized disease, while metastatic patients suffer considerable impairment in most domains. The aim was to identify significant clinical and HRQL predictors of long-term survival in prostate cancer patients. The sample consisted of 1138 (81%) unselected prostate cancer survivors (mean age at diagnosis ¼ 72.3 years) assessed 1.5 to 3.5 years after diagnosis included in the regional cancer registry over a well-defined catchment area of western Sweden (1.6 million inhabitants). Subjects completed the SF36 Health Survey, the EORTC QLQ-C30 and a prostate cancer-specific module using a mailout/mailback procedure. Ten-year survival statistics were obtained from the registry. HRQL variables, particularly SF-36 Physical Functioning and Social Functioning, were significant independent predictors of 1-, 5- and 10-year survival and were better predictors than commonly used clinical prognostic variables such as stage, grade, PSA-level. Combined HRQL and clinical variables accurately discriminated between short- and long-term survival. HRQL assessments may potentially enhance the accuracy of survival prognosis in prostate cancer. # 1412/FRONT-LOADED DARBEPOETIN ALFA PROVIDES QUICKER AND SUSTAINED IMPROVEMENT IN PATIENT REPORTED OUTCOMES COMPARED TO EPOETIN ALFA John J. Isitt, Health Economics, AMGEN INC., Thousand Oaks; John A. Glaspy, Oncology, UCLA, Los Angeles; N.S. Tchekmedyian, Oncology, Pacific Shores Medical Group, Long Beach; M.H. Erder, Joel Kallich, Health Economics, AMGEN INC., Thousand Oaks, CA, USA Anemia-related fatigue occurs during chemotherapy with negative effects on HRQOL. We investigated the change in HRQOL over time during front loaded darbepoetin alfa or standard rHuEPO in patients receiving cancer chemotherapy. An exploratory combined analysis of two randomized studies with anemic patients (hemoglobin (hgb) O11 g/dl) compared darbepoetin alfa accelerated correction and maintenance regimen (ACM; n ¼ 74) to conventional epoetin alfa (n ¼ 104). ACM used 4.5 mcg/kg during correction, and a lower/less frequent dose in maintenance. Patients completed the Functional Assessment of Cancer Therapy (FACT)-Anemia at monthly intervals. To assess differences in fatigue between patients dosed with ACM darbepoetin alfa and conventional epoetin alfa, the total number of severe fatigue-free days was measured in patients achieving a FACTFatigue subscale score higher than 30. On average patients receiving darbepoetin alfa had a clinically meaningful improvement in their FACT-Fatigue subscale score (>3 point change) by week 4 compared to 12 weeks for epoetin alfa. Early and sustained improvement was observed in all HRQOL measures. Patients dosed with ACM darbepoetin alpha had on average 6.12 (95% CI: 5.1, 7.2) fatigue free days compared to 4.36 (95% CI: 3.5, 5.2) fatigue free days in patients receiving conventional epoetin alpha over 12 weeks of treatment. Patients on darbepoetin alfa had on average significantly fewer days with severe fatigue than those who received conventional epoetin alfa therapy (p ¼ 0.0112, obtained from two tailed t-test).

794 # 1229/CHILDREN’S LEG PAIN – DEVELOPMENT AND VALIDATION OF A PARENTAL QUESTIONNAIRE FOR ASSESSMENT OF PREVALENCE AND QUALITY OF LIFE (QOL) ISSUES IN CHILDREN AGED 4–6 YEARS Angela M. Evans & Sheila D. Scutter, School of Health Science, University of South Australia, Adelaide, South Australia, Australia; Linda M. Lang, School of Health Science, University of Hallam – Sheffield, Sheffield, UK Purpose of research: To assess the prevalence and investigate QOL implications of children aged 4–6 years who suffer from recurrent, benign leg pains. Subject sample and statement of methods: The sample was derived from all schools and child centers in South Australia using systematic randomized selection. The total population was 77132. The sample size of 2321 was calculated [CLs (±) 2; 95% CI]. Questionnaire development involved triangulation of parent interviews, childrens focus group and the literature. (Questionnaire reliability was 82.4% and internal validity 93%). Summary of results: The prevalence of children’s leg pain based on the study sample was estimated at 36%, which is higher than previously reported. Analysis revealed many significant QOL relationships: Conclusion/s: This study is the first to estimate the prevalence of leg pain (growing pains) using a validated instrument in a well-defined sample. The 36% prevalence and QOL associations show the impact of this benign condition on children. The health professional consulted, treatment/intervention, frequency/features of leg pains and activity levels were all reported as factors associated with reduced childrens QOL. Primary health care providers need to be aware of the QOL effects of this too readily dismissed complaint. Table: Childrens QOL was estimated and reported by the parents. Significant association between QOL and other reported factors are shown. # 1810/PATIENT AND PROVIDER DIFFERENCES IN PERCEPTIONS OF PATIENT QUALITY OF LIFE: AN ANALYSIS OF PATIENTS USING COMPLEMENTARY AND ALTERNATIVE MEDICINE Amy R. Eisenstein, Center on Outcomes, Research and Education (CORE), Evanston Northwestern Healthcare (ENH), Evanston, Il.; David T. Eton, CORE, Karen Koffler, Integrative Medicine, ENH, Northwestern University, Evanston, Il. Healthcare provider’s understanding of their patient’s quality of life (QoL) can inform treatment and enrich the patient–provider relationship. Individuals seeking complementary and alternative forms of medicine (CAM) often report that they do so to enhance QoL. We determined the degree of concordance between patient and provider assessments of patient QoL. Fifty CAM-receiving patients (mean age ¼ 56.4 years; 72% female) and 12 licensed CAM practitioners (mean cumulative CAM experience ¼ 14.9 years) completed a battery of measures including: the Functional Assessment of Chronic Illness Therapy-Spirituality (FACIT-SP), the Brief Profile of Mood States (BPOMS), the Brief COPE, and the Patient Reactions Assessment (PRA) – a measure of patient–provider relationship. Providers completed the measures as if they were a typical patient receiving CAM therapy. In general, patient reports of their own well-being exceeded provider estimates of patient well-being. Patient reports of their own QoL exceeded providers’ estimates (p’s < 0.01). Patients reported less total mood disturbance than what provider’s estimated (p < 0.01). Patient reports of positive coping (taking action, acceptance) exceeded providers’ estimates, whereas provider estimates of negative coping (denial, behavioral disengagement) exceeded patient reports (p’s < 0.05). There were few differences in perception of the patient– provider relationship. Our results underscore the importance of relying on CAM patients to assess their own QoL and well-being. Implications for specifying the unique QoL aspects of patients using CAM therapies will be presented.

# 1171/CORRELATION BETWEEN PATIENT AND PROXY HQL SCORES IN A RECURRENT BRAIN CANCER TRIAL Wayne Weng, Medical & Scientific Affairs, Schering-Plough Corporation, Kenilworth, NJ, USA Background: Health-related quality of life (HRQOL) was evaluated using the EORTC QLO-C30 questionnaire in a clinical trial in which two chemotherapy treatments for recurrent brain cancer were compared. In this trial, in addition to the patient HRQOL data, a proxy was to complete the HRQOL questionnaire based on the patient’s HRQOL status throughout the trial. Objective: To evaluate the correlation between patient and proxy HRQOL scores to determine the appropriateness of using the proxy HRQOL scores to replace the missing patient HRQOL data. We evaluated both the absolute and change from baseline scores. Results: Two hundred and four patients with HRQOL data were included in the analysis. Patients were treated until tumor progression. More than 50% of patients had tumor progression by the end of the 3rd visit (84 days). On average, patient HRQOL scores were about 6–10 points higher (better) than the proxy HRQOL scores (range from 0 to 100) at baseline and the gaps remained the same at the end of visits 1 and 2. The Pearson correlation coefficients between patient and proxy scores at baseline ranged from 0.40 to 0.62 in the six general physical, social, and emotional functioning domains of the QLO-C30 and improved to 0.46 to 0.68 after two cycles of chemotherapy. The Pearson correlation coefficients between patient and proxy change from baseline scores ranged from 0.20 to 0.40 at cycle 1 and remained in this range at cycle 2. Furthermore, changes from baseline patient HRQOL scores between two treatment groups were quite different from those on proxy scores, especially in Physical Functioning and Role Functioning. The differences in change from baseline scores between the two treatment groups were less than 3 points based on patient evaluation and more than 10 points on proxy evaluation. Conclusion: The disparity between patient and proxy HRQOL absolute scores is sufficiently large and the correlation between patient and proxy change from baseline HRQOL scores is low in this trial. These results do not support the use of proxy HRQOL data in place of missing patient HRQOL evaluation. # 1757/PATIENT-INFORMANT AGREEMENT ON A MULTIDIMENSIONAL MEASURE OF COGNITIVE IMPAIRMENT AND ITS IMPACT ON HEALTH-RELATED QUALITY OF LIFE Louis S. Matza, MEDTAP International, Inc., Bethesda, MD; Jennifer A. Flynn, Eli Lilly and Co., Indianapolis, IN; Lori Frank & Leah Kleinman, MEDTAP International, Inc., Bethesda, MD; Lee Bowman, Eli Lilly and Co., Indianapolis, IN, USA; Roger Bullock, Kingshill Research Centre, Victoria Hospital, Swindon, UK The purpose of this pilot study was to assess patient-informant agreement on a multidimensional instrument designed to assess deficits related to cognitive impairment and their impact for individuals with mild cognitive impairment (MCI) and mild-to-moderate dementia of the Alzheimer’s type (DAT). Participants were 18 patient-informant dyads: 16 probable mild DAT (NINCDS-ADRDA criteria) and 2 MCI (memory complaints with informant corroboration; no ADL impairment; normal cognitive function; no dementia). Patients completed the Patient Reported Outcomes in Cognition instrument (PROCOG), a psychometrically validated 55-item questionnaire designed to capture symptoms and impact of cognitive impairment. Informants rated patients using an informant version of the PROCOG. Patient-informant agreement was assessed with t-tests, intra-class correlation coefficients (ICC), and Kendall’s tau. There were statistically significant differences between patient and informant reports on subscales assessing affect, skill loss, semantic memory, memory for recent events, cognitive function, and social impact (p < 0.05). ICC’s were in the small-to-moderate range ()0.08 to 0.39). In these domains, informants reported greater impairment than patients. The one domain (or subscale) without a significant patient-informant difference was a single long-term memory item. These pilot results from a mildly impaired sample are consistent with previous research with more severe DAT samples, which suggests that MCI and mild DAT patients may underestimate cognitive deficits and their impact on HRQL domains such as affect and social functioning. Use of this instrument may add a unique contribution to the characterization of cognitive impairment from the patient and informant perspectives.

795 # 1386/DIFFERENCES ON THE EVALUATION OF THE HEALTH RELATED QUALITY OF LIFE (HRQOL) BETWEEN DIALYSIS PATIENTS AND THEIR CAREGIVERS (FAMILIAR, NURSES AND PHYSICIANS) Fernando Alvarez-Ude, Medicina Interna, Hospital General Segovia, Segovia; Covadonga Valdes, Nefrology I, Hospital Central Asturias, Oviedo; Carmen Estebanez, Medicina Interna, Hospital General Segovia, Segovia; Pablo Rebollo, Nefrology I, Hospital Central Asturias, Oviedo, Spain When the patients on dialysis cannot express their HRQoL, it is necessary to use the carer’s assessment of patients’ QoL. The aim of this study was to assess the agreement between patients and carers: (family carer-FAM, nurse-NUR, physician-PH) on patient’s. A sample of 221 pairs of patient-family carer stratified by age and gender was randomly selected out of 14 dialysis units: 152 hemodialysis and 69 peritoneal dialysis. Patients’ QoL was evaluated by the patient and FAM, NUR and PH, using both components of the EQ-5D: the five dimensions (Mobility – M, SelfCare – SC, Usual Activities – UA, Pain – P, Anxiety/Depression – AD) and their Tariff (T), and the Visual Analogue Scale (VAS). Patients and FAM answered the SF-36 (physical and mental component summary: PCS, MCS). Physicians scored the comorbidity Index of patients and nurses scored the Barthel Scale (BS) of patients. The correlation coefficients for the VAS of the patients and carers were 0.42 (FAM), 0.49 (NUR) and 0.30 (PH); NUR and PH scored VAS higher than patients did (p < 0.01). For the T, the correlation coefficients were 0.66 (FAM), 0.68 (NUR) and 0.57 (PH); FAM scored T lower than patients did (p < 0.01). The agreement (Kappa) between the EQ-5D varied between moderate for M (0.56 FAM; 0.55 NUR; 0.47 PH) and for SC (0.48 FAM; 0.50 NUR; 0.42 PH) and low or insignificant for P (0.32 FAM; 0.40 NUR; 0.18 PH) and for AD (0.26 FAM; 0.28 NUR; 0.14 PH). The variables associated to the size of differences found for VAS were: FAM: patient comorbidity (+0.61) and MCS of the FAM ()0.48); NUR: BS ()0.23); PH: age of the physician ()0.41). So, the agreement between patients and carers for objective dimensions of QOL (M and SC) was moderate, and it was low for subjective dimensions (P and AD). The agreement in all the EQ-5D dimensions was lower for PH than for FAM and NUR, and the variables associated to a greater difference were: (1) FAM: higher comorbidity of the patient and worse mental health state of family carer; (2) NUR: greater degree of dependency of the patient; (3) PH: lower age (experience) of the physician. # 1401/IS THERE A BEST SELF-ADMINISTERED INSTRUMENT FOR EMPHYSEMA AND COPD APPROPRIATE FOR CLINICIAN USE? Deborah L. Cullen, Respiratory Therapy Program, Indiana University, Indianapolis, IN, USA The Scientific Advisory Committee for the Medical Outcomes Trust has generated standards as a litmus test for quality of life measures. Eight categorical attributes have been defined: conceptual model, reliability, validity, responsiveness, interpretability, burden of administration, alternative modes and cultural/language adaptation. Instruments that can determine treatment effectiveness may be of tremendous value. However, clinicians are uncertain regarding the best measure to augment the patient interview and physiological data. The purpose of this study was to evaluate instruments designed for Emphysema and COPD against the new international standards to clarify which measure(s) can be confidently utilized in clinical practice. The Pulmonary Function Status Scale (PFSS), Pulmonary Function Status and Dyspnea Questionnaire-Modified (PFSDQ-M), and the St. George’s Respiratory Questionnaire (SGRQ) were evaluated. For each instrument attribute, literature-based psychometric support was examined. Evidence was rated strong (+++), moderate (++), limited (+) or not available/applicable (NA). Strong evidence was available defining the methods utilized for validity, reliability and burden of administration for these instruments. Less literature was evidenced for responsiveness or interpretability. Very little evidence was available for alternative modes or cultural adaptation. The conceptual model differed for each instrument but was well described in the literature. Clinical adoption is recommended for any of these measures to supplement treatment outcomes. The SGRQ is the most commonly utilized measure.

# 1208/CHRONIC ALLERGIC RHINITIS: QUALITY OF LIFE IMPACT Charles Taieb & Eric Myon, Pharmaco Economics & Quality of Life Department, IRPF, Boulogne Billancourt, France Context: Allergic rhinitis is a real problem for public health insofar as it generally affects young and economically active subjects, whose quality of life is altered. The WHO has ranked it sixth in its concerns. Objective: The objective is to assess the quality of life impact of treatment of rhinitis in adults depending on whether the patient is treated homeopathically or allopathically. Method: A generic scale (SF-12) is completed over a sufficiently long period of time and at regular intervals (Inclusion and 3 months). The completed questionnaires were returned by post through prepaid envelopes. Results: The results consisted of 2 scores: mental (MCS-12) and physical (PCS12). The norm observed in a standard American population is 50 for each dimension. At inclusion the population under homeopathic treatment scores were: PCS-12 ¼ 49.9 (6.8) MCS-12 ¼ 42.4 (10.3), 3 months later the patients scores were: PCS-12 ¼ 51.8 (6.4) MCS12 ¼ 47.6 (8.2) p-values PCS-12 ¼ 0.13 MCS-12 ¼ 0.004. At inclusion the population under allopathic treatment scores were: PCS12 ¼ 48.7 (8.5) MCS-12 ¼ 46.5 (7.8), 3 months later the patients scores were: PCS-12 ¼ 47.9 (8.7) MCS-12 ¼ 47.3 (9.2) p-values PCS-12 ¼ 0.7 MCS-12 ¼ 0.7. The population under homeopathic treatment scored better on the physical scale than the group under allopathic treatment. Conclusion: With regard to the evaluation of quality of life as recorded on the SF-12, the population under homeopathic treatment scored better on the physical scale than the group under allopathic treatment, the score of the two groups had been identical at the start of the study. It had therefore improved for the former group. # 1466/QUALITY OF LIFE OF COPD PATIENTS USING BOTH WESTERN MEDICINE AND TRADITIONAL CHINESE MEDICINE Jian-Jung Chen, Department of Chinese Medicine, Chinese Medicine College Hospital; Li-Jin Huang, Department of Nursing, China Medical College; Te-Chen Hsia & Liang-Wen Hang, Medical Department, China Medicine College Hospital; Wen-Miin Liang, Department of Public Health, China Medicine College; Pi-Chu Hsieh, Department of Chinese Medicine, China Medical College Hospital, Taichung; ChinYen Wu, Division of General Thoracic Surgery, Changhua Christian Hospital, Changhua, Taiwan The purpose of this study was to understand the attitudes of Taiwanese COPD patients who have been regularly taking western medicine (WM) and traditional Chinese medicine (TCM) for at least 1 year, and to assess the combined therapies affected their quality of life (QOL). A between-method triangulation design including qualitative and quantitative data was used. Qualitative data were collected by a focus group in 2003. Quantitative data were collected by using WHOQOL-BREF for study patients and 262 COPD patients from western medicine hospital comprised the control group. COPD patients receiving regular treatment in the department of Chinese medicine in a medical center were invited to participate in the focus group and interviewed by the WHOQOL-BREF. Five male COPD patients participated in the study with age from 58.5 to 78.8 years old. The patients in focus group were asked: ‘Why did you choose to take TCM regularly for your COPD?’ and ‘What has changed in your life as a result of taking TCM?’ Some concrete concepts were summarized as follows. The attitudes related to taking TCM were: less side effect, traditional thoughts and feeling efficacy. The QOL related problems with which the subjects were most concerned were: coughing hindering interpersonal relationships, coughing influencing traveling, symptoms limiting dietary intake. Patients in the focus group liked exercising as a way to relieve symptoms. The results from WHOQOLBREF showed similar results to those obtained using the focus group. For examples, QOL were clearly better in the focus group compared to the control group with regard to participating for leisure activity (focus group, 3.8; control group, 2.8), positive feeling (4.2 vs. 3.6). QOL was markedly worse in the focus group compared to controls with regard to concept of ‘face’ (3.0 vs. 3.7), liberal attitude to food (3.8 vs. 4.4). Overall, despite being older and having a later stage of disease, the focus group patients taking both WM and TCM appeared to have employed various practical and psychological strategies to master their disease and live as normal a life as possible.

796 # 1293/THE IMPACT OF WHEEZING AND DYSPNEA ON HEALTHRELATED QUALITY OF LIFE IN A RANDOM GENERAL POPULATION SAMPLE OF PRE-SCHOOL CHILDREN Ashna D. Mohangoo, Public Health, Erasmus MC, University Medical Center Rotterdam, Rotterdam, ZH; Marie-Louise Essink-Bot, Public Health; Henrie¨tte A. Moll, Pediatrics, Erasmus MC, Rotterdam, ZH, The Netherlands; Elizabeth F. Juniper, Clinical Epidemiology, McMaster University, Hamilton, Ontario, Canada; Harry J. de Koning & Hein Raat, Public Health, Erasmus MC, Rotterdam, ZH, The Netherlands Wheezing and dyspnea have a high, still increasing, prevalence in preschool children. As asthma in this age group is often reversible, one of the suggested aims of treatment is to improve the current health-related quality of life (HR-QoL). In this study we assessed the impact of parent-reported current (prior 4 weeks) wheezing and dyspnea on generic HR-QoL in pre-school children. Generic HR-QoL was measured by the 12 scale TNO-AZL Pre-school Children Quality of Life (TAPQOL) questionnaire for parents that covers physical, social, cognitive and emotional functioning. A random general population sample of 500 parents of 2 months–3 years old children was mailed the TAPQOL and a respiratory symptoms questionnaire. Response was 82% (n ¼ 410; 50% boys). The prevalence of combined wheezing and dyspnea during the past 4 weeks (asthma symptoms) was 10% (n ¼ 41; 51% boys). For five TAPQOL scales (‘lung problems’, ‘sleeping’, ‘appetite’, ‘positive mood’ and ‘communication’) we observed significantly lower mean scores in the subgroup children with asthma symptoms as compared to the subgroup without these symptoms (p < 0.05). Cohen’s effect sizes were large (defined as P0.80) for ‘lung problems’ (2.06) and ‘sleeping’ (0.80). The impact of asthma symptoms on HR-QoL was more profound in 2–12 months old children than in 1–3 year olds. In the subgroup with asthma symptoms there were no significant differences with regard to mean TAPQOL scale scores between boys and girls (p P 0.05). In conclusion, we established a negative impact of current asthma symptoms on generic HR-QoL, especially in infants younger than 1 year old. At this moment, an asthma specific HR-QoL measure as well as empirically established norms for ‘minimal important differences’ (MIDs) are not yet available for pre-school children. We suggest the development of such an instrument and empirically based MIDs to further enhance measurement of the consequences of asthma and the effects of asthma management in pre-school children. # 1614/HEALTH-RELATED QUALITY OF LIFE ASSOCIATED WITH CHRONIC BRONCHITIS IN THREE EUROPEAN COUNTRIES Darren J. Clayson & Helen A. Doll, Outcomes Research, Oxford Outcomes Ltd, Oxford, Oxon, UK; Marc Miravitlles, Barcelona, Catalonia, Spain; Joachim Lorenz, Luedenscheid, Westphalia, Germany; Jacques Gaillat, Service des Maladies Infectieuses, Centre Hospitalier d’Annecy, Annecy Cedex; Isabelle Duprat-Lomon, Health Economics and Outcomes Research, Bayer Pharma S.A., Puteaux, Cedex, France; Pierre Philippe Sagnier, Global Health Economics and Outcomes Research, Bayer Plc, Slough, Berkshire, UK The aim of this study was to explore the health related quality of life (HRQoL) associated with chronic bronchitis (CB) in three European countries to enable Europe-wide assessment of CB-associated impairments. Collectively 1138 patients aged 40 years and over from studies conducted in France (n ¼ 413), Germany (n ¼ 283), and Spain (n ¼ 442) were analysed at acute exacerbation of chronic bronchitis (AECB) and AECB-free period. Patients were recruited prospectively from those presenting with AECB in general practice (France and Germany) or stable chronic obstructive pulmonary disease (COPD) in thoracic medicine departments (Spain). The St George’s Respiratory

Questionnaire (SGRQ) was used to assess HRQoL. The patients were aged 66.2 (SD 10.0) years on average; 78% were male. The mean number of AECBs reported during the previous year differed among the countries, with the German patients reporting the highest number and the Spanish the lowest, although Spanish patients had particularly poor lung function in line with recruitment criteria. The impact of CB on HRQoL was indicated by very high (i.e. poor) SGRQ scores at both AECB and AECB-free period compared with general population norms. Factors independently associated with HRQoL impairment included AECB severity and previous AECB number. Spanish patients reported notably better SGRQ symptoms scores (at AECB-free period) than German patients (mean 49.5 vs. 63.0, respectively), but poorer SGRQ Activities (64.1 vs. 49.9) and Impacts (39.0 vs. 32.7) scores. The poorer SGRQ Symptoms scores in the German patients, even though Spanish patients had more severe basic CB, reflects the greater number of exacerbations reported by the German patients in the previous year. This study demonstrates the high level of HRQoL impairment associated with CB in Europe. In particular, the study highlights the importance of controlling AECBs, both in number and severity, in an attempt to improve HRQoL. The number of previous AECBs should be taken into account when interpreting SGRQ Symptom scores, which are highly dependent on both previous AECB number and severity. # 1637/ASSESSING CLINICALLY MEANINGFUL DIFFERENCES: RESPONSIVENESS OF THE PATIENT ASSESSMENT OF UPPER GASTROINTESTINAL SYMPTOM SEVERITY INDEX (PAGI-SYM) Anne M. Rentz & Dennis A. Revicki, Center for Health Outcomes Research, MEDTAP International, Bethesda, MD, USA; Dominique Dubois, HE & Pricing, Johnson & Johnson Pharmaceutical Services, LLC, Beerse, Belgium Purpose: Evaluate the responsiveness of a new patient self-report instrument, the Patient Assessment of Upper Gastrointestinal Disorders-Symptom Severity Index (PAGI-SYM) in subjects with gastroesophageal reflux disease (GERD) and dyspepsia. Statement of methods: Subjects with GERD (n ¼ 810) or dyspepsia (n ¼ 767) participated in this multi-country, observational study. All subjects completed the PAGI-SYM, a 20-item, six subscale questionnaire, a global symptom relief question, and a measure of patient-rated change in GIrelated symptoms, the Overall Treatment Effect (OTE) scale. Responsiveness was evaluated at 8 weeks by comparing groups by disease, symptom relief, and OTE (improved, stable, and worsened). Summary of results: Subjects reporting symptom relief reported significantly lower (better) PAGI-SYM scores than those reporting no symptom relief (p < 0.0001 to p < 0.0005). The largest differences in mean scores were seen for fullness/early satiety (1.15–1.16), bloating (1.06–1.08) and upper abdominal pain (1.02–1.05) subscales; large differences were also seen for heartburn/regurgitation in the GERD sample (1.17). Subjects with improvements in overall GI symptoms demonstrated significant decreases in PAGI-SYM subscale scores compared with those who remained the same or worsened (all p < 0.0001). Effect sizes ranged from 0.21 to 1.28 and standard errors of measurement ranged from 0.29 to 0.63, depending on subscale and disease sample. Based on these results, a treatment difference of 0.30 to 0.55 points is recommended as the minimally important difference for heartburn/regurgitation scores in GERD studies and a treatment difference of 0.30 to 0.70 points is recommended as the minimally important difference for the fullness/early satiety, bloating, and upper abdominal pain scores in patients with dyspepsia. Conclusions: The PAGI-SYM is a brief symptom severity instrument measuring common upper GI symptoms with demonstrated responsiveness and sensitivity to change in clinical status in subjects with GERD or dyspepsia.

797 # 1112/IS THERE A CONSTANT RELATIONSHIP BETWEEN THE MINIMALLY IMPORTANT DIFFERENCE AND HEALTH STATE UTILITY VALUES? Stephen J. Walters & John E. Brazier, ScHARR, University of Sheffield, Sheffield, S. Yorkshire, United Kingdom The SF-6D and EQ-5D are both preference-based measures of health. Empirical work is required to determine what the smallest change is in utility scores that can be regarded as important and whether this change in utility value is constant across measures and conditions. The objectives of this study were to use anchor-based methods to determine and compare the Minimally Important Difference (MID) for the SF-6D and EQ-5D for various datasets. The SF-6D and EQ-5D can be regarded as continuous outcomes. The SF-6D is scored on a 0.3 to 1.0 scale and the EQ-5D on a )0.6 to 1.0. For both measures a score of 1.0 indicates ‘full health’. Patients were followed for a period of time, then asked, using question 2 of the SF-36 as our anchor or global rating scale, (which is not part of the SF-6D), if there general health is much better (5), somewhat better (4), stayed the same (3), somewhat worse (2) or much worse (1) compared to the last time they were assessed. We considered patients whose global rating score was 4 or 2 as having experienced some change equivalent to the MID. In patients who reported a worsening of health (global change of 1 or 2) the sign of the change in the utility score was reversed (i.e. multiplied by minus one). The MID for the SF-6D and EQ-5D was then taken as the mean change of the patients who scored (2 or 4). This paper describes and compares the MID for the SF-6D and EQ-5D from longitudinal studies in 10 patient groups which used both instruments. From the reviewed studies the MID for the SF-6D ranged from 0.011 to 0.097, mean 0.043. The mean MID for the EQ-5D was 0.076 (range 0.001– 0.140). The paired mean difference between the EQ-5D and SF-6D MIDs was 0.04 (95% CI: 0.02–0.07). There is some evidence that the MID for these two utility measures is not the same, and that the MID varies between conditions. Further empirical work is required to see whether or not this holds true for other utility measures, patient groups and populations. # 1210/CADI: CULTURAL AND LINGUISTIC VALIDATION INTO FRENCH OF AN ACNE DISABILITY INDEX Francois Verriere & Therese Nocera, Medical Direction, Laboratoires Dermatologiques Ave`ne; Charles Taieb & Eric Myon, Pharmaco Economics & Quality of Life Department, IRPF, Boulogne Billancourt, France Background: The Cardiff Acne Disability Index (CADI) is a scale devised by Professors R.J. Motley and A.Y. Finlay of Cardiff [1] to evaluate the impact of acne on the life of the patient. This questionnaire has been used in some studies to evaluate the consequences of the disease in daily life or the effects of different treatments. Objective: The objective of our work was to translate the Cardiff Acne Disability Index (CADI) into French, with careful attention to the linguistic aspects and the cultural context of the French patients. Method: Following international methodological recommendations, the CADI scale was translated in a standardized way consisting of forward translation, quality control, backward translation, and a pilot test (see Figure 1). Highly experienced bilingual translators and reviewers were involved. Results: Many difficulties of compatibility between the cultural background of United Kingdom and France were identified and resolved by consensus. A pre-test in volunteers demonstrated clarity and understand ability across social classes and ages no changes were therefore needed in the instrument. A preliminary test–retest comparison of the final scale showed sufficient reliability with a correlation coefficient of 0.90 for the French version concerning the total CADI score (n ¼ 16) (see Figure 2). Moreover a good internal consistency of scale was demonstrated with an Cronbach a equals to 0.87 (n ¼ 22). Investigations on different kinds of validity are underway. Conclusion: The CADI scale is a valuable tool for assessing the impact of acne on the life of the patient. This scale can nowadays be used in French (tested only in France and not for all the French speaking countries – see Figure 3 for the final French version of the scale) [1] R.J. Motley, A.Y. Finlay. Practical use of a disability index in the routine management of acne. Clinical Exp Dermatol 1992; 17: 1–3.

# 1554/TRANSLATION OF QUALITY OF LIFE QUESTIONNAIRES: IMPLICATIONS FOR IRANIAN POPULATION Hamid R. Javadi, Department of cardiology, Qazvin University of Medical Sciences, Avesina Hosp, Qazvin, Iran; Mohsen Asadi-Lari, Division of Cardiovascular Medicine, University of Nottingham, University Hospital (QMC), Nottingham, UK; Ali A. Sayyari, Ministry of Health, Tehran, Iran; David Gray, Division of Cardiovascular Medicine, University of Nottingham, University Hospital (QMC), Nottingham, UK The trend towards globalisation has resulted in growing international collaboration to conduct multinational and cross-cultural health-related quality of life assessment. Finding the ideal bilingual candidate to translate the questionnaire who is fluent in the target language, quite familiar with the culture, and the content of the instrument, is the first obstacle. Despite applying the standard approach of forward and backward translation, test–retest on five bilingual respondents to confirm language and cultural equivalence, review of the translation by an editorial board and finally reliability and validity performance test on a pilot population, some problems remain. First, the literacy rate in our study population was low at 50%, perhaps reflecting the age group studied. Second, cultural barriers may play an important role in crosscultural investigations, for example while questions relating to sexual function may be difficult in western countries, among populations with strong religious beliefs similar questions are forbidden. Third, people are not familiar with self-administered tools; training appropriate interviewers is not straightforward, and whilst interview administration may facilitate the process, it has apparent disadvantages. Moreover matching the 7-point interval scale wording found in MacNew or similar tools can be difficult. This presentation discusses the barriers we encountered in preparing a range of generic and specific quality of life tools for administration in Iran. # 1300/SIMULTANEOUS GLOBAL DEVELOPMENT: A NEW METHODOLOGY FOR QUESTIONNAIRE DEVELOPMENT (AN EXAMPLE FROM RHEUMATOLOGY) David U. Himmelberger, Health Outcomes Group, Palo Alto, CA, USA Many of the questionnaires in use today are strongly biased towards the English language and western culture. These biases are usually only revealed through translation and linguistic validation as the questionnaires are prepared for use in different languages. Researchers try to minimize bias inherent in the traditional development process by ‘cultural adaptation’. By taking a multinational approach where questionnaire development is conducted simultaneously throughout the world in linguistic families most of these biases will be eliminated and the resulting questionnaire will be easily translated between languages with no change to the content. We propose that in the early stages of questionnaire development, when determining what areas of a disease are most important to the patient, that a multinational approach be used. Data from focus groups and interviews with patients are collected simultaneously on each continent in countries of prime importance and then compared to ascertain the similarities and differences. This allows one to determine the most relevant aspects of the disease on a worldwide basis, not just from a western perspective. The goals of this development process are twofold; first, to construct a questionnaire with a universal content of equal importance to patients throughout the world, and second, to identify items that are significant to local areas. Once the primary items of importance have been agreed upon, questions are developed with the aid of linguists representing the major language families who work together to construct questions that do not contain the colloquialisms and ambiguities often found in current questionnaires. This process will permit easy translation into many languages. The simultaneous, global method of questionnaire development results in a questionnaire with a content that is relevant and equally important to all cultures of the world. This methodology will also produce a slightly larger questionnaire that has a content highly relevant to a specific culture. Examples will be provided from rheumatology.

798 # 1265/PSYCHOSOCIAL CONSEQUENCES OF FALSE POSITIVE SCREENING MAMMOGRAPHY – AN ADAPTATION OF THE PSYCHOLOGICAL CONSEQUENCES QUESTIONNAIRE (PCQ) INTO DANISH John Brodersen & Hanne Thorsen, Department of General Practice, Institute of Public, University of Copenhagen, Copenhagen, Denmark The aim of the study was to adapt and validate the PCQ into Danish. Four bilinguals in English/Danish first translated the Australian 12-item version of the PCQ. A lay panel of four female public school teacher, evaluated the Danish translation. Six focus group interviews, each with 5–7 participants were carried out. All participants were women who previously had a false positive mammography screening. The consequences of false positive mammography were discussed and the women were asked to complete the questionnaire. They were then asked whether or not they understood each item, if they found them unambiguous, jargon-free, and if the response options were relevant. If not, they were asked to suggest alternative wordings. They were also asked if they found the items relevant and covering the feelings, thoughts, and emotional, physical, social and sexual consequences they had themselves experienced. The design and easiness to complete was also discuss. Fifteen additional women took part in telephone interviews after having completed the questionnaire. The purpose of these interviews was to see if the questionnaire was acceptable for women who had not yet been screened or for women who had never taken part in a mammography screening. Results from all the interviews showed that three items were ambiguous and two were double-barrelled. None of the original items were found to be irrelevant but the interviewees found that not all the consequences of a false positive screening mammography were covered by the questionnaire. Therefore all together 15 new items were added. All together 600 women have completed the extended version of the questionnaire together with the Danish version of the Nottingham Health Profile. One-third of the women had an abnormal screening mammography and two third had a normal. Analyses of these data to estimate the reliability, validity, dimensionality, additivity and order of the questionnaire is ongoing and will be presented at the conference. # 1561/VALIDATION OF RUSSIAN VERSIONS OF BRIEF PAIN INVENTORY (BPI-R), BRIEF FATIGUE INVENTORY (BFI-R) AND M.D. ANDERSON SYMPTOM INVENTORY (MDASI-R) TO ASSESS SYMPTOMS IN CANCER PATIENTS T. I. Ionova & A. A. Novik, Multinational Center of Quality of Life Research; G. V. Manikhas, City Oncological Center; X. S. Wang, T. R. Mendoza, Houston, S. A. Kalyadina, M. O. Ivanova, O. Y. Uspenskaya & A. V. Kishtovich, Multinational Center of Quality of Life Research, St. Petersburg, Russia; C. S. Cleeland, Anderson Cancer Center, Houston, USA Despite the distress that symptoms can cause, comprehensive symptom assessment is rarely part of routine cancer care. However, it can provide the basis for detecting symptoms, grading their severity, and evaluating treatment adequacy. In Russia, lack of valid measures has led to lack of symptom assessment. We recently completed translation and cultural adaptation of BPI-R to assess pain, BFI-R to assess fatigue, and MDASI-R to assess major symptoms in cancer patients. This study provides validation of the questionnaires. The study sample consisted of 221 advanced cancer patients (38% males,

62% females, mean age 62). Cronbach coefficient as established reliability. Construct validity was determined using confirmatory factor analysis. Correlation analysis between questionnaire scales and the SF-36 was used for convergent validity analysis. Discriminant validity was established correlating questionnaire scales with ECOG performance status (PS). Cronbach coefficient as were: 0.9 for BPI-R and BFI-R for both severity and interference; 0.8, 0.7, and 0.9 for three MDASI-R factors. Confirmatory factor analysis of BPI-R and BFI-R resulted in a two-factor solution; such analysis of MDASI-R resulted in a three-factor solution, satisfying reproducibility and interpretability criteria. Significant Spearman rank correlations indicated that symptom severities are associated with poorer PS: 0.41 (p < 0.01) between BPI pain severity and ECOG PS correlates severe pain with poorer PS; 0.491 (p ¼ 0.00) between BFI-R fatigue severity and ECOG PS correlates severe fatigue with poorer PS; 0.481 (p < 0.01) between the mean of symptom severity items and ECOG PS correlates severe symptoms with poorer PS. Satisfactory convergent validity was proved by correlation analysis between the questionnaires scales and SF-36. In sum, the BPI-R, BFI-R and MDASI-R are reliable, valid instruments for rapid assessment of cancer symptoms in Russia. # 1136/VERIFICATION OF PSYCHOMETRIC PROPERTIES OF THE WHOQOL-BREF POLISH VERSION Krystyna Jaracz, Krystyna Gorna & Grazyna Baczyk, Department of Nursing, University of Medical Sciences, Poznan, Poland The aim of this study was to evaluate validity and reliability of the short form WHOQOL instrument (WHOQOL-Bref). The sample consisted of 1109 patients and healthy subjects. The mean age for the patients group and the healthy group was 44.6 and 25.1 years, respectively. The sick group included more males and was more poorly educated. WHOQOL was translated into Polish according to the standard procedure provided by the WHOQOL group. The instrument produces a profile with four domain scores ((1) physical, (2) psychological, (3) social relationships, (4) environment) and two separately scored items about the individual’s overall perception of quality of life (Q1) and health (Q2). We evaluated: (1) internal consistency by Cronbach’s a, (2) concurrent validity by Spearman’s correlation analysis between the domain and the general questions scores and self-rating health, (3) discriminant validity by comparing healthy and unhealthy subjects on the domain scores, (4) construct validity by exploratory factor analysis and (5) importance of individual domain in assessing overall perception of quality of life and health by multiple linear regression. The results were following: (1) the Cronbach’s a of the four domains ranged from 0.66 in domain 3 (social relationships) to 0.79 in domain 1 (physical), (2) there were significant correlation between health status, and the domain scores as well as Q1 and Q2 (p < 0.01), (3) the domain 1, 2 and 3 scores showed significant differences between patients and healthy subjects (p < 0.001), (4) a factor analysis confirmed the different subscales in a large extend, however the proportion of the explained variance was a little low (48%), (5) in regression analysis when Q1 was used as the dependent variable, all the four domains made significant contribution in explaining the variance. The psychological domain had the highest input. When Q2 was analysed, two domains (physical and psychological) made significant contribution but the physical domain contributed most. In conclusion, these preliminary data suggest that the WHOQOL-Bref is reliable and valid, however further analyses are needed.

799 # 1047/LINGUISTIC ADAPTATION AND VALIDATION OF THE BENIGN PROSTATIC HYPERPLASIA-PATIENT IMPACT MEASURE (BPH-PIM) INTO IBERIAN SPANISH Carlos Lorente, Department of Urology, Hospital Fundacio´n Alcorco´n, Alcorco´n; Miguel Ruiz, Department of Methodology, College of Psychology; Manuel Esteban, Department of Urology, Hospital Prı´ncipe de Asturias, Alcala´ de Henares; Fernando Villasante, Primary Care, Health Center ‘Orcasitas’; Javier Rejas, Health Outcomes Research, Medical Unit, Pfizer Spain, Alcobendas, Madrid, Spain; Asha Hareendran, Health Outcomes Research, Central Research Sandwich, Pfizer Ltd., Sandwich, United Kingdom Aim: The BPH-PIM is a self-reported disease specific Quality of Life questionnaire to measure the impact of urinary/prostate symptoms on the Quality of Life of subjects with BPH (Benign Prostatic Hyperplasia). The aim of the present study was to adapt the original instrument to Iberian Spanish and to assess the psychometric properties of the new version. Methodology: Linguistic adaptation was performed using the standard processes for establishing conceptual equivalence. Psychometric properties were assessed on a convenience sample of 131 patients with BPH, who completed the BPH-PIM at baseline, 1 week later and after 4 weeks of therapy with doxazosin GITS. Results: All the standardised items had response rates above 99%, there were no floor or ceiling effects. Reliability in terms of Cronbach’s a was 0.96 and intra-class correlation coefficient was 0.97. Construct validity was confirmed by factor analysis and showed two factors corresponding to the hypothesised Psychological Impact and General Function Interference domains, explaining 67% of the variance and a third factor corresponding to Specific Function Interference. Convergent validity was confirmed in terms of the relationship to the SF-12 (p < 0.01), HUI-3 (p < 0.01) and the IPSS QoL item (p < 0.01). The instrument showed discriminant validity, in terms of IPSS (International Prostate Symptom Score) score groups. There was a mean difference of 23.5 (SE ¼ 3.58) points (0–100 scale) between moderate and severe BPH patients (p < 0.001). The instrument was sensitive to change in terms of the correlation of the domain scores of the PIM to improvements in symptom scores (p < 0.001). Significant differences were observed for those who improved 3 or more points in the IPSS scale (p < 0.001). Conclusion: A psychometrically valid and conceptually equivalent version of the BPH-PIM questionnaire has been produced to be used in patients with BPH in Spanish speaking countries. # 1879/LINGUISTIC VALIDATION OF A SLEEP QUALITY SCALE IN 32 LANGUAGES Katrin Conway, Mapi Research Institute, Lyon, France; Sandy Lewis, Merck & Co. Inc., Philadelphia, NJ, USA, Marion Chaput, Mapi Research Institute, Lyon, France Objectives: Measuring the quality of sleep of individuals included in an international clinical trial involving 32 languages required the linguistic validation of the single item questionnaire developed in US English. The single item asks patients to rate their overall sleep quality. The instructions section directed patients to think of sleep quality in terms of the number of hours slept, ease in which they fell asleep, amount of awakenings and how refreshing their sleep is. Methods: The translation process of this sleep quality scale was co-ordinated by a specialist in each target country and was conducted according to the following

methodology: (1) two forward translations by professional translators, native speakers of the target language and fluent in English; (2) comparison and reconciliation of the translations by the specialist in the target country and translators; (3) backward translation by a native English speaker; (4) comparison of source and backward version; (5) comprehension test in five healthy subjects; (6) international harmonisation to check the comparability of translations across countries. Results: Linguistic and conceptual issues emerged when translating idiomatic English phrases. No linguistic equivalent was available for the idiomatic US expression ‘go to the bathroom’ that is listed in the instructions section. To translate the intended concept language versions were required to be more specific. Similarly the time reference expressed as ‘the majority of nights during the last 7 days’ was confusing through the juxtaposition of ‘night’ and ‘day’ and appropriate alternatives had to be found in the Latin languages. Conclusions: The 32 language versions were established according to a rigourous translation process to ensure conceptual equivalence and cross-cultural acceptability of translations to facilitate the international comparison and poling of data. Psychometric testing will be conducted to ensure the reliability and validity of each translation. This work highlights the value of international feedback prior to finalizing the original questionnaire. # 1634/RELIABITLITY AND VALIDTY OF THE KANSAS CITY CARDIOMYOPATHY QUESTIONNAIRE (KCCQ) IN PATIENTS WITH PREVIOUS MYOCARDIAL INFARCTION Kjell I. Pettersen & Knut Stavem, Foundation for Health Services Research, Nordbyhagen; Asmund Reikvam, Department of Pharmacotherapeutics, University of Oslo, Oslo; Arnfinn Rollag, Akershus University Hospital, Nordbyhagen, Norway The KCCQ is a 23-item disease-specific HRQL instrument for use in heart failure. We report on the reliability and validity of a Norwegian version in patients with previous myocardial infarction (MI). KCCQ items are scored on 5–7 point Likert scales and grouped into six domains and two summary scores. Domain scores are reported on a 0– 100 (highest level of functioning) scale. Internal consistency reliability was assessed using Cronbach’s a, test–retest reliability over 4 weeks with an intraclass correlation coefficient (ICC), and construct validity with a multitrait-multimethod approach, investigating association of KCCQ domains with scales of the SF-36 questionnaire. Additionally, we compared patient groups selected according to left ventricular ejection fraction (LVEF), normal (>50%), intermediate and reduced (<40%) LVEF in 252 patients. We mailed the KCCQ to 548 survivors of MI, 2–3 years after their MI and 408 (74%) responded. Mean domain scores ranged from 51.8 (Symptom stability) to 81.5 (Symptoms). Scores after adjusting for age, gender and education with ANCOVA are shown in patients according to LVEF (see table). Cronbach’s a was >0.80 for all domains except Self-efficacy (0.66). In the test–retest, the ICC for the domains/summaries ranged 0.44–0.82 with five out of eight domains/summaries having ICC > 0.70. The correlations between KCCQ domains and SF-36 scales largely supported a priori hypothesized associations. All domains except Symptom stability and Self-efficacy discriminated well between patients according to LVEF (see table). In conclusion, the KCCQ showed acceptable reliability and validity in patients with previous MI.

800 # 1777/THE INTERNATIONAL HEALTH-RELATED QUALITY OF LIFE OUTCOMES DATABASE (IQOD). REFERENCE VALUES OF THE PGWBI ON 8532 SUBJECTS Olivier Chassany, Delegation a la Recherche Clinique – AP-HP, hospital Saint-Louis, Paris, France; Dominque Dubois, Janssen Pharmaceutica, Beerse, Belgique; Elof Dimenas, Astrazeneca, Goteborg, Sweden; Martine Mouly & Letitia Lobo-Luppi, Mapi Research Institute, Lyon, France; Albert Wu, Johns Hopkins University, Baltimore, MD, USA For the IQOD Group, IQOD aims at improving the lack of cross-cultural reference values. It is based on the analysis of multinational item responses to three questionnaires and their translations: WHQ, MLwF, and Psychological General Well-Being Index (PGWBI). The PGWBI is a 22-item scale to measure affective or emotional states reflecting a sense of subjective well-being or distress, developed in English US, and translated into 36 languages. The IQOD-PGWBI database includes 8532 eligible patients from 12 studies (19 countries, 16 languages). Results showed substantial support for the construct validity of the six PGWBI dimensions and for cross-cultural equivalence (results reported elsewhere). We report some PGWBI values according to clinical and socio-demographic data. The global PGWBI score was 78 ± 21 [0, 110 (best HRQoL)] on the whole sample. Score was lower among women than men: 75 ± 22 vs. 83 ± 17, p < 0.0001. Scores differed with the marital status: divorced or widowed subjects had a lower score than if married (p < 0.0001). There was no correlation between global score and age (r ¼ 0.018): highest at 83 ± 18 (<20 years), dropped to 77 ± 21 until the age of 59 years, raised to 80 ± 20 (60–79 years) possibly related to a switch phenomenon, and decreased to 77 ± 21 (80–99 years). The global score differed among diseases: digestive problems such as GERD (82 ± 17), colorectal cancer (81 ± 18), menopause symptoms (64 ± 19), and chronic constipation (55 ± 34). Differences existed also through countries, e.g. within a large study of digestive problems, global score was highest in Switzerland (86 ± 16) and lowest in Italy (79 ± 17) and Netherlands (79 ± 18) [ANOVA, p < 0.0001]. The PGBWI database may help the interpretation of scores of this commonly used questionnaire. Acknowledgment: European Commission support under contract no. QLRI-CT-2000-00551. # 1887/HEALTH RELATED QUALITY OF LIFE QUESTIONNAIRES IN SPAIN Xavier Badia & Sandra Garcia, Health Outcomes Research Europe Group, Barcelona, Spain A review was done in order to know those instruments to measure HRQL that had been elaborated in Spain or adapted to Spanish language and it’s psycometric development level. A bibliographic research was carried out of all those studies related to the development, validation or use of HRQL in Spain published until March 2002. Medline, Indice Me´dico Espan˜ol, conferences abstracts, ascending bibliography and the catalogue of PhD disertations in Barcelona University were exhaustively searched. The key words used was quality of life or health care coincidents with questionnaires more health-related quality of life, health state, validity, reliability and checklist coincidents with spain. A total of 89 HRQL were found, 22.47% had been developed in Spain, and 77.53% had been adapted and/or validated to Spanish. Of all the instruments found in the revision 52.8% had good

psycometric properties, being the generic instruments the ones with more developed psycometric properties in relation to the specific ones and the rest of them. The sensitivity to change happened to be the less studied property, only was studied in less of half of the questionnaires. Psycometric development (criteria: translation, validation/weights adaptation, construct validity, convergent/discriminant validity, validation/ cut-off point adaptation, sensitivity in different populations, internal reliability, test–retest reliability, reliability between observers, sensitivity to change) level of questionnaires: High (5 criteria) Moderate (1– 5 criteria) Low (£ 2 criteria) Generic 58.3, 33.3, 8.3%, Specific 52.7, 34.5, 12.7%. Psychological wellbeing, Functional incapability, Social health and Pain questionnaires 50, 36.4, 13.6%. In Spain there is a significative number of HRQL questionnaires with good psycometric properties, that can be used either in research or in clinical practice. They allow us to know the patient’s perceived health status and use it as a measure of treatment effectiveness. # 1696/CHILDREN WITH SPECIAL HEALTH CARE NEEDS (CSHCN) AND QUALITY OF LIFE - VALIDATING THE CSHCNSCREENER IN 7 EUROPEAN COUNTRIES Angela Gosch, Research Unit Child Health and Quality of Life, Robert Koch Institute, Berlin, Germany; Luis Rajmil, Barcelona, Spain, Ulrike Ravens-Sieberer, Research Unit Child Health and Quality of Life, Robert Koch Institute, Berlin, Germany; Silke Schmidt, Center for Psychosocial Medicine, University of Hamburg, Hamburg, Germany; Ute Thyen, Klinik fu¨r Kinder- und Jugendmedizin, Universita¨tsklinikum SH, Campus Lu¨beck, Lu¨beck, Germany, and the European KIDSCREEN Group The Children with Special Health Care Needs Screener (CSHCN, Bethell et al., 2002) for parents to identify children with chronic conditions is widely used in US-American national surveys (Van Dyck et al., 2002). The objective of the presented European cross-sectional study is to validate the instrument in Europe and to explore the relationship between the health status of children and their health related quality of life (HrQoL). In seven European countries (A, CH, D, E, F, NL, UK) 2375 parents filled in the translated and harmonised CSHCNScreener, a 5-item screening tool using current health consequences or service use criteria to identify children with special health care needs (SHCN). Furthermore, 2375 children between 8 and 18 years of age (52% females) filled in the KIDSCREEN questionnaire in schools, a generic HrQoL instrument with 10 dimensions. Results indicate that 11.4% of the students in this European population show a chronic condition, while the data from the American national survey are slightly higher (15.3%). In addition the percentages of children with SHCN varied in the seven European countries. Children with SHCN showed a significantly lower HrQoL in comparison to healthy children in seven KIDSCREEN domains: physical, psychological and bullying (with effect sizes between 0.4 and 0.54), as well as with the quality of the parental relationship, self-perception and school functioning (effect sizes between 0.35 and 0.37). No significant differences were found with regard to the financial well-being, autonomy and social support scales. In conclusion, a large number of children with special health care needs were identified based on parental assessment and this data is comparable to the US. Differences between the seven European countries will be discussed in more detail. Overall children with SHCN reported a lower HrQoL except on the financial well-being, autonomy and social support scales.

801 # 1095/CROSS-CULTURAL ASSESSMENT OF THE MLHF QUESTIONNAIRE Olatz Garin & Montse Ferrer, Health Services Research Unit, Institut Municipal d’Investigacio´ Me`dica, Barcelona, Catalonia, Spain; Ingela Wiklund, Behavioural Medicine, Astra Ha¨ssle Research Laboratoris, Mo¨elndal, Sweden; Eric Van Ganse, Guilhem Pietri, Unite´ de Pharmacoe´pide´miologie, Centre Hospitalier Lyon-Sud (CHLS), Lyon, Pierre Be´nite Cedex, France; Martine Mouly, Research and Education Department, MAPI Research Institute, Lyon, France; Jordi Alonso, Health Services Research Unit, Municipal Institute for Medical Research, Barcelona, Spain The aim of this study (part of the IQOD Programme) is to assess the cross cultural equivalence of the Minnesota Living with Heart Failure Questionnaire (MLHFQ) across 13 countries. With this propose, Health Related Quality of Life, clinical and sociodemographical data from eight studies were pooled (n ¼ 2780 patients with heart failure). The MLHFQ contains two domains (physical and emotional) with eight and five items. It was originally developed in USA, and has been widely translated and used in many countries. Data came from: Australia, Brazil, Canada, Denmark, France, Germany, Great Britain, Hungary, Israel, Italy, Poland, Sweden and USA. In two of the countries (Great Britain and Denmark) full methodology translation was used to produce the new version. The metric properties were compared across countries. Twelve percent of patients presented at least one missing item, without differences by country. Item-scale correlations were above 0.4 for all countries. In both domains, 100% of items were higher correlated with its scale than with other, for all countries except Denmark, Hungary and Italy’s physical domain (87, 87 and 75%) and Canada, Hungary, Israel, and USA ‘s emotional domain (80, 60, 80 and 60%). Floor effects were very low (0–4.3%) and ceiling effect was higher than 15% only for emotional domain in five countries. Cronbach’s a internal consistency was: above the standard proposed for individual comparisons (0.9) for all countries total score; between 0.87 and 0.94 for physical score and a bit lower for the emotional score (0.77–0.92). Correlations with the New York Heart Association classes were studied in those countries that had more than 10 patients per class (Germany, France and Great Britain). The coefficients indicated moderated correlations; higher for physical (0.47–0.6) and total (0.4– 0.57) scores than for emotional (0.21–0.44). In summary, the MLHFQ shows satisfactory metric properties similar across the studied countries. These suggest that the questionnaire would be useful in international multicentric studies. Supported by the EU Commission (QLRICT-2000-00551 (IQOD)) and the CIRIT (2001SGR 00405). # 1725/INTERNATIONAL PSYCHOMETRIC VALIDATION OF A PATIENT SELF-ADMINSTERED QUESTIONNAIRE TO ASSESS QUALITY OF LIFE IN UPPER GASTROINTESTINAL DISORDERS: THE PAGI-QOL Christine De La Loge, Lyon, France; Dominique Dubois, Pharmaceuticals Group Strategic Marketing Health Eco, Janssen Pharmaceutica, Beerse, 2340, Belgium, Patrick Marquis, Boston, MA, USA The Patient Assessment of upper gastrointestinal disorders-Quality of Life (PAGI-QOL) was developed for international use. The psychometric validation has been performed in patients with dyspepsia, GERD or gastroparesis for US, Dutch, French, German, Italian, and Polish versions. The reliability and validity results are presented here. Participants completed the pilot PAGI-QOL at baseline and 8 weeks, with a subsample also completing it at 2 weeks. Other assessments

administered included the Patient Assessment of upper gastrointestinal disorders-Symptom Severity (PAGI-SYM), SF-36, and number of disability days. In total 1736 patients completed the PAGI-QOL at baseline. The PAGI-QOL was reduced and finalized as a 30-item measure covering five domains: Daily Activities, Clothing, Diet and food habits, Relationship and Psychological well-being and distress. Cronbach’s a ranged from 0.83 to 0.96 demonstrating excellent internal consistency reliability. Test–Retest reproducibility was good: Intraclass Correlations Coecients were over 0.70 except for the Relationship scale (0.61) in stable patients. Moderate (0.52) to strong (0.72) correlations between the PAGI-QOL total score and all SF-36 subscale scores provided a good level of concurrent validity. PAGIQOL scores showed excellent discriminant properties, with patients who had spent some days in bed, patients who had missed some days at work and patients who were kept from usual activities having much lower PAGI-QOL scores than patients who did not (p < 0.0001). The psychometric validation of the PAGI-QOL in all languages and all conditions allows confidence in using this instrument in international clinical settings. The PAGI-QOL is a valid and reliable instrument assessing QOL in patients with dyspepsia, GERD or gastroparesis. # 1178/EFFECTS OF PATIENT EMPOWERMENT ON HEALTH OUTCOME AND PATIENT SATISFACTION AMOUNG SUGERY PATIENTS Ursula F. Trummer & Peter Nowak, Vienna, Austria; Juergen M. Pelikan, Ludwig Boltzmann-Institute for the Sociology of Heal, Vienna, Austria The influence of so called context effects quality of interaction between physician and patient, patient and provider expectations have been acknowledged as having an important therapeutic effect. Systematic reviews of RCTs underline this theory (Di Blasi, Harkness et al. 2001) but also point out that there is little evidence so far and further studies are needed. Recent discussions use the concept of empowerment of patients (Levin-Zamir, Peterburg 2001; Nutbeam 2000; Wallerstein, Bernstein 1988) and emphasise communication between patients and professionals as a main tool and quality factor. In 2001 three Austrian Hospitals worked in a model project to survey the effectiveness of interventions to improve the quality of communication between hospital staff and patients (improvement of doctor–patient communication and information, involvement of patients in decision making) on patient satisfaction and health outcome after surgery in the phase of inpatientcare. Effects of interventions were evaluated in a before-and-after study (baseline n ¼ 257 patients: 113 general surgery, 100 cardiac surgery, 44 thorax surgery; evaluation n ¼ 208 patients: 63 general surgery, 99 cardiac surgery, 46 thorax surgery) with data from a patient questionnaire (satisfaction with care and communication, self reported health status) and documentation of post-surgical health outcome (complications, length of stay in hospital). Results show that improvement of communication has effects on patient satisfaction as well as on postsurgical health outcome especially in cardiac surgery, where communication with patients was improved by interventions covering the whole inpatient care process from admission to discharge (increase of patient satisfaction by 15%, reduction of postoperative cardiac disrhythmia by 15%, reduction of postoperative length of stay by 1 day). The paper will present data and discuss measures to improve communication between surgery patients and hospital staff, their effects and necessary preconditions for successful implementation.

802 # 1534/MEASURING SATISFACTION WITH HORMONAL CONTRACEPTIVES: A CONCEPTUAL FRAMEWORK FOR A GENERIC MEASURE OF SATISFACTION Susan D. Mathias, Hilary H. Colwell & Tricia Cimms, Ovation Research Group, Highland Park, IL; Debra L. Karvois, John M. LoCoco, Raritan, NJ; Andrew J. Friedman, Ortho-McNeil Pharmaceutical, Inc., Raritan, NJ, USA Introduction: There has been recent interest in the development of measures assessing treatment satisfaction. Typically these measures are developed with the goal of differentiating one product vs. another. To our knowledge, a standard, validated questionnaire to evaluate treatment satisfaction with hormonal contraception does not currently exist. In fact, many of the questionnaires that are currently being used contain only single items (e.g., How satisfied are you with your method of birth control?). Objective: We developed a comprehensive generic measure of satisfaction with hormonal contraceptives which is suitable for use in studies of women using any one of a variety of hormonal contraceptives, including birth control pills, injections, vaginal rings, and the ORTHO EVRA patch. Methods: Items were generated based on information from available literature, results from six focus groups supplemented with in-depth telephone interviews conducted across the US of women using hormonal contraception (n ¼ 55), and in-depth interviews with gynecologists (n ¼ 6) who regularly prescribe hormonal contraceptives. The questionnaire was revised based on findings from in-depth interviews of a subsequent group of women using hormonal contraception (n ¼ 24). The final questionnaire contains 52 items. Results: Our current hypotheses suggest that the questionnaire contains several domains, each containing multiple items and/or scales. The domains include eectiveness, ease of use, convenience, symptom relief, lifestyle issues, compliance, and overall satisfaction. The factor structure of the questionnaire will be finalized based on data from a validation study of several hundred women currently underway. Conclusions: Patient satisfaction is a subtle but essential component of a patient’s willingness to take a medication. The development of a generic questionnaire to measure treatment satisfaction with hormonal contraception is novel and such a questionnaire would be a useful tool to help dierentiate amongst the myriad of contraceptive options. # 1360/ADJUVANT CHEMOTHERAPY FOR BREAST AND COLORECTAL CANCER: WHAT DO PATIENTS WANT, AND WHY? Sylvia J. Jansen, Wilma Otten, Monique C. Baas-Thijssen & Anne M. Stiggelbout, Medical Decision Making, Leiden University Medical Center, Leiden, The Netherlands Background: Adjuvant chemotherapy may have a considerable impact on quality of life. Previous studies have assessed the amount of benefit that patients with cancer want to offset the side effects of treatment. However, the determinants of these preferences are yet unknown. Therefore, our goal is to explore the determinants of preferences for adjuvant chemotherapy qualitatively. Methods and patients: Semi-structured, in-depth, interviews were conducted with patients with colorectal (n ¼ 27) and breast cancer (n ¼ 25). Patients for whom adjuvant chemotherapy was planned (n ¼ 16) were interviewed before and after chemotherapy. Other patients were interviewed at similar points in time. Patients with progressive disease (n ¼ 10) were interviewed only once. Results: Qualitative analyses, using the program N*Vivo, revealed that the determinants could be classified under six major themes: (1) chance of cure or prolongation of life, (2) specialist recommendation, (3) clinical characteristics, (4) cognitive and emotional factors, (5) socio-demographic characteristics, (6) side effects of treatment. In addition, some remarkable findings were: the utmost confidence in the specialist (e.g., ‘the doctor knows what’s best for me’) and the large impact of emotional factors (e.g., ‘willingness to do everything out of fear for recurrence’). Another important point was the incomprehension with regard to adjuvant treatment e.g., patients advised to undergo adjuvant chemotherapy believed that the disease would definitely recur without chemotherapy. Finally, many patients stated that the impact of chemotherapy on quality of life was temporary and of less importance. Conclusions: These findings show that patients’ preferences may be the result of emotional processes and the tendency to merely follow the specialist’s recommendation. Furthermore, a good understanding of the meaning of ‘adjuvant’ is lacking. This knowledge can help doctors to commu-

nicate benefits and side effects of adjuvant chemotherapy in such a way that patients are able to make decisions that truly represent their well-considered point of view. # 1520/STUDY ON THE LINK BETWEEN SATISFACTION AND QUALITY OF CARE FOR INPATIENTS L. Boyer, S. Antoniotti, C. Sapin, Perceived Health Research Unit, Public Health Department; P.A. Thomas, Thoracic Surgery Service; D. Raccah, Endocrinology Service; P. Auquier, Perceived Health Research Unit, Public Health Department School of Medecine, Marseille, France Since 1996, all the French hospitals have had to evaluate the patients’ satisfaction. We have developed a French self-administered instrument based on the patient’s perceptions (QSH in 1998). But some questions still remain such as the link between satisfaction and quality of care. To look for the relations between the patient’s satisfaction (evaluated by the QSH) and the quality of care, according to two diseases (diabetes, lung cancer). Quality of care is measured by some objective indicators, in accordance with the recognised guidelines. The scientific committee has chosen in cooperation with the medical staff the items for the evaluation of quality of care. Our instrument is divided in two parts: a common part to both pathologies based on criteria resulting from the accreditation handbook; and a specific part for each pathology according to the literature review. The patients’ satisfaction is evaluated by the QSH. The quality of care is assessed retrospectively from the patients’ files. We offer a specific score, an accreditation score, some scores according to dimensions and a global score. No simple relation between satisfaction and quality of care is highlighted in this pilot study including 119 subjects (average comparison, correlation coefficient and multivariate logistic regression). The least satisfied patients are not those who have the worst quality of care. Satisfaction related to the hospital staff is negatively correlated with the quality of care ‘accreditation’ (r ¼ )0.23; p < 0.05). This result is also found with the logistic regression (OR ¼ 0.57; p < 0.05). A link has been highlighted between the ‘specific’ quality of care and the satisfaction with the nurse (OR ¼ 1.16; p < 0.05) but not with the doctors (p > 0.05). In spite of the current tendency to assimilate satisfaction and quality of care, it does not seem relevant to evaluate a quality whose only reference is the patients’ opinion. # 1175/TREATMENT SATISFACTION: A REVIEW OF RANDOMIZED CONTROLLED DRUG TRIALS Alessio Degl Innocenti, Outcomes Research, AstraZeneca R&D, Mo¨lndal; Linda B. Hassing, Department of Psychology, Gothenburg University, Go¨teborg; Anders Ingelgard, Ka´roly R. Kulich & Ingela Wiklund, Outcomes Research, AstraZeneca, Mo¨lndal, Sweden Patient satisfaction with medical care can be crucial for compliance and must be considered as an important factor in treatment decisions. However, there is no consensus on the conceptual framework of neither patient satisfaction nor treatment satisfaction, which is probably related to the multidimensional nature of these constructs. The objective of this study was to critically review the clinical trials literature using treatment/patient satisfaction as an outcome, with special focus on the instrument used. We explored four aspects: (1) type of instruments used, i.e. single-item or questionnaire, (2) generic or disease specific instruments, (3) assessment by the patient or clinician, and (4) psychometric documentation. In the Embase (1980-) and Medline (1966-) databases, respectively, we identified 133 studies. Search terms included: consumer satisfaction, patient satisfaction, treatment satisfaction or drug preference together with controlled study, controlled clinical trials, or randomized clinical trials, or drug therapy. Fiftytwo percent of the studies assessed patient satisfaction by single-item question whereas 41% used a questionnaire; and 7% did not report type of instruments used. Of those using a questionnaire 89% were disease specific. Further, 93% of the studies used patient-reported assessment. Only 30% of the questionnaires were psychometrically validated. The frequent use of single-item questions indicates that assessments of treatment satisfaction in clinical trials are inadequate. The lack of instruments assessing the multidimensional nature of treatment satisfaction shows that the importance of a precise assessment is often overlooked. In conclusion the assessment of treatment satisfaction needs to be improved in terms of construct consensus and psychometric validation.

803 # 1373/VALIDATION OF A SURVEY TO EVALUATE PATIENT SATISFACTION WITH HOSPITALISATION Nerea Gonzalez, Jose Maria Quintana, Felipe Aizpuru, Antonio Escobar, Cristobal Esteban, Emilio de la Sierra & Jose A. San Sebastian, Unidad de Investigacio´n, Hospital de Galdakao, Galdakao, Bizkaia, Spain There is an increasing interest in the area of Health Service Research and Quality of Care Improvement in developing Patient satisfaction questionnaires (PSQ) as a way of collecting information about the effect of an intervention, from the patient point of view. Patient satisfaction questionnaires are not usually validated. The goals of our study were to develop a questionnaire to evaluate patient satisfaction with hospitalisation and to assess its construct validity and reliability. The questionnaire was developed based in the information obtained from the bibliography research and focus groups. The survey was pilot tested before the field study. In this validation study, 2600 patients randomly selected from four different hospitals were contacted to fill out the questionnaire. A follow-up was done to increase the response rate. The exploratory factor analysis technique, using principal component analysis with varimax rotation, was employed to evaluate the structure of the questionnaire. The internal consistency of the scales was measured using Cronbach’s a coefficient. Scores for each of the domains that arised from the Factor Analysis were developed. Those scores ranged from 0 to 100, the later meaning greater satisfaction. Finally, 1910 (73.5%) patients answered the questionnaire. Report items, those with dichotomous response scales and items with more than 10% of no response were excluded from the analysis. Those items with a saturation rate less than 0.35 were also excluded. Finally, 34 items were included in the final solution. Six factors entitled ‘Information’, ‘Human touch’, ‘Comfort’, ‘Visits’, ‘Intimacy’ and ‘Cleanliness’ emerged, with a cumulative variance of 49.2%. Cronbach’s a coefficients ranged from 0.60 to 0.85. Domains mean scores ranged from 66.7 for Confort, to 90.6 for Intimacy. These data support the construct validity and the reliability of our questionnaire. More analysis would be needed to definitely validate this instrument, in terms of its discriminant and predictive validity. # 1387/A NEW MEASURE OF PATIENT SATISFACTION WITH OCULAR HYPOTENSIVE MEDICATIONS: TREATMENT SATISFACTION SURVEY - INTRAOCULAR PRESSURE (TSS-IOP) Mark J. Atkinson, Outcomes Research, Pfizer, Kalamazoo, MI; William C. Stewart, Pharmaceutical Research Network, University of South Carolina School of Medicine, Charleston, SC; Ravinder Dhawan, Pfizer Global Pharmaceuticals, Pfizer, Peapack, NJ; Jeanette A. Stewart, Clinical Project Management, Pharmaceutical Research Network, Charleston, SC; Joel M. Fain, Pfizer Global Pharmaceuticals, Essy Mozaffari, Pfizer Global Pharmaceutical, Pfizer, New York City, NY, USA Purpose: To validate the treatment-specific TSS-IOP. Methods: Item content was developed by four heterogeneous patient focus groups (n ¼ 32). Instrument validation involved 250 patients on ocular hypotensive medications recruited from ophthalmology practices in the Southern USA. Participants responded to demographic and test questions during a clinic visit. Standard psychometric analyses were performed on the resulting data. Sample: Of the 412 patients screened, 253 consented and 250 provided complete datasets. The sample included 44% male (n ¼ 109), 44% Black (n ¼ 109) and 57% brown eyed (n ¼ 142) participants, with a mean age of 64.6 years (SD 13.1) and a history of elevated IOP for an average of 8.4 years (SD 7.8). A majority were receiving monotherapy (60%, n ¼ 151). Results: A PC Factor analysis (w/varimax rotation) yielded five factors (Eigenvalues >1.0) explaining 70% of the total variance. Weak items were removed and the remaining 15 items reanalysed. The satisfaction factors were; Eye Irritation (EI; 4 items), Convenience of Use (CofU; 3 items), Ease of Use (EofU; 3 items), Hyperemia (HYP; 3 items), and Medication Effectiveness (EFF; 2 items). Cronbach a’s ranged from 0.80 to 0.86. Greater distributional skew was found for less common experiences (i.e., HYP & EI with 65% and 48.4% ceilings) than for more common experiences (i.e., EofU, CofU, EFF with 10.8, 20.8 and 15.9% ceilings). TSS-IOP scales converged with previously validated scales on the TSQM (r ¼ 0.36 to 0.77). Evidence of clinical criterion-related validity was found. Patients’ symptomatic rat-

ings of eye irritation, hyperemia and difficulties using the medication correlated with satisfaction on these dimensions (r ¼ 0.30–0.56, all p < 0.001). Clinicians’ ratings of IOP control, severity of side effects and problematic medication use correlated with patients’ satisfaction scores on these dimensions (r ¼ 0.13–0.26, all p < 0.01). Conclusions: This study provides initial evidence that the TSS-IOP is a reliable and valid measure, assessing patients’ satisfaction with ocular hypotensive medications. # 1749/PERCEPTION OF EQUITY AND SOCIAL SUPPORT – DIFFERENCE BETWEEN HEALTHY POPULATION AND PEOPLE WITH ASTHMA BRONCHIALE Hana Janeckova & Helena Hnilicova, Institute of Medical Humanities, 1st Medical School, Prague, Czech Republic, Czech Republic The research activities of the Institute of Medical Humanities of the 1st Medical School, Charles University, Prague in the field of the quality of life are oriented to various health, psychological, social and even philosophical aspects of the quality of life. In the frame of this larger project a research group realised and inquiry on a representative sample of more than 1000 Prague inhabitants and also – among others – on the sample of 200 people with asthma bronchiale. Both quantitative and qualitative items were included in the questionnaire to describe relations between subjective health status and various psycho-social indicators of the quality of life, including individual’s values, attitudes, life events and standard of living. In this paper we will analyse quantitative and qualitative data about subjectively perceived health status of ‘healthy’ people and people with asthma bronchiale in relation to some indicators of social cohesion like social support, perception of social equity and injustice (we used questions like ‘Have you experienced any injustice in recent past? Give some examples’.) We shall also assess the impact of micro- and macro-processes in the Czech society on the feeling of injustice, social support and subjectively perceived health. Comparison of ‘healthy’ people and people with ‘chronic illness’ will be done. Differences among age groups as well as differences between ‘healthy’ and ‘ill’ are expected. # 1207/PSYCHOMETRIC PROPERTIES OF A NEW INDIVIDUALISED MEASURE OF QUALITY OF LIFE (QOL) FOR TEENAGERS WITH DIABETES MELLITUS: THE ADDQOL-TEEN Carolyn McMillan, Psychology Department, Royal Holloway, University of London, Egham, Surrey; Jessica Datta & Nicola Madge, Research Department, National Children’s Bureau, London; Clare Bradley, Psychology Department, Royal Holloway, University of London, Egham, Surrey, UK This study evaluated the psychometric properties of a new individualised, patient-centred questionnaire, the ADDQoL-Teen, measuring perceived impact of diabetes mellitus on QoL of teenagers. Respondents rate personally applicable life domains for impact of their diabetes and importance to their QoL. ADDQoL-Teen design was based on the ADDQoL (for adults with diabetes). Young people [mean age 16.4 (sd 2.4) years, n ¼ 152], attending diabetes clinics at six UK hospitals, completed the questionnaire that included 30 specific domain items and two overview items. Five domains were found to detract from the reliability and factor structure of the measure and were analysed separately. The 25-domain ADDQoL-Teen had high internal consistency reliability [Cronbach’s a coefficient ¼ 0.91, (N ¼ 133)] and could be summed into an overall Average Weighted Impact score. There were 2 subscales: a 10-item Impact-Self subscale (measuring impact of diabetes and its treatment on the individual) and a 15-item ImpactOther subscale (measuring impact on interactions with others and the external world). Both subscales had good internal consistency reliability, [Cronbach’s a coefficients of 0.82 (n ¼ 142) and 0.88 (n ¼ 138) respectively]. Domains reported as most severely (and negatively) impacted by diabetes were [mean weighted impact (sd)]: lie in bed [)3.68 (3.41)], interrupting activities [)3.5 (3.23)] and worry about the future [)3.45 (3.28)], (maximum range )9 to +9). The majority (62%) felt that having diabetes had a negative impact on QoL, but 36% considered it had no impact on QoL. In conclusion, the ADDQoL-Teen is a new measure of perceived impact of diabetes and its treatment on QoL of teenagers. It will help health professionals consider QoL as well as medical outcomes when caring for young people with diabetes. It may be used in clinical trials and for routine clinical monitoring.

804 # 1206/STRUCTURAL RELATIONSHIPS BETWEEN BELIEFS, SELF-EFFICACY, ADHERENCE TO SELF-CARE REGIMENS, BODY MASS INDEX AND HRQL IN TYPE 2 DIABETES Sheri L. Maddigan & Jeffrey A. Johnson, ACHORD, Institute of Health Economics, Edmonton, Alberta, Canada

# 1173/CONCEPTUAL ISSUES AND ASSESSMENT OF PRODUCTIVITY IN DIABETES Meryl Brod, Global Health, The BROD GROUP, Mill Valley, CA, USA; Soren E. Skovlund, Global Health, Novo Nordisk, Bagsvaerd, Denmark

The purpose of this analysis was to determine the structural relationships between diabetes management beliefs (MB), self-efficacy (SE), diet and exercise at baseline and body mass index (BMI) and healthrelated quality of life (HRQL) at a 6 month follow-up in type 2 diabetes. The sample included 250 individuals who had been recruited into a clinical trial to improve quality of care in type 2 diabetes. A theoretical model of the relationships between MB, SE, diet, exercise, BMI and HRQL was proposed. Determinants of health, clinical parameters and the intervention itself were also controlled for in the model. HRQL was assessed using the RAND-12 and Health Utilities Index Mark 3, while adherence to diet and exercise recommenations was assessed using the Summary of Diabetes Self-Care Activities. The Diabetes Lifestyle Form was used to evaluate SE and MB. Goodness of fit of the proposed model was assessed using LISREL 8.52. The theoretical model accounted for 30.0% of the variance in subsequent HRQL. Statistically significant direct relationships were observed between subsequent HRQL and SE (b ¼ 0.32, t ¼ 2.43), BMI (b ¼ )0.23, t ¼ )3.21), insulin use (b ¼ )0.14, t ¼ )2.05) and income (b ¼ 0.20, t ¼ 2.59). MB and SE were both directly associated with adherence to dietary self-care, while adherence to exercise recommendations was associated with MB, but not SE. Exercise adherence (b ¼ )0.16, t ¼ )1.96), MB (b ¼ )0.52, t ¼ )3.53), SE (b ¼ )0.34, t ¼ )2.43, and had significant direct relationships with BMI. Hypothesized indirect relationships between exogenous variables and HRQL were not supported in the analysis. Overall model fit was acceptable (RMSEA ¼ 0.035, GFI ¼ 0.92, AGFI ¼ 0.87). This theoretical model highlighted the relationships between SE and BMI to subsequent HRQL in type 2 diabetes. A positive association of SE, MB, exercise and healthier BMI was observed. Thus, improving self-efficacy, enforcing positive beliefs and promoting adherence to self-care regimens may make a positive contribution to overall HRQL.

Productivity, the ability to be effective or produce a desired effect, is a critical component of quality of life. Unfortunately, our understanding and assessment of productivity is generally conceptualized in relation to job performance. However, in chronic conditions like diabetes, productivity issues exist in both the work place as well as everyday life. Additionally, not all diabetics are employed and so productivity issues for this population are overlooked. A broader, more clinically relevant definition of everyday life productivity is needed which is applicable to non-working persons as well as the employed and cover productivity issues both in everyday life as well as the work environment. This presentation will (1) present the findings from diabetes patient focus groups (US and UK) regarding issues of productivity using this broader definition of productivity (2) describe a new disease specific measure of productivity for diabetes applicable to working and nonworking diabetics and (3) present data from a survey of 136 Type I diabetics, examining key productivity issues. For example, it was found that the most frequent impairments in life productivity for diabetics are difficulties in accomplishing as much as they would like due to feeling of being tired or lacking energy and trouble getting up and being active in the morning. Additionally, diabetics who do not work have greater life productivity impairments than those who do work (p ¼ 0.08) and even for those who work, life productivity impairments are greater than work productivity issues (p ¼ 0.0001). There is a trend that women have more life impairments than men (p ¼ 0.08) although there is no difference between men and women in degree of productivity impairment at work. Both life and work productivity impairments are strongly related to each other (r ¼ 0.78) and related to number of diabetes symptoms (p ¼ 0.001). Age is not a significant factor for either life or work productivity issues. The clinical relevance and research implications for defining and assessing productivity issues in diabetes will also be discussed.

# 1644/VALIDATION OF THE CHINESE VERSION DIABETES IMPACT MEASUREMENT SCALES AMONG PEOPLE WITH DIABETES Tsai-Chung Li, Institute of Chinese Medical Science, China Medical College, Taichung, Taiwan, Republic of China; Cheng-Chieh Lin, Department of Family Medicine; Chia-Ing Li, Department of Medical Research, China Medical College Hospital, Taichung, Taiwan, Republic of China Objectives: Diabetes Impact Measurement Scales (DIMS) is a welldocumented instrument for health-related quality of life (HRQoL) for people diabetes. The aim of this study was to provide results of the reliability and validity of the Chinese version DIMS among people with diabetes. To test the construct validity of DIMS, we hypothesized that diabetic patients with poor glucose control or complications had worse HRQoL. Methods: DIMS was translated following the guidelines proposed by The International Quality of Life Assessment Project for cross-cultural adaptation by translating English-version ones. A crosssectional study consisting of 266 diabetes patients using face-to-face interview was conducted. The DIMS domains include symptoms, diabetes-related morale (attitude towards managing the disease), social role fulfillment and well-being. Diabetic control is measured by hemoglobin A1C (glycosylated hemoglobin) with a cutoff point of 7.0. Complications consisted of stroke, heart disease, visual impairment, amputations, kidney disease, cognitive impairment, and incontinence. Results. All scales of DIMS pass tests for item-internal consistency and item-discriminant validity. Estimates of internal consistency coefficients ranged from a low of 0.51 to a high of 0.84 across scales. Validation by confirmatory factor analysis yielded results remarkably similar to those proposed by the authors. After adjusting for the effect of age, gender, and complication, those who had good diabetic control had a better status of well-being and overall score than those who did not. After adjusting for the effect of age, gender, and diabetic control, those who had any complications had a worse status of social role fulfillment, well-being and overall score than those who had not. These results suggest that Chinese DIMS is a reliable and valid instrument and appropriate in clinical settings for Chinese with diabetes.

# 1191/QUALITY OF LIFE AND PHYSICAL ACTIVITY FOR TYPE 2 DIABETIC PATIENTS: PSYCHOMETRIC VALIDATION OF AN EXISTING DISEASE SPECIFIC SCALE Judith Chwalow, INSERM Department of Diabetology, Hotel Dieu Hospital, Paris; Mounir Mesbah, Department of Statistics, University of Bretagne-Sud, Vannes, France The purpose of this study was to change physician and patient behavior relating to the management of type 2 diabetes by following an educational program that taught general practitioners (GPs)and visiting nurses to teach diabetic patients in small groups. This was a randomized clinical trial with 214 patients divided into two arms, control n ¼ 122, experimental n ¼ 92. Patients were recruited by their GPs and the experimental group participated in seven small group sessions. Data were collected using self administered patient questionnaires, including quality of life (QOL). QOL was measured at: T0, 1 and 2 years using the French versions of the already validated SF36 and the Diabetes Health Profile (DHP). A score was established for the level of physical activity. Three types of analyses were conducted: (1) descriptive, (2) univariate, and, (3) multivariate.The population was 52% male, patients were 59 ± 8 years at T0 and had diabetes for a mean of 9 years. There were no significant differences between the two groups at T0. At the end of the study, patients were well informed about the impact of physical activity on their level of glycemic control. Most patients (80%) indicated that their GPs had recommended that they exercise. The difficulty expressed related primarily to lack of time and motivation rather than lack of knowledge. When the physical activity score was divided into sedentary vs. any activity, there was a statistically significant impact on patient reported QOL. The QOL scale (DHP) was validated using a Rasch model (partial credit model) within the conceptual framework of the SAS procedure NLMIXED. Patients and GPs are well informed about the benefits of physical activity for NIDDM patients. The barriers to physical activity for most patients do not include a lack of knowledge. When patients engage in even slight activity, there are significant improvements in their QOL as measured by this now standardized, validated, disease specific QOL scale, in French. This information might be used as an incentive to otherwise sedentary diabetic patients.

805 # 1317/TEN-YEAR FOLLOW-UP OF WEIGHT LOSS AND QUALITY OF LIFE AFTER OBESITY TREATMENT. SWEDISH OBESE SUBJECTS (SOS) - AN INTERVENTION STUDY OF OBESITY Jan Karlsson, Health Care Research Unit, Lars Sjo¨stro¨m, Department of Body Composition and Metabolism; Marianne Sullivan, Health Care Research Unit, Sahlgrenska University Hospital, Go¨teborg University, Go¨teborg, Sweden Purpose: To examine the effects of obesity treatment on health-related quality of life (HRQL) in a prospective controlled clinical trial of surgical vs. conventional weight reduction treatment. Subjects and methods: The first 367 surgical cases (68% women) and 357 conventionally treated matched controls followed for 10 years were evaluated. Inclusion criteria were age at accrual (37–60 years) and a body mass index (BMI) of at least 34 for men and 38 for women. The SOS Quality of Life Questionnaire was used for HRQL assessment. HRQL was assessed prior to treatment and at 6 mo, 1, 2, 3, 4, 6, 8 and 10 years after intervention. Response rates at 10-year follow-up were 84% in the surgically treated group and 79% in the conventional treatment group. Main results: Average weight loss at 10-year follow-up was 15.8% in the surgical group, whereas the conventional cases had gained 1.4% compared to their baseline weight. Great variation in weight loss was observed in the surgical group: 37% had lost more than 20% of their baseline weight, 28% had weight losses between 10–19.9%, and 35% had weight losses below 10%. Poor HRQL was improved after obesity surgery, while stable HRQL ratings over time were observed in the conventionally treated group. Powerful shortterm HRQL changes (at 6 and 12 months) in the surgical group were followed by a longitudinal decrease on average. The long-term HRQL improvements were generally related to the magnitude of weight loss. More stable improvements were observed in patients with substantial and enduring long-term weight loss (30+%), while a regress over time was observed in patients with less weight reduction. Patients with minor weight loss (<10%) returned to their baseline levels. Conclusion: Poor HRQL in the severely obese is improved after weight reduction and the beneficial effects are related to the magnitude of weight loss. Substantial long-term weight reduction is required for durable HRQL improvements. # 1892/DISABILITY AND QUALITY OF LIFE IN PATIENTS WITH MIGRAINE: DETERMINING FACTORS Otman Fernandez Conception, Leonides Canuet Delis, Institute of Neurology and Neurosurgery, Havana, Cuba In the period between August, 2001 and July, 2002; 116 patients with a diagnosis of migraine and suffering from headache who consecutively entered the Emergency Unit of the Institute of Neurology and Neurosurgery underwent an observational non experimental study with the objectives of describing the behaviour of the disability and quality of life in patients with migraine, identifying the factors associated to these variables and determining the concordance between the measurements of disability and quality of life. The main variables of the study were: (1) Disability provoked by the migraine, (2) Quality of life affected by the headache, and (3) General assessment of the health status. The remaining variables of the study included: Demographic variables (age, sex, color of the skin). Clinical variables of the migrainous crises (with or without aura, pain location, pain nature, duration of the crises, presence of associated symptoms, presence of precipitating factors and their number, and the number of days with headache). For the evaluation of the disability we used the questionnaire known as (MIDAS) Migraine Disability Assessment , while for the quality of life we applied the questionnaire for the evaluation of needs provoked by the headache namely, the Headache Needs Assessment (HANA). The mean of the total score of MIDAS was of 40.8 and 58.6% of patients were in IV degree of disability. The values for the median number of days with headache in three months and the median intensity of headache were of 32.25 and 8.50 respectively. HANA’s median score was of 76.37. The factors associated to the disability were the number of days with headache and the co-morbidiy with arterial hypertension while, the quality of life was associated to the number of days with headache, pain intensity and female sex. The results obtained through both questionnaires were highly correlated as well as the simple subjective assessment of the health status.

# 1585/CHRONIC PAIN, DEPRESSION, AND QUALITY OF LIFE: CORRELATIONS WITH GENDER, AGE, AND NUMBER OF PAIN TYPES Thomas E. Elliott, Colleen M. Renier & Jeanette A. Palcher, Division of Education and Research, St. Mary’s/Duluth Clinic Health System, Duluth, MN, USA Previous research demonstrated very high prevalence of major depressive disorders [MDD] and very low health-related quality of life [HRQoL] as measured by the SF-36 in chronic pain patients [CPP]. The relationships among age, gender, number of types of pain, depression, and HRQoL in CPP has not been studied. The purposes of this study were to determine the correlation of number of pain types with SF-36 scores in a population of CPP, and the correlation of age and gender with MDD and SF-36 scores in a population of CPP. A convenience sample of 323 consecutive CPP completed the SF-36 and was assessed by a pain medicine physician at entry into a pain management program. The study sample was 64% female with mean age 45 (range 19–83). Using DSM-IV criteria, 54% had MDD, 33% minor depression, and 13% no depression. Many patients had multiple pain types: back pain 53%, fibromyalgia/myofascial pain 46%, neuropathic pain 29%, headache 26%, arthritis 15%, and other pains 8%. The number of pain types per patient in the study sample was 1 (18%), 2 (37%), and 3 (45%). Pearson’s r method measured the correlation between SF-36 cores and variables of interest. Continuous data were analysed using two-tailed t-tests and ANOVA. Categorical data were analysed using v2. The number of pain types per CPP was significantly correlated with lower SF-36 PF, VT, SF, RE, MH, and MCS scores. Younger CPP had higher prevalence of MDD, with age 18–44 prevalence of MDD 59% (p ¼ 0.01) and age was significantly positively correlated with SF-36 BP, GH, SF, RE, MH, and MCS scores. Females had a higher prevalence of MDD, 70% (p ¼ 0.02), and lower SF-36 VT score (r )0.115, p < 0.05) than males. The number of pain types in CPP is highly correlated with lower SF-36 scores. Age and gender have significant correlations with MDD and SF-36 scores. These results are consistent with clinical experience. HRQoL may become an important outcome measure in chronic pain management. # 1130/VISION-RELATED QUALITY OF LIFE IN PATIENTS WITH SUBFOVEAL CHOROIDAL NEOVASCULARIZATION DUE TO OCULAR HISTOPLASMOSIS OR IDIOPATHIC CAUSES Pa¨ivi H. Miskala & Barbara S. Hawkins, Department of Ophthalmology, The Johns Hopkins University, Baltimore, MD, USA Because individuals with subfoveal choroidal neovascularization (CNV) due to the ocular histoplasmosis or idiopathic causes are typically working age, they are at risk for decrements in quality of life due to visual disability. We present baseline vision-related quality of life findings of patients with CNV due to these conditions who enrolled in the Submacular Surgery Trials. The effect of bilateral CNV as compared to unilateral CNV on quality of life also was assessed. Patients had subfoveal CNV and best-corrected visual acuity of 20/20 to 20/320 inclusive in the better-seeing eye. The 39-item National Eye Institute Visual Function Questionnaire (NEI-VFQ) and the 36-item Medical Outcomes Study Short Form Health Survey (SF-36) were administered centrally by telephone by trained interviewers prior to randomization. Demographic and clinical information were collected as part of the baseline examination. Two-hundred and twenty-five patients were enrolled; 170 had unilateral CNV and 55 had bilateral CNV. Median age was 48 years (range: 18–79); 56% were women and 96% were non-Hispanic white. Median overall NEI-VFQ score was 75 points; the general vision (median: 55) and mental health (median: 65) subscales had the worst scores. The NEI-VFQ overall score and nine of 11 subscales differed between patients with unilateral CNV and bilateral CNV (p < 0.0001). Scores on the NEI-VFQ near activities and distance activities were on average 15 and 10 points worse, respectively, in bilateral cases than in unilateral cases after adjustment for visual acuity and reading speed of the better seeing eye, age, gender, and scores of SF-36 mental and physical component summary scales. Despite younger age, better visual acuity, and better short-term prognosis, these patients had NEI-VFQ scores that were similar to those published for patients with age-related macular degeneration. This study stresses the importance of measuring quality of life in this patient population and provides information on disability of this condition to caregivers.

806 # 1673/IS QUALITY OF LIFE (QOL) AN APPROPRIATE OUTCOME MEASURE FOR THE EVALUATION OF EXERCISE PROGRAMS FOR CHRONICALLY ILL INDIVDUALS? Anke Hanssen-Doose & Klaus Schuele, Institute of Rehabilitation & Sport for the Disabled, German Sport University Cologne, Cologne, Germany The interdependency of Quality of Life (QOL), this being understood as a multidimensional construct, and Physical Activity (containing physical, psychological and social components) has not yet been proven. Although there is no clear relationship between Quality of Life (QOL) and Physical Activity levels, QOL-measurements are widely used for the evaluation of exercise programs for people with chronic diseases. Some studies with healthy individuals show no improvement of QOL that could be attributed to exercise. The Objectives of this paper are, firstly, to give a systematic overview over existing exercise programs evaluated with generic QOL-instruments on the basis of a literature review and, secondly, to discuss conceptual issues pertaining to the suitability of QOL-measurement for the evaluation of exercise programs. Methods: The Literature review was carried out by using special keywords. The study designs were controlled studies. The studies were analysed within a framework of different parameters, such as kinds of chronic diseases, sample sizes, instruments, exercise programs, QOL results. Results: Twenty three relevant exercise programs for chronically ill individuals were extracted. Twenty one studies (91%), however, showed a significant increase of QOL after chronically ill individuals had participated in corresponding exercise programs. The following factors influenced the suitability and the accuracy of generic QOL-measurement for the evaluation of exercise programs: (1) Evaluation of disease stages and symptoms in order to choose sensitive QOL-instruments which help avoiding ceiling effects, (2) evaluation of activity levels (everyday life and sports) at baseline and during the studies in order to control the impact of the total physical activity of the study population. Conclusions: The results suggest that generic questionnaires are indeed suitable outcome criteria to measure the effectiveness of special exercise programs because due to their multidimensional structure they best reflect the holistic nature of the effects of exercise programs. # 1397/OUTCOMES AND WELL-BEING IN TYPE 2 DIABETIC PATIENTS Jose L. Pais-Ribeiro & Clara Garcia, Psicologia, Universidade do Porto, Porto, Portugal Well-Being is considered an important outcome for diabetic patients and other chronic diseases. The aim of the present study is to identify the relationships between well-being and characteristics of type 2 diabetes. The participants in the study were 389 outpatients, with a mean age of M ¼ 60.93 years, 35% female, 85% married, with type 2 diabetes, with no less than18 years of age; literate; no less than 6 months after the first contact with the medical doctor; with no other severe disease; and with no mental disease. Patients completed Bradley’s12 item Well Being questions (W-BQ12) developed for diabetic patients. Principal components analysis of the Portuguese WBQ12 identified two factors, Positive Well-Being and Negative WellBeing (Cronbach a 0.80 for the seven items of the NWB, and 0.70 for

the five items of the PWB (correlation between the two scales r ¼ 0.45, higher score means worst results for both scales). Evaluation also included clinical outcomes measures – Medical Perception of Patient Adherence, Patient Self-Perception Adherence, Glycated Hemoglobin (HbA1c), and other disease-related data – the duration of the disease, the number of complications due to disease (retinopathy, nephropathy and/or neuropathy), severity of the disease evaluated by a medical doctor, and BMI. Results show statistically significant correlations between NW-BQ and complications due to disease (r ¼ 0.22) severity of disease (r ¼ 0.19) (both p < 0.0001), and BMI (r ¼ 0.10, p < 0.05), and between PW-BQ and complications due to disease (r ¼ 0. 23) and severity of disease (r ¼ 0.21) (both p < 0.0001). No relationship was found between WB and adherence indicators. It seems that WB is not an important issue with regard to adherence for the Portuguese population. WB seems to be more closely related with the consequences of disease # 1613/QUALITY OF LIFE AND BODY IMAGE OF PATIENTS WITH ACROMEGALY Rueya D. Kocalevent & Matthias Rose, Department of Psychosomatic Medicine; Charite´ M. Ventz, Department of Endocrinology; Herbert Fliege, Janine Becker, Otto B. Walter & Burghard F. Klapp, Department of Psychosomatic Medicine, Humboldt University, Berlin, Germany Little empirical research has been conducted regarding the quality of life and body image of patients with acromegaly. The present study examines what aspects of these two constructs have changed as a result of the illness. Ninety one patients were examined (age 56 ± 12yrs., 42.9% men) using the WHO Quality of Life questionnaire (WHOQOL-Bref) and the Frankfurter Body Concept Scales (FKKS). Two representative group samples (n ¼ 2073/1344) served as controls. The patients IGF-1 and STH values were also assessed. The quality of life of patients was significantly reduced in psychological (64 ± 19 vs. 77 ± 18), physiological (65 ± 18 vs. 74 ± 19) and social aspects (66 ± 18 vs. 72 ± 19) (each p < 0.001) when compared to the normal population. The regression analysis showed (R2 corr ¼ 0.32, p < 0.001) that the IGF-1 values are associated with physical wellbeing (beta ¼ )0.21). No differences could be found between the forms of therapy administered, gender, or age. The patients assessed their social/life circumstances more positively than the general population (74 ± 12 vs. 70 ± 14 p < 0.05). They also have a significantly altered body image in all aspects in comparison to the general population (each p < 0.001). Not only do they report less acceptance of their own bodies (20.9 ± 5.6 vs. 24.5 ± 5.1), but also considerable restrictions with respect to their sexuality (18.8 ± 5.6 vs. 25.3 ± 4.7) and fears of exuding an unpleasant body odour (11.1 ± 3.2 vs. 17.4 ± 3.0). Other’s acceptance of their bodies was, however, assessed more positively than in the general population (14.8 ± 3.3 vs. 14.2 ± 3.2, p < 0.05). All aspects of body image can be seen to be closely connected to quality of life (R2 corr ¼ 0.32–0.38, p < 0.001). Patients can be shown to have a considerably reduced quality of life. This is connected not only with the therapeutic regulation of the IGF-1 values, but is essentially co-dependent on the extent of alteration of the subjective body image.

807 # 1447/HEALTH RELATED QUALITY OF LIFE AND LONG TERM SUPPLEMENTATION WITH NUTRITIONAL DOSES OF ANTIOXIDANT VITAMINS AND MINERALS. THE PLACEBO-CONTROLLED RANDOMIZED ‘SU.VI.MAX’ TRIAL SergeBriancon&StephanieBioni,PublicHealth SchoolEA3444,Faculty ofmedecineUniversityNancyI,VandoeuvrelesNancy;SandrineBertrais, U 557, INSERM, Paris; Francis Guillemin, Public Health School EA 3444, Faculty of medecine University Nancy I, Vandoeuvre les Nancy; Serge Hercberg, U 557, INSERM, Paris, France Purpose: We tested the efficacy of supplementation with a combination of antioxidant vitamins and minerals, at nutritional doses, in improving Health related Quality of Life (HRQoL) in a healthy French population. Methods. The SU.VI.MAX study is a randomized double-blind, primary prevention trial; 7551 subjects (3151 men aged 45–60 and 4400 women aged 35–60) took a single capsule daily, filled in twice HRQoL questionnaires SF36, Duke Health Profile, ghq12- and were free of major health events during the two measures interval: 3809 received a combination of 120 mg vitamin C, 30 mg vitamin E, 6 mg beta-carotene, 100 lg selenium and 20 mg zinc; 3742 received the placebo. Mean time between the two measures was 76.0 (±4.2) months. Results: The HRQoL scores changed significantly in both gender and with each of the instruments. Exceptions were the SF36 Physical Role and Emotional Role, the SF36 General Health among men, and the Duke general health due to the deterioration for physical dimension and to the improvement for mental and social dimensions. These changes remained of weak magnitude, ranging from 0.1 for the Duke general health to )2.9 for the Duke physical dimension. About 60% of the subjects reported no health change during the follow-up period, while less than 20% declared an improvement. No major differences appeared between placebo and intervention groups: among men, change in SF36 Physical functioning was significant ()1.1 vs. )0.2). Among women, differences were observed in SF36 Mental Health (+2.2 vs. +1.2; p < 0.07), vitality (+2.2 vs. +0.9; p < 0.01), mental summary scale (+1.4 vs. +0.8; p < 0.05) and GHQ12 (+1.7 vs. +0.9; p < 0.08). All changes were consistent with the answer to the health change question which did not differ between groups. Conclusions: While nutritional doses of a combination of antioxidants may have a protective effect on the development QoL is very modest in itself and mainly mediated by the prevention of major morbid events. # 1014/THE IMPACT OF ASTHMA SELF-MANAGEMENT EDUCATIONAL PROGRAMS ON THE HEALTH OUTCOMES IN PEDIATRIC AND ADULT POPULATION OF UNITED STATES OF AMERICA: A META-ANALYSIS (SYSTEMIC REVIEW)OF RANDOMIZED CONTROLLED TRIALS Surender Gddam & Jospeh Lovett, Health Science An Human Ecology, California State University, San Bernardino, San Bernardino, CA, USA Objective and design: The study was designed to examine the impact of asthma self-management educational programs on the health outcomes in pediatric and adult subjects of United States.The quality of the studies was assessed by Jadad’s validated quality scale. Trials were critically appraised and evidence tables created with the key information in the studies. Effect sizes (standard mean differences) were calculated using standard statistics. The pooled effect size was calculated using inverse variance weight method both in categorical and the continuous data. Principal findings and conclusions: The literature search had retrieved 60 clinical trials but only 17 were included in the study. Ten of the 17 studies were of poor quality. On pooling the effects of the individual studies though there was an improvement in health outcomes it was only a negligible to small effect {(hospitalizations: ES ¼ )0.13()0.30, 0.04); hospital days: ES ¼ )0.21()0.56, 0.14); subjects requiring ED visits: OR ¼ 0.67(0.35, 1.30); ED visits (number): (ES ¼ )0.16()0.28, )0.04); unscheduled doctor visits: (ES ¼ )0.17()0.31, )0.03); days lost from school: (ES ¼ )0.05()0.26, 0.16); asthma attacks: ES-0.23 ()0.52, 0.06); AM and PM asthma attacks: ES ¼ 0.04()0.32, 0.40), (ES ¼ )0.37()0.72, )0.02); daily average AM and PM PFER measurements: ES ¼ 0.04()0.25, 0.33), (ES ¼ 0.14()0.15, 0.43)}. In overall the educational interventions in adults were more effective than in children but only with a negligible to small effect {(hospitalizations: (ES ¼ )0.28 ()0.85, 0.29), (ES) ¼ )0.12()0.30, 0.06); ED visits: (ES ¼ )0.22

()0.42, )0.02), (ES ¼ )0.11()0.27, 0.05); unscheduled doctor visits: (ES ¼ )0.36()0.56, )0.16), (ES ¼ )0.03()0.20, 0.15)}. The same was the case when individualized education compared with the group education.Self-management teaching programs have a positive impact on the health outcomes but with anegligible effect. The educational programs seem to be more effective in adults than in pediatric age group, in individualized education than group education. # 1884/REHABILITATION AND SPORTS: EFFECTS OF SPORTS AND DAILY PHYSICAL ACTIVITY PROMOTION Kitty Streppel, Roessingh Research and Development, Rehabilitation centre Het Roessingh, Enschede; Hidde v d Ploeg, Department of Social Medicine, Institute for Researc, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands Sports participation and a physically active lifestyle are expected to have a positive effect on functionality, mental well being, and health. For this reason several rehabilitation centres started offering a standardised sports promotion programme – Rehabilitation & Sports (R&S). In addition an Active after Rehabilitation programme (AaR) was developed to stimulate daily physical activity. The purpose of this study is to investigate the effects of the R&S and AnR programmes on sports participation, daily physical activity and health status. A quasi-experimental intervention study is currently conducted. In four rehabilitation centres patients are randomly assigned to the R&S programme with or without the supplementary AaR programme. In six control centres patients receive ‘usual care’ only. Measurements take place about 6 weeks before (t0), 9 weeks (t1) and 1 year after discharge (t2). Measurements exists of an interview (ALDS) and questionnaires (i.e. SF36, PASIPD, SCL) Preliminary data (n ¼ 880) of t0 show that sports participation 1 year before rehabilitation is similar in the intervention (57%) and control group (58%), while sports participation at t0 is different; 56% vs. 40% respectively. Besides this the percentage of patients who are regular daily active at t0 is 58% vs. 47% respectively. Further only the physical function score of the SF36 at t0 is a little lower (44.5) in the intervention patients than the controls (50.5) at t0. Results of these preliminary data are as expected. Intervention centres pay more attention to sports during treatment, which has consequences for sport participation and daily physical activity. Besides this, intervention centres treat more severe patients with lower physical function scores. Soon preliminary data of t1 and t2 can be presented. # 1253/QUALITY OF LIFE IN RESIDENTS WITH HEALTH PROMOTION PROGRAM Shizuko Sugimoto, Michiko Ogushi, Rie Yamashita, Michiko Kobayashi, Yukiko Ueda & Takashi Mandai, Japanese Society of Quality of Life Research, Kobe, Hyogo, Japan Health promotion projects named ‘Health 21’ had begun since 2000 in Japan, so Sakido town in Nagasaki prefecture has also begun the factfinding survey on health problem as the health promotion projects. 1670 residents over 20 years old participated in this QOL study. Both personal interviews and original self-administered questionnaire including 40 questions divided into 13 categories were used. Cronbach’s a coefficients of our questionnaire were high enough to accept for practical use: 0.88 in environmental problems, 0.88 in dietary problems, 0.84 in medical service, 0.84 in psychological problems, 0.82 in social particiaption, 0.82 in well-being and 0.81 in economy, etc, respectively. Our questionnaire contained 10 main factors and cumulative contribution was 0.70. Excellent QOL scores were found in civilized areas, but deteriolated ones were indicated in isolated island. Hypertension, cataract, and musculo-skeletal disease were most common three diseases in this district. Compared with normal subjects, the patients with cataract indicated the significantly excellent QOL scores (p < 0.05). On the other hand, the significant deteriorations were found in the female patients with musculo-skeletal disorders (p < 0.01) and diabetes mellitus (p < 0.05) compared with female healthy subjects. The residents who have more than 20 original teeth showed the significantly excellent total QOLs compared with the ones who have less than 10 original teeth (p < 0.01). The residents who have more attention to oral health care showed the significantly excellent QOLs compared with the ones who have less attention to oral health care. These findings indicate that the good control for health care shows the protency to establish the high QOL daily life, so we would like to apply these findings to health promotion program in future.

808 # 1421/QUALITY OF LIFE AND MEDITATION PRACTICE Charoen Treesak, Xin Ye, Valerie G. Russas & Cynthia R. Gross, Social & Administrative Pharmacy, University of Minnesota, Minneapolis, MN, USA Meditation is a complementary therapy posited to improve health and quality of life (HQOL). The purpose of this study was to determine if HQOL and meditation practice are related in the general population. A convenience sample of 265 persons in the Twin Cities area were surveyed at six sites (university classes, yoga programs, clinics and religious centers) where persons were likely to know what meditation is, regardless of their actual practice. Subjects were classified based on self-reported meditation practice into three groups: (1) no practice; (2) short-term (current regular practice but less than 1 year); and (3) long term (current regular practice for more than one year). Differences in practice levels by demographic variables were examined using v2 tests. The relationships between practice level and health status (Rand Short Form 12 PHC and MHC scores) adjusting for potential confounders (age, gender, education, religion, marital status, and income) were tested using GLM. Long-term, regular practice was associated with higher age (p ¼ 0.002), being female (p ¼ 0.008), being Buddhist (p < 0.001), and higher education (p ¼ 0.013). After adjusting for these confounders, current practice levels were associated with higher PHC (p ¼ 0.034) and MHC (p < 0.001) scores. The short term (p ¼ 0.029) but not long term practice group had significantly higher PHC scores than those with no practice. Differences in MHC scores among groups were more prominent. The long term practice group had higher MHC scores, on average, than both the short term practice group (p ¼ 0.023) and the no practice group (p < 0.001) and the short term practice group had higher MHC scores than those with no practice (p ¼ 0.001). Our study shows being female, older, having more education and being Buddhist were associated with meditation practice. We found that meditation practice is associated with higher HQOL, especially mental health. These findings are consistent with the hypothesis that meditation practice improves HQOL, but interventional studies are needed to confirm this claim. # 1335/SELF-REPORTED LIPODYSTROPHY (LD) AND MENTAL HEALTH IN HIV-INFECTED PATIENTS TAKING HAART Rita Murri & Massimo Fantoni, Infectious Diseases, Catholic University of Rome; Cosmo Del Borgo, Infectious Diseases, INMI L Spallanzani, Rome, Italy HIV-infected people with LD may experience changes in self-image and eventually in self-esteem. Purpose of research was to assess the impact of self-reported LD on health-related quality of life of HIV+ pts in the new antiretrovirals (HAART) era. Methods: Prospective, multicenter, cohort study on Italian HIV+ pts at any stage of disease. Consecutive patients were enrolled in seven Italian Centers. Visits were planned every 6 months. The MOS-HIV was used as tool of the study. Two summary scores for physical (PHS) and mental health (MHS) were obtained. A single-item on signs of body fat redistribution was added at the fourth visit of the study (Jan–Jun 2000) (‘Have you experienced any abnormal fat accumulation in abdomen, breast or dorsal site or a fat loss in cheeks, legs or bottom?’). Response options

were not at all, a little, a fair amount, and a lot. Outcome of the study was considered self-reported LD at any degree (from a little to a lot). Results: Five hundred and nineteen patients were enrolled into the study. Mean 39 years (+8): 32% were females, 48% intravenous drug users, 32% CDC’s group C. Mean CD4+ cells count: 239/mm3 (+284); mean log HIV RNA 3.0 (+1.3). 38% had undetectable HIV RNA at the moment of the survey. 90% were taking HAART (61% with a protease inhibitor [PI]-containing regimen, 25% with a nonucleoside reverse transcriptase inhibitor [NNRTI] and 14% with both). Mean time on PI was 24 months (+18) while mean time on NNRTI was 4 months (+7). 40% self-reported signs of LD (26% a little, 9% a fair amount, and 5% a lot). On multivariate analysis being not IDU (OR 1.98; 95% CI 1.36– 2.89), time on PI (every 6 months: OR 1.09; 95% CI 1.02–1.16), MHS (OR for each point increment 0.97; 95% CI 0.95–0.98) were independently associated with self-reported LD even when adjusted for age, gender, HIV disease stage, log HIV RNA, and time on NNRTI. Conclusions: LD is negatively associated to mental health even when adjusted for demographic and clinical variables. # 1543/QUALITY OF LIFE OF HCV-INFECTED PATIENTS IN URBAN METHADONE AND HIV CLINICS Bruce R. Schackman & Paul A. Teixeira, Department of Public Health, Weill Medical College of Cornell University, New York, NY, USA Hepatitis C (HCV) treatment is associated with transient declines in quality of life and increased incidence of depression. This study describes quality of life and depression among urban HCV-infected patients who are connected to medical care but have not initiated HCV treatment. Subjects were recruited from one methadone maintenance (MMT) clinic (n ¼ 15) and two HIV clinics (n ¼ 14) in New York City. The survey was administered in English or Spanish using a touchscreen computer. Measurements included the SF-12, CES-D 10-item depression scale, visual analogue scale (VAS), and standard gamble (SG). SF-6D estimates of community-derived SG values were also calculated from SF-12 responses. Subjects were 52% African-American (n ¼ 15), 28% Hispanic (n ¼ 8), and 31% women (n ¼ 9). Mean (SD) SF-12 physical health composite score was 33.35 (11.40), with no significant difference between MMT and HIV patients. Mean SF-12 mental health composite scores were 35.96 (1.87) for MMT patients vs. 47.90 (3.20) for HIV patients, p ¼ 0.03. 50% of all subjects scored 15 or above on the CES-D 10, indicating severe depression symptoms. Mean CES-D 10 scores were higher (more depression symptoms) for MMT patients than for HIV patients: 17.40 (6.88) vs. 11.92 (5.35), p ¼ 0.028. Mean VAS scores were 0.60 (0.25) overall, 0.50 (0.27) for MMT patients, and 0.66 (0.18) for HIV patients. Mean SG scores were 0.63 (0.35) overall, 0.74 (0.24) for MMT patients, and 0.56 (0.40) for HIV patients. Two HIV patients appeared to be risk seeking in gambles for perfect health, accounting for the lower mean SG for this group. Mean SF-6D scores were 0.62 (0.14) overall, 0.58 (0.10) for MMT patients, and 0.66 (0.16) for HIV patients. Urban HCV-infected patients at MMT and HIV clinics have poor quality of life. Mental health distress and depression are common and more frequent among MMT patients. Depression symptoms and future quality of life expectations of patients need to be addressed prior to initiating HCV treatment that could cause further deterioration in these outcomes.

809 # 1077/A RANDOMIZED TRIAL OF THE IMPACT OF A PROGRAMMABLE MEDICATION REMINDER DEVICE ON QUALITY OF LIFE IN HIV/AIDS PATIENTS Claire F. Snyder & Albert W. Wu, Health Policy & Management; Adriana S. Andrade, School of Medicine, Johns Hopkins University, Baltimore, MD; Henraya F. Davis, Division of Adult and Community Health, CDC, Atlanta, GA; Shivaun A. Celano & Richard L. Skolasky, School of Medicine, Johns Hopkins University, Baltimore, MD; Alan M. Letzt, Adherence Technologies, Adherence Technologies Corp., Purcellville, VA; Justin C. McArthur, School of Medicine, Johns Hopkins University, Baltimore, MD, USA

stigma (21%) limited their social life. Forgetfulness was the only affected cognitive function. 63% felt depressed, helpless and unhappy intermittently. Over 70% patients reported irritation, anger and tension while 62% patients experienced bodily pain. Sleep was affected, primarily by emotional problems. Although 69% reported good appetite, 41% lost weight. Only 37% had sexual relationships in the last month.Conclusion: HIV infection affected social activities, sexual life, and food intake among study participants, but did not significantly affect physical health, work & earning, daily activities, cognitive functions, feelings and emotions, pain, sleep. Prospective changes need to be studied.

This 6-month prospective RCT evaluated the impact on QOL of a medication reminder device for patients with HIV/AIDS. Patients with HIV/AIDS were eligible if they were <3 HAART or naive. The intervention was a Disease Management Adherence System (DMAS), a prompting device that verbally reminds patients at medication times and electronically records dosing times, and a monthly 30 min adherence educational session. Controls received adherence education only. QOL was measured at baseline and 6 months using the CES-D, IADLs, and the MOS-HIV (10 subscales and physical and mental summary scales (PHS,MHS)). Cronbach’s a was >0.70 for all scales. At baseline, 62 patients completed QOL surveys (31 control, 31 DMAS), and at month 6, 48 patients completed QOL surveys (25 control, 23 DMAS). Patients differed significantly at baseline on seven scales, primarily those measuring mental health (CES-D, IADLs, MHS, General Health Perceptions, Quality of Life, Mental Health, and Health Distress). At month 6, there was a trend towards improved medication adherence for the DMAS group; this difference was statistically significant in participants who had memory impairment. QOL change scores were significantly different between groups, favoring the control arm, for seven scales that primarily measure physical health (CES-D, IADLs, PHS, General Health Perceptions, Pain, Quality of Life, and Role Function). The differences in change scores generally reflected improvement in the control arm, with some deterioration in the DMAS arm. These differences in change scores persisted after controlling for demographics, baseline CD-4 count, and medication adherence. Thus, DMAS was associated with improved adherence but decreased QOL. This may have been due to increased adverse drug effects in people receiving the intervention.

# 1321/QUALITY OF LIFE IN HIV-POSITIVE BRAZILIANS: A COMPARISON BETWEEN WHOQOL-100 AND SF-36 Roge´rio R. Zimpel & Marcelo P. Fleck, Psychiatry Department and Legal Medicine, Faculdade de Medicina, Universidade Federal do RGS, Porto Alegre, Rio Grande do Sul, Brazil

# 1655/QUALITY OF LIFE OF PEOPLE ATTENDING THE HIV REFERENCE CLINIC IN PUNE, INDIA Rewa M. Kohli, Social & Behavioural Science; Ramesh S. Paranjape, Immunology, Sanjay M. Mehendale, Epidemiology, National AIDS Research Institute, PUNE, Maharashtra, India Purpose of research: The present study assessed QOL of HIV infected individuals seen at the HIV Reference Clinic in Pune. Methods: MOS core instrument adapted to collect information on QOL including 11 domains pertaining to physical health, functioning and activities, work, earnings, social activities, cognitive function, feelings, food intake, sex, and attitudes was administered to 100 HIV infected persons between Feb 02 and Mar 03. Open-ended questions on stigma, sleep, sexual behavior, coping, future concerns and planning were also included. Results: Mean age was 34 ± 8.2 years, 34% were female, 65% married, 24% widowed, 70% secondary education. Cronbach’s a ranged from 0.43 to 0.95. Mean CD4 count was 325/mm3. Physical health, social activities, food intake and appetite were significantly correlated with CD4 counts (r ¼ )0.22, )0.36, )0.29, all p < 0.05). Significant correlation was found between viral load [n ¼ 13 patients] and social activities, feelings and emotions (r ¼ )0.71, )0.62 and p < 0.01, p < 0.05 resp). For symptomatic patients daily activities, pain and cognitive function were affected less than other domains. Physical health was good for 52% and although 15% were hospitalized in last 3 months, 89% did not require help outside of the home. Most were pleased with personal life [87%]. Although 84% were satisfied with their ability to work, 42% mentioned that work was affected. Fear of

To compare the performance of the generic instruments World Health Organization Quality of Life instrument (WHOQOL-100) and Medical Outcomes Study 36-Item Short-Form Health Survey (SF-36) in a sample of HIV-positive Brazilians. In Porto Alegre/RS, quality of life was evaluated using the WHOQOL-100 and the SF-36 in a sample of 308 HIV infected men and women selected by convenience (131 symptomatic, 91 symptomatic and 86 with AIDS). Depression and anxiety symptoms were measured by the Beck Depression Inventory (BDI) and by the State-Trait Anxiety Inventory (STAI), respectively. Sociodemographic characteristics were also obtained. Both the WHOQOL-100 and the SF-36 aptly discriminated quality of life among stages of infection in the expected direction. In BDI correlations both revealed Pearson coefficients of moderate to large magnitude; STAITrait subscale magnitude values were lower, ranging from small to moderate. In direct correlation between the instruments themselves, the eight SF-36 domains correlated the strongest with three WHOQOL-100 domains (Physical, Psychological and Level of Independence) The SF-36 confirms its capability as a ‘functional status’ evaluator, whereas the WHOQOL-100 proves itself as a quality of life tool of more comprehensive construct. # 1319/QUALITY OF LIFE, DEPRESSION AND ANXIETY IN HIVPOSITIVE BRAZILIANS Roge´rio R. Zimpel & Marcelo P. Fleck, Psychiatry Department and Legal Medicine, Faculdade de Medicina, Universidade Federal do RGS, Porto Alegre, Rio Grande do Sul, Brazil Mental symptoms may interfere with quality of life of persons living with HIV/AIDS (PLHAs). This study: (1) Evaluates the presence of symptoms of depression and anxiety in HIV-positive Brazilians; (2) analyzes the correlation between depression and anxiety scores with the World Health Organization Quality of Life instrument HIV module (WHOQOLHIV), a generic quality of life instrument modified for the HIV/AIDS disease. In Porto Alegre/RS, QOL was evaluated using the WHOQOLHIV in a sample selected by convenience of 308 men and women infected by HIV (131 asymptomatic, 91 symptomatic and 86 with AIDS). Depression and anxiety symptoms were measured with the Beck Depression Inventory (BDI) and the State-Trait Anxiety Inventory (STAI), respectively. Sociodemographic characteristics of the sample are reported as well. Phases of infection did not reveal differences in anxiety scores, yet depression scores were higher in AIDS patients when compared to asymptomatic and symptomatic patients. In correlation with the WHOQOL-HIV domains, Pearson coefficient values were of a moderate to very large magnitude for the BDI but small to moderate for the STAI-Trait sub-scale. After adjusting in accordance to disease states and clinical and sociodemographic variables in a multiple regression model, the BDI showed, in relation to the remaining variables, significantly higher beta coefficient values. Data in this study indicates that QOL of PLHAs is affected by other variables besides disease stage, depression in particular.

810 # 1852/THE MULTIPLE SCLEROSIS IMPACT SCALE: VALIDATION IN EIGHT EUROPEAN COUNTRIES USING RASCH TECHNOLOGY Jeremy C. Hobart, Clinical Neurosciences, Peninsula Medical School, Plymouth, Devon; Stefan J. Cano, Alan J. Thompson & Rory O’Connor, Neurological Outcome Measures Unit, Institute of Neurology, London, England Clinical trials are frequently conducted in multiple countries, and increasingly using patient-based outcome measures. It is therefore essential that these rating scales are proven stable across different languages and cultures. The examination of cross-cultural stability of rating scales lends itself to Rasch technology for two specific reasons. First, when the data fit the Rasch model parameter estimates are separated. That is, estimates of item difficulty (calibrations) are independent of the distribution of the sample from which they are generated. Second, there is good evidence that useful estimates can be derived from small samples. This study investigated the use of Rasch technology to examine the stability of the Multiple Sclerosis Impact Scale (MSIS-29) across eight European countries. The MSIS-29 was administered to 50 MS patients in Finland, France, Holland, Italy, Spain, Sweden, and Ireland. Clinical settings and sample characteristics differed. First, we determined if the MSIS-29 satisfied Rasch criteria for objective measurement in each country. Second, we examined cross-cultural differential item functioning (DIF) by computing MSIS-29 item calibrations for each country and comparing these with those generated by the Rasch analysis of a large diverse UK sample (n ¼ 1725). Third, we examined the impact of any DIF by computing, and scatterplotting Rasch linear person measures for each country using: (i) their own item calibrations; (ii) item calibrations from the UK sample; (iii) item calibrations from each of the other 6 countries. The MSIS-29 satisfied Rasch criteria in all countries. There was some DIF across countries but in the vast majority of cases these were within 0.5 logits (considered acceptable) of the UK sample. The DIF had little impact on person measures, all scatterplots of measures for any one country generated from different item calibrations (i.e. any other country) yielded near straight lines. This study suggests that Rasch technology may have a useful role in the evaluation of cross-cultural validity. # 1423/QUESTIONNAIRE ITEM REDUCTION USING RASCH ANALYSIS AND PRINCIPAL COMPONENT ANALYSIS Makiko Meguro & Paul W. Jones, Division of Respiratory Medicine, St George’s Medical School, London, UK The objective of the study is to compare Rasch Analysis (RA) with Principal Component Analysis (PCA) based techniques for excluding items during questionnaire development. We used five sets of St George’s Respiratory Questionnaire (SGRQ) data. Each was tested using Rasch Analysis (RA) and Principal Component Analysis (PCA). The item exclusion criterion for RA was a v2 value above that acceptable for the number of items in the model. Items were excluded using PCA if their factor loading was less than 0.49. The mean number of subjects in each dataset was 538 and their mean age 64. In total 15 items met both exclusion criteria in one or more datasets. All five datasets excluded two items, four datasets excluded three other items, and three datasets excluded further two items. The remaining eight items were excluded by both technique in only one dataset. We conclude that whilst RA and PCA analyses identifies similar groups of items for exclusion. Use of both analyses together may strengthen decisions about which items during questionnaire development.

# 1768/MEASURING PATIENTS’ EXPERIENCES OF TREATMENT FOR MULTIPLE SCLEROSIS (MS) RELAPSES Afsane Riazi, Neurological Outcome Measures Unit, Institute of Neurology, London; Jeremy C. Hobart, Peninsula Medical School, Derriford Hospital, Plymouth; Bernie Porter & Jeremy Chataway, National Hospital for Neurology and Neurosurgery, London; Alan J. Thompson, Neurological Outcome Measures Unit, Institute of Neurology, London, UK Rigorous measures of appropriate constructs are needed to compare different treatment modalities. However, empirical evidence demonstrates that treatment satisfaction surveys tend to elicit very positive ratings that are not sensitive to specific problems in the quality of care delivery. More pertinent is the measurement of patients’ experiences of health care. We are developing such a measure for multiple sclerosis (MS) for a randomised controlled trial comparing home vs. outpatient administration of intravenous steroids for relapses to determine whether steroid treatment could be delivered at home in the majority of cases. This could lead to a paradigm shift in the way that people with MS in the UK are treated for their relapses. Qualitative interviews were conducted with 21 people with MS recently treated for an MS relapse in an outpatient or inpatient setting, or in their own home. Statements were extracted and grouped in the following diverse dimensions: access to care, continuity of care, coordination of care, physical comfort, technical aspects, involvement of family and friends, interpersonal care, and attitude of health care professionals. An initial pool of 154 items was generated from these statements. A number of different response options were required for the items. Data analysis must account for this, offer the flexibility to examine modifications to the response options, and enable the construction of measures that are applicable to both treatment groups from diverse items. These needs can be satisfied by the Rasch partial credit model. The first field test is in progress and the 154-item long form questionnaire has been sent to 300 people with relapsing-remitting MS. This paper will present the results of the item reduction process and address these methodological issues. # 1331/ITEM SELECTION IN QUALITY OF LIFE ASSESSMENT WITH THE APPLICATION OF PARAMETRIC IRT MODELS Ting Hsiang Lin, Department of Statistics; Grace Yang, Psychology, National Taiwan University, Taipei, Taiwan The WHO has devised WHOQOL-BREF containing 26 items plus optional national items as an assessment tool to measure quality of life. However, the psychometric properties of individual items of WHOQOL-BREF have not been examined yet. Using Item Response Theory (IRT), we studied the item properties of the WHOQOL-BREF Taiwan version. Data were collected in the 2001 NHIS, and respondents were general population aging from 20 to 65. Samejima’s graded response model (SGRM) with natural metric of the logistic response function was fitted. These items displayed several patterns of discrimination. First, items with negative natures were less discriminating than items phrased towards positive directions. Second, the descriptions of some items were written so specific that they are more suitable to assess certain subgroups and they were less discriminating. Third, the national item measuring if being respected by others has above average discrimination power, while the other national item assessing eating provides little information. Fourth, questions measuring object specific issues performed worse than items assessing general aspect of quality of life. We also generated several datasets with subset of items by omitting items with lower discrimination power. The same SGRM were fitted and individuals, were scored for the subsets and 28item full scale. With 20% reduction of items, the correlation of ,s between the subset and the full scale is as high as 0.994. The correlation remained strong even with a 50% reduction (r ¼ 0.968). We also present test information function across different levels of latent traits for both the full scale and the subsets. The maximum amount of information spaced over two ends of the theta continuum, and this suggested that the WHOQOL-BREF provided more information for groups with either lower or higher satisfaction of quality of life, while it is less discriminating for individuals in the middle range. In summary, IRT are useful in item evaluation and a subset of optimal items can be chosen in developing quality of life instrument.

811 # 1827/COMPARING TWO ITEM-RESPONSE THEORY BASED ON STRATEGIES FOR THE DEVELOPMENT OF A GLOBAL QUALITY OF LIFE INDEX USING ADOLESCENTS DATA Michael Erhart & Ulrike Ravens-Sieberer, Child & Adolescent Health, Robert Kock Institute, Berlin, Germany; Michael Power, Department of Psychiatry, Edinburgh University, Edinburgh, UK; Janine Becker, Matthias Rose & Otto Walter, Psychosomatic Medicine, University Hospital Charite´, Berlin, Germany; Pascal Auquier, Ste´phane Robitail, University Hospital of Marseille, Marseille, France and the European KIDSCREEN group Today there is a broad consensus to think about health related quality of life (QoL) as a multidimensional construct which contains physical, psychological, social and behavioral elements of well-being and functional capacity from the view of the individual. However QoL assessment in health surveys and health-economic studies often demands the usage of indices and global scores. The aim of this paper is to compare two IRT approaches for the construction of global QoLindexes in performing the development of the scoring algorithm for the short-version of the KIDSCREEN Qol instrument. The data of 3019 children and adolescents (8–18 years) from the pilot test of the ECfounded European KIDSCREEN Qol project was used for the analysis. In the first strategy factor analytic methods were used to test unidimensionality and local independence of items. The ordinal consistency of response choices was examined followed by the estimation of itemparameter and calculations of itemfit according to the the partial-creditmodel. Some items had to be dropped. A global index was created which proofed to be unidimensional and rasch-scaled. Physical, psychological and social QoL-aspects were covered. Using the second strategy 5 so called Rasch-homogenous-latent-groups can be identified which corresponds to five QoL-types detected with means of clusteranalysis. This mixed-Rasch-model fitted the data well (p ¼ 0.120). An algorithm can be developed that at the same time classifies the responders according to their group-membership and estimates their individual QoL-parameter-value. Both strategies were promising and yielding psychometrically robust results, the first approach offers a more practical solution. # 1822/CROSS-CULTURAL EXAMINATION OF THE PSYCHOMETRIC PROPERTIES OF THE YQOL-S WITH MEANS OF ITEM-RESPONSE THEORY Michael Erhart & Ulrike Ravens-Sieberer, Child & Adolescent Health, Robert Koch Institute, Berlin, Germany; Donald Patrick, Health Services, University of Washington, Seattle, WA, USA; Michael Power, Department of Psychiatry, Edinburgh University, Edinburgh, UK; Luis Rajmil, Catalan Agency, Barcelona, Spain; Jeanet Bruil, Child Health Division, TNO Prevention and Health, Leiden, The Netherlands; Bernhard Cloetta, University of Bern, Bern, Switzerland; Wolfgang Duer, Ludwig Boltzmann Institute, University of Vienna, Vienna, Austria The YQOL is a generic QoL instrument for youths developed with inductive grounded theory qualitative methods: in-depth interviews and focus groups with youths with and without disabilities. Although the YQOL originally was not constructed using IRT methods, an attempt is made to examine the psychometric properties of the YQOL-S across seven European countries using CTT methods as well as means of IRT. Using the pilot data of the European KIDSCREEN Project (n ¼ 2470 adolescents aged 12–18) the cross-cultural acceptance of the 13 questions was investigated accompanied by basic item analyses. In addition it was tested if the eight perceptual items are raschscaled and display no differential item functioning across the countries. The YQOL-S items showed reasonable psychometric properties across the seven European countries. Seven of the eight perceptual items fitted the data well according to the probabilistic partial credit model (p ¼ 0.058 to p ¼ 0.971) and furthermore displayed no DIF. Only one perceptual item displayed significant deviation (p ¼ 0.012) from the rasch-model and furthermore showed sizeable DIF which can account for the significant itemfit-statistic. This study shows that the YQOL-S enables a satisfying cross-cultural-assessment of adolescents QoL and indicates that the summation of the YQOL-S perceptual items to an overall score is statistically justified – only one item had to be corrected for cross-cultural-DIF. IRT methods are a reasonable supplement for the psychometric analysis of CTT instruments.

# 1579/QUALITY OF LIFE IN SYSTEMIC SCLEROSIS Erik J. Groessl, Health Outcomes Assessment Program, University of California San Diego, La Jolla; Vanessa L. Malcarne, Psychology, San Diego State University; Ingunn Hansdottir, Joint Doctoral Program, SDSU/UCSD; Mahasin Abdullah, Psychology, San Diego State University, San Diego; Robert M. Kaplan, Family and Preventive Medicine, University of California San Diego, La Jolla, CA, USA The study describes health-related quality of life (QOL) in systemic sclerosis. The self-administered version of a widely used, generic, preference-based QOL instrument, the Quality of Well-Being Scale (QWB-SA), was administered to systemic sclerosis patients at two points in time approximately 1 year apart. Subjects were 73 adults (61 women and 12 men; 75% Caucasian) with confirmed diagnoses of systemic sclerosis. Mean age was 49 years, and mean disease duration was 5 years. Subjects were recruited from UCLA and UCSD Medical Centers in southern California, and the Virginia Mason Research Center in Seattle, Washington. As part of a longitudinal study documenting changes in quality of life and associated outcomes, subjects completed the QWB-SA at entry into the study (T1) and approximately 1 year later (T2). This widely used and well-validated measure has been used to document health-related quality of life in a variety of medical populations, but has not previously been used for systemic sclerosis. The QWB-SA yields an index of quality-adjusted life years (QALYs) that can range from 1 (perfect health) to 0 (death). Mean QWB score at T1 was 0.45 and at T2 was 0.41. This represents a very low health-related QOL, compared to a healthy community sample (QWB ¼ 0.72) and to other patient populations as reported in the literature (e.g., fibromyalgia QWB ¼ 0.56, COPD ¼ 0.66, advanced cancer ¼ 0.63). QWB scores at T1 and T2 were significantly correlated (r ¼ 0.54, p < 0.001) but were also significantly different (T2 QALY score was significantly lower than T1), suggesting a decrease in QOL over the 1-year study period. Some evidence of differential effects for sex and ethnicity emerged but small subgroup sample size limits interpretation. Health-related QOL appears to be particularly low in systemic sclerosis compared to other patient groups, and to significantly decrease over a one-year period. # 1779/DOES THE IMPACT OF VISUAL IMPAIRMENT SCALE CAPTURE VISION DYSFUNCTION IN PATIENTS WITH MULTIPLE SCLEROSIS? Monika L. Baier, Biostatistics, Cooper Institute, Golden, CO; Deborah M. Miller, Mellen Center/Neurology, Cleveland Clinic, Cleveland, OH; Gary R. Cutter, University of Nevada at Reno, Reno, NV, USA The Multiple Sclerosis Quality of Life Inventory (MSQLI) contains 138 questions, excluding the three short-version subscales, covering all aspects of functioning affected by multiple sclerosis (MS). The MSQLI is composed of the SF-36 and nine symptom-specific scales. How sensitive the individual scales are to change is currently being studied by various authors. The focus of this investigation will be only on the Impact of Visual Impairment Scale (IVIS), which consists of five questions. In the IMPACT trial, 324 English speaking secondary progressive MS patients completed the MSQLI. Of these, 44% did not record any difficulties with their vision. At the end of the trial 71% of these patients still claimed no difficulties with their vision. In a sub-study to the IMPACT trial, 65 patients had contrast letter acuity (CLA) assessed. CLA was measured similar to a usual visual acuity Snellen chart, using 100, 5, and 1.25% contrast. The measure is number of letters read correctly (maximum score of 60). The 100% chart is equivalent to the Snellen visual acuity, which has not provided estimates of differential change over the course of a clinical trial. We examined Spearman Rank correlations between each question of the IVIS and CLA at baseline and 24 months. No significant correlations were observed at baseline between CLA and any of the questions. Reading of books, magazines, etc., and identifying house numbers, street signs, etc. were significantly correlated at 24 months with 1.25% contrast (r ¼ )0.70 and )0.67, respectively) in the treatment group. Reading of books, magazines, etc., and reading of dials on stoves, etc. were significantly correlated with 5% contrast (r ¼ )0.61 and )0.64, respectively) in the same group. No correlations were observed with the 100% chart. Change in CLA from baseline to month 24 did not correlate with change for any questions in treatment or placebo groups. The individual questions in the IVIS may not be sensitive enough to capture the broad spectrum of visual dysfunction and other more sensitive visual quality of life indices most likely need to be evaluated.

812 # 1671/CONCURRENT VALIDITY AND SENSITIVITY TO CHANGE OF THE MULTIPLE SCLEROSIS QUALITY OF LIFE INVENTORY COMPARED TO OBJECTIVE MEASURES OF DISABILITY, DURING RECOVERY FROM EXACERBATIONS OF MULTIPLE SCLEROSIS Francois A. Bethoux & Deborah M. Miller, The Mellen Center for Multiple Sclerosis/U10, The Cleveland Clinic Foundation, Cleveland, OH The Multiple Sclerosis Quality of Life Inventory (MSQLI) is a diseasespecific quality of life measure, which includes the SF-36 and nine symptom severity scales. The Expanded Disability Status Scale (EDSS) and Incapacity Status Scale (ISS) are widely used diseasespecific measures of impairments and activity limitations. During a controlled trial of outpatient rehabilitation after exacerbation of MS, the EDSS, ISS, and MSQLI were administered before and after intervention. We are reporting data obtained prior to randomization, at the time of enrollment (T0) and 4 weeks later (T1). During this period, subjects were expected to experience recovery from their exacerbation. Statistical analyses included Spearman correlation coefficient, Wilcoxon Signed Ranks Test, and effect size calculation. Significance level was set for p < 0.05. One hundred and seven patients were enrolled. At T0, EDSS scores correlated with SF36-PCS (r ¼ )0.40, p ¼ 0.0001), SF36-MCS (r ¼ 0.22, p ¼ 0.037), Pain Effects Scale (PES) (r ¼ 0.21, p ¼ 0.045), and Bladder Control Scale (BLCS) (r ¼ 0.43, p ¼ 0.0001). ISS scores correlated with SF36-PCS (r ¼ )0.51, p ¼ 0.0001), Modified Fatigue Impact Scale (MFIS) (0.391, p ¼ 0.0001), PES (r ¼ 0.362, p ¼ 0.0001), BLCS (r ¼ 0.43, p ¼ 0.0001), Bowel Control Scale (r ¼ 0.34, p ¼ 0.001), Perceived Deficits Questionnaire (r ¼ 0.36, p ¼ 0.001), Social Support Scale (r ¼ )0.32, p ¼ 0.002), Sexual Satisfaction Scale (r ¼ 0.43, p ¼ 0.0001), and Impact of Visual Impairment Scale (r ¼ 0.301, p ¼ 0.004). Similar correlations were observed at T1. There was significant improvement of scores between T0 ad T1 for EDSS (p ¼ 0.008, effect size ¼ 0.16), ISS (p ¼ 0.009, ES ¼ 0.20), SF-36 MCS (p ¼ 0.01, ES ¼ 0.28), MHI (p ¼ 0.021, ES ¼ 0.28), MFIS (p ¼ 0.001, ES ¼ 0.31), PES (p ¼ 0.001, ES ¼ 0.33), and BLCS (p ¼ 0.038, ES ¼ 0.16). The MSQLI demonstrated significant correlations with objective measures of disability and was responsive to clinical change in our sample of MS patients. These results suggest that the MSQLI is a valid measure of perceived health status and symptom severity in MS. # 1220/QUALITY OF LIFE IN WOMEN WITH BREAST CANCER AND THEIR FAMILY ENVIRONMENT Tomoko Takayama, Faculty of Health Sciences, Okayama University Medical School, Okayama-shi, Okayama; Sumie Ikezaki & Michiko Kato, Health Sociology, The University of Tokyo, Bunkyo-ku; Yumiko Inoue & Atsushi Fukuuchi, Mitsui Memorial Hospital, Chiyoda-ku, Tokyo; Izou Kimijima, Breast Center, North-Fukushima Medical Center, Date-cho, Date-Gun, Fukushima; Miyako Takahashi, Social Gerontology, The University of Tokyo, Bunkyo-ku, Tokyo, Japan Purpose: With the aim of measuring and evaluating QOL in breast cancer patients receiving endocrine treatment, this study examines the way psychosocial factors, such as family relationships, influence patient QOL and discusses issues related to the design of future quantitative research in this area. The study was conducted in order to develop a new QOL measurement for patients undergoing endocrine therapy that will be used in upcoming clinical trials (N-SAS BC 03). Method: Semi-structured interviews were conducted with 34 breast cancer patients living in urban (Tokyo) and suburban (Fukushima) areas of Japan and currently receiving hormone therapy (mean age,

57.8). Each interview was transcribed and utterances that related to family were examined for factors influencing patient QOL. The average household size was 3.5 persons (Tokyo, 2.5; Fukushima, 4.5). Findings and conclusion: In order to adequately measure the QOL of cancer patients under endocrine therapy, especially at a national level, this study indicates the importance of formulating research programs with the following three issues in mind: ‘Family member roles’. In large extended families, a patient held responsibilities (such as tending to parents or grandchildren) which tended to influence her quality of life as well as her cancer symptoms. ‘Relationships with family members over time’. The onset of breast cancer redefined relationships between family members and reshaped family cohesion. These changes had an influence not only in terms of a patient’s mental and physical support, but also in terms of her disease and her attitudes toward life and death. This indicates the importance of examining ties between QOL and family relationships from a longitudinal, as well as crosssectional, perspective. ‘Reciprocity between family members’. The onset of breast cancer in post-menopausal women occurs at middleage, at a time when their husbands are often themselves suffering from disease. This aspect of spousal relationships influences a patient’s mental and physical condition. # 1108/HIP-P: THE PREOPERATIVE HOSPITAL INFORMATION PROGRAM FOR CANCER PATIENTS: FIRST RESULTS Maria Berend & Janna Bents, Reference Center Quality of Life in Oncology; Thomas Birkner & Regine Sagermann, Department of General and Thoracic Surgery; Thomas Ku¨chler, Reference Center Quality of Life in Oncology; Bernd Kremer, Department of General and Thoracic Surgery, Universita¨tsklinikum Schleswig-Holstein, Campus Kiel, Kiel, Schleswig-Holstein, Germany Purpose of research: Aim of this ongoing study is to evaluate a preoperative hospital information program for cancer patients and their relatives awaiting surgery. Background to this clinical initiative: (1) The high pressure to economize hospital processes over the last few years in Germany led to dramatic changes in surgical hospital politics: almost all of the preoperative diagnostics is now done in outpatient settings. Since the patients arrive on the surgical wards just the day before the operation is scheduled new strategies for preoperative information management have to be tested. (2) A prestudy showed that the subjective perception of received preoperative information has a strong impact on QoL and patients overall treatment satisfaction. Subject sample and statement of methods: All patients with a (prelim.) diagnosis of GI or lung-cancer scheduled for further preop. diagnostics are invited to attend a 2-hour-session on general aspects of cancer, surgical treatment and psychosocial support. The information is given by a surgical resident and a psychotherapist. About half of the 2 hours is received for patients questions. Evaluation: Additional to the general assessment of QoL and treatment strain qualitative data (postoperative interviews with all invited patients) are collected. The comparison is between participants and non-participants. Summary of results: Of the invited sample about 40% attended the informational sessions. The first and spontaneous reaction of the participants and their relatives (95% spouses, 5% children) after the sessions were 100% positive. The postoperative interview showed in detail, that the participating patients felt less helpless, more relaxed and participated more frequently in the psychosocial support program. Furthermore the subjective level of ‘feeling informed about all aspects of cancer surgery’ was higher in the participants. The first results of the program evaluation are very encouraging, since patients rate such a relatively simple intervention (2-hour-session) as extremely helpful in coping with their cancer diagnosis and the subsequent surgical treatment.

813 # 1494/HEALTH-RELATED QUALITY OF LIFE FIVE-YEARS AFTER DIAGNOSIS OF LARYNGEAL CARCINOMA Eva B. Hammerlid & Ewa Silander, Department of Otorhinolaryngology Head&Neck Surgery, Sahlgrenska University Hospital, Go¨teborg; Mats Nordgren, Magnus Jannert, Department of Otorhinolaryngology Head&Neck Surgery, Malmo¨ University Hospital, Malmo¨, Sweden; Kristin Bjordal, Department of Radiation Oncology, The Norwegian Radium Hospita; Morten Boysen, Department of Otolaryngology Head&Neck Surgery, Rikshospitalet, Oslo; Helmut Abendstein, Department of Otolaryngology Head&Neck Surgery, St Olavs Hospital, Trondheim, Norway Purpose: To evaluate the health-related quality of life (HRQL) of patients with laryngeal carcinoma in a prospective longitudinal multicenter study at diagnosis, after 1 and 5 years in relation to tumor location and treatment modality. Subjects and method: Eighty six patients (mean age 66 years; 72% males) with laryngeal carcinoma were evaluated with standardized HRQL questionnaires; the European Organization for Research and Treatment of Cancer, Quality of Life Questionnaire Core-30 (EORTC QLQ-C30), the EORTC QLQ-Head and Neck Cancer Module (EORTC QLQ-H&N35), and the Hospital Anxiety and Depression Scale (HADS). Results: Some significant changes in HRQL were found between diagnosis and 5 years after diagnosis, depending on the treatment given. The patients ability to speak improved significantly while five scales and single items showed both clinical and statistical significant deterioration (physical functioning, role functioning, problems with dyspnoea, dry mouth and sticky saliva). When comparing HRQL at 1 and 5 years after diagnosis it appears that most values at the 1-year follow-up assessment persist until five years, but a few deteriorated significantly (physical functioning, role functioning and problems with social eating). Patients who survived 5 years had better HRQL at diagnosis than those who died during the study (21 out of 28 scales and single items showed a both clinically and statistically significant difference at diagnosis). The HRQL at diagnosis seems to be associated with survival rate after 5 years and the global quality of life scale at diagnosis tends to predict HRQL after 5 years. Conclusions: The use of HRQL questionnaires is valuable when comparing different treatments and as an aid in predicting treatment side-effects. Evaluation of HRQL at diagnosis for patients with laryngeal carcinoma seems to be of value for the prognosis of HRQL over time and for the prognosis of survival. # 1365/EORTC QLQ-C30 vs. FUNCTIONAL LIVING INDEX FOR EMESIS (FLIE) IN SYMPTOM CONTROL STUDIES Benny Zee, Tony Mok, Maria Lai, Karen Chak, Jam Jun Lee, Zania Li & Winnie Yeo, Department of Clinical Oncology, Chinese University of Hong Kong, Shatin, Hong Kong Background: European Organization of Research and Treatment of Cancer (EORTC) Quality of Life Core Questionnaire (QLQ-C30) has been widely used in symptom control trials. The sensitivity of this instrument for chemotherapy induced nausea and vomiting (CINV) has not been fully assessed. FLIE, an emesis-specific questionnaire, was used to assess how nausea and vomiting affects patients’ quality of life (QOL). Objectives: (1) To assess if individual FLIE items correlate with the corresponding QLQ-C30 domains change scores; (2) To determine if overall QOL as measured by EuroQoL Visual Analogue Scale (EQVAS) correlates with FLIE items and with corresponding QLQ-C30

domains. Methods: A total of 46 cancer patients were entered. Daily diary for 5 days was used to obtain EQVAS. A change score from baseline to day 5 was used for QLQ-C30 and FLIE. Spearman rank correlations were used to assess the associations among QOL domains and FLIE items. Results: The QLQ-C30 and FLIE nausea (N) (#1) and vomiting (V) (#10) has correlations of 0.87 (p < 0.0001) and 0.86 (p < 0.0001) respectively with QLQ-C30 NandV domain. The four FLIE appetite items (#4,5 for N and #13,14 for V) have correlations of 0.49 (p ¼ 0.0006), 0.52 (p ¼ 0.0003) and 0.54 (p ¼ 0.0002), 0.51 (p ¼ 0.0004), respectively with QLQ-C30 appetite loss. The FLIE emotional functioning items (#9 for N) has a correlation of 0.36 (p ¼ 0.0137) with QLQ-C30 emotional functioning. QLQ-C30 global QOL significantly correlates with all nine FLIE N items (r ¼ )0.33 to )0.53) and with six out of nine FLIE V items. The QLQ-C30 NandV domain correlates with all 18 FLIE items (p < 0.0001). The EQVAS correlates with QLQ-C30 N&V (r ¼ )0.38, p ¼ 0.0103) and global QOL (r ¼ 0.41, p ¼ 0.0051). The EQVAS also correlates with 17 out of 18 FLIE items (r ¼ )0.34 to )0.67, p < 0.0262) except item #11 with r ¼ )0.30 (p ¼ 0.0511). Conclusion: The QLQ-C30 N&V, appetite loss and global QOL were sensitive to the effect of CINV as detected by FLIE. The overall QOL as measured by EQVAS also pointed to a similar conclusion. (Study supported by Merck Sharp & Dohme (Asia) Ltd.). # 1236/THE IMPACT OF ROUTINE ASSESSMENT OF INFORMATION NEEDS ON HRQOL OF PATIENTS WITH CANCER A. Al-Mohammadi & M.S. Salek, WSP Centre for Socioeconomic Research; T. Maughan & M. Mason, UWCM; P.J. Nicholls, Welsh School of Pharmacy, Cardiff University, Cardiff, UK Communicating medical information is one of the crucial foundations that have an effect on a patient’s treatment and care package. Several studies have shown that patients are diffident and different in their information seeking behaviours. Moreover, studies showed a discrepancy between patients with cancer and their clinicians views of patients information needs and the resultant outcome. A systematic and practical measurement of patient preferences/expectations from the communicated information is warranted. Therefore, the aim of this study was to assess the impact of routine assessment of patients’ needs for detailed medical information on patient QoL. In this RCT, all patients (n ¼ 105) at baseline completed a Patient Communication Profile (PCP) immediately after an outpatient visit to assess information needs, understanding and satisfaction and mental distress. For the study group (n ¼ 64), the counselling physicians were shown their patients responses prior to each encounter. The PCP was repeated after the next two encounters. Health-related quality of life (HRQoL) was assessed at baseline and final encounter using the revised McGill QoL Questionnaire (rMQOL). Ninety-eight patients remained at the final follow-up visit. There were no significant differences between the control and study group in any factor at the baseline assessment. Patients whose doctors received feedback showed greater satisfaction with communication at the final encounter (p ¼ 0.017). Also, within group analysis showed remarkable improvement in the psychological domain (p < 0.01), existential domain (p ¼ 0.08) and support domain (but not reaching statistical significance) of the rMQOL. Routine assessment of patients’ information needs improves the QoL of patients with cancer.

814 # 1831/ASSESSING THE SUPPORTIVE CARE NEEDS OF PATIENTS WHO ATTEND A CANCER CENTRE Judith R. Davidson, Leanne Findlay, Lisa Zetes-Zanatta & Michael Brundage, Cancer Care and Epidemiology, Queen’s Cancer Research Institute; Anne Smith, Kingston Regional Cancer Centre, Kingston, Ontario, Canada Background: In order to ensure that patients receive appropriate resources to meet their emotional, informational, psychological, spiritual, and practical needs, their unmet or under-met needs in these areas must first be identified. Existing questionnaires used for this purpose were found to either lack empirical validation or to have inconsistent response categories. Purpose: To systematically develop a questionnaire to assess the supportive care needs of cancer centre patients, and to test its feasibility. Methods: First, the possible domains of needs were recorded. This included consideration of the need domains reported in the literature, and areas that the investigators and administrators anticipated would be of interest. A list of possible items was prepared covering each of the domains. Second, these items were reviewed by patients, cancer centre staff (representing 17 disciplines or types of work), and cancer care providers in the community, who checked for relevance and suggested additions or deletions. Third, the content was reviewed and synthesized by a panel whose members had expertise in cancer care, health services research, and in questionnaire design. Fourth, the revised questionnaire underwent three phases of pilot testing to assess the ease of completion, completion time, item comprehension, and the efficiency of mail-out methods, data retrieval, data entry, and data analysis. The final questionnaire has 52 items which cover five areas: (1) needs for symptom control; (2) needs for information; (3) needs for support services; (4) impressions of care/services at the cancer centre; and (5) impressions of coordination of care. The last page contains demographic items. Time to completion is 12–20 min. The questionnaire is undergoing validation testing. Response to the survey: The questionnaire was mailed (using the Dillman method) to 1740 outpatients between December 2002 and January 2003. By March 31, 2003, the response rate was 81%, and the completion rate was 75%. Conclusion: The questionnaire appears to be a feasible instrument for assessing the supportive care needs of cancer centre patients. # 1758/THE IMPACT OF PULMONARY ARTERIAL HYPERTENSION (PAH) ON PATIENTS: RESULTS FROM QUALITATIVE PATIENT INTERVIEWS Natalie Doughty, PVD Unit, Papworth Hospital, Cambridge; Stephen P. McKenna, Lynda C. Doward & David M. Meads, Galen Research, Manchester Science Park, Manchester; Joanna Pepke-Zaba, PVD Unit, Papworth Hospital NHS Trust, Cambridge, Cambridgeshire, UK Few patient-reported or clinical outcome measures are available for use in studies of PAH. Two new complementary PAH-specific instruments are in development; (1) assessing symptoms and functioning and the other needs-based QoL. The first stage of the study involved patient interviews, the results of which are reported below. Qualitative, unstructured interviews were conducted with 35 relevant patients recruited from a major clinical centre in the south of England. The aim of the interviews was to explore the impact of the condition and its treatment on all aspects of the interviewees’ lives. An unstructured format was adopted with interviewees encouraged to talk at length about issues of importance to them. The interviews were audio-recorded, transcribed verbatim and then content analysed independently by four experienced researchers. A consensus meeting was then convened. Interviews were conducted with nine male and 26 female patients aged 20–80 (mean 50.0, SD ¼ 16.5) years. Interviews varied in length from 30 min to 2 hours. In addition to the physical impact of the disease (breathlessness on exertion, fatigue, oedema, palpitations, problems with mobility and carrying items, etc.) the following issues were considered particularly important by patients: Communication; loss of social contacts, lack of understanding from others about the nature of the disease. Personal relationships; worries about dependency on family and friends, sexual activity, difficulty joining in social activities, feelings of isolation, concerns about appearance. Difficulty leaving the home; necessity for forward planning and limitations imposed by the disease and its treatment. The interviews provided a rich source of information about the nature of the impact of

PAH and the concerns of affected individuals. This forms the basis of the content of the new instruments which are currently being validated with new samples of PAH patients. # 1838/BEING ELDERLY AND HAVING QUALITY OF LIFE: REPRESENTATIONS OF ELDERLIES LIVING IN CITIES OF THE SOUTH MINAS GERAIS STATE – BRAZIL Jjose Vitor Da Dilva, Medical Surgical Nursing, School of Nursing of Itajuba´, Itajuba´, Minas Gerais; Miako Kimura, Medical Surgical Nursing, School of Nursing – University of Sa˜o Paulo, Sa˜o Paulo; Fernando Lefevre, Public Health School – University of Sa˜o Paulo, Sa˜o Paulo, Brazil This is a qualitative research whose objectives were to identify the representations of the elderly regarding the meaning of ‘being an old person’ and the concept of ‘quality of life’. Fifty elderlies, both male and female, were interviewed. They were from three cities in the South of the estate of Minas Gerais-Brazil: Itajuba´, Pouso Alegre and Santa Rita. Data collection was carried out through a semi-structured interview, which was tape-recorded and literally transcribed. The methodological framework of the Collective Subject’s Speech, proposed by Lefe`vre, were used for the selection of the Central Ideas and of the corresponding key-expressions, from which the elderlies’ speeches were extracted. The following representations in relation to ‘being an old person’ were identified: a person presenting both functional capacity and inability; an old-aged person who has a good health and physical disposition to work; someone who has considerable life experience; a person who has lived for many years and fulfilled the goals of his/her life. The representations regarding the concept of ‘quality of life’ were structured within the following dimensions: physical health; physical independence; economic, psychological and social; autonomy; family; religiousness and transcendence; environment and citizenship. This study confirms the multidimensional and multidetermined character of the phenomena, related to several objective and subjective aspects of human life. # 1294/QUALITY OF LIFE IN OLD ADULTS: A PERSPECTIVE FROM FOCAL GROUPS IN BRAZIL Clarissa Trentini, Eduardo Chachamovich & Marcelo P. Fleck, Psiquiatria e Medicina Legal, Hospital de Clinicas de Porto Alegre, Porto Alegre, Rio Grande do Sul, Brasil Attention towards conditions that allow a satisfactory quality of life in elderly, as well as variations or restrictions brought by ageing, are issues of scientific and social relevance. Objectives: To describe the WHOQOL-OLD project and present the results from the focal groups in Brazil, as well as the methodology applied in those groups. Methods: The methodology and the strategy used in WHOQOL-OLD project are based on several steps defined and applied in the development of others WHOQOL modules in Brazil. Five focal groups were conducted: healthy old adults aged between 60 and 80, unhealthy old adults aged between 60 and 80, healthy old adults aged more than 80, unhealthy old adults aged more than 80, and careers. It was a convenience sample. Eighteen old adults and five careers took part in the focal groups. The subjective perception regarding one own health status was used to classify him\her as healthy or unhealthy. The methodology followed is presented. Results: The notion that QOL is a multidimensional construct is clear. There is a trend regarding the association between quality of life and well-being or good feelings. Spontaneously, responses included all the six domains proposed by WHOQOL-100. Nineteen out of 24 original facets of this instrument were cited as relevant for the quality of life of old adults, and the five facets not remembered did not concentrated in one only domain. When questioned about the importance of the WHOQOL-100 facets to quality of life in old people, all of them were considered relevant. However, the facets body image and appearance, work capacity, negative feelings, sexual activity and dependence of medication or treatment received suggestions of alterations. Additional items were also examined and considered relevant to the assessment of quality of life in old adults by the old adults themselves and by the careers. Conclusions: Results strength the hypothesis suggesting that old adults are a particular group, and therefore have specific issues which are relevant for their quality of life. An adequate instrument to assess quality of life in old adults should unavoidably investigate these issues.

815 # 1180/FACTORS AFFECTING ADHERENCE TO ASTHMA TREATMENT: PATIENT AND PHYSICIAN PERSPECTIVE Ann-Christin Mork, Clinical Science, AstraZeneca R&D Lund, Lund, Sweden; Eric van Ganse, Unite´ de Pharmacoe´pide´miologie, Centre Hospitalier Lyon-Sud (CHLS), Pierre-benite cedex, France; Elisabeth Stahl, Clinical Science, AstraZeneca R&D Lund, LUND, Sweden Adherence to asthma treatment is not optimal and may be a significant factor contributing to adverse outcomes, reduced health-related quality of life and high costs of the disease. The aim was to identify important factors affecting treatment adherence of patients with asthma that can be used in patient–physician communication to improve adherence. Information from literature and interviews with 12 pulmonologists (France, Spain and the UK) and 46 asthma patients was obtained. Factors affecting adherence to asthma treatment were identified, reviewed and a flow chart developed to indicate the relationship between key factors. Main factors influencing adherence included: the patient– physician relationship; the patient’s understanding of the disease and its treatment; the patient’s beliefs and perception of the disease and its treatment, and, importantly the patient’s willingness to take active part in his/her asthma management. The results confirm the necessity for patients to appreciate the benefits of regular controller treatment and to make the medications simple and convenient to use. For optimal usage of medicines, it is desirable to achieve a situation in which physicians and patients work in partnership to manage a disease. In conclusion, patient adherence to asthma treatment can be improved, and the likelihood of treatment success increased, by paying attention to the factors that influence patients’ willingness to participate in their treatment. # 1464/QUALITATIVE ANALYSIS OF APPETITE LOSS AFTER SURGERY FOR OESOPHAGO-GASTRIC CANCER Jane M. Blazeby, University Division of Surgery, Bristol Royal Infirmary, David M. Wainwright, Department of Social Medicine, University of Bristol; Vas Kavadas, University Division of Surgery, Bristol Royal Infirmary; Helen Cramer, Department of Social Medicine, University of Bristol; Derek Alderson, University Division of Surgery, Bristol Royal Infirmary; Jenny Donovan, Department of Social Medicine, University of Bristol, Bristol, Avon, United Kingdom Potentially curative surgery for oesophago-gastric cancer has a major impact on patients’ quality of life and frequently leads to decreased food intake and weight loss. This qualitative study explored patients’ experiences of appetite loss after surgery. Consecutive patients between 3 and 12 months after oesophagectomy or gastrectomy were screened for appetite loss with the EORTC QLQ-C30. Fourteen patients with scores of ‘quite a lot’ or ‘very much’ were selected for semistructured interviews conducted at the patients’ home. Interviews were tape recorded, transcribed and their content was coded. Analytical themes were identified through constant comparison. Maintaining or increasing body weight post-operatively was a key objective for many informants, both as a means of returning to pre-illness levels of physical performance and of rejecting the identity of the terminal cancer patient. Despite this impetus, the objective was difficult to achieve because physiological sensations of hunger, satiety, swallowing, the taste and smell of food were radically changed necessitating a re-mapping of the body and the adoption of new patterns of eating. Successful adaptation and coping were influenced by beliefs about body image and support from the family. The QLQ-C30 may be used to screen for problems after surgery and provision of timely and accurate information from clinicians and fellow sufferers may aid recovery. # 1189/LIFE WELL-BEING FOR OUTPATIENTS WITH CANCERVALIDATION OF A QUALITY OF LIFE (QOL) QUESTIONNAIRE, CARE NOTEBOOK Kunihiko Kobayashi & Mari Shimonagayoshi, Outpatient Clinic, Saitama Cancer Center, Kita-adachi, Saitama; Noriaki Kanemoto, Ryu Kasai, Yoichi Itoh & Yasuo Ohashi, School of Health Science and Nursing; Joseph Green, Student Exchange Division, University of Tokyo, Bunkyo, Tokyo, Japan; Carolyn C. Gotay, Cancer Research Center of Hawaii, University of Hawaii, Honolulu, HI, USA

Although we have developed Japanese versions of QLQ-C30 and FACT-G in cooperation with EORTC and CORE, respectively (Eur J Cancer 1998; 34: 810; QOL Res 2001; 10: 701), it is somewhat difficult to answer them every week in the fields of daily clinic and palliative care. In this study, a QOL questionnaire, Care Notebook, was developed and validated. The Care Notebook has 24 items, which are considered universally to be important questions consisted of one word or a phrase, and employs 11-point linear analogue scale, to seek good acceptability. Of the 266 cancer patients (pts) who used our outpatients clinic in March 2001, 249 pts (Male/female/missing:139/ 101/9 pts; Age of –49/50–69/70–: 47/152/41 pts; ECOG PS 0/1/2– :177/55/16 pts; Cancer disclosure ±: 238/8 pts; Breast/lung/gastrointestinal cancers: 70/50/40 pts) answered Care Notebook, QLQ-C30 and FACT-Sp twice at 4-week interval. Care Notebook was well accepted in patients with PS 0-1. The VARCLUS procedure on SAS program showed a dendrogram, which was constructed by three major scales (Physical Well-Being, Mental Well-Being and Life Well-Being) at a level of 0.55 of proportion of variance explained. At 0.75 of that, Life Well-Being was divided into three subscales of Daily Functioning, Social Functioning, and Global QOL Functioning. Multitrait scaling analysis confirmed convergent validity and discriminant validity of these scales and subscales. Internal consistency by Cronbach’s a and test–retest reliability showed favorable results. Known-groups comparison showed the expected clinical validity with PS. Concurrent validity with QLQ-C30 and FACT-Sp was satisfied. In conclusion, through analyzing the VARCLUS procedure, we found a major scale of Life Well-Being, which was constructed by Daily Functioning, Social Functioning, and Global QOL Functioning. Care Notebook was found to be valid and useful to outpatients with cancer. # 1657/EVALUATING THE LIFE QUALITY OF CHILDREN WITH EPILEPSY BASED ON PEDSQL Dorota Talarska, Faculty of Health Sciences, University of Medical Sciences, PoznaD, Poland Children with epilepsy are exposed to many difficulties resulting both from the illness process and from the therapy. The usage of antiepileptic medicines influences various life spheres, especially behavior, cognitive functions, and the mood. The main aim of the research was to achieve an answer to the following question: What is the life quality of children and youth with epilepsy? The research included 87 children with epilepsy treated at the Clinic and Institute of Development Age Neurology at the University of Medical Sciences in PoznaD. The research tool was a two-part questionnaire: PedsQL – Pediatric Quality of Life Inventory. One part is devoted to children, whereas the other one is for parents. The questions of both parts are equal. The questionnaire analyses six spheres: biological functioning, emotional functioning, social functioning, functioning at school, mood, a general evaluation of health condition. In order to verify the collected data, tests U by Mann–Whitney and Wilcoxon were used. Two subgroups were separated out of the researched group: age of 8–14/50%/ and 15–20/50%/. The ratio of girls and boys was 1:1. In the PedsQL questionnaire children evaluated the sphere concerning emotions and functioning at school at the lowest level. Their best evaluation concerned the sphere of well-being and the physical activity. Parents evaluated children’s physical activity at a equally high level. The discrepant evaluations concerned contacts with peers and functioning at school (in parents’ opinion it was much higher). In the analysis of life quality the following independent variables were also distinguished: age, sex, the frequency of seizures and the influence of all the spheres on the global evaluation of life quality. In the biological aspect, the life quality of children with epilepsy is influenced by the frequency of seizures psychosocial difficulties in studies and acceptance by the peers.

816 # 1640/VALIDITY OF THE SPANISH VERSION OF THE MINI ASTHMA QUALITY OF LIFE QUESTIONNAIRE Gonzalo Nocea, Outcomes Research, Merck, Sharp & Dohme de Espan˜a, Madrid; Luis Prieto, Departamento de Alergologı´a, Hosp. Univ. Doctor Pesset, Valencia; Miguel Hinojosa, Departamento de Alergologı´a, Hosp. Ramo´n y Cajal, Madrid; Fernando Florido, Departamento de Alergologı´a, Complejo Hosp. Ciudad deJaen, Jaen; Tamin Malek, Departamento de Alergologı´a, Hosp. Gral de Castello´, Castello´; Teresa Caloto, Outcomes Research, Merck, Sharp & Dohme de Espan˜a, Madrid, Spain The objective of the study was to analyze psychometric properties of the Spanish version of the Juniper Mini Asthma Quality of Life Questionnaire (Mini-AQLQ) for use in clinical practice. In this study 275 patients were included (61% women, mean age 37 years) with a diagnosis of mild to moderate asthma (according to the GINA classification) treated at 116 Spanish hospitals in a standard clinical practice setting. The patients’ socio-demographic and clinical characteristics were recorded and all patients were administered the Mini-AQLQ and EuroQoL 5D (EQ-5D) questionnaires twice: at recruitment and after 2 months in patients with uncontrolled asthma (patients with symptomatology and/or need for short-acting b2 agonists), and at recruitment and after 15 days in patients with controlled asthma. Montelukast was added to the patients with uncontrolled asthma at the baseline visit. The validity, reliability and sensitivity to change of the Mini-AQLQ were assessed. The ceiling effect for the different domains ranged from 1.5 to 7.7% and there was no floor effect in any domain. The Mini-AQLQ showed a high internal consistency (Cronbach’s a 0.85 for total score) and a moderate 2-week reproducibility (Intraclass correlation coefficient 0.63 for total score). In univariate analyses, significant differences in Mini-AQLQ scores were observed according to gender, disease severity according to GINA, symptoms and use of on-demand short acting b2 agonists. The cross-sectional and longitudinal correlations were strong between Mini-AQLQ and EQ-5D as well as between Mini-AQLQ and symptoms. The Pearson correlation between Mini-AQLQ and FEV1 (% of predicted value) was weak (0.08). Changes in Mini-AQLQ scores were significantly different in patients who improved, compared with patients who either remained stable or deteriorated. The effect size between both visits was 0.91 for total score and ranged from 0.5 to 1.0 for the four domains. The Spanish version of the Mini-AQLQ is reliable, valid and sensitive to changes, making it an appropriate way to assess quality of life in adults with mild to moderate asthma. # 1489/CONSTRUCT VALIDITY OF THE HEALTH UTILITIES INDEX MARK 3 (HUI3) IN PATIENTS WITH SCHIZOPHRENIA Boon-Kheng Seng, Department of Medical Social Work, Institute of Mental Health; Nan Luo & Shu-Chuen Li, Department of Pharmacy, National University of Singapore; Julian Thumboo, Department of Rheumatology and Immunology, Singapore General Hospital, Singapore The Health Utilities Index Mark 3 (HUI3) is a widely used utility-based health-related quality of life (HRQoL) instrument, whose construct validity in schizophrenia has not been evaluated. We therefore investigated the construct validity of the HUI3 in Singaporeans with schizophrenia. A consecutive sample of outpatients with schizophrenia seen in a tertiary-level mental health hospital in Singapore were interviewed using the HUI3 health classification system (8-item version) and for assessing psychiatric symptoms. Subjects also completed an identical English or Chinese questionnaire containing the Schizo-

phrenia Quality of Life Scale (SQLS, a disease-specific HRQoL instrument), the Short Form 36 Health Survey (SF-36, a generic HRQoL instrument) and assessing socio-demographic status. Construct validity of the HUI3 was assessed by examining correlations between HUI3 multi-attribute utility scores and other variables and by comparing HUI3 mean scores between the subjects and Singaporean general population. Subjects (n ¼ 200, 104 female, 139 completing the English HUI3) were mainly ethnic Chinese (83.5%, Malay: 10.0%, Indian: 5.5%) with a mean (SD) age of 37.9 (10.2) years. HUI3 utility scores were moderately or strongly correlated with SQLS scale scores (Spearman’s rho: )0.56 to )0.36, p < 0.001), SF-36 physical and mental health summary scale scores (Spearman’s rho: )0.30 and )0.51, respectively, p < 0.001) and the number of psychiatric symptoms (Spearman’s rho: )0.50, p < 0.001). Reductions in mean scores of the subjects from (unadjusted) Singaporean population norms were 0.07 points for multi-attribute utility scores and 0.03, 0.09, and 0.04 points for speech, emotion and cognition single-attribute utility scores respectively (p < 0.001 for all, two-sample t-tests). Differences in mean scores between the two groups were no larger than 0.02 points for other single-attribute utility scores. This study provides preliminary evidence for the construct validity of the HUI3 utility scores in patients with schizophrenia in Singapore. # 1501/SENSITIVITY AND VALIDITY OF THE NEW EATING DISORDERS RELATED QUALITY OF LIFE QUESTIONNAIRE Carlota Las Hayas, Begon˜a Urresti, Arantza Madrazo & Francis E. Cook, Research Unit, Galdakao Hospital, Galdakao, Bizkaia, Spain The objective of this study was to assess the sensitivity and validity of the Eating Disorders Quality of Life Questionnaire (ED-QLQ) using the method of known-groups validation. The ED-QLQ was applied to a sample of 358 eating disorders patients and to 305 students to conform the control group. Psychiatrists rated the severity of the disorder using the Clinical Global Index and diagnosed the eating disorder subtype according to the DSM-IV criteria for each patient. Differences in quality of life scores were hypothesised among groups who differed in presence/absence of the disorder, severity of the disorder and eating disorder diagnosis. Mean ED-QLQ scores for each subscale were calculated and compared using T-test and ANOVA (Scheffe´). Scores for each domain were standardised to range from 0 to 100. Differences between the control group and patients aroused in all domains. Meanwhile control group mean scores were 6.6 on the Restrictive Behaviors domain and 35.2 on the Mental Health domain, patients’ mean scores were 29.8 and 51.4, respectively on the same domains. With regard to the discrimination among the three severity groups, differences were also found in all domains. The domain mean scores for the minor severity patients ranged from 11 to 39, for the moderately severe patients ranged from 29 to 57, and finally for the most severe patients, mean scores ranged from 37 to 67. When comparing the scores of the patients diagnosed with Bulimia Nervosa (BN) with the Anorexia Nervosa (AN) and with the Eating Disorder Not Otherwise Specified (EDNOS) patients, it was confirmed that patients diagnosed with Bulimia Nervosa had a statistically significant worse quality of life perception in the Physical Symptoms (BN mean score of 53.6 vs. 42.8 (AN) and 41.1 (EDNOS)) and body image domains (BN mean score of 70 vs. 57.7 (AN) and 61.9 (EDNOS)). Bulimia Nervosa patients had higher scores, meaning worse health status, in all the other domains though these differences were not statistically significant. The results confirm the sensitivity of the instrument to differentiate groups of patients and add psychometric validity to it.

817 # 1483/FEASIBILITY OF RETROSPECTIVE PSYCHOMETRIC VALIDATION: EXAMPLE OF THE INDEVUS URGENCY SEVERITY SCALE (IUSS) FOR PATIENTS WITH OVERACTIVE BLADDER Annabel Bowden, The Lewin Group, Quintiles UK Ltd, Bracknell, UK; Shoshana Colman & Hany Zayed, Quintiles Late Phase, San Francisco, CA, USA This study established the feasibility of a retrospective psychometric validation of the IUSS, a new patient-reported outcome assessment instrument for patients with overactive bladder (OAB). The IUSS required a ‘quality assessment’ to ensure it is a valid, reliable and responsive assessment tool. A feasibility assessment was undertaken to determine the extent to which data collected during a clinical trial could be used for the psychometric assessment of the IUSS. It was concluded that the construct, content, and known-groups validity, test– retest reliability, respondent burden and responsiveness of the IUSS could be investigated retrospectively. The psychometric assessment of the IUSS was performed on data collected during a 12-week clinical trial of trospium chloride in patients with OAB associated with predominant urge incontinence. It was concluded that this analysis had the benefit of utilizing data collected within the careful guidelines of the clinical trial study protocol, ensuring that data was of the highest standard. Data of this quality is not always available to those developing and testing new patient reported outcome measures. However, a limitation of the retrospective approach is the lack of opportunity for those undertaking the psychometric assessment to provide input into the design of the data collection. As a consequence, retrospective validation is not a ‘perfect’ approach, but the best that can be achieved with the available data. However, the use of clinical trial data in this study enabled key psychometric properties to be assessed. The IUSS demonstrated good known groups validity, responsiveness to change (effect size 0.49–1.39), good test–retest reliability (0.80), and moderate correlation with frequency of toilet voids, urge incontinence episodes and with IIQ scores. The content validity of the IUSS was established as well as its’ low respondent burden. Thus this study demonstrated the IUSS to be psychometrically sound in terms of construct, content and known groups validity, test–retest reliability and to be responsive to change. # 1042/PSYCHOMETRIC VALIDATION OF THE NOCTURIA QUALITY OF LIFE QUESTIONNAIRE (N-QOL) Lucy Abraham & Asha Hareendran, Outcomes Research; Ian W. Mills, Clinical Development, Pfizer Ltd, Sandwich, UK; Mona Martin, Health Research Associates, Seattle, WA, USA; Paul H. Abrams, Bristol Urological Institute, Southmead Hospital, Bristol; Ruaraidh P. MacDonagh, Urology Department Taunton & Somerset Hospital, Taunton; Marcus J. Drake, Urology Department, Freeman Hospital, Newcastleupon-Tyne; Jeremy G. Noble, Urology Department, Acland Hospital, Oxford, UK Nocturia is a common symptom among the elderly, characterised by the need to wake up at night to void. An increase in the number of times that a person has to interrupt their sleep to get up to urinate will have a negative impact on their Quality of Life. The aim of this study was to develop and assess the psychometric properties of a nocturiaspecific QoL questionnaire. To explore the impact of nocturia, focus groups were held with men with this condition. Based on the transcripts of the focus groups, a 31-item Nocturia Quality of Life Questionnaire (N-QoL) was developed. The items covered impacts on psychological well-being, sleep, relationships, activities and travel, physical energy and concentration. To psychometrically validate the N-QoL, a methods study was conducted with 107 men with nocturia from four UK urology clinics. Item reduction analyses for overlap and redundancy suggested that 18 items be dropped. Factor analysis of the remaining 13-items revealed two underlying domains sleep/energy and bother/concern, all loading at 0.5 and above. Excellent internal consistency was demonstrated, with Cronbach’s as for the total score and domains ranging from 0.84 to 0.90. The scale also showed excellent test–retest reliability with intraclass correlation coefficients ranging from 0.74 to 0.82. Convergent validity was good, with scores being highly related to sleep quality as measured by the Pittsburgh Sleep Quality Index and energy/vitality as measured by the SF36. The N-QoL also demonstrated excellent discriminant validity with those with more severe nocturia (having to get up at night to void three times

or more) having significantly poorer scores on the N-QoL than those with mild and moderate nocturia. These analyses provide support for the psychometric validity of the N-QoL for use in a male population with nocturia. # 1590/KOREAN TRANSLATION AND VALIDATION OF BATH ANKYLOSING SPONDYLITIS DISEASE ACTIVITY INDEX (BASDAI): A PILOT TEST Hye J. Park, Nursing; Sehyun Kim, Preventive Medicine; Jong E. Lee, Nursing, College Of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do, Korea; Jae B. Jun & Sang C. Bae, Rheumatology, Hanyang University, Seoul, Korea Ankylosing spondylitis (AS) is the form of rheumatoid arthritis affecting the spine. It occurs predominantly in young males and produces pain and stiffness as a result of inflammation of the sacroiliac, intervertebral, and costovertebral joints. The objective of this study was to validate a Korean translation of the Bath Ankylosing Spodylitis Disease Activity Index (BASDAI) instrument. Two bilingual Korean native speakers translated the original English version of BASDAI with the permission of the authors. To evaluate the conceptual equivalence and consistency of the items and scales, reconciliation and focus group discussion were performed. After completing forward translation, a pilot test was conducted for cognitive testing. The Korean translated BASDAI instrument was administered to 15 outpatients with AS. Of the 15 subjects, 13 (80.0%) were men with a mean(±SD) age of 27.7 (±6.28). Cronbach’s as for the internal consistency reliability of five items (fatigue, axial pain, peripheral pain, discomfort, stiff/swelling) were 0.82, 0.81, 0.82, 0.84 and 0.77, respectively. The mean duration of morning stiffness was 53.33 (SD ¼ 37.50) min. The stiff/swelling item was highly correlated with fatigue/tiredness, axial pain, overall level of peripheral pain, and discomfort (r ¼ 0.69, p < 0.01; r ¼ 0.64, p < 0.01; r ¼ 0.52, p < 0.05; r ¼ 0.60; p < 0.01). The score of fatigue positively correlated with axial pain (r ¼ 0.85, p < 0.0001). The score of peripheral pain positively correlated with discomfort (r ¼ 0.69, p < 0.01). These results of the cognitive testing support the applications of the Korean BASDAI for the field test evaluating the Korean AS patients. # 1586/KOREAN TRANSLATION AND VALIDATION OF THE KIDNEY DISEASE QUALITY OF LIFE INSTRUMENT (KDQOLTM-36) Hye J. Park, Nursing; Sehyun Kim, Preventive Medicine, College of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do, Korea; Jin S. Yong & Sung S. Han, Nursing, Catholic University, Seoul; Dong H. Yang & Kyeong S. Kim, Nephrology, College of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do, Korea Assessment of quality of life with end-stage renal disease will provide information to evaluate the care delivered. The objective of this study was to validate a Korean translation of the KDQOLTM-36 instrument. We have been following rigorous translation steps guided by the KDQOL Working Group. Forward translation into Korean, reconciliation to establish the conceptual equivalence and consistency, pilot test, panel discussion and backward translation into the original English version were performed. The translated Korean KDQOLTM-36 was validated by analyses of test–retest reliability and internal consistency reliability. Construct validity was examined by the correlations of kidney-disease-targeted scales and SF-12 scales with 164 dialysis patients. Test–retest correlation coecients (n ¼ 30) of SF-12 scales was 0.78 (p < 0.001), and that of the kidney-disease-targeted scales ranged from 0.75 to 0.81 (p < 0.001). The Cronbach’s as of SF-12 scales ranged from 0.54 to 0.89, and the kidney-disease-targeted scales ranged from 0.74 to 0.86. Overall health rating correlated with burden of kidney disease, symptoms/problems, effects of kidney disease, SF-12 PCS and MCS (r ¼ 0.23, p < 0.01; r ¼ 0.35, p < 0.0001; r ¼ 0.42, p < 0.0001 r ¼ 0.56, p < 0.0001; r ¼ 0.23, p < 0.01). Burden of kidney disease, symptoms/problems and effects of kidney disease positively correlated with the items of physical functioning, physical role limitation, emotional role limitation, pain, and emotional well-being (p < 0.05), and negatively correlated with energy (p < 0.001).These results support the applications of the Korean KDQOLTM-36 in studies evaluating Korean dialysis patients.

818 # 1441/SYMPTOM SEVERITY AND PREVALENCE IN ADVANCED CANCER PATIENTS: THE RUSSIAN EXPERIENCE Tatiana I. Ionova & A.A. Novik, St.Petersburg, Russia; X.C. Wang & T.R. Mendoza, Houston, TX, USA; S.A. Kalyadina, M.O. Ivanova, O.A. Uspenskaya, A.V. Kishtovich, St. Petersburg, Russia; C.S. Cleeland, Houston, TX, USA Cancer patients experience symptoms related both to the disease and to its treatment; undertreating these symptoms can cause serious health problems. In Russia, data on symptom prevalence, severity and treatment are lacking. This study used standardized assessment tools to provide baseline measures of symptom severity and prevalence in Russian patients with advanced cancer. Two hundred twenty-six advanced cancer patients (87 males, 139 females, mean age 61) from four St. Petersburg oncology centers were enrolled in the study. Primary cancers were gynecological 20%, leukemia 19%, gastrointestinal 17%, breast 14%, lymphoma 13%, myeloma 7% and others 10%. Multiple symptoms were measured by the Russian version of M.D. Anderson Symptom Inventory. The most severe and frequent symptom reported was fatigue (see Table 1), occurring in 94% of patients. The other four severe symptoms reported – sleep disturbance, pain, distress, and sadness – were present in two thirds of patients. More than half of the patients reported other, less severe symptoms. In conclusion, the majority of advanced cancer patients are burdened by multiple symptoms. Symptom assessment is of much value for clinical decision-making. Further studies monitoring symptoms during treatment and follow-up are worthwhile. # 1519/CROSS-CULTURAL VALIDATION OF THE HUNGARIAN TRANSLATION OF THE ILLNESS INTRUSIVENESS RATING SCALE Eszter Vamos, Institute of Behavioral Sciences, Semmelweis University, Budapest, Hungary; Kenneth Mah, Princess Margaret Hospital, University Health Network, University of Toronto, Toronto, Ontario, Canada; Marta Novak, Miklos Z. Molnar, Gabor Csepanyi & Istvan Mucsi, Institute of Behavioral Sciences, Semmelweis University, Budapest, Hungary; Gerald M. Devins, Department of Psychiatry, Toronto General Hospital, University Health Network, Toronto, Ontario, Canada The main aim of this study was to validate the Hungarian version of the Illness Intrusiveness Ratings Scale (IIRS) testing internal validity and using simultaneous confirmatory factor analysis. The translation procedure was carried out according to current recommendations. Following pilot testing the scale was completed by 365 maintenance haemodialysis patients. Hungarian data were compared with IIRS data from North American ESRD patients undergoing maintenance haemodialysis to evaluate item bias (Group · Item ANOVA). Confirmatory factor analyses indicated a good fit between the previously hypothesized 3-factor model (‘relationships and personal development,’ ‘intimacy,’ and ‘instrumental’ life domains) of the original English version and the Hungarian translation. Although statistically significant (p < 0.05), the effect size for the Groups · Items interaction was not substantial. Acceptable reliability was observed for each of the subscales (Cronbach’s a 0.64–0.67) and for the composite score (0.80). Test–retest reliability was also acceptable. The Hungarian translation of the Illness Intrusiveness Ratings Scale, therefore, meets many of the requirements of cultural equivalence. Supported by grants from ETT 240/2000, NKFP 1/002/2001, OTKA TS040889 and T038409 (M.N.). I.M. is a Bekesy postdoctoral fellow. M.Zs.M. is a scholar of the Hungarian Kidney Foundation. G.M.D. has been supported by the National Health Research and Development Program (Health Canada) and the Canadian Institutes of Health Research. # 1148/LINGUISTIC VALIDATION OF THE FACT-GASTRIC (FACTGa) IN JAPANESE AND ENGLISH Sonya Eremenco, John Cashy & Kimberly Webster, CORE, ENH, Evanston, IL, USA; Yasuo Ohashi, Faculty of Medicine, University of Tokyo, Tokyo, Japan; Gershon Y. Locker, Department of Medical Oncology, ENH, Evanston, IL, USA; Shinichi Uemura, Outcomes Research, Pharmacia KK, Tokyo, Japan; Guy Pelletier, Department of Psychosocial Resources, Tom Baker Cancer Center, Calgary, AB, Canada; David Cella, CORE, ENH, Evanston, IL, USA

We developed a disease-specific subscale for gastric cancer to complement the FACIT Measurement System. Because gastric cancer is more common in Asia than North America, we concurrently developed it using input from both cultures. The FACT-Ga subscale underwent a standard development process: item generation (17 pts and 12 experts), scale construction, initial validation (30 pts) and scale refinement. The item generation phase yielded 65 candidate items, which were reduced to 36 after a rigorous qualitative data review. Next, 13 experts from the US, Canada, and Japan rated the items on relevance and importance. Results led to the removal of 10 items rated below an accepted median rating. The 26-item FACT-Ga was translated into Japanese using the FACIT translation methodology, followed by linguistic validation in Japan and content validation of English in Canada. A total of 30 patients, 20 Japanese and 10 Canadian, participated. The Japanese and Canadian samples had mean ages of 57 and 66 years, respectively. Patients completed the FACT-Ga and an Item Rating Survey to evaluate relevance and importance of each item. Interviews were also conducted to elicit feedback on appropriateness of item wording and content. The FACT-Ga performed well in this initial evaluation, with high reliability in both patient groups. The combined data had a Cronbach’s a of 0.90 (0.87 JPN and 0.94 ENG). Patients commented about the wording of 15 items in the Japanese, and none in the English. However, patients in both groups who had undergone total gastrectomy found it difficult to answer items with the word ‘stomach’ or related to nausea and vomiting. Seven items were removed from the scale because they were rated relatively unimportant or due to redundancy with other items. This resulted in a 19-item questionnaire (Cronbach’s a ¼ 0.84 JPN; 0.93 ENG). The FACT-Ga in English and Japanese shows acceptable initial psychometric characteristics and performs well with patients groups in diverse countries, providing evidence of conceptual equivalence in both languages. Further validation studies will assess the full psychometric performance of the FACT-Ga. # 1228/THE RELIABILITY AND VALIDITY OF THE MALAY VERSION OF THE MCGILL QUALITY OF LIFE QUESTIONNAIRE – CARDIFF SHORT FORM (MMQOL-CSF) IN PATIENTS WITH ADVANCED CANCER P.L. Lua, M.S. Salek, WSP Centre for Socioeconomic Research, Cardiff University; I. Finlay, Velindre NHS Trust, UWCM, Cardiff, UK; A.G.I. Boay & R.M. Said, Queen Elizabeth Hospital, Kota Kinabalu, and Ministry of Health, Malaysia By and large HRQoL instruments used in patients receiving palliative care have been created in English for Western cultures. The aim of this study was to determine the reliability and validity of MMQOL-CSF in the Malaysian palliative care population. This study was carried out in a Malaysian general hospital recruiting in-patients with advanced cancer. The original McGill Quality of Life Questionnaire (MQOL) and its Cardiff Short Form were translated and cross-culturally adapted into Malay using the iterative forward–backward method to produce Malay MQOL (MMQOL) and MMQOL-CSF. Eligible patients were approached and those who gave informed written consent were randomised into two groups: (1) patients who were asked to complete the long form MMQOL followed by MMQOL-CSF; and (2) patients who completed them in reverse order. The MMQOL contains 16 items (assessing five domains), one global question, an open-ended question and a tick box question if a patient required help. The MMQOLCSF contains seven items (three domains), one global question and an open-ended question. Altogether, 10 in-patient wards were involved in this 12-week study with a total of 65 patients (mean age ¼ 40.6, range ¼ 18–78; male ¼ 31, female ¼ 34) of 104 approached (response rate ¼ 62.5%). There was no significant difference in the two groups’ age and gender. The internal consistency of MMQOL-CSF ranged from 0.39 to 0.84. Items from MMQOL-CSF correlated moderately to highly (0.39–0.86; p < 0.01) with domains from the MMQOL (criterion validity). Except for the existential items, all other items correlated either moderately/strongly with their own domains (0.47– 0.80), demonstrating convergent validity. Known-group validity was shown by the differences in HRQoL scores for patients with differing haemoglobin levels (p < 0.05 for the physical symptoms and existential domains and total score) and white cell counts (p < 0.05 for the global question, existential domain and total score). The findings confirm the reliability and validity of the MMQOL-CSF.

819 # 1226/THE FEASIBILITY AND RELIABILITY OF THE MALAY VERSION OF THE MCGILL QUALITY OF LIFE QUESTIONNAIRE – CARDIFF SHORT FORM (MMQOL-CSF) IN PALLIATIVE CARE POPULATION P.L. Lua & M.S. Salek, WSP Centre for Socioeconomic Research, Cardiff University; I. Finlay, Velindre NHS Trust, and UWCM, Cardiff, UK; A.G.I. Boay & R.M. Said, Queen Elizabeth Hospital, Kota Kinabalu, and Ministry of Health, Malaysia Many established health-related quality of life (HRQoL) instruments in palliative care have been developed and tested in the West. Crosscultural comparisons of HRQoL and palliative care services can be greatly enhanced if a tool can be adapted and applied in another culture. This study aimed to investigate the feasibility and reliability of the Malay McGill Quality of Life Questionnaire-Cardiff Short Form (MMQOL-CSF) in the Malaysian palliative care population. This study was carried out in an out-patient clinic in a Malaysian general hospital. The original McGill Quality of Life Questionnaire (MQOL) was translated and cross-culturally adapted into Malay using the iterative forward–backward method to produce both the Malay MQOL and the MMQOL-CSF simultaneously. The latter was envisaged to facilitate routine HRQoL assessment with minimal completion burden. Eligible out-patients were approached, given information about the study and invited to participate while waiting to see their physicians. Following written consents, patients completed MMQOL-CSF (an 8-item tool assessing global quality of life, physical symptoms, psychological and existential concerns) and responded to five questions on the practicality aspects of the instrument. Over the 12-week study period, 51 patients (mean age ¼ 50.6 years, range ¼ 19–82 years; male ¼ 18, female ¼ 33) consented to participate out of the 65 approached (response rate ¼ 78.5%). The average completion time was 5.35 min (SD ¼ 1.62). All the patients agreed that the questions and instructions were either ‘very clear’ or ‘clear’. Ninety-eight percent (n ¼ 50) agreed that the form was comprehensive. There were no ‘difficult, unsuitable or distressing’ questions. No missing data was recorded. Internal consistency (Cronbach’s a) ranged from moderate to excellent for all domains (physical symptoms ¼ 0.761, psychological ¼ 0.914, existential ¼ 0.514, total score ¼ 0.834). These results confirm that the MMQOL-CSF is a feasible and reliable instrument, useful for measuring HRQoL in the Malaysian palliative care population. # 1675/USING THE NOTTINGHAM HEALTH PROFILE IN THE HOSPITALIZED FRAIL ELDERLY Eva Baro´, Outcome Research, 3D-Health Research; Montse Ferrer, Health Services Research Unit, Institut Municipal d’Investigacio´ Me`dica; Olga Vazquez, Ramo´n Miralles, Ascensio´n Esperanza & Antoni Maria Cervera, Centre Geria`tric – UFISS, Hospital de l’Esperanc¸a; Jordi Alonso, Health Services Research Unit, Institut Municipal d’Investigacio´ Me`dica, Barcelona, Barcelona, Spain Low completion rates of Health Related Quality of Life (HRQL) questionnaires, specially for the SF-36, have produced concerns about its suitability in hospitalised older adults. The high prevalence of disability and cognitive impairment among these patients could limit suitability of measuring generic HRQL. We assessed the feasibility and internal consistency of the Nottingham Health Profile (NHP) among 134 hospitalized frail elderly (over 64 years) presenting problems at discharge to return home. Cognitive impairment (CI) was measured by means of Mini Mental State Examination (MMSE). A trained interviewer administered the NHP (38 items with Yes/Not response options), which scores range between 0 (best health) to 100 (worst). Over a third (36.6%) of the patients showed none or minimal CI (MMSE > 20); 30.6%, moderate CI (MMSE ¼ 15–20); and 32.8%, severe CI (MMSE <15 or unable to perform MMSE due to aphasia, severe mental status deterioration or decreased awareness). The NHP could be administered to 70% of the overall sample: 98% of patients without CI, 90% of those with moderate CI, and only 20.5% (n ¼ 9) of those with severe CI (p < 0.01). Frequency of missing items, score distribution and Cronbach’s a were evaluated for the six NHP dimensions and the

global score. The percentage of patients with at least 1 missing item was less than 5% for all dimensions except for Emotional reactions and Pain (8 and 65%, respectively), with no differences by CI. The Cronbach’s a was above the standard proposed (0.7) for all scores, except for Energy, Physical mobility and Social isolation (0.48–0.61), with no statistically significant differences between patients with moderate and with no CI. When interviewer-administered, the NHP presented high completion rates and acceptable internal consistency, suggesting that it could be useful among hospitalized frail elderly patients with normal and moderate CI. Further studies to assess validity and responsiveness are needed. Funded by: CIRIT 2001SGR 00405; AATRM 038/06/02. # 1763/IMPACT VARIABLES ASSOCIATED WITH QUALITY OF LIFE IN OLD ADULTS Clarissa M. Trentini, Eduardo Chachamovich, Milena F. Costa, Daniela H. Muller & Marcelo P. Fleck, Programa de pos-graduacao em Psiquiatria, UFRGS, Porto Alegre, RS, Brasil Background: Variations on the perception of one’s Quality of life may due to several variables. Particularly in old adults, this perception may be determined by age and gender among others. Investigating factors which may contribute to a better quality of life seems to be extremely relevant in old adults as populational data indicate a growing participation of old adults in the population as a whole. Objectives: To identify potential variables which may alter elderly’s Quality of Life. Methods: Old adults (60 years old or older) were interviewed. Sampling method was convenience. Three hundred thirty nine subjects were included, stratified by gender, age and subjective perception of his/her healthy status. Social-demographic information was collected, as well as quality of life perception (WHOQOL-100) and depression symptomatology (Beck Depression Inventory). Results: T-test and Multivariable Linear Regression were performed. Analysis of overall perception of QOL showed association with depression levels (p ¼ 0.000), subjective perception of health status (p ¼ 0.000) and gender (p ¼ 0.022). The individual analysis of each domain concluded that depression levels are correlated to all of them, while health status was associated with physical (p ¼ 0.001), psychological (p ¼ 0.026), independence levels (p ¼ 0.000) and social relationships (p ¼ 0.018) domains. Different variables such as gender, age and social class were also examined. Conclusions: Elderly’s perception of his/her QOL is strongly associated with depression levels. Further studies are required. # 1766/CARER’S PERCEPTION OF OLD ADULT’S QUALITY OF LIFE Eduardo Chachamovich, Clarissa M. Trentini, Michelle Figueiredo & Marcelo P. Fleck, Psiquiatria e Medicina Legal, UFRGS, Porto Alegre, RS, Brazil Background: It is not know to what extent carers are reliable to estimate QOL of his caree. Objectives: To investigate the potential association between the perception of an old adult about his/her quality of life and the perception of his/her carer about the caree’s (old adult’s) quality of life. Methods: Sampling method used was convenience. Twenty seven sets of old adult and carer were interviewed. Socialdemographical data were collected, as well as the perception of quality of life (WHOQOL-100) and depression symptomatology (Beck Depression Inventory). Carers also responded to a adapted WHOQOL100, which addressed questions to his/her perception of the old adult’s quality of life. A comparison of the two angles of evaluation was run in order to identify converging and diverging areas. Results: Pearson correlation was performed. Statistically significant correlation was found in all domains and in overall QOL perception scores. Partial correlation showed no influence of the carer’s depression levels on his/ her perception of elderly’s QOL. Other refined analysis are being carried out. Conclusions: There is a strong association between the perception of an old adult about his/her quality of life and the perception of his/her carer about it.

820 # 1379/DEPRESSION AND QUALITY OF LIFE IMPAIRMENTS IN THE ELDERLY Isabel P. Leal & Catarina F. Rodrigues, Psychology, Instituto Superior De Psicologia Aplicada, Lisboa, Lisboa, Portugal Background: Late-life depression is a major public health problem. As our elderly population grows, the number afflicted increases. With this work we intended to verify the magnitude and specificity of quality-oflife (QOL) impairments in non-depressed, mildly depressed and severely depressed elderly. Methods: We examined the variation of 272 elderly (aged 65–84) with none to severe depression. Geriatric Depression Scale (GDS) and generic Medical Outcomes Short Form-36 Health Survey (SF-36) QOL ratings obtained at baseline were analysed. Results: Compared with non-depressed subjects, depressed subjects showed significant QOL impairments in all of the eight baseline SF-36 items (p < 0.01). Women rated their QOL lower than men on physical functioning, social functioning and vitality (p < 0.01) and physical role and mental health (p < 0.05) and showed similar trends n all other QOL items. Compared with younger subjects, subjects older than 75 years, reported lower QOL on physical functioning, physical role, social functioning, vitality, pain and general health (p < 0.01) and mental health (p < 0.05). Depression accounted for 63.7% of the variance observed. Conclusions: Our findings confirm the disabling nature of late life depression and enhance the Importance of targeting depression as well as QOL outcomes in interventions with elderly populations. # 1626/SUBJECTIVE QUALITY OF LIFE IN DEMENTIA: A MEASURABLE CONSTRUCT? Saskia Teunisse, Psychology, University of Aberdeen, Aberdeen, United Kingdom Dementia is a disorder that has a profound impact on the lives of afflicted individuals and their families (Teunisse et al., 1991; Logsdon et al., 1999). In the absence of a cure, care is directed at a slowing of the progression of the disease and optimising quality of life of dementia patients and their families. The term Quality of Life (QL) refers to the subjective perception of an individual of his or her position in life and for that reason, whenever possible, should be evaluated by asking the person involved (Fayers and Machin, 2001). While the quality of life of family members of dementia patients has been extensively studied, patient self-reports of quality of life have long been avoided because of the possible impact of memory impairment, loss of verbal skills, and other cognitive impairments on the completion of measures (Albert et al., 1999). Instead, assessment has focused on more objective indicators of QL in dementia: the cognitive impairment, and the functional limitations and behavioural and psychological symptoms as reported by carers, clinicians or other professional health care providers (Teunisse, 1997). Patient rated QL, however, would contribute to evaluation of care and new interventions, end-of-life decision making, and the ethical and political debate with regards to health care utilization in dementia. Recent research has shown that QL ratings can be obtained from patients with dementia. Indeed, this has been found possible for dementia-specific measures that have been adapted to the cognitive limitations of dementia, such as the QOL-AD (Logsdon et al., 1999, 2002), the DQOL (Brod et al., 1999) and the QOL-LSD (Mills and Teunisse, 2001), as well as for generic measures that have been widely used in other populations, such as the Duke Health Profile (Novella et al., 2001). An overview of the recent developments with regards to subjective QL measurement in dementia will be presented and critically evaluated. # 1606/SOCIAL SUPPORT AND HEALTH-RELATED QUALITY OF LIFE IN ELDERS RETREATING FROM MAINLAND CHINA TO TAIWAN AFTER WORLD WAR II Chia-Ing Li, Department of Medical Research, China Medical College Hospital; Tsai-Chung Li, Institute of Chinese Medical Science, China Medical College; Cheng-Chieh Lin, Department of Family Medicine, China Medical College Hospital, Taichung, Taiwan, Republic of China Introduction: After World War II, hundreds thousands of mainlanders followed Chiang Kai-Shek, retreating from Mainland China to Taiwan. They lived in communities built by government around the island. Chung-Shing-Shin-Tseun (CSST), one of these communities but dif-

ferent from the others, is the location of Taiwan Provincial Government. Using a theoretical framework that divided social factors measures into interaction, dependability, and affective dimensions, this study will determine whether these mainlanders differ from the Taiwaners in terms of social support, health-related quality of life (HRQoL). Methods: The study recruited all residents aged 65 and over in CSST in 1998 with an overall response rate of 89.13% (n ¼ 921). Face-to-face interviews were used to collect data. The Short Form 36 (SF-36) is a short questionnaire which measure eight multi-item variables: physical functioning, social functioning, role limitations due to physical and emotional problems, mental health, vitality, pain, and general health. Results: This study showed that the mainlanders were at older age, more likely to be male, retired, and living alone, with higher level of education and perception of loneliness, and with less availability of social resources, but no difference in marital status, and availability of confidant. After controlling for age and gender, the mean number of chronic conditions for these mainlanders was higher than that for Taiwaners, but no difference for the ADL and IADL, subjectively assessed income adequacy, number of financial dependents and eight scales of SF-36. After controlling for sociodemographics and number of chronic conditions, the measures of social resources were associated with different aspects of HRQOL. # 1255/GENERIC AND DISEASE-SPECIFIC QUALITY OF LIFE IN JAPANESE COLOSTOMATES Hitomi Kataoka, Nobuyoshi Mori, Makoto Nagasaka, Taku Harada, Naoyoshi Minami, Hajime Kurosawa Masayuki Kanazawa & Masahiro Kohzuki, Department of Internal Medicine and Rehabilitation S, Tohoku University Graduate School of Medicine, Sendai, Miyagi, Japan Purpose of research: Colostomates with stoma performed large intestine confront many problems, both physical and psychological. Although QOL in colostomates is reported to be low, the health-related generic and disease-specific QOL has not been fully elucidated. The purpose of this study is to evaluate the health-related generic and disease-specific QOL in Japanese colostomates. Subject sample and statement of methods: The generic QOL and disease-specific QOL were evaluated by SF-36 and Olbrisch ostomy adjustment scale (OAS), respectively. Those questionnaires were sent by mail with stoma appliances when ostomates ordered them. We were also elucidated the factors that influence the QOL of Japanese colostomates. Date analysis was performed T-test, Pearson’s correlation coefficient and stepwise multiple regressions. Summary of results: One hundred and thirty one colostomates (age: 67 ± 11 years, 5.0 ± 4.8 years after surgery) who underwent ostomy surgery replied. The SF-36 scores of colostomates were below the average scores of Japanese general population except bodily pain (BP). The mean total score of OAS were lower than that reported by Olblisch (1983), Burckhardt (1990) and Brydolf (1994), 130.6 compared to 155.7, 168 and 178.6. The factor of skin irritation may deteriorate the QOL of Japanese colostomates, however the other factors which sex, age, marital status, interval after the surgery, independence in changing of a stoma appliance, experiences of stoma clinic and the enrollment to an ostomy association do not affect the QOL. In the present study, the generic QOL of Japanese colostomates was lower than that of Japanese general population. The mean value of total OAS of Japanese colostomates was lower than that which has been already reported in Western countries. As skin irritation may deteriorate QOL, care to the management of skin irritation postoperatively should be carefully planned. Longitudinal prospective studies need to be done to elucidate whether the increase in physical and psychological status can improve QOL in colostomates.

821 # 1370/QUALITY OF LIFE IN PATIENTS WITH ULCERATIVE COLITIS: THE IMPORTANCE OF SYMPTOMS Elaine McColl, Centre for Health Services Research, University of Newcastle upon Tyne, Newcastle upon Tyne; Seong W. Han, Mark R. Welfare, Roger Barton & Peter James, North Tyneside General Hospital, Northumbria Healthcare NHS Trust, North Shields; Nick Steen, Centre for Health Services Research, University of Newcastle upon Tyne, Newcastle upon Tyne, Tyne & Wear, United Kingdom Establishing predictors of quality of life in patients with inflammatory bowel disease could help clinicians to identify those most likely to experience poor quality of life, and thus to target interventions appropriately. Our aim was to investigate the relationship between disease-specific quality of life and demographic, disease and physiological variables, cognitive representations of illness, and perceived general health status. One hundred and eleven patients completed the disease-specific Inflammatory Bowel Disease Questionnaire (IBDQ), the generic SF-36 and the Illness Perception Questionnaire (IPQ). Extent of disease was recorded from records and disease activity was determined by a self-reported symptom index. Bivariate analyses and multi-variate regression models were used to identify predictors of disease-specific quality of life. Bi-variate analyses showed that symptom-related disease activity, elements of illness representation and physical and mental health as measured by the SF-36 were the only variables that were strongly or moderately correlated with the total IBDQ score. In addition, multi-variate regression modelling showed that duration of disease was also significant in explaining variation in the total IBDQ score. Symptom-related disease activity was found to be the major explanatory variable for each of the four domains of the IBDQ. This study highlights the strong relationship between patients symptoms and all aspects of their health-related quality of life but little association with age, gender, physiological markers of disease activity or anatomical disease extent. These findings suggest that the best strategy to improve patients quality of life may be to find ways to reduce their symptoms.

# 1843/QUALITY OF LIFE AND COPING OF PEOPLE WITH TEMPORATY AND PERMANENT STOMAS Vera Lucia Conceicao de G. Santos, Miako Kimura & Elaine C. Chaves, Medical Surgical Nursing, School of Nursing, University of Sa˜o Paulo, Sa˜o Paulo, Sa˜o Paulo, Brazil The aims of this study were to evaluate and compare coping strategies use and quality of life of people with temporary and permanent stomas. Forty two persons with temporary stomas (Group A) and 72 persons with permanent stomas (Group B) from Outpatient Ostomy Care Centers were interviewed. For data collection we used the Coping Strategies Inventory by Folkman and Lazarus, and Ferrans and Powers Quality of Life Index, both of them validated for Brazilian culture. The results showed patients aged 44.69 (SD ¼ 16.85) and 59.64 (SD ¼ 12.62) years old in average, respectively for Groups A and B. Other demographic characteristics were similar among patients. Trauma and cancer were the most frequent causes for stoma creation, respectively for temporary and permanent stomas. Colostomies predominate among stoma types. In relation to coping strategies escapeavoidance (2.50), seeking social support (2.12), planful problem solving (2.10) and positive reappraisal (2.06) were the most frequent for persons with temporary stomas; and planful problem solving (1.91), seeking social support (1.86) and positive reappraisal (1.80) were the most frequent for persons with permanent ones. Significant differences were obtained for escape-avoidance (p < 0.001), confronting (p ¼ 0.011) and positive reappraisal (p ¼ 0.027), with the major scores for Group A. Quality of life scores did not show significant differences between groups, and they were: 22.52 and 21.74 for Health-Functioning domain; 20.38 and 20.92 for Social-Economic; 24.69 and 24.16 for Psychological/Spiritual; 25.81 and 24.10 for Family domain, and 23.60 and 22.73 for Total Index. The study revealed that people with ostomies use more passive coping strategies and pointed out for a more positive quality of life scores, being Health-Functioning domain the most affected with the lowest scores.

# 1773/ADDED VALUE OF THE PATIENT’S PERSPECTIVE IN IRRITABLE BOWEL SYNDROME Olivier Chassany, Delegation a la recherche clinique – AP-HP, Hopital Saint-Louis, Paris, France, Philippe Le Jeunne, Thales, Boulogne Billancourt, France

# 1726/QUALITY OF LIFE IN ADULTS AFTER CORRECTION OF OESOPHAGEAL ATRESIA Jacqueline A. Deurloo & Seine Ekkelkamp, Pediatric Surgical Center Amsterdam (EKZ/AMC, VUMC); Esther E. Hartman & Mirjam Sprangers, Department of Medical Psychology, Academic Medical Center/ UvA; Daniel C. Aronson, Pediatric Surgical Center Amsterdam (EKZ/ AMC, VUMC), Amsterdam, Netherlands

Under the behalf of ALFIS (Association des Laboratoires et des Firmes de Sante´) Irritable bowel syndrome (IBS) is an exemplary condition for which the patients perspective should be regarded as a major outcome. There is no objective marker of the disease activity. IBS is expressed by many digestive and extra digestive symptoms, even though abdominal pain or discomfort are the most frequent. IBS is a chronic disease, which impacts daily life by its repeated symptomatic flares over years. A cross-sectional survey included 239 IBS patients recruited by 163 general practitioners (GP). Patients had to complete a specific HRQL questionnaire, the Functional Digestive Disorders Quality of Life (FDDQL), validated in French. The FDDQL contains 43 items gathered into eight domains [0–100 (best HRQL)]. Patients scored their abdominal pain on a 10-cm visual analog scale [0–10 (maximal pain)]. GP were asked to estimate the pain intensity of their patients as well as the HRQL score for each of eight domains similar to the FDDQL. Patients were 57.5 ± 16 years old, and 64% were women. The pain intensity perceived by patients was 3.90 ± 2.49, and estimated by physicians was 3.04 ± 2.10 (p < 0.0001). The global FDDQL score was 56.1 ± 11.6. The correlation between the perception of patients and of physicians for the pain intensity was moderate at r ¼ 0.47 (p < 0.0001). The correlation was logically higher between the patients’ perception of pain and of HRQL, r ¼ 0.63 (p < 0.0001). There was some moderate correlation between patients and clinicians regarding the HRQL, only for daily activities (r ¼ 0.43) and diet domains (r ¼ 0.30). Clinicians’ and patients’ perspectives although overlapping, are not similar. Clinicians tend to underestimate the pain intensity of their patients. Thus, perception of abdominal pain cannot be accurately inferred from the clinician’s point of view. Similarly, patient’s perception of pain cannot completely reflect the impact of HRQL. Patient’s perspective is a major outcome in the evaluation of therapies in IBS.

Over the last decades, the mortality of patients with oesophageal atresia (OA) has decreased to approximately 5%. Little is known about the influence of OA and its long-term sequellae on the quality of life (QoL) of the surviving patients. The aim of our study was to investigate the generic and disease specific QoL after correction of OA in a large adult population, and to compare the generic QoL to that of a healthy population. All surviving patients treated for OA between 1947 and 1985 (n ¼ 119) were sent a questionnaire. The questionnaire assessed sociodemographic characteristics, generic (SF-36) and disease specific (GIQLI, EORTC-OES24) QoL. It also contained three open-ended questions about the daily consequences of the disease. Clinical characteristics were collected from patient charts. For reasons of multiple testing, the level of significance was defined as p < 0.01. We received 97/119 completed questionnaires (82% response rate). When comparing the generic QoL of OA patients to that of healthy subjects, we found no differences in the ‘overall’ physical and mental functioning. However, patients do report impaired QoL on ‘general health’ and ‘vitality’. The type of atresia did not influence generic or disease specific QoL. The presence of concomitant congenital anomalies did not influence generic QoL, these patients only reported impaired QoL in the ‘indigestion’-scale (EORTC-OES24). A third of the patients reported negative consequences of the OA in their daily life. Gastro-intestinal symptoms were mentioned most often (23%). These patients also reported impaired QoL on several domains of the questionnaires. Conclusion: In general, after correction of OA, patients perceive their generic and disease specific QoL to be good. The type of atresia or the presence of concomitant congenital anomalies does not influence generic QoL. However, a third of the patients report negative consequences of the disease.

822 # 1795/DEVELOPMENT OF CORE SETS FOR MUSCULOSKELETAL CONDITIONS TO APPLY THE INTERNATIONATIONAL CLASSIFICATION OF FUNCTIONING DISABILITY, AND HEALTH (ICF) IN CLINICAL PRACTICE AND RESEARCH Gerold Stucki & Alarcos Cieza, Physical Medicine and Rehabilitation, University of Munich, Munich, Germany; Rob de Bie, Department of Epidemiology, University of Maastricht, Maastricht, The Netherlands; Karsten Dreinho¨fer, Orthopedic clinic, University of Ulm, Ulm, Germany; Jan Schouten, Department of Epidemiology, University of Maastricht, Maastricht AZ, The Netherlands; Somnath Chatterji, CASGroup, World Health Organization, Geneva 27, Switzerland; Nicolas Walsh, Health Science Center, The University of Texas, San Antonio, Texas, USA The purpose of this study was to establish ICF Core Sets (lists of ICF categories relevant to most patients with a health condition) for clinical practice (long ICF Core Sets) and for research (short ICF Core Sets) for low back pain (LBP), rheumatoid arthritis (RA), osteoporosis (OP), and osteoarthritis (OA). A decision-making and consensus process with 80 experts from 28 countries took place during the 1st ICF Core Set Conference based on the results of preliminary studies using empirical data from patients, international Delphi surveys, and systematic reviews. The long ICF Core Sets for LBP, RA, OP and OA encompass 79, 95, 69 and 55 and the short, 35, 23, 23 and 13 ICF categories, respectively. The body functions b280 Sensation of pain, b730 Muscle power and b710 Mobility of joint, the activity d450 Walking and the environmental factor e110 Immediate Family are contained in all four short ICF Core Sets. Additionally, the four large ICF Core Sets have seven body functions (b134, b152, b440, b740, b765, b770, b780), 16 activities and participations (d410, d415, d430, d445, d455, d470, d475, d510, d540, d620, d640, d770, d850, d859, d910, d920) and 12 environmental factors (e110, e120, e135, e150, e155, e225, e355, e410, e450, e460, e540, e570) in common. The body functions b765, b545 and b755 are specific to OP, b260, b750 and b126 for LBP, b760 for OA and b510 and b430 for RA. The large ICF Core Set for RA also differentiates among generalized pain, pain in head and neck, pain in back, pain in upper and in lower limbs and in joints. There are also categories that are common to two or three of the health conditions, such as b735 Muscle tone, which is represented in LBP and OA ICF Core Sets for musculoskeletal conditions are both intuitive and practical and link the ICF to the condition-oriented perspective. The ICF Core Sets will be tested in different countries and settings within the scope of a collaboration project by the University of Munich and WHO. # 1474/QUALITY OF LIFE AMONG RHEUMATOID ARTHRITIS PATIENTS Grazyna Baczyk, Faculty of Health Sciences, Department of Nursing, Karol Marcinkowski University of Medical Sciences, Poznan, Poland It has been reported that rheumatoid arthritis (RA) affects 0.5–1% of the adult population in Poland. More than a half of RA patients suffer from at least one additional chronic disease. Important goal of a health care system is to improve quality of life of patients with RA. However, there is no research on quality of life among RA patients in Poland. The aim of this study was to assess the quality of life of the RA patients treated in Rheumatoid Special Clinic in Poznan, Poland. The specific questions were: (1) Does the quality of life of RA patients depend on demographic variables (gender, age) and duration of the disease? (2) Does the quality of life of RA patients depend on clinical symptoms and pain? The study sample consisted of 120 RA patients, including 90 women (75.0%) and 30 men (25.0%). The mean age of the treated women was 50.8 years, 56.0 for men and 52.2 years for the whole group. The Polish version of the Arthritis Impact Measurement Scales 2 (AIMS)was used to assess the quality of life. Ritchie Articular Index and 10 point VAS scale were used to evaluate clinical symptoms and pain severity. Grip strength, duration of the morning stiffness were also measured. The results showed that the mean score on the AIMS was 4.24, which is within the medium section of the average measurement of the quality of life. The quality of life depended on the sex of the patient. Women scored significantly lower in the emotional area than men (p < 0.05). Also, younger patients demonstrated higher evalua-

tion of quality of life in the area of physical state and mobility (p < 0.05). Life satisfaction was significantly higher among the patients suffering longer than 5 years (p < 0.05). The assessment in most of the subscales of the AIMS correlated significantly with Grip Strength Measurement symptoms and VAS scale (p < 0.01). In conclusion, this study showed that quality of life of RA patients depends on gender, age and disease variables. Quality of life measures are important in the evaluation of the impact of RA on the patients’ life. # 1144/THE EFFECTS OF ANKYLOSING SPONDYLITIS ON HEALTH-RELATED QUALITY OF LIFE J. Michael Woolley, Global Health Economics, Amgen Inc., Thousand Oaks; John C. Davis, Clinical Trials Center, Division of Rheumatology, University of California San Francisco, San Francisco, CA, USA Ankylosing spondylitis (AS) is a systemic rheumatic disorder characterized by inflammation of the axial skeleton and large peripheral joints. Subjects with AS may experience significant pain and disability. The objective of this study was to evaluate the effects of AS on healthrelated quality of life (HRQOL). A randomized, double-blind clinical trial of etanercept therapy was conducted in 40 subjects with AS who had evidence of moderate to severe disease activity despite stable doses of currently accepted treatments. HRQOL was assessed using the Short Form-36 Health Survey (SF-36) at baseline and after 16 weeks of therapy. The SF-36 can be summarized into eight scales measuring physical functioning (PF), role limitations due to physical problems (RP), body pain (BP), general health perceptions (GH), vitality (VT), social functioning (SF), role limitations due to emotional problems (RE), and mental health (MH). The subjects with AS who were treated with etanercept (n ¼ 20) were compared to a nationally representative sample (US Pop, n ¼ 2470) from the Medical Outcomes Study both before (baseline) and after 16 weeks of therapy. Hypotheses were tested using the t-test in a regression analysis adjusting for age and gender. Results are shown in the table. At baseline, subjects with AS had significantly lower HRQOL for each SF-36 scale. Differences between the US Pop sample (n ¼ 2470) and those with AS (n ¼ 20), were very large (exceeding 30 for all but two scales) and statistically significant (p < 0.001 for each scale). The magnitude of the difference in physical functioning, for example, is similar to the effect of a healthy person aging about 40 years. After 16 weeks of treatment SF-36 scores had improved, reducing the baseline difference from US Pop by about 20%. In conclusion, prior to therapy, subjects with ankylosing spondylitis had markedly reduced health-related quality of life. Etanercept therapy for 16 weeks improved it. These results are consistent with the hypothesis that moderate to severe ankylosing spondylitis significantly reduces health-related quality of life. # 1761/THE ADAPTATION OF SATISFACTION WITH INCOME IN GERMANY AFTER UNIFICATION Peter Krause, GSOEP, German Institute for Economic Research, Berlin, Berlin, Germany The German unification may be regarded as a big coincidence for quality of life research as East German countries with lower welfare levels and less income inequality adapted stepwise to the more wealthy and more unequal distributed countries in West Germany. The German Socio-economic Panel study (GSOEP) – which started in 1984 in the West and even before unification in June 1990 in the East – provides unique opportunities to monitor the whole adaptation process on an individual level year by year from the beginning. Using income satisfaction measures (i.e. individual evaluations of the someone’s own income on a 0–10-scale) we are able to illustrate the increasing expectation levels on income in East Germany towards the higher standards in the West especially in the first half of the 90s. However these results give us not only a more detailed description of the development of living conditions in East Germany. They provide us also with insights in the meaning of quality of life measures themselves. Indifference curves on income satisfaction illustrate, that these income evaluations follow absolute as well as relative standards of living. Further more there is some empirical evidence, that the evaluation of income by satisfaction scores includes also aspects, which are in line with Rawl’s theory of social justice.

823 # 1480/PERCEIVED HRQL AMONGST THE KAMBA OF EASTERN KENYA: SOCIO-ECONOMIC AND DEMOGRAPHIC VARIATION IN KENQOL SCORES Julia Fox-Rushby, London, UK, Annabel Bowden, Bracknell, UK This study investigated the variation in perceived HRQL of a Kenyan population by socio-economic and demographic characteristics using the Kenya Quality of Life (KENQOL) structured survey. A household survey was conducted in E. Kenya during an 8-week period in 1999 using a two-stage random sampling strategy. The KENQOL instrument consisted of 112 questions with a ‘yes/no’ response. Questions were grouped into five conceptual areas: Completeness (n ¼ 3), Capacity (n ¼ 19), Co-operation (n ¼ 22), Cleanliness (n ¼ 6) and Contentment (n ¼ 57). Total and dimension scores had a potential range of 0–100% (100% ¼ highest HRQL). The survey achieved an 81.3% response rate. The mean KENQOL total score was 58.5 (SD 19.1), mean scores for each conceptual dimension ranged from 45.1 (Completeness) to 73.9 (Cleanliness). Bivariate analyses found that KENQOL total and dimension scores varied by age ()ve), number of children parented ()ve) and asset index scores (+ve). Gender (+ve males), marital status (varied), unmarried motherhood (+ve), polygyny (varied), number of children died ()ve), education level (varied), occupation (varied), home and land ownership (varied) were statistically significantly related to specific dimensions and/or total scores. The final OLS multiple regression model for KENQOL total score included the asset index score, number of children born, semi/unskilled work, selling crops, age, and the gender of the head of household. The model explained 21.0% of the variance and was significant (F 6,440 ¼ 20.8, p < 0.01). Comparison with HRQL studies worldwide revealed that some of the variation in HRQL by demographic and socio-economic qualities is transferred across context e.g. decreased HRQL with increased age, but that this is not always the case. On the whole, demographic and socio-economic information collected in the comparator studies was rather bland, not reflecting the diversity of the people compared to their counterparts in Western societies. The procedure used in this study to develop demographic and socio-economic questions resulted in a range of culturally specific indicators. # 1185/A TWOFOLD STRUCTURE OF QUALITY OF LIFE: CAUSAL AND EFFECT INDICATORS OF HEALTH-RELATED QUALITY OF LIFE IN CANCER PATIENTS Sonja Boehmer, Department of Health Psychology, Freie Universitaet Berlin, Berlin, Germany; Aleksandra Luszczynska, Faculty of Psychology, Warsaw University, Warsaw, Poland Health-related quality of life is regarded to be a ‘latent construct’ that can be measured indirectly. According to Fayers et al. (1997), two kinds of indicators of quality of life (QoL) may be differentiated within questionnaires: some items like assessments of physical symptoms may be called causal indicators because their occurrence can cause a change in QOL, but a poor QOL need not necessarily imply that a patient suffers from these symptoms. Other items can be regarded as effect indicators. They reflect the level of QOL and a patient with a poor QOL is likely to have low scores on all effect indicators. The aim of the study was to test a twofold structure of QoL concept. A confirmatory factor analysis (CFA), based on data from 223 gastrointestinal cancer patients (57.4% male) whose QOL was assessed on up to three occasions (i.e., 1, 6 months and 1 year after surgery), was performed on the items of the EORTC-QLQ-C30. Maximum likelihood estimation was employed to estimate the hypothesized model. The items referring to physical symptoms like shortness of breath, constipation, diarrhea and that item asking about financial difficulties were selected as causal indicators of QOL in gastrointestinal cancer patients whereas all other items were found to be effect indicators. This model was shown to be stable over time. The model-data fit was satisfactory, v2 ¼ 554.2, df ¼ 326, TLI ¼ 0.92, CFI ¼ 0.93, RMSEA ¼ 0.05, SMC ¼ 0.39. The results support the hypothetical twofold structure of QoL concept. The EORTC-QLQ-C30 mainly comprises eect indicators of quality of life, i.e., items which represent this latent construct. Further analyses should be made to investigate if these eect indicators might form a summary scale for QOL.

# 1482/INTERRELATIONSHIP BETWEEN HEALTH INDICATORS AND QUALITY OF LIFE: RESULTS FROM THE CROSS-CULTURAL ANALYSIS OF THE EUROHIS FIELD STUDY Silke Schmidt & Monika Bullinger, Institute of Medical Psychology, University Clinic of Hamburg Eppendorf, Hamburg, Germany; Mick J. Power, Edinburgh, UK The aim of the cross-cultural analysis of the Eurohis field study was to determine the performance of various health indicators to predict quality of life, mental health and general health. The Eurohis study includes a broad range of health care and health behaviour related indicators, such as preventive care, health care utilisation, use of medicine, physical health, mental health, alcohol consumption, physical activity and quality of life. Data were based on the Eurohis field study which is a WHO coordinated health interview survey co-sponsored by the EC. Data on various health indicators and quality of life were collected from 10 countries amounting to a sample size of 4849 (2750 females and 2099) males. Rather than taking a descriptive approach, an analytical approach was taken to investigate the interrelationships between indicators of each particular indicator set (such as alcohol consumption) and between conceptually different indicator sets. Both regression analyses as well as path analyses were employed. Analyses were conducted pooled across all countries as well as separately for different groups of countries (western/central European states, Baltic states, south/eastern European states). In regression analyses, results showed strong and consistent effects of 18 health behaviour indicators and psychosocial determinants on quality of life (R2 ¼ 0.43), the MHI (R2 ¼ 0.50) or general health (R2 ¼ 0.45), however a dierential eect of socio-demographic variables, in particular education, and health behavioral determinants was found in dierent groups of countries. In the structural equation modelling, very good fit indices were observed for the models determining physical and mental health. However the model fit was quite low when quality of life was considered as the primary outcome criterion. Findings suggest that quality of life is a mediator of mental and physical health implying that these are eect indicators of QoL in this particular set of health indicators and this European sample. # 1343/QUALITY OF LIFE – INFLUENCING FACTORS BEING PRESENT WITHIN THE PAST 20 YEARS Ute Ellert, Ba¨rbel M. Kurth, Friedrich Tiemann & Jutta Wirz, Epidemiology and health reporting, Robert Koch-Institute, Berlin, Germany Several factors are known to influence health related quality of life. The aim of the presented study is the determination of influencing factors on todays quality of life and an occurrence within during the past 20 years. The study results are based on the ‘Spandauer Gesundheitstest’, a prospective cohort study of the Robert Koch Institute in Berlin, Germany. The eleventh examination wave started this year. Approximately 7000 adults have been interviewed and examined in the context of the study over a time period of up to 20 years. In the tenth wave, among other things the short form 36 questionnaire (SF-36) was included for the first time to assess health related quality of life. To include the information of all observation periods in the analyses, a retrospective index was built. The values of this index are higher the closer an event of interest is to the present. The underlying thought of the index building was the assumption that an event, which dates far back, has less influence on the present than on that dates back a short time only. The index variables have been condensed into four characteristics: never had, formerly had, acute and chronic. The means of the SF-36-scales have been compared for these four characteristics. Because of the age dependence especially of the physical scales, the analyses have been carried out controlled for age. The most obvious effect can be seen in the self-rated state of health. Subjects, who rated their state of health very good or good continuously, reached the highest values in all SF-36-scales. The lowest values in all SF-36scales were obtained for people who rated there state of health to be bad during the whole observation period. Furthermore, we have calculated linear regression models for the eight SF-36-scales for women and men separately. Among others, the self-rated state of health, frequent hospital admissions, diseases of the locomotor system, pathologic blood or urine test results or continuous sick leave of more than 6 weeks were significant negative influencing factors regarding health related quality of life.

824 # 3008/GENDER DIFFERENCES IN HEALTH REPORTED BY CHILDREN AND ADOLESCENTS A. Riley, G. Rebok & J. Robertson, Department of Health Policy and Management, Johns Hopkins University, Baltimore, Maryland, USA Women and men experience their health differently. Young women are more likely to report physical and emotional symptoms and young men are more likely to report behaviors that pose a risk to their health. However, almost no research has focused on when over the life course these gender differences emerge. In a group of 1680 children and 842 adolescents, we compared gender differences in health on five domains of the Child Health and Illness Profile. Children 6– 11 years reported on the Child Report Form, adolescents 11–17 on the Adolescent Edition, and parents of the younger children reported on the Parent Report Form. Girls ages 6–17 were significantly healthier than boys in terms of their Risk Avoidance and Achievement. Conversely, girls had significantly worse health on Comfort by their own reports throughout ages 6–17 years; parents of younger children 6–11 did not report this difference. Adolescent girls, but not young girls, had significantly lower satisfaction than boys. No differences in socioeconomic status exist between the girls and boys that might account for these findings. These results document that the beginning of gender differences in perception of symptoms, achievement of social roles, and engagement in risky behaviors begins early in childhood and is quite consistent through adolescence. Further, cognitive interviewing data from an earlier study suggest that girls and boys selectively exaggerate symptoms and risk-taking activities, respectively. # 3007/EVIDENCE OF GENDER DIFFERENCES IN ADULTS BY CONDITION, TYPE OF STUDY AND OUTCOME MEASURE Michael Herdman, 3D Health Research, Barcelona, Spain Gender differences in scores on HRQOL instruments have been documented in various studies, but there have been few systematic reviews of these differences. The relationship between gender and results on HRQOL instruments is complex. In health surveys in the general population using generic HRQOL instruments, women tend to have lower (worse) scores on many HRQOL dimensions, though this is not necessarily true of all age groups and all dimensions. Results may also vary depending on the instrument used. In patient populations, the impact of gender on HRQOL scores varies considerably between conditions. In cancer patients, being male has been shown to predict higher HRQOL scores, but only on physical domains; in patients with heart failure, recent research shows very few differences between men and women, with women showing statistically significant worse HRQOL only in terms of emotional functioning, and improvement over time was similar in men and women; in patients with inflammatory bowel disease, being male tended to predict better HRQOL. In determining the nature and extent of gender differences in HRQOL scores, therefore, it is important to take into account several aspects, including: the type of study performed (health surveys, clinical research, other); the population addressed (e.g. general population or patient populations); the type of measure used (generic, diseasespecific, etc); whether global scores or dimension scores are being considered; whether static or change scores are being considered; degree of control for confounding factors. The aim of this part of the symposium will be to review the way in which gender impacts on results using different types of HRQOL instruments and in different measurement conditions, focussing particularly on the interplay between gender, HROQL dimensions and underlying medical condition, though the impact of gender on changes in HRQOL scores will also be analysed. The focus will be principally on adult populations.

# 1374/ARE PERSONALITY, PAIN AND STRESS PREDICTORS OF IMPORTANCE FOR SELF-RATED HEALTH? A 30-YEAR FOLLOW UP STUDY Pia Svedberg, Carola Bardage & Nancy L. Pedersen, Department of Medical Epidemiology & Biostatistics, Karolinska Institutet, Stockholm, Sweden Several investigators have studied the predictive value of self-rated health on future morbidity, functional capacity and mortality. Less well understood is what predicts self-rated health. The purpose of this prospective study was to explore the importance of personality (neuroticism and extraversion), pain and stress including shift work, unemployment and overtime for self-rated health. Data were collected 25–30 years ago in the Swedish Twin Registry and concurrently in a recently finished study. The design of this effort was of longitudinal nature, and provides insight into long-term predictors of self-rated health in a subset of the registry sample that is currently 42 years or older (n ¼ 24,452). Three self-rated health items developed for the Older Americans Resource Survey were included: (1) How would you rate your general health status? (2) How would you rate your health status compared to 5 years ago? (3) Do you think your health status prevents you from doing things you would like to do? Items were standardized separately (M ¼ 0, SD ¼ 1) and then summed. A high score indicates a more positive health rating. Results of linear regression analyses show that people high on the neuroticism scale, suffering from back-, neck- or shoulder pain or those who have recurrent headaches generally scored lower on the self-rated health scale. Extraverted people, on the other hand, had a better self-rated health status, although neuroticism was a stronger predictor of health than extraversion. Women, older persons, and those who experienced perceived stress, shift work and unemployment scored lower on selfrated health. Suffering from migraine and working overtime was not associated with self-rated health. We conclude that personality as well as chronic pain and stress measured 30 years ago predict future selfreported health status. # 3006/EVIDENCE OF GENDER INEQUALITIES IN SPANISH TEEN HEALTH PROFILE-TYPES (CHIP-AE) V. Serra-Sutton, L. Rajmil, J. Alonso, A. Riley & B. Starfield, Catalan Agency for Health Tech Assess and Research, Barcelona, Spain The CHIP-AE is a generic health status measure for teens that proved to be reliable and valid in Spain. The CHIP-AE includes six domains: Satisfaction, Discomfort, Resilience, Risks, Achievement and Disorders. This measure allows creating health profile-types with the combination of scores in four domains. The aim of this study was to assess gender inequalities using the CHIP-AE domains and health profiletypes as an outcome measure. The Spanish CHIP-AE was administered to a sample of 902 adolescents in Barcelona. The 13 profiletypes were developed and validated in three US samples. Cut-off scores were defined as 0.6 SD of a standardized mean of 20 in each domain: e.g. low (<17), medium (17–23), and high satisfaction (>23). For instance, an adolescent in profile B (good health) scored between 17 and 23 in all domains; while a teen in profile M (worst health) scored below 17 in three or four domains. Following the original US results, gender differences were expected on subdomain and domain scores and also health profile-types. The probability of being in a profile of excellent/good health and also worst health by gender was calculated using logistic regression models. Girls scored lower (worse) on satisfaction with health and self-esteem and also presented worse scores on emotional discomfort. Boys showed worse scores on threats to achievement and negative peer influences and academic achievement. Comparing US and Spanish samples, some differences were found, girls in the Spanish sample presented worse scores on Risk domain than girls in the US. Similar to the US, girls had a higher probability of being in profile M (odds ratio, OR: 2.1 [95%CI: 1.4–3.3] 5%CI: 0.28–0.84). Results from this study evidence a similar pattern of gender inequalities measured with the CHIP-AE in the US. Risk behaviors might be more socially accepted in Spain than in the US and for this reason risk patterns are different.

825 # 1346/COPING AND PREVIOUS FUNCTIONING AS PREDICTORS OF QUALITY OF LIFE IN SIBLINGS OF CHILDREN WITH CANCER Bregje A. Houtzager & Martha A. Grootenhuis, Paediatric Psychosocial Department, Academic Medical Center, University of Amsterdam, Amsterdam; Josette J. Hoekstra-Weebers, Medical Psychology, Academic Hospital, University of Groningen, Groningen; Bob F. Last, Paediatric Psychosocial Department, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands Aim of the present study is to describe the quality of life (QoL) of siblings of a child with cancer shortly after diagnosis, and to predict sibling QoL according to ways of coping with the illness and the sibling’s functioning before diagnosis. Participants were 83 siblings from 56 families, 46 girls and 37 boys, aged 7–18. The siblings’ QoL was assessed 4–8 weeks after the diagnosis in the ill child with the TNO/ AZL children’s Quality of Life Questionnaire (TACQOL). Coping strategies were assessed with the Cognitive Coping Strategies Scale for siblings (CCSS-s). Physical problems and problems sleeping and eating previously to the ill child’s diagnosis were determined in a structured interview with the parents. A substantial number of siblings (26–56%) reported impaired QoL, compared to the reference group (approximately 20%). Cognitive functioning, negative emotions and positive emotions were impaired, compared to peers. Besides, school aged siblings (7–11) reported more trouble with motor functioning than their peers. Predictive cognitive control, or the ability to maintain positive expectations regarding the illness, and functional problems before diagnosis, predicted the sibling’s QoL. The tendency to seek instead of to avoid information about the illness co-occurred with more negative emotions in this acute phase of the illness. It remains unclear, however, whether distressed siblings seek more information, or whether information distresses the sibling. Shortly after the diagnosis of cancer in a brother or sister, children experience lower emotional and and cognitive QoL. In young children, distress may be expressed more indirectly in motor problems. Functional complaints before diagnosis may be indicative of vulnerability for later psychosocial problems. The ability to remain optimistic about the illness course seems a relevant protective factor. The importance of timing and of the amount and kind of information about the illness offered to siblings needs further exploration. # 1792/THE VALIDITY AND UTILITY OF BPI INTERFERENCE MEASURES ADDRESSING THE IMPACT OF PAIN Valerie Williams, RTI-HS, Research Triangle Park, NC; Meredith Y. Smith, Health Economics and Outcomes Research, Purdue Pharma, LP, Stamford, CT; Sheri E. Fehnel, RTI-HS, Research Triangle Park, NC, USA In preparation for upcoming clinical trials in osteoarthritis (OA), the psychometric properties of two components of the BPI, a well-validated measure of pain and its impact, were assessed using data from two previously conducted trials of pain medications. Specifically, the total BPI pain-related interference with functioning subscale (7 items) and the pain-related interference with sleep item were examined. Study #1 (n ¼ 133) included patient-reported ‘number of night awakenings with pain’ and ‘quality of sleep’ in addition to the two BPI measures. Results showed that ‘night awakenings with pain’ was positively and significantly correlated with the total BPI score (r ¼ 0.36–0.47, p < 0.0001) and the sleep interference item (r ¼ 0.55– 0.57, p < 0.0001). Both BPI measures correlated negatively with ‘quality of sleep’ (r ¼ )0.72 to )0.53, p < 0.0001). In Study #2 (n ¼ 107), a single item on the WOMAC – degree of pain experienced ‘at night while in bed’ – was found to correlate more highly with the BPI sleep interference item (r ¼ 0.55–0.86, p < 0.0001) than with the BPI subscale (r ¼ 0.42–0.68, p < 0.0001), confirming that pain ‘at night while in bed’ is more closely connected to the extent to which pain disrupts sleep than other activities. Intraclass correlation coefficients of 0.75 and 0.63 for the interference subscale (for Study #1 and #2, respectively) and 0.56 and 0.52 for the sleep-interference item denote adequate test–retest reliability. Guyatt’s statistics provided strong evidence of the responsiveness of the BPI measures: Moderate and large effect sizes were found, ranging from 0.46 to 1.14 and signifying that the treatment effects, as assessed by the two measures of the impact of pain, were considerably larger in the treatment arms than the placebo groups. The results of these analyses supplement the existing

wealth of knowledge and experience with the BPI, confirming the validity and sensitivity of both the BPI Interference subscale and sleep interference item as potential endpoints in trials of pain medications involving patients with OA. # 1500/IMPROVING OUR HAPPINESS AND SATISFACTION WITH THE SEIQOL – A QUESTION OF WEIGHTING Jonathan Todman, Saskia Teunisse & Louise Phillips, Department of Psychology, University of Aberdeen, Aberdeen, Scotland The Schedule for the Evaluation of Individual Quality of Life-Direct Weighting (SEIQoL-DW) is a measure of individual quality of life (QL) and asks respondents to name the five most important areas of their life, to rate their satisfaction with each area and to weight the importance of each area. A total QL score is obtained by summation of the products of the satisfaction ratings and weightings for each area. Previous research using other QL measures showed low correlations between satisfaction ratings and weightings, suggesting an independent contribution of ratings and weightings to QL. However, it was also found that total QL scores based on ratings and weightings were not necessarily better predictors of global QL responses than the summed rating scores representing no weighting (Hsieh, 2003). Indeed, there has been some suggestion that weighting by importance may reduce the effectiveness of QL scales (Trauer and Mackinnon, 2001). To date little is known about the actual contribution of the weighting procedure within the SEIQoL. The current study, therefore, investigates (1) the relationship between satisfaction ratings and weightings in the SEIQoL, and (2) whether a total SEIQoL score based on a summation of the product of satisfaction ratings and weightings is a better predictor of global QL than a total SEIQoL score based on a summation of the satisfaction ratings only. Data have been collected in a general population sample (n ¼ 80). Results will be presented, and the implications of the findings for QL measurement will be discussed. # 1539/THE ITALIAN CORONARY REVASCULARISATION OUTCOME QUESTIONNAIRE (CROQ-PTCA): A PRELIMINARY PSYCHOMETRIC EVALUATION Sara Schroter, BMJ Editorial, BMA House, London, Great Britain; Plinio Pinna Pintor & Salvatore Colangelo, Arturo Pinna Pintor Foundation, Turin; Beatrice Dilaghi, Medical Clinic, University of Florence, Florence; Bruna Catuzzo & Dora Fonzo, Arturo Pinna Pintor Foundation, Turin, Italy The Coronary Revascularisation Outcome Questionnaire (CROQ) is a validated instrument measuring symptoms, physical, psychosocial and cognitive functioning, adverse effects and satisfaction with treatment for coronary artery bypass graft surgery (CABG) and percutaneous transluminal coronary angioplasty (PTCA). It was originally validated for use with English speaking patients and this paper reports findings from the preliminary psychometric evaluation of the Italian translated version of the pre-revascularisation CROQ-PTCA. Methods of translation and cross-cultural adaptation have previously been reported. Standard analyses were used to evaluate acceptability, reliability and validity in a sample of 47 patients before PTCA. The Italian translated pre-revascularisation questionnaire has similar properties to the English version. The 33-item CROQ-PTCA was acceptable to patients (missing data <5%), satisfied scaling assumptions (maximum endorsement frequency <80%, aggregate endorsement frequencies <10%) and showed good internal consistency (Cronbach’s a coefficients for all scales P0.80, item–total correlations >0.20), construct validity (high correlations between subscales and total scores, moderate correlations between subscales. Further confirmation of the psychometric properties of the Italian versions of the CROQ-PTCA and CROQ-CABG (including an extensive validation against other disease-specific measures) are currently being undertaken in large samples of patients recruited from three hospitals across Italy as part of a larger study evaluating clinical and psychosocial risk factors for predicting quality of life and clinical outcome after CABG and PTCA.

826 # 1240/PSYCHOMETRIC VALIDATION OF THE BENIGN PROSTATIC HYPERPLASIA PATIENT IMPACT MEASURE (BPH-PIM) Mona L. Martin & Donald M. Bushnell, Health Research Associates, Inc., Mountlake Terrace, WA, USA; Asha V. Hareendran, Outcomes Research, Pfizer, Sandwich, Kent, UK Objective: Patient reporting of the impact of Benign Prostatic Hyperplasia (BPH) symptoms on their lives are important for treatment decision making and evaluating outcomes of new treatments. The BPHPIM is a unique measure benefiting from the strengths of both standardized as well as individualized items. Our objective is to provide the psychometric properties of this newly developed measure. Methods: Eligible subjects (recruited from a clinical trial) were at least 40 years of age, with documented evidence of benign prostatic enlargement and an International Prostate Symptom Score (IPSS) score of >12. Cross-sectional validation was conducted at the baseline visit, test– retest reproducibility between screening and baseline (2-weeks), and responsiveness 4 weeks after baseline (using a global rating of change). Internal consistency was measured using Cronbach’s a, reproducibility using intraclass correlation coefficients (ICC), and responsiveness with effect sizes. Results: A total of 588 men with a mean age of 63.2 were assessed. Item reduction statistics showed no items: 1) with > than 5% missing, 2) exhibiting a ceiling effect, 3) with item-to-total correlations < 0.50, 4) with item-to-item corrections > 0.70. The measurement model (total score plus 3 subscales) was confirmed using factor analyses. Alpha coefficients were all above 0.72 (total ¼ 0.92), and ICC’s were above 0.81 (total ¼ 0.87). The BPH-PIM total score showed a strong association with the IPSS score (0.46, p < 0.001) and with the American Urological Association QoL item (0.53, p < 0.001). BPH-PIM scores were able to discriminate between levels of severity (F ¼ 30.6, p < 0.001) and QoL levels (F ¼ 28.8, p < 0.001). Minimally important differences were found to be 8.5 points (effect size ¼ 0.55). Conclusion: Psychometric performance of the BPH-PIM is consistently high and it appears that it is capable of demonstrating statistically meaningful improvement in patient impact related to urinary symptoms due to BPH. # 1121/RELIABILITY, RESPONSIVENESS, LONGITUDINAL AND CROSS-SECTIONAL CONSTRUCT VALIDITY OF THE PEDIATRIC ASTHMA QUALITY OF LIFE QUESTIONNAIRE (PAQLQ) IN A DUTCH POPULATION CHILDREN WITH ASTHMA Hein Raat, Public Health; Herman J. Bueving, General Practice, Erasmus MC University Medical Center Rotterdam, Rotterdam, ZH, The Netherlands; Elizabeth J. Juniper, Health Sciences, McMaster University, Bosham, West Sussex, UK; Marion H. Grol, Pediatrics, Groningen University, Groningen, G, The Netherlands; Johannes C. van der Wouden, General Practice, ErasmusMC, Rotterdam, ZH, The Netherlands In a large sample of pediatric asthma patients we evaluated the Dutch version of the 23-item Pediatric Asthma Quality of Life Questionnaire (PAQLQ) that was developed by Juniper et al. in Canada. The presence of ceiling effects, the internal reliability, responsiveness, and the longitudinal and cross-sectional construct validity with regard to the three domains (‘Symptoms’, ‘Activities’, ‘Emotions’) and the overall PAQLQ-scores were assessed. Our study group of 238 6–18 year olds was a subgroup of participants in an influenza vaccination trial. All had used maintenance therapy for asthma or frequent relief medication in the previous year. During one winter season the participants completed symptom diaries and participated in a ‘baseline’ PAQLQ with regard to 7 days with diary-symptom scores below a predefined level, which was contrasted with an ‘episode’-PAQLQ regarding 7 days with diary-symptom scores above a predefined level. ‘Baseline’/’episode’ measurements were compared. Additionally, in a subgroup of 101 children scores regarding two subsequent episodes were compared. Overall, the properties of the Dutch PAQLQ were found to be similar to those originally established in Canada by Juniper et al. The domain ‘Emotions’ showed a ceiling effect. ‘Activities’ was the only domain with Cronbach’s a below 0.70. Mean PAQLQ-score changes between ‘baseline’/’episode’ (n ¼ 238) were significantly different (p < 0.01) from changes between two consecutive ‘episodes’ (n ¼ 101) indicating sufficient responsiveness. (Changes in) lower respiratory tract diary-symptom scores, indicating asthma severity, correlated better with (changes in) PAQLQ scores than (changes in)

upper respiratory tract diary-symptom scores, which supports the longitudinal and cross-sectional construct validity. In conclusion, the data showed satisfactory psychometric properties, excellent responsiveness, and longitudinal and cross-sectional construct validity. Issues to be explored are test–retest reliability and the reliability of ‘Activities’. # 1716/PSYCHOMETRIC PROPERTIES OF THE MOS-SLEEP USING INTERNATIONAL CLINICAL TRIAL DATA Christine de la Loge & Muriel Viala, Mapi Values, Lyon, France, Susan Martin, Global Outcomes Research, Pfizer, Ann Arbor, MI; Ron D. Hays, UCLA Division of General Internal, Los Angeles, CA, USA The reliability and validity of the English-language version of the MOS Sleep scale has been evaluated, but the psychometric properties of translated versions have not been reported. Translations of the MOSSleep scale for use in six countries (Germany, Hungary, Poland, Australia, the United Kingdom and South Africa) are evaluated using data from a double-blind, randomised, multicenter clinical trial conducted to evaluate the efficacy and safety of pregabalin compared to placebo (n ¼ 396). The MOS Sleep scale was collected along with the SF-36 and clinical data at baseline and termination visits. Internal consistency reliability estimates for the 9-item sleep problems scale ranged from 0.76 in Hungary to 0.90 in the United Kingdom. The level of correlation between the MOS Sleep scale and the SF-36 vitality score supported the construct validity of the sleep measure (r ¼ 0.53 for the MOS-Sleep Index). Significant differences on MOS-Sleep scores (except for snoring) were obtained between severity groups (p O 0.0001 for the Index score), indicating a high discriminative power of the scale. In addition, the MOS Sleep problems scale was found to be highly responsive to clinical changes as measured through a variety of criteria (ES ranging from )0.56 to )1.42 for improved patients). Results of this study provided preliminary support for the reliability and validity of the six translations of the MOS Sleep scale. # 1425/INTEGRATING IRT- AND UTILITY-BASED ASSESSMENT OF HEALTH: OVERVIEW OF METHODOLOGY Jacek A. Kopec, Health Care and Epidemiology, University of British Columbia; Eric C. Sayre, ARC, Arthritis Research Centre of Canada, Vancouver, BC; Aileen M. Davis, Research, Toronto Rehabilitation Institute; Elizabeth M. Badley, Public Health Sciences, University of Toronto, Toronto, ON; Michal Abrahamowicz, Epidemiology and Biostatistics, McGill University, Montreal, PQ; Aslam H. Anis, Health Care and Epidemiology, University of British Columbia, Vancouver, BC; Jack I. Williams, Research, Toronto Rehabilitation Institute, Toronto, ON; John M. Esdaile, Rheumatology, University of British Columbia, Vancouver, BC, Canada We developed a new health questionnaire for people with musculoskeletal conditions. Innovative features of this instrument include (1) adaptive paper-and-pencil format and (2) utility-based scoring formula. Using data on 851 subjects with arthritis, 5 item banks were developed to measure the following domains: ambulation (35 items), handling objects (36), daily activities (36), pain (33) and emotional function (42). Muraki’s generalized partial credit model (GPMC) was employed to derive item properties. Items were selected by examining item information, location, dimensionality, and differential item functioning. The final adaptive questionnaire is completed by selecting one of four sets of items per domain, depending on the answer to a 4-level initial question, with five multi-option items in each set. The items for each health level were chosen based on item and test information and other properties. A model was developed to transform maximum likelihood (ML) score estimates from the GPCM into single-attribute preferences by measuring standard gamble (SG) preferences for 15 response patterns per domain, corresponding to different ML estimates. In a parallel study, we measured SG utilities and rating scale values for 256 complex states with multiple deficits, defined by five domains and four health levels per domain. We also measured values for all states in which only one domain was affected. Data from these studies allowed us to model overall utility as a function of single-attribute preferences and provided a link between IRT- and preference-based assessments of health status. Preliminary evidence suggests that our approach is feasible and combines the advantages of adaptive assessment with the interpretability of health utilities.

827 # 1750/RESPONSIVENESS AND GUIDANCE FOR INTERPRETATION OF A QUALITY OF LIFE QUESTIONNAIRE SPECIFIC TO UPPER-GASTROINTESTINAL DISORDERS: THE PAGI-QOL Elyse Trudeau & Christine De La Loge, Mapi Values, Lyon, France; Dominique Dubois, Pharmaceuticals Group Strategic Marketing Health Eco, Janssen Pharmaceutica, Beerse, Belgium The Patient Assessment of upper GastroIntestinal disorders-Quality of Life (PAGI-QOL) has been developed and validated to assess QoL in gastroesophageal reflux disease (GERD), dyspepsia and gastroparesis. The aim of this work was to assess responsiveness of the PAGI-QOL to clinical changes and provide guidance for the interpretation of score changes. The analysis was based on two observational multicenter, longitudinal 8-week observation studies, one in the US and one in five European countries. The Overall Treatment Effect scale completed by patients and clinicians at 8 weeks was used to define groups of improved, worsened and stable patients. Responsiveness was assessed by describing the change in PAGI-QOL scores for each of these groups, comparing these changes over time and calculating Effect sizes and standardized Response Means. These data were used to define the minimal clinically important difference (MID). A total of 1736 patients suffering from UGID were analysed. High sensitivity to change over time was demonstrated with Effect Sizes in improved patients for the PAGI-QOL total score ranging from 0.65 to 1.07. Effect sizes at dimension level were greater than 0.36. A change of 0.4 for the PAGI-QOL total score may be considered as the minimal clinically important difference when comparing pre and post treatment PAGI-QOL total scores. The PAGI-QOL questionnaire is a responsive and clinically relevant instrument for assessing quality of life in patients with upper gastrointestinal disorders. # 3003/DEVELOPMENT AND VALIDATION OF A HEALTH-RELATED QUALITY OF LIFE SURVEY FOR INDIVIDUALS WITH SCHIZOPHRENIA LIVING IN THE COMMUNITY Judy T. Barr, School of Pharmacy, Northeastern University Phase I: We have developed a 51-item, self-administered, health-related quality of life survey for use with individuals with schizophrenia living in the community, the SOAP-51 (Schizophrenia Outcomes Assessment Project). Eight factors and 51 items were identified through factor analysis and item reduction based on nine administrations with 443 clients in Massachusetts. Cronbach’s a for factors ranged from 0.624 to 0.881; factor test/retest reliability, 0.781–0.868. Phase II: Validity was evaluated in 1535 clients in five geographically diverse states over four administrations. Trend analysis indicated that factor scores decreased as patients reported more impact of the condition using an independent, global factor assessment item (p < 0.001). Staff assigned clients to four levels of functionality; all but two factor scores trended upward as the level of functionality improved (p < 0.001). Phase III: Based on results from two administrations with 193 clients in Massachusetts, item weighing did not contribute to significant differences in either factors or total scores. Works in progress: (1) clients’ and clinicians’ focus groups to convert research survey format for everyday use and (2) examination of survey’s clinical utility to clients and clinicians. Client input has been incorporated throughout its development: initial item identification, face validity, item weighting mechanism and relative importance of items, and survey design. # 3004/THE DEVELOPMENT AND PSYCHOMETRIC VALIDATION OF THE SCHIZOPHRENIA QUALITY OF LIFE SCALE-REVISION 4 (SQLS-R4) Diane Wild, Oxford Outcomes Ltd. The original SQLS was a 30-item self-completion questionnaire comprising three domains (Psychosocial, Energy/Motivation, Side Effects). Changes made following further developmental work including additional item generation and cognitive debriefing interviews and a

translation and cultural adaptation process resulted in a number of revisions culminating in the 47-item SQLS-R3. The objective of the current study was to assess the psychometric properties of the SQLSR3. Two hundred and three patients with a diagnosis of schizophrenia completed the SQLS-R3. Over half the patients also completed the Short Form-36 (SF-36), Psychosis Evaluation Tool for Common use by Caregivers (PECC) and the SQLS-R3 for a second time. Exploratory factor analysis of the SQLS-R3 resulted in a 3-factor structure after the removal of seven items that did not load on any factor. The three domains were named Psychosocial (20 items), Vitality (13 items), and Side effects (seven items). All three domains had good internal reliability consistency. However, based on a number of findings including the results of Structural Equation Modelling (SEM) confirmatory factor analysis, the Side effects domain was removed from the SQLS-R3 and a 2-factor model was taken forward for further psychometric evaluation. This 33-item measure was renamed the SQLS-R4. Convergent and Divergent validity of both domains of the new measure were found to be excellent. The sensitivity/discriminant validity of both domains was very good in relation to both disease severity measured by the PECC and type of medication. Both domains also displayed good test–retest reliability. The SQLS-R4 is a 33-item, 2 domain (Psychosocial 20 items; Vitality 13 items) disease specific health related quality of life (HRQOL) measure. It has excellent psychometric properties and will be available in the near future for use in clinical practice and research. # 1324/QUALITY AND QUANTITY IN ITEM GENERATION AND REDUCTION IN DEVELOPING A SPECIFIC MEASURE FOR WOMEN IN MATURE LIFE (THE S.M.I.LE.) Mauro Niero, Education, University of Verona, Verona, Italy; Thomas Kohlmann, Social Medicine, University of Greifswald, Greifswald, Germany; Anne Marciniak, Pfizer, Sandwich, UK; Sally Shumaker, Wake Forest University, Wake Forest, NC, USA; Hanne Thorsen, General Practice, University of Copenhagen, Copenhagen, Denmark; Linda Abetz, Mapi Values, Bollington, Cheshire, UK QoL questionnaires related to mature life conditions often focus on expert-selected symptoms of diseases related to ageing. In contrast, the SMILE project sees mature life in women as encompassing both positive and negative experiences. Qualitative (Focus Groups, In Depth Interviews and Nominal Group Techniques) and quantitative psychometric strategies were combined in item generation and reduction so as to capture the typical nuances of this population in US, Denmark, Germany, Spain, France and Italy. Phase 1: 208 items were drawn from literature, existing questionnaires and FGs in participating countries. Phase 2: NGT selected 72 items. Phase 3: IDI with 20 US women resulted in a 60 item test version. Phase 4: The test version was administered to 572 US women to assess its ability to discriminate between health states. Phase 5: FGs and IDIs aimed at increasing the questionnaire’s sensitivity and cross-cultural applicability. Adding new items and rewording produced a 118-item questionnaire. Phase 6: This questionnaire was tested on 823 women in a cross-sectional (healthy/with disease) and a 3-month cohort (women taking HRT) study. The conceptual structure of the SMILE determined in phases 1– 3 included items on feelings on ageing, specific symptoms, cognitive problems and daily-life. In the Phase 4 survey sensitivity resulted as equivalent to existing questionnaires, hence the need for further improvement. The version produced in Phase 5 included 10 scales: happiness, mood, perspectives about the future, social interaction, appearance, comfort/mobility, cognition, sleep, energy and genitourinary health. Analysis after Phase 6 yielded a 61-item and 10-scale version. Sensitivity, responsiveness, floor/ceiling effect and scaling properties were the main criteria for item reduction. Shorter versions (30 and 15 items) are now being developed. By integrating qualitative and quantitative strategies the women’s voice was heard throughout the development of the questionnaire, by making item-reduction decisions based on psychometric properties with constant face-validity support.

828 # 1289/URINARY URGENCY: CAN ‘GOTTA GO’ BE MEASURED? Karin S. Coyne, Louis Matza, Center for Health Outcomes Research, MEDTAP International, Bethesda, MD; Eboo Versi, Pfizer, Peapack, NJ, USA Urinary urgency, the strong, sudden desire to urinate, is a central and particularly bothersome hallmark symptom of overactive bladder (OAB) due to its unpredictable nature and consequent impact on daily living. There are currently no standardized measures of urgency; thus the purpose of this study was to evaluate the psychometric properties of a newly developed measure of urinary urgency. Based on qualitative patient interviews and clinician reviews, the Urgency Questionnaire (UQ) was developed with 20 Likert-items and four visual analog scales. The UQ was administered at baseline and end of study (10 days) in a tolterodine trial of women with OAB. Micturition diaries, global assessments, and perception of treatment benefit questions were collected. Exploratory factor analysis was used to examine the item and subscale structure. Internal reliability, construct validity, and responsiveness were assessed using Cronbach a, ANOVA, correlations, t-tests, and effect size. A total of 974 women completed the baseline UQ with a mean age of 48.9 years; 84.6% were Caucasian. Approximately 74% rated their bladder problems as moderate or severe. After item reduction, the final UQ consisted of 15 items with four subscales (Fear of Incontinence, Time to Void, Impact and Nocturia) and four VAS. Cronbach a’s ranged from 0.79 (Fear) to 0.91 (Nocturia). The UQ subscale correlations were small to moderate with the micturition diary variables ranging from 0.01 to 0.57. All UQ subscales but nocturia significantly discriminated among levels of patient-rated bladder condition, urge rating, incontinence episodes, and number of pads used. The nocturia subscale significantly discriminated among the levels of urinary frequency and nocturia. Effect sizes ranged from 0.43 to 0.76 for the subscales and from 0.70 to 1.27 for the VAS. The UQ has excellent reliability, construct validity, and responsiveness. Patients experiencing urinary urgency can quantify this symptom and perceive change in urgency. The UQ appears to be a useful and valid measure for urinary urgency. # 1123/DEVELOPMENT AND VALIDATION OF A DISEASE-SPECIFIC QUESTIONNAIRE FOR THE EVALUATION OF HEALTH RELATED QUALITY OF LIFE IN PATIENTS WITH BENIGN PROSTATE HYPERPLASIA Alessandro Bertaccini, Urology; Roberta Ceccarelli, Urology, SAMUR onlus; Marco Urbinati, Francesco Costa, Giovanni Vitullo & Antonio Martinelli, Urology; Deborah Marchiori, Urology, SAMUR onlus; Giuseppe Martorana, Urology, University of Bologna, S. Orsola Malpighi Hospital, Bologna, Italy The present study proposes the development and the validation of an instrument to evaluate the Quality of life (QoL) of patients affected by benign prostatic hyperplasia (BPH). Among the available questionnaires, up to now known, no one contemplates the subjective satisfaction. The BSP-BPH, was built on the basis of previous validated ones and on the experienced acquired by the use of the Bononian Satisfaction Profile-Prostate Cancer (BSP-PC) a similar disease specific instrument concerning prostate cancer. After three successive meetings with experts (n ¼ 10) and patients (n ¼ 20) we developed a 31 items questionnaire, that we administered to 125 patients, in addition to the EuroQol. Further meetings and statistical analyses defined the 18 items of the BSP-BPH. This one plus the EuroQol questionnaire was administered to 435 patients that came under our

observation during a campaign for the early detection of prostatic diseases among the male population of Bologna. The mean age of the participants in the development phase was 63 years; 47.1% of the sample had diagnosis of BPH. The 18 items were selected on the basis of the following criteria: (1) r > 0.50, p < 0.05 (correlation test– re-test), (2) p < 0.05 (ANOVA presence vs. absence of BPH), (3) r > 0.50, p < 0.05 (correlation with EuroQol). The participants mean age, in the validation phase, was 63 years. Sixteen patients with a history of any kind of neoplasia were excluded from the sample (21 were missing data). Fourty five percent of the sample (n ¼ 398) had diagnosis of BPH. Principal component analysis identified fivepercent components (1) satisfaction about sexual functionality (Cronbach a ¼ 0.94), (2) satisfaction about social functionality (a ¼ 0.80), (3) satisfaction about cognitive/emotional functionality (a ¼ 0.82), (4) satisfaction about urinary functionality (a ¼ 0.87) (5) satisfaction about physical functionality (a ¼ 0.66); total Cronbach a was 0.88. The BSPBPH questionnaire can be used as an instrument that evaluates the HQoL in BPH patients by principally considering their subjective satisfaction. # 1638/A CULTURALLY SPECIFIC MEASURE OF VISUAL OUTCOME IN INDIA: DEVELOPMENT AND VALIDATION OF THE INDIAN VISUAL FUNCTION QUESTIONNAIRE-33 (IND-VFQ33) S.K. Gupta, Rajendra Prasad Centre for Opthalmic Sciences, All India Institute of Medical Sciences, New Delhi; K. Viswanath, Sarojini Devi Eye Hospital, Hyderabad 28, Andhra Pradash; R.D. Thulasiraj, Lions Aravind Institute of Community Opthalmology, Madurai, Tamil Nadu; G.V. Murthy, Rajendra Prasad Centre for Opthalmic Sciences, All India Institute of Medical Sciences, New Delhi; Donna L. Lamping & Sarah C. Smith, Health Services Research Unit, Martine Donoghue, Health Policy Unit; Astrid E. Fletcher, Epidemiology Unit, London School of Hygiene & Tropical Medicine, London, UK The current and predicted burden of blindness is highest in low income countries, yet few patient-reported outcome measures have been developed within these countries. To address this issue we developed and validated a new measure of vision-related outcome in India. Extensive qualitative work ensured high content validity relevant to the local context. We developed an item pool of 45 items representing three a priori domains (general functioning, psychosocial symptoms and visual symptoms), then field tested the IND-VFQ45 with a large sample of patients with cataract (n ¼ 120), glaucoma (n ¼ 120), retinal disease (n ¼ 120) and normal controls (n ¼ 120). Psychometric item reduction (based on missing data, maximum endorsement rates, adjacent endorsement rates, item redundancy and item convergent/discriminant analyses) produced a 33-item questionnaire (IND-VFQ33). Psychometric evaluation showed that two of the three scales (psychosocial impact and visual symptoms) had good acceptability, and all three scales showed high internal consistency (a > 0.80; item–total correlations 0.54–0.86) and test–retest reliability (>0.89). As expected, inter-correlations between the three scales were moderately high (>0.65). All three scales showed moderate evidence of convergent and discriminant validity. All three scales significantly differentiated between the three clinical groups and the normal comparison group. Responsiveness, assessed in cataract patients (n ¼ 120) pre-post surgery, was good for all three scales (effect sizes >1), but needs further evaluation in other vision-related conditions. With its strong psychometric properties, the IND-VFQ33 will provide a valuable outcome measure for use in assessing the impact of and treatments for blindness in India.

829 # 1654/THE CERVICAL DYSTONIA IMPACT PROFILE (CDIP-58): 1 – DEVELOPMENT AND VALIDATION Tom T. Warner, Stefan J. Cano, Department of Clinical Neurosciences, Royal Free & University College Medical School, London; Ray Fitzpatrick, Division of Public Health & Primary Health Care, University of Oxford, Oxford; Khailash Bhatia, Private Consulting Rooms; Alan J. Thompson, Neurological Outcome Measures Unit, Institute of Neurology, London; Jeremy C. Hobart, Department of Clinical Neurosciences, Peninsula Medical School, Plymouth, UK Cervical dystonia, also known as spasmodic torticollis, is a chronic progressive neurological disease with substantial public health impact. The availability of a range of medical (e.g. botox injections) and surgical (e.g. selective peripheral denervation) treatments has emphasised the need for rigorous patient-based measures of disease impact, and highlighted the limitations of existing scales. The aim of this study was to develop a cervical dystonia rating scale suitable for clinical trials, epidemiological studies, and routine data collection. There were three stages. First, patient interviews (n ¼ 25), expert opinion, and a literature review were used to generate a pool of items (n ¼ 150) representing the health impact of cervical dystonia. Second, the item pool was pretested (n ¼ 19), modified accordingly, and sent to 556 patients. Analyses of these data led to the construction of a rating scale, the 58-item Cervical Dystonia Impact Profile (CDIP-58), measuring the health impact of CD in eight domains. Third, the psychometric properties of the CDIP-58 and four other rating scales (SF-36, Euroqol, General Health Questionnaire, Hospital Anxiety and Depression Scale) were administered to an independent sample of 391 patients and the full range of psychometric properties evaluated. Two psychometric techniques for scale development and evaluation were used (classical test theory and Rasch technology). The impact of this is discussed in the second abstract. The CDIP-58 satisfied criteria for rigorous measurement as defined by both psychometric techniques suggesting it may be a promising health measure. Further evaluations are required, in particular its ability to detect change. # 1174/DEVELOPMENT OF A NEW QUALITY OF LIFE INSTRUMENT FOR BREAST CANCER PATIENTS IN JAPAN Shozo Ohsumi, Surgery, National Shikoku Cancer Center, Matsuyama, Ehime; Kojiro Shimozuma, Hospital and Welfare Services, University of Marketing and Distribution Sciences, Kobe, Hyogo; Mitsue Saito, Metabolic Care and Endocrinological Surgery, Graduate School of Medicine, University of Tokyo, Tokyo; Naoto Shikama, Radiology, Graduate School of Medicine, Shinshu University, Matsumoto, Nagano; Takanori Watanabe, Surgery, Sendai National Hospital, Sendai, Miyagi; Michihide Mitsumori, Chikako Yamauchi, Radiology, Graduate School of Medicine, Kyoto University, Kyoto; Katsumasa Kuroi, Surgery, Toyosu Hospital, Showa University, Tokyo, Japan Development of a New Quality of Life Instrument for Breast Cancer Patients in Japan Shozo Ohsumi (1), Kojiro Shimozuma (2), Mitsue Saito (3), Naoto Shikama (4), Takanori Watanabe (5), Michihide Mitsumori (6), Chikako Yamauchi (6), Katsumasa Kuroi (7), for the Task-force Committee of the Japanese Breast Cancer Society (1) National Shikoku Cancer Center, Matsuyama, (2) University of Marketing and Distribution Sciences, Kobe, (3) University of Tokyo, Tokyo, (4) Shinshu University, Matsumoto, (5) Sendai National Hospital, Sendai, (6) Kyoto University, Kyoto, (7) Showa University, Tokyo, Japan, The Task-force Committee of the Japanese Breast Cancer Society has developed a new quality of life (QOL) instrument for breast

cancer (BC) patients, or QOL-ACD-B. This instrument has been designed to complement the QOL-ACD (general QOL instruments developed in Japan) with items specific to QOL of BC patients in Japan. In this paper, we report on the development and validation of this new QOL instrument. First, more than 50 different concerns were collected through a semi-structured interview from 67 BC patients and 22 medical professionals. Provisional version 1 of the QOL-ACD-B consisting of 26 items was developed from the interview. Second, a preliminary pilot testing was done for 78 patients to verify feasibility, reliability and validity. Provisional version 2 was developed after elimination of one item (pain from recurrent site). Third, another pilot testing for more detailed validation was performed for 274 patients using the version 2 in conjunction with the QOL-ACD and the FACT-B. One item (fear of recurrence) was eliminated because a high correlation with an item in the QOL-ACD. Several other items with a high correlation among them were abbreviated. Factor analyses revealed that the final version of the QOL-ACD-B with 21 items includes commonly used domains such as BC- or treatment-specific symptoms, body image and sexual function as well as other unique domains, such as coping with problems related to the disease/medical staff and nursing/pregnancy issues. In conclusion, the QOL-ACD-B will be helpful to understand the QOL of BC. # 1431/DEVELOPMENT OF A HEALTH-RELATED QUALITY OF LIFE QUESTIONNAIRE FOR THAI PATIENTS WITH RHINOCONJUNCTIVITIS Watcharee Lermankul, Clinical Pharmacy, Faculty of Pharmacy, Silpakorn University, Nakhon Pathom; Chaweewan Bunnag, Otolaryngology, Panida Kosrirukvongs, Ophthalmology, Faculty of Medicine & Siriraj Hospital, Bangkok, Thailand The objective of this study was to develop a disease-specific questionnaire for patients with rhinoconjunctivitis. All patients were recruited at the out-patients clinic at Siriraj Hospital. Item generation was gathered from several sources, obtaining a list of 63 items. In Part 1, the draft questionnaire was completed by 363 patients. Forty-eight items were identified by clinical impact analysis in item removal process and two more questions were added, giving the total number of 50. Part 2 involved another 243 patients, filling a 50-item intermediate version. The average time taken to complete the questionnaire was 6.38 (±4.36) min. Item removal process in Part 2 encompassed a multi-step process i.e. clinical impact analysis, missing data and response distribution, Cronbach’s a if item deleted, correlation coefficient analysis and experts opinion. The total number of the final version questionnaire was 36. It consisted of six dimensions with two independent items, i.e. Symptoms (17 items), Physical functioning (3 items), Role limitations (3 items), Sleep (3 items), Social functioning (3 items), Emotions (5 items), General Health (1 item), with a five-point scale, and Absenteeism (1 item). The scores ranged from 1 to 5. A lower score indicates a better quality of life. Data from the selected 36 items was extracted to test the validity and reliability of the final version. The floor and ceiling effects of the scores on each dimension were low. Multitrait multi-item analysis was conducted to examine construct validity (convergent and divergent). The scaling success of convergent and divergent validity were 100 and 94%, respectively. Internal consistency determined by Cronbach’s a coefficient was satisfactory (0.79–0.87). In conclusion, the questionnaire is applicable and practical to use in settings. Results of the testing are encouraging, however further work is to be done on the test–retest reliability and responsiveness.

830 # 1416/VALIDATION OF A NEW PATIENT REPORTED OUTCOMES INSTRUMENT FOR MILD COGNITIVE IMPAIRMENT Lori Frank, MEDTAP International, Inc., Bethesda, MD; Jennifer A. Flynn, Eli Lilly and Co, Indianapolis, IN; Mary Kay Margolis, Leah Kleinman, Louis Matza, MEDTAP International, Inc., Bethesda, MD; Lee Bowman, Eli Lilly and Co., Indianapolis, IN, USA Existing neuropsychological measures have limited sensitivity for mild cognitive impairment (MCI) as well as mild forms of dementia of the Alzheimer’s type (DAT). Given evidence that MCI frequently progresses to DAT over time, measurement of MCI is important to the identification of patients who may benefit from early intervention. The patient perspective of the condition, its symptoms and their impact is particularly relevant for a disorder like MCI, where standard measures may not detect deficits. A new instrument, the 55-item Patient Reported Outcomes in Cognition instrument (PROCOG), was developed and evaluated to obtain the patient perspective on MCI and mild DAT. The 5-point Likert scaled items derive from clinical input, patient and informant focus groups in the US and Europe, and the literature. Total score and scores for seven subscales can be calculated: memory for recent events, semantic memory, long term memory, cognitive functioning, emotional impact, and social impact of cognitive impairment. Psychometric properties were tested in a sample of 60 MCI subjects, 72 mild DAT subjects, and 34 controls, age 64+, recruited through outpatient neurology and memory clinics in the US. Item distributions indicated acceptable response spread for most items; most control subjects endorsed the lowest severity options and most DAT subjects endorsed the highest severity options. Scores for MCI subjects were intermediate between DAT and control subjects. Internal consistency reliability was excellent for total and subscale scores, ranging from 0.83 to 0.93. Test–retest reliability across 2 weeks was extremely high; intraclass correlations ranged from 0.60 to 0.91. Concurrent validity was supported by moderately high correlation with the QOLAD (0.56 for all groups combined, p < 0.05), a quality of life instrument developed for dementia of the Alzheimer’s type (Logsdon et al., 2002). Correlation to the MMSE was )0.39 (p < 0.05) for skill loss but lower for the remaining subscales. Results indicate that the PROCOG is a reliable and valid measure for collection of the patient perspective on mild levels of cognitive impairment. # 1106/DEVELOPMENT OF AN EATING DISORDER RELATED QUALITY OF LIFE QUESTIONNAIRE Carlota Las Hayas, Jose M. Quintana, Angel Padierna & Pedro Mun˜oz, Research Unit, Galdakao Hospital, Galdakao, Bizkaia, Spain Even though there is a wide variety of instruments developed to measure different aspects related to eating disorders, it was identified an absence of a quality of life instrument for these type of disorders. Our group developed and validated a self-reported measure of eating disorders related quality of life, the Eating Disorders Quality of Life Questionnaire (ED-QLQ), to examine the impact of the eating disorder on the physical, social and psychological domains of a person’s life. The items of the questionnaire were developed taking as reference the issues considered relevant by eating disorders patients, relatives and clinical experts during personal interviews. The questionnaire was validated in a sample of 358 eating disorders patients. The 12-item Short Form Health Survey (SF-12, 1995) and the Eating Attitudes Test -26 (EAT-26,1982) were also applied to examine the convergent validity. The final questionnaire consists of 65 items from which five collect demographic information. Exploratory Factor Analysis resulted in the identification of eight subscales: Physical Symptoms, Restrictive

Behaviours, Bulimic Behaviours, Body Image, Mental Health, Role Emotional, Role Physical, Personality Traits and Social Relations. Cronbach a was greater than 0.7 in all subscales. Item to scale correlations, correcting for overlap, were also performed to examine construct validity. Pearson correlation coefficients were calculated to assess the convergent validity of the questionnaire with the SF-12 components and the EAT-26 factors. The Physical Component from the SF-12 had a correlation of 0.54 and 0.63 with the Physical Symptoms and Role Physical domains respectively. The Mental Component of the SF-12 showed correlation coefficients of 0.75, 0.72 and 0.55 with the Mental Health, Role Emotional and Social Relations domains of the ED-QLQ. Finally, the EAT-26 Dieting Concern subscale correlated with a coefficient of 0.73 with Restrictive Behaviours domain and 0.76 with the Body Image Domain. The preliminary results support the content, convergent and construct validity of the measure to evaluate quality of life in eating disorders patients. # 1493/THE QUALITY OF LIFE ASSESSMENT IN PATIENTS WITH TYPE-1 DIABETES (DM1) AFTER SIMULTANEOUS PANCREASKIDNEY TRANSPLANTATION (SPK) Radomira Koznarova, Diabetes Center, Instit for Clin and Exp Med, Prague 4; Eva Dragomirecka, Psychiatric demography, Psychiatric Center, Prague 8; Tamara Hrachovinova, Frantisek Saudek, Petr Boucek, Diabetes Center, Instit for Clin and Exp Med, Prague 4; Pavel Trunecka, Milos Adamec, Libor Janousek, Transplant Center, Inst for Clin and Exp Med, Prague 4, Czech Republic Background: Successful SPK provides the prerequisites for improving the QoL. Evidence is often difficult because of the effect of simultaneous kidney Tx. Aim: To assess the possibility of using the SQUALA and SF36 questionnaires in SPK recipients. Methods: We used two questionnaires: (1) The SQUALA Questionnaire determines the difference between expectations and reality (results: partial score of life domain (PS), total score ¼ sum of all PS). (2) The SF36 Questionnaire (results: Transformed Scales(TS) Score; range 0–100) assesses eight health concepts. We evaluated group of 172 pts with DM1 divided into two subgroups: (1) Pts after successful SPK (group SPK, n ¼ 155), (2) Pts after kidney Tx alone (group KTA, n ¼ 17). Study I. We completed the questionnaires in SPK group before Tx, 6 months post-Tx. Study II. We compared the QoL in 17 recipients from the SPK group vs. a matched KTA group recipients 5–10 years post-Tx. Results: Study I: In SPK recipients, SQUALA evaluation pre- and 6 months post-Tx showed significant improvement in most parameters [general health p < 0.001; physical autonomy p < 0.001; psychic well-being p < 0.01; food p < 0.05]. Total score was also higher post-Tx (p < 0.05). When using the SF36 questionnaire, the SPK group showed a higher TS score at 6 months post-Tx in domains: physical functioning, general health, limitation due to physical and emotional (RE) problems (p < 0.05). Study II: When compared the QoL in SPK recipients vs. a matched group KTA recipients, the SQUALA questionnaire demonstrated a higher PS as well as the higher total score in the SPK group in all parameters (general health p < 0.001, sexuality p < 0.05, truth p < 0.05). The SF36 questionnaire revealed a difference in RE only (p < 0.05). Conclusion: Using both questionnaires, the post-Tx assessment of QoL in SPK recipients showed a better score in most parameters when comparing status pre-Tx and post-Tx, as well as when compared with the KTA control group. This is the first report on the use of the SQUALA questionnaire in pancreas graft recipients. Supported by grants VZ/CEZ:L17/98:00023001 and VZ/CEZ MZ 00000023752.

831 # 1426/THE EFFECT OF PERSONALITY ON HEALTH-RELATED QUALITY OF LIFE (HRQL) IN LUNG TRANSPLANT PATIENTS Xin Ye, Charoen Treesak, Barbara J. Matthees, Val R. Russas & Cynthia R. Gross, Social & Administrative Pharmacy, University of Minnesota, Minneapolis, MN, USA Increasingly, self-reported HRQL are being used to provide assessments of health, health care needs, and outcomes of care. However, the accurate interpretation of HRQL measures requires an understanding of the contextual factors that might influence patients’ assessments. It has been found that HRQL evaluations may be influenced by personality factors, social relations, and life engagement. The purpose of this study was to investigate the effect of personality facets (both positive and negative affectivity) on HRQL in lung transplant patients. Data were collected from 166 lung transplant candidates and recipients using a mail survey, which included the SF36, Positive and Negative Affect scales (PA and NA), CES-D and other HRQL measurements. Multivariate regression models were used to estimate the effect of personality on each of the eight SF-36 domains and the physical and mental component summary scores, adjusted for transplant status, age, gender, illness intrusion and depression. There were significant negative associations between NA and the SF-36 mental component summary score (b ¼ )0.640, p ¼ 0.007), and the mental health subscale (b ¼ )1.576, p < 0.0001) after adjustment. PA was positively associated with the SF-36 physical component summary score (b ¼ 0.401, p ¼ 0.045), the general health subscale (b ¼ 0.975, p ¼ 0.018), the role-physical subscale (b ¼ 2.171, p ¼ 0.006), the vitality subscale (b ¼ 1.487, p ¼ 0.001) and the mental health subscale (b ¼ 0.669, p ¼ 0.012) after adjustment. Personality (PA and NA) significantly explained an additional 3.1% of the variance in the SF-36 mental component summary score (p ¼ 0.0097), 3.0% in the role-physical subscale (p ¼ 0.024), 4.0% in the vitality subscale (p ¼ 0.002) and 6.8% in the mental health subscale (p < 0.0001), after adjustment for other covariates. These results suggest that personality plays an important role in self-reported HRQL, especially in mental health reports, and should be taken into account when evaluating and interpreting self-reported HRQL. # 1738/COMPARISON OF MODALITY-SPECIFIC QUESTIONNAIRES FOR MEASURING QUALITY OF LIFE (QOL) IN AUTOLOGOUS PERIPHERAL BLOOD STEM CELL TRANSPLANT (ASCT) PATIENTS Mary E. Morris & James C. Lynch, Preventive and Societal Medicine; Gregory Bociek, Stefano Tarantolo, Philip J. Bierman, Julie M. Vose & James O. Armitage, Internal Medicine, University of Nebraska Medical Center, Omaha, NE, USA Two bone marrow transplant (BMT)-specific QOL questionnaires are being utilized in an ongoing, pilot longitudinal study of QOL of ASCT patients at the University of Nebraska Medical Center. Currently, participants complete the Functional Assessment of Cancer Therapy (FACT)-BMT and City of Hope (COH) QOL in BMT Survivors questionnaires at pre-ASCT and at 100 days, 6 months, and yearly intervals post-ASCT. To assess the merit of continuing to utilize both questionnaires, preliminary FACT and COH results have been compared. Twenty patients, transplanted between September 2001 and March 2002, for lymphoma (n ¼ 13.65%) or multiple myeloma (n ¼ 7.35%) have completed their one year questionnaires. Twelve

(60%) are male, 19 (95%) are White, non-Hispanic, and average age at transplant is 53 years (range 27–68). All patients received carmustine, etoposide, cytarabine, and melphalan (BEAM). Three patients received a monoclonal antibody in addition to BEAM. Total QOL scores are shown below and generalized estimating equations was used for longitudinal modeling of the QOL scores. The COH scores at both 100 days and 6 months were significantly lower than pre-ASCT values (p < 0.01 and p < 0.05, respectively). QOL appears to improve after 6 months, but remains significantly below pre-ASCT levels at 1 year. Using the FACT, statistically significant declines in QOL were not observed at either 100 days or 6 months and QOL is significantly greater at 1 year than at pre-ASCT (p < 0.05). Different trends in post-ASCT QOL were identified by the FACT and COH, suggesting that either the FACT does not capture early post-ASCT declines in QOL or that the questionnaires are measuring different aspects of QOL in this patient population. # 1514/SLEEP DISORDERS AND QUALITY OF LIFE AFTER KIDNEY TRANSPLANTATION Barotfi Szabolcs, QUINTILES Hungary Ltd.; Marta Novak, Institute of Behavioral Sciences, Semmelweis University; Miklos Z. Molnar, 2nd Department of Internal Medicine, St. Margaret Hospital; Agnes Z. Kovacs, Eszter P. Vamos & Rezso Zoller, Institute of Behavioral Sciences; Adam Remport, Department of Transplantation and Surgery; Adrienn Marton & Istvan Mucsi, 1st Department of Internal Medicine, Semmelweis University Budapest, Budapest, Hungary A high prevalence of sleep disorders (Restless Legs Syndrome (RLS) and Sleep Apnea Syndrome (SAS)) have been reported in patients with chronic renal disease. The aim of our study was to analyze the effect of these sleep disorders and health-related quality of life (QoL) after renal transplantation. We report data obtained in 398 patients followed in a single transplant center in Budapest. Mean age of the patients was 47 years (18–78 years), 57% male, 16% diabetics. Mean serum creatinine level was 156 mol/l, mean hemoglobin level was 135 g/l. Mean time since transplantation was 54 months (3– 242 months). The patients were administered a battery of questionnaires including diagnostic questionnaires for RLS and SAS. Also included in the battery was the Kidney Disease Quality of Life Questionnaire (KDQoL). The prevalence of RLS was 4.8% that is much lower than it had been reported for dialysis patients. 22.6% of the patients scored high risk for SAS on the Berlin questionnaire. There was statistically and clinically significant difference in numerous QoL domains between patients with or without sleep disorders: for example the SF-12 Physical Composite score (RLS: 36.9 ± 9.2, SAS: 39.8 ± 9.8, no sleep problem: 44.8 ± 10.3, p < 0.001 ANOVA) and the SF-12 Mental Composite score (RLS: 43.6 ± 14.4, SAS: 46.5 ± 11.6 no sleep problem: 50.7 ± 11.2, p < 0.05 ANOVA) was significantly lower in patients with sleep problems. The prevalence of sleep disorders was lower in our kidney transplant patients than published previously for patients on maintenance dialysis. The presence of sleep disorders was associated with poorer QoL. Supported by grants from ETT 240/2000, NKFP 1/002/2001, OTKA TS040889 and T038409 (M.N.). I.M. is a Bekesy postdoctoral fellow. M.Zs.M. is a scholar of the Hungarian Kidney Foundation. The study was also supported by unrestricted grants from Sanofi-Synthelabo Ltd. and Janssen-Cilag/Division of Johnson-Johnson Ltd.

832 # 1511/HEALTH RELATED QUALITY OF LIFE OF PATIENTS ON MAINTENANCE DIALYSIS vs. KIDNEY TRANSPLANTED PATIENTS IN HUNGARY Agnes Z. Kovacs, Institute of Behavioral Sciences, Semmelweis University; Miklos Z. Molnar, 2nd Department of Internal Medicine, St. Margaret Hospital; Judit Pap, Institute of Behavioral Sciences, Csaba Ambrus, FMC-SE Dialysis Center; Adam Remport, Department of Transplantation and Surgery; Lilla Szeifert, Istvan Mucsi, Eszter Vamos & Marta Novak, Institute of Behavioral Sciences, Semmelweis University, Budapest, Hungary Quality of Life (QoL) as perceived by patients with end stage renal disease (ESRD) is recognized as an important outcome measure for renal replacement therapy. In this study, using the Kidney Disease Quality of Life questionnaire, we compared QoL of hemodialysis patients and patients after successful kidney transplantation. Data for 398 patients after kidney transplantation (Tx) and 283 patients on maintenance hemodialysis (HD) treated in five dialysis units in Budapest were entered into an electronic database. Mean age of the Tx patients was 47 years (57% male, 16% diabetics, the mean hemoglobin level was 135 g/l). Mean time since transplantation was 54 months (3–242 months). In the dialysis group the mean age was 57 years, the mean time on HD was 28 months (50% male, 23% diabetics, mean hemoglobin level 112 g/l). The patients were administered the Kidney Disease Quality of Life Questionnaire (KDQoL); laboratory data, disease history and medications were also recorded. In the Tx group there was a weak but statistically significant correlation between QoL and renal function (GFR calculated with the Cockroft– Gault formula) (r ¼ 0.254 and r ¼ 0.207, for SF12 physical and SF12 mental composite scores, respectively; p < 0.01). Women and diabetic patients tended to report poorer quality of life along most of the domains. Tx patients reported better QoL in most of the domains and this difference was statistically and clinically significant (for example: burden of kidney disease: 44 ± 26 vs. 69 ± 26 ; emotional well being: 60 ± 25 vs. 72 ± 23 ; SF12 physical composite: 37 ± 10 vs. 43 ± 10 for HD vs. Tx patients, respectively; p < 0.01 for all cases). We conclude that successful kidney transplantation brings significant improvement of all aspects of QoL for patients with ESRD. Supported by grants from ETT 240/2000, NKFP 1/002/2001, OTKA TS040889 and T038409 (M.N.). I.M. is a Bekesy postdoctoral fellow. M.Zs.M. is a scholar of the Hungarian Kidney Foundation. The study was also supported by unrestricted grants from Sanofi-Synthelabo Ltd. and Janssen-Cilag/Division of Johnson-Johnson Ltd. # 1473/THE HEALTH RELATED QUALITY OF LIFE (HRQOL) IMPACT CAUSED BY STARTING HEMODIALYSIS IS LOWER IN ELDERLY THAN IN YOUNGER PATIENTS Monica Garcia-Mendoza, Covadonga Valdes, Pablo Rebollo, Teresa Ortega, Dario Moreno & Francisco Ortega, Nephrology I, Hospital Central Universitario Asturias, Oviedo, Spain This is a prospective study in which HRQoL of elderly (P65 years) and younger patients on renal replacement therapy (RRT) are compared and several variables are studied. All patients starting RRT in our region during last two years (n ¼ 284) were included. Interview was carried out with each patient at 3 months from start hemodialysis using the SF-36 Health Survey, with scores standardized using the Spanish

general population norms, the Physical Symptom scale (PSS) of the Kidney Disease Questionnaire, sociodemographic and clinical data, the Karnofsky Scale and a comorbidity index. Elderly patients showed higher standarized scores than younger patients in all SF-36 dimensions, being statistically significant (p < 0.05) except in Bodily Pain and Mental Health, so lower loss of HRQoL was found.The scores in the Physical and Mental Component Summary (PCS), (MCS) of SF-36 were also higher in elderly people: PCS in elderly 46.61 ± 7.46 vs. 39.90 ± 12.60 in younger (p ¼ 0.001); MCS in elderly 48.69 ± 10.50 vs. 44.12 ± 13.08 in younger (p ¼ 0.028). No difference was found for the PSS. Variables associated to HRQoL in Elderly patients were (p < 0.05): female gender, with lower scores on Physical Functioning (PF), General Health (GH), Vitality (VT), Role Emotional (RE) and Mental Component Summary (MCS); Monthly income above 900 euros, with higher scores in all dimensions except Role Physical (RP) and Bodily Pain (BP); and the higher Comorbidity Index, with lower scores on PF, VT and GH. No statistically significant association was found in younger patients. In conclusion, elderly patients showed lower loss of HRQoL than younger patients at 3 months from start hemodialysis. Variables associated to worse HRQoL in the aged group were female gender, lower monthly income and higher comorbidity index. # 1723/QUALITY OF LIFE IN SELF-CARE DIALYSIS PATIENTS Carole Loos, Epide´miologie et e´valuation cliniques (EA 3444), CHU, Nancy; Luc Frimat & Miche`le Kessler, Service de Ne´phrologie, CHUBrabois, Vandoeuvre-les-Nancy; Serge Brianc¸on, Epide´miologie et e´valuation cliniques (EA 3444), CHU, Nancy, France In the Lorraine area (France), dialysis centers propose an educational program to improve patients ability to perform dialysis by themselves. Thus, patients are treated with home dialysis (haemodialysis or peritoneal dialysis) or with self-care dialysis (in center but unsupervised). Many studies have documented the lowest Quality of Life (QoL) level in dialysis patients compared to renal transplant patients. But none of them have compared QoL between in-center dialysis (ICD) and home/ self-care dialysis (HSCD) patients. The aim of the present prospective study was to assess QoL in HSCD patients and to compare it with QoL in ICD patients. All end stage renal disease patients aged between 18 and 70 and having started their first dialysis between June 1997 and June 1999 in the Lorraine area were included. Patients with acute renal failure or returning to dialysis following kidney graft failure were not included. Socio-demographic, medical data and QoL (KDQoL) were assessed for each patients at the enrolment and at 12 months. The educational program occurred during the first 12 months. So the transition between ICD and HSCD occurred during the same period. QoL was compared at 12 months. Multivariate analysis was adjusted on baseline QoL mean scores and on confounding factors. At 12 months, 194 patients were in dialysis, 132 in ICD and 62 were in HSCD. These two groups had significant differences on mean age (respectively 56 ± 12 vs. 52 ± 14, p ¼ 0.05), on body mass index (27 ± 6 vs. 25 ± 4, p ¼ 0.03) and on cardiovascular diseases (65% vs. 47%, p ¼ 0.02). At enrolment, there was no significant difference on mean QoL scores. At 12 months, only the KDQoL burden of disease dimension was significantly and independently higher in the HSCD group (44.3 vs. 36.0, p ¼ 0.04). Patients entering in educational program are better able to cop the strength of their disease and treatment.

833 # 1224/QUALITY OF LIFE AND HEALTH LOCUS OF CONTROL IN PATIENTS RECEIVING RENAL DIALYSIS Maria Coccossis, Psychiatry, Medical School, University of Athens; Mary Lela, Psychology, Panteion University; Dimitra Alexandra Lorentzatou & George N. Christodoulou, Psychiatry, Medical School, University of Athens, Athens, Greece The purpose of the study is to investigate in end-stage renal disease (ESRD) patients, the interaction of health locus of control beliefs with quality of life (QOL) as a treatment outcome indicator. The HLC beliefs include ‘internal–external control’, ’powerful others’ and ‘chance’ .The present sample consists of 52 ESRD patients with a mean age of 54.23 years (SD ¼ ±16.35) and a mean age of starting dialysis 46.94 years (SD ¼ ±16.25). In terms of gender, 36.53% were women,while in terms of family status, 59.61% were married. Further, 55.76% did not hold any work appointment. The patients, in the first 30 min of dialysis, were administered the following inventories (in semistructured interviews): (1) the WHO Quality of Life Inventory, WHOQOL-BREF (validated in Greece and presenting satisfactory psychometric properties), (2) the Multidimensional Health Locus of Control Inventory of Wallston and Wallston (under validation in the Greek population) and (3) the Goldberg’s General Health Questionnaire, GHQ-28 (validated in Greece). The data were analysed by the SPSS program. The results of analysis indicated that male patients assessed their psychological well being as being higher than that of female patients. Also, patients with higher education tended to consider their health condition as being less influenced by external factors, such as chance. Working patients indicated better perceived somatic health and a more satisfactory level of independence. Further, male patients who were widowers, assessed their level of somatic and psychological well being as less favorable. For the total sample, it was found that patients with internal HLC beliefs presented a higher level of perceived QOL, particularly concerning somatic health, level of independence, psychological well being, as well as satisfaction with social relations. On the contrary, patients holding external HLC beliefs, such as chance, presented a lower level of psychological well being. The implications of HLC beliefs on treatment outcome and QOL are discussed. # 1772/ASSOCIATION OF SIX MEASURES OF UTILITY AND DEPRESSION Denise R. Globe, Joanne Wu & Michael B. Nichol, Pharmaceutical Economics and Policy, University of Southern California, Los Angeles, CA, USA The purpose of this study was to estimate the association of estimated utility and depression. The data set consisted of survey and pharmacy data collected from 6923 members of a large managed care organization. Subjects who completed a global visual-analogue scale (VAS), the SF-36, and the HUI were included. Depression was defined by use of anti-anxiety or anti-depressant medications, or self-reported (SF-36 downhearted and blue item). Two utility score estimates were calculated from the SF-36; (1) a previously published Ordinary Least Squares algorithm (NSG-OLS) and (2) the SF-6D, which derived one utility score (SG). The other two estimates (SF-12 HUI and SF-12 VAS) were calculated from the SF-12. Utility scores were compared using T-test or one-way ANOVA to assess the association of utility and depression in the cohort. Mean estimated utility scores ranged from 0.57 (downhearted and blue all of the time) to a high of 0.81 (no anxiety and no depression). For patients with depression, the estimated utility scores were moderately correlated with each other (r: 0.40–0.97, p < 0.0001); in addition, utility scores were negatively associated with the Chronic Disease Score. Lower utility scores were associated with (1) taking medications for anxiety or depression; (2) self-reported depression. Similar to findings reported for other diseases (e.g. cardiovascular) each method orders the impact of depression on the utility score similarly, although the range and distribution differs significantly between estimation methods.

# 1453/THE VALUATION OF ATOPIC DERMATITIS HEALTH STATES IN CHILDREN John E. Brazier & Katherine J. Stevens, School of Health and Related Research, The University of Sheffield, Sheffield, South Yorkshire; Stephen McKenna, Lynda Doward, Galen Research, Manchester, UK Torrance and colleagues (1996) asked parents of school age children to value HUI2 states using the standard gamble (SG) technique by imagining they are 10 years old and live in that health state for the rest of their life. This methodology was applied to value 16 childhood health states derived from a condition-specific measure for atopic dermatitis, but using a general population sample. In addition, three variants of the methodology were explored: (1) Using a sample of parents (2) Reducing the starting age to 5 years 3) Changing the duration to 10– 16 years in the health state, followed by full health in adulthood. The main survey interviewed 150 people who were divided into two groups, each valuing 10 health states. Each group valued the best, worst, two common and six other states. For the three variants surveys the samples achieved were 36, 68 and 51 and each valued eight states. The primary analysis was the estimation of mean SG health state values and a comparison across variants. For variant 3, the primary analysis estimates a health state value by assuming the QALY assumptions of zero time preference, constant proportional trade-off and linearity over time. Regression analyses were also undertaken to estimate the impact of different aspects of quality of life on the preference score. Mean values across the 16 health states range from 0.36 (SD ¼ 0.36) for the worst state to 0.84 (0.19) for the best state and are logically consistent with the classification. Parents were found to give higher values across the states overall (at 5% level), though this was explained entirely by the age difference between the samples. The earlier starting age of the state was found to make no significant difference. The shorter duration produced a wider dispersion in mean health state values of 0.12–0.95. These results show the perceived importance of atopic dermatitis in quality of life and that the format of the question can have significant implications for the values obtained for childhood states. # 1458/ADAPTATION OF THE HUI-MARK3 TO ELDER POPULATION Miguel A. Ruiz, Irene Rebollo, Methodology, Universidad Auto´noma; Fe´lix Bermejo, Neurology, Hospital 12 de Octubre, Madrid; Javier Rejas, Health Outcomes Research, Pfizer Spain, Alcobendas, Madrid, Spain The HUI-MARK3 is a generic instrument specifically developed to estimate population utilities towards health estates. The instrument is applied to patients in a very similar way as other QoL questionnaires, but it allows to estimate a utility score (ranged 0–1) by means of a Multi-Attribute Utility Function (MAUF). The original instrument was adapted to Spanish Iberian language (Ruiz M, Rejas J, et al. 2001), and small but significant differences were found between the original Canadian utilities and the Spanish utilities. Moreover, it is suspected that perceived social utilities might be influenced by socio-demographic variables (such as age), or other personal variables like previous experience with disease. In this work, a new estimation of the Spanish MAUF is developed with a representative sample of elder people. A sample of 165 elder people with mean age 79 years old (SD ¼ 4.92) to estimate a new MAUF for this particular population. Significant differences (p < 0.001) between both MAUFs are found. Elder people assign lower utilities to the same health states. Interaction is also present (p < 0.001), so that differences in mean utility values are bigger for worse health states, between elder and normative valuations. It should be noted that direct statistical comparison of the MAUF weights is not possible, since MAUFs are multiplicative (non linear) in nature, in order to allow interaction between simple attributes. Indirect procedures are used. When simple attribute functions are compared with the normative sample, additional differences are found. Deterioration in Dexterity is perceived as the biggest loss in utility for health states. As it happened with the normative validation sample, a nonmonotonic simple attribute function for the Cognition attribute is obtained. Our results suggest that elder people can be considered a particular subpopulation, and traditional instruments should be adapted to their situation.

834 # 1371/INCORPORATING SOCIAL PREFERENCES IN THE MEASUREMENT OF OUTCOMES IN ADULT GROWTH HORMONE DEFICIENCY Paul Kind, Outcomes Research Group, Centre for Health Economics, University of York, York, England; Maria Koltowska-Haggstrom, KIGS/ KIMS Outcomes Research, Pharmacia AB, Stockholm, Sweden Growth hormone deficiency in adults usually arises as a result of pituitary disease, radiation therapy or trauma and may occur at any time of life. The impact of adult-onset growth hormone deficiency is both physical (lipid disorders, changes in body composition and bone mineralisation) and psychological (fatigue and loss of energy, inability to concentrate, poor memory, depression, sleep disturbance, general loss of well-being and decreased social functioning). Condition-specific measures such as Quality of Life Assessment of Growth Hormone Deficiency in Adults (QoL-AGHDA) have been developed to record change in patients with growth hormone deficiency. However, for the purposes of economic evaluation and specifically for use in cost-utility analysis, outcomes must be measured on a scale of social preferences that assigns values of 1 and 0 to full health and dead. Conditionspecific instruments sometimes generate summary index scores but typically these lack legitimacy for use as measures of quality of life outcomes in economic evaluation. Questionnaires containing an extended version of the 25 items of QoL-AGHDA together with EQ-5D, a generic measure of health-related quality of life (HrQoL) were distributed to 1000 individuals registered with the National Population Preference Panel, yielding 729 questionnaires for analysis. QoL-AGHDA item ‘yes/no’ responses were entered as dummy variables in regression models with EQ-5D as the dependent variable. The resultant regression weights indicate the dominant role of four QoL-AGHDA items (struggling to finish jobs, feeling lonely even with others, having to read things several times and feeling a burden) as contributory elements to HrQoL. An adjusted r2 of 0.533 suggests an adequate fit for the principal model. Mapping a condition-specific measure onto a generic single index measure calibrated with pre-existing social preference weights is a viable method for estimating the social value of items and can be used to present data from clinical studies in a form suitable for economic evaluation. # 1292/A COMPARISON OF ADDITIVE AND MULTIPLICATIVE MODELS FOR A NEW PREFERENCE-BASED QUESTIONNAIRE IN PATIENTS WITH MUSCULOSKELETAL CONDITIONS Eric C. Sayre, ARC, Arthritis Research Centre of Canada; Jacek A. Kopec, Health Care and Epidemiology, University of British Columbia, Vancouver, BC; Michal Abrahamowicz, Epidemiology and Biostatistics, McGill University, Montreal, Quebec; Aslam H. Anis, Health Care and Epidemiology, University of British Columbia, Vancouver, BC; Elizabeth M. Badley, Public Health Sciences, University of Toronto, Toronto, Ontario; Aileen M. Davis & Jack I. Williams, Research, Toronto Rehabilitation Institute, Toronto, Ontario; John M. Esdaile, Rheumatology, University of British Columbia, Vancouver, BC, Canada We previously reported (ISOQOL 2002) on a new adaptive questionnaire to measure five domains of health in arthritis (ambulation, handling objects, daily activities, pain, and emotional function). Here we present a model to combine domain-specific scores into overall utilities. Data were obtained from a random sample of 128 subjects in Vancouver, Canada. There were four levels of health for each domain. Each subject provided rating scale (RS) values for all health states in which only one domain was affected. We applied a fractional factorial design to obtain standard gamble (SG) utilities for 256 complex states with multiple deficits. The data were used to transform RS values to SG utilities and to fit the overall utility models. We employed least squares to fit multiplicative, simple additive, and multi-linear models. The simple additive model and multiplicative model produced very similar R2: 0.529 and 0.527. Inclusion of all two-way interactions in the additive model increased the R2 to 0.607. The models were also fit using maximum likelihood. The ratios of scaled deviance/d.f. were 1.78 for the multiplicative model, 1.77 for the simple additive model, and 1.50 for the multi-linear model. Residual plots showed no dierences amongst models in our data. However, the additive models predicted counterintuitive overall utilities for very low levels of health. We conclude that the multiplicative model is a preferred model for our questionnaire.

# 1245/A TEST OF PROSPECT THEORY David H. Feeny & Ken Eng, Institute of Health Economics, Edmonton, Alberta, Canada Background: In prospect theory people judge the quality of health states relative to a reference point, often their current health. States better than the reference point are valued on a concave portion of the utility function while states worse than the reference point are valued on a convex portion. Prospect theory is one prominent explanation for response shift. Methods. An observational prospective cohort study in which osteoarthritis (OA) patients undergoing total hip arthroplasty (THA) were followed from the time the primary-care physician referred them to the surgeon until 3 months after surgery. Patients periodically provided standard gamble (SG) utility scores for three OA marker states (mild, moderate, severe) as well as their subjectively defined current state (SDCS). Hypothesis #1: Patients deteriorated while waiting for surgery. If their reference point – score for SDCS – fell below their previous score for a marker state, prospect theory hypothesizes that scores for marker states previously >SDCS but now SDCS should be lower. Results: Eighteen patients declined surgery but only three patients met the conditions outlined for hypothesis #1. Observed changes in marker state scores conformed to prediction in 1 case (total change in utility score ¼ 0.10); changes were in the direction opposite that predicted in two cases (0.40). Forty six patients improved after surgery; 42 met the conditions for hypothesis 2. Predictions were confirmed in four cases (1.5); there was no change in marker state scores in seven cases; observed scores moved in the direction opposite to that predicted in 31 cases ()7.0). For hypothesis #2, the net change for all 42 patients was )0.13 per person. Discussion: For the most part the direction of the changes in marker state scores among THA patients do not conform to the predictions of prospect theory. Further, the magnitude of change observed is substantial; the lack of support for prospect theory is unlikely to be due to measurement noise. # 1568/THE NEGATIVE IMPACT OF LOW FAMILY INCOME ON HEALTH-RELATED QUALITY OF LIFE AND PHYSICAL FUNCTION IN PATIENT WITH RHEUMATOID ARTHRITIS IN A PUBLICLY FUNDED HEALTH CARE ENVIRONMENT Carlo A. Marra, Aslam H. Anis & Larry D. Lynd, Health Care and Epidemiology; John M. Esdaile, Arthritis Research Centre, University of British Columbia, Vancouver, British Columbia, Canada Objective: To assess the relationship between measures of socioeconomic status (SES) and quality of life (QoL)/functional status in rheumatoid arthritis (RA). Methods: Three hundred and thirteen RA patients between 19 and 90 years (61.5 ± 13.5) were evaluated. QoL was quantified using the Short Form 6D (SF-6D) and the Health Utilities Index 3 (HUI-3). Functional status was determined using the Health Assessment Questionnaire Disability Index (HAQ). SES was measured using both self-reported annual household income and education and census data (median neighbourhood income and education level). Statistical analyses involved multiple linear regression with the SF-6D, HUI-3 or the HAQ as the dependent variable, adjusted for potential confounders. For the SF-6D and the HUI-3, lower scores reflect poorer QoL whereas for the HAQ, higher scores represent worse functional status. Results: We identified a consistent association between low household income and poorer QoL and functional status, adjusting for gender, duration of RA, self-reported pain, selfreported disease severity and number of tender/swollen joints. Relative to those with incomes >$50,000, those with incomes between $20,000 and $50,000 (b ¼ 0.35 ± 0.10) and <$20,000 (b ¼ 0.26 ± 0.09) had significantly worse functional status (p ¼ 0.001). Similarly, for the SF-6D, relative to those with incomes >$50,000, those with incomes between $20,000 and $50,000 (b ¼ )0.03 ± 0.02) and <$20,000 (b ¼ )0.04 ± 0.02) had significantly worse QoL (p ¼ 0.04). Similar results were found for the HUI-3. However, the same associations were not present for self-reported education or for the census tract SES variables. Conclusions: There is a strong and significant relationship between self-reported household income and QoL and functional status despite adjustment for clinical variables. Thus, even in a society with universal access to health care, the impact of a welldefined chronic disease seems to be closely linked SES.

835 # 1312/THE EVALUATION OF HRQOL (HEALTH RELATED QUALITY OF LIFE) OF DIABETIC OUTPATIENTS BY USING TIME TRADE-OFF METHOD Hisako Adachi, Nursing Course, School of Medicine, Gifu University, Gifu; Oyamada Takaaki, Gifu City Educational and Cultural Foundation, Gifu City, Gifu, Japan Purpose: This study aimed to evaluate the preference (utility) value of the present health state of diabetic outpatients by using Time TradeOff (TTO) method. Method: As an index of the health-preference value, TTO score was measured by TTO method that Torrance et al. developed in 1972. TTO score was a ratio of more brief life (years) without disease desired by the patient to remaining life(years) of the patient with disease. Eighty-five diabetics (4 of type1 and 81 of type2) were recruited from diabetic outpatients in University Hospital. Mean age of subjects was 59.2 ± 14.9 years old and average HbA1c% was 6.92 ± 1.15. This research was approved by the ethics committee of School of Medicine in Gifu University (Japan). Results: (1) Mean TTO score was 0.79 ± 0.31 for all patients, 0.64 ± 0.41 for 21–40 age group and 0.86 ± 0.27 for 41–60 age group. A significant difference was found between those two groups on mean TTO score(p < 0.05). (2) TTO score increased significantly with age of patients (r ¼ 0.23, p < 0.05). (3) Mean TTO score was 0.69 ± 0.35 for the patient group with physical disabilities, but was 0.90 ± 0.21 for the group without physical disabilities. There was a significant difference between those two groups on mean TTO score (p < 0.01). TTO score became lower when diabetic patients had some complications of diabetes (r ¼ )0.35, p < 0.01). (4) TTO score was 0.67 ± 0.35 for the patient group with negative feelings, but was 0.88 ± 0.24 for the group without negative feelings. There was a significant difference between two groups on mean TTO score (p < 0.01). (5) There was no significant difference on mean TTO score between two groups of insulin-dependent and noinsulin patients. (6) The correlation between TTO score and HbA1c% was 0.013. Conclusion: The HRQOL of diabetic patients were evaluated with TTO score by TTO method. The remaining life (years) physical and psychological disabilities, and complications of diabetes were negative effect factors on HRQOL of diabetic outpatients. # 1875/ANAMNESTIC COMPARATIVE SELF-ASSESSMENT (ACSA) TO MEASURE QOL. II. ADVANTAGES OF AN INTERNAL STANDARD Jan L. Bernheim, Human Ecology; Peter Theuns & Mehrdad Mazaheri, Psychology, Vrije Universiteit Brussel, Brussels, Belgium; Matthias Rose, Psychosomatik, Charite´ Klinikum – Campus Mitte, Berlin, Germany Background: Conventionally, subjective QOL (SQOL) is assessed by one global question about happiness, wellbeing or life satisfaction. Scoring is on a scale between e.g. ‘extremely happy’ and ‘extremely unhappy’. However, some problems of these scales include trivialisation of responses, proximate, peer- and cultural relativity, and response shifts. Methodology: The originality of ACSA is using the respondent’s memories of his best and worst periods in life (given ratings +5 and )5 respectively) as the extremes of the scale. The elicited scores for SQOL are relative to these anchors. Respondents typically attribute +5 to periods of love experiences, births or professional achievements. Experiential nadirs are typically loss of a loved person, imprisonment, bankruptcy or a serious disease. Hypothesis: Compared to the conventional global question on QOL, ACSA offers: (1) solemnity, which discourages trivial or socially desirable responses, (2) concreteness of the (extremes of the) scale rather than the artificiality or abstraction of the ‘best’ or ‘worst’ possible states, (3) a more internal standard, with less exogenous bias, more sensitivity to internal changes, and less proximate, peer and cultural relativities and shifts in expectations, (4) lesser sensitivity to personality traits. Empirical study. ACSA and the conventional global life-satisfaction question were administered in parallel to 2235 patients suffering from a wide range of psychiatric and somatic diseases. Results: ACSA was less influenced by individually invariable socio-demographic variables such as sex, age, nationality and marital status. It discriminated better between illnesses, and, in a retest situation at a later time proved more sensitive to objective evolutions of health state. Conclusion: Depending on study objectives, ACSA should be considered as an alternative or addition to conventional SQOL instruments.

# 1546/SELF-ASSESSMENT OF HEALTH STATUS: ARE INDIVIDUALS USING THE SAME MEASUREMENT SCALES? Pedro L. Ferreira & Oscar D. Lourenc¸o, Faculty of Economics, University of Coimbra, Coimbra, Portugal It is not uncommon to use health status perceptions to compare countries or regions within the same country. Eurostat is one of these examples. However, apart from cultural aspects, these comparisons have to take into account the distributions of some relevant sociodemographic indicators. This is so because these indicators do influence not only the true health status but also the subjective view of health status. In fact, gender, age, education and place of residence have an impact on the location of the thresholds used to define the different scales – anchors used to assess health status. So, the purpose of this presentation is to evidence that people use different scales – thresholds when the perceptions of their own health status are asked. Using several thousands of perceptions of health status of a representative sample of the Portugal mainland we applied generalized ordered probit models with variable thresholds. In this study, thresholds are function of individuals characteristics. By using a simple ordered probit model, among the major findings we may state that older individuals show lower health status scores, as well as female, low educated and rural individuals. However, using a generalized ordered probit model, the found differences are not so evident. As a conclusion, scores provided by indicators of health status perceptions have to be calibrated tanking into account these differences in the thresholds locations. # 1135/CAUSAL MODELING OF RELATIONSHIPS BETWEEN EMOTIONAL WELL-BEING, COMMUNICATION IMPAIRMENT, AND RELATED PARTICIPATION RESTRICTIONS Patrick J. Doyle, Joseph M. Mikolic & William D. Hula, Geriatric Research Education & Clinical Center, VA Pittsburgh Healthcare System; Malcolm R. McNeil, Communication Science and Disorders, University of Pittsburgh, Pittsburgh, PA, USA Path analysis was used to examine two competing models of the relations among emotional well-being, communication impairment, and the activity limitations, participation restrictions, and psychological distress associated with communication impairment. Independent measures of these constructs were obtained from 132 communicatively disordered stroke-survivors who participated in the validation of the Burden of Stroke Scale. Both models were grounded in the ICF which proposes a dynamic interaction among impairments, activity limitations, participation restrictions, and personal and environmental factors. The models were constructed to examine alternative accounts of the causal relationships among these variables. One model posited that communication impairment caused patient-reported communication activity limitations which in turn causally impacted both communication associated psychological distress and general emotional wellbeing, with each of the latter two factors causally impacting the reported participation restrictions. This model can be summarized as one which gives patient-reported activity limitations temporal priority over communication associated psychological distress and emotional well-being, with the latter two constructs mediating the causal relationship between activity limitations and participation restrictions. The alternative model posited that communication impairment and emotional well-being interacted to cause communication-associated psychological distress, which in turn caused individuals to report communication activity limitations and consequent participation restrictions. An omnibus v2 test of the first model found it to be plausible, while the second model was found by the same test to be implausible.

836 # 1660/A COMPARISON OF SF-36 AND THE HOSPITAL ANXIETY AND DEPRESSION SCALE (HADS) IN PATIENTS WITH ANGINA PECTORIS Bjørg Ulvik, Department of Public Health and Primary Health Care, University of Bergen, Bergen; Tove A. Hanssen, Tore Wentzel-Larsen, Centre for Clinical Research; Ottar K. Nyga˚rd, Department of Heart Disease, Haukeland University Hospital; Berit Rokne Hanestad, Department of Public Health and Primary Health Care, University of Bergen, Hordaland, Norway Patients with angina pectoris often experience impaired health related quality of life, anxiety and emotional distress. There is no agreement on what is the best instrument to measure psychological wellbeing in patients with ishemic heart disease. The purpose of this study was to compare the scores the Hospital Anxiety and Depression Scale (HADS) with scores from The Short Form- 36 (SF-36), and between HADS-A (anxiety) and HADS-D (depression) and the Mental (MCS) and Physical Component Summary (PCS) of the SF-36. The crosssectional study is part of the large West Norwegian B-vitamin Intervention Trial (WENBIT). The patients were consecutively recruited from August 2000 until February 2002 from the cardiac outpatient clinic of the Haukeland University Hospital. The sample consisted of 800 men (74%) and women, mean age 62 (SD 10.3), with clinical evidence of angina pectoris, and admitted to have a cardiological investigation with coronary angiography. The questionnaires were filled in at home about two days before the investigation: (1) HADS consisting of HADS-D (7 items) and HADS-A (7 items) with higher score indicating more serious problems (2) SF-36 consisting of 36 items measuring the following domains: Physical Functioning (PF), Role Physical (RP), Bodily Pain (BP), General Health (GH), Vitality (VT), Social Functioning (SF), Role Emotional (RE) and Mental Health (MH), higher scores indicate better health status. Both HADS-A and HADS-D correlate significantly (p < 0.001) with all the SF-36 domains. The highest association was between HADS-A and the Mental Health (MH) r ¼ )0.73 and the HADS-D and MH r ¼ )0.67. The corresponding correlation coefficients between HADS-A and the MCS were r ¼ )0.68, and between HADS-D and MCS r ¼ )0.65. # 1502/PATIENTS IMPROVED THEIR HRQOL BY MAKING THEIR PHYSICAL FUNCTION AND GENERAL HEALTH BETTER THAN BEFORE HEART TRANSPLANTATION WITH DIFFERENT EFFECT SIZES Teresa Ortega, Nephrology I; Juan C. Alameda, Maria A. Montoliu & Blanca Rodriguez, Ergonomics Unit; Covadonga Valdes, Pablo Rebollo & Francisco Ortega, Nephrology I, Hospital Central de Asturias, Oviedo, Asturias, Spain Several reports in the medical literature document significant improvements after heart transplantation in physical, social and mental functions. While survival data have been published, sparce information is available regarding the effect size in the health related quality of life (HRQoL) of patients before and after heart transplant. The aim of this study was to evaluate the effect size in the HRQoL of patients on the cardiac transplant waiting list and after being transplanted. There were 103 patients in heart transplant waiting list, decreasing to 21 who received a transplant. Evaluation of HRQoL over time was obtained using The Sickness Impact Profile (Physical – PHYS, Psychosocial – PSY dimension and the total score – TSIP) and Quality Of Life Profile For The Chronically Ill (PLC) as HRQoL questionnaires. The change in scores of the SIP dimensions and PLC was studied before (when patients were included in heart transplant waiting list) and 1 year after transplantation, using the Student’s T test for paired samples. The Effect Size was also assessed for SIP and PLC scores, dividing the difference between the mean score in the first interview and that of the last one by the standard deviation of the mean score in the first interview. The Effect Size was considered as small if it was under 0.2; moderate near 0.5; and large over 0.8. The mean scores improved, throughout the first year of evolution, with statistical significance in the PHYS (p ¼ 0.000) and TSIP (p ¼ 0.000), however, not in the PSY and in the PLC. The Effect Size calculated for each dimension were large (0.81) for PHYS and for TSIP (0.73) and small for the PSY dimension

(0.35), which hardly varies during the follow up; and it was moderate for PLC (0.63). Conclusions: HRQoL measured by SIP improved after transplantation at least during 1 year, but the trajectories of the dimensions were not the same. The effect sizes calculated for studied patients were large for the physical area and small for the psychosocial area. So heart transplantation improves the physical domain of HRQoL. # 1498/POST-INTERVENTION QUALITY OF LIFE IN CARDIOVASCULAR PATIENTS Henk F. van Stel & Erik Buskens, Julius Center for Health Sciences and Primary Care, UMC Utrecht, Utrecht, Netherlands Introduction: We compared the health-related quality of life post-intervention in groups with various cardiovascular diseases. In addition, we assessed the heterogeneity between groups and the level of impairment compared to the general population. Methods: Post-intervention SF-36 scores from several randomised controlled trials with cardiovascular patients were gathered. The included studies were Octopump (on- vs. off-pump coronary artery bypass grafting, CABG; n ¼ 263), Octostent (off-pump CABG vs. percutaneous transluminal angioplasty, PTA; n ¼ 232), DIST (PTA vs. PTA with stenting in patients with claudication; n ¼ 245), and BOA (antiplatelet or anticoagulant therapy after infrainguinal bypass grafting surgery; n ¼ 525). Heterogeneity was tested with ANOVA, Levene-test (for equality of error variance) and Box-test (for equality of covariance matrices). Results: Both univariate and multivariate ANOVAs were highly significant (p < 0.001), as were both heterogeneity tests. Post-hoc comparisons showed that each study differed in one or more domains from all other studies. Conclusion: Patients return to normal functioning levels after coronary revascularisation and after PTA for claudication. HRQL remains impaired in patients undergoing infrainguinal bypass grafting surgery. Further combined analysis with these data will require multi-level modelling, to overcome the problem of heterogeneity. # 1764/HEALTH-RELATED QUALITY OF LIFE IN PULMONARY ARTERIAL HYPERTENSION: RESULTS FROM A POSTAL SURVEY Stephen P. McKenna, Galen Research, Manchester Science Park, Manchester; Natalie Doughty, PVD Unit, Papworth Hospital NHS Trust, Cambridge; David M. Meads & Lynda C. Doward, Galen Research, Manchester; Joanna Pepke-Zaba, PVD Unit, Papworth Hospital NHS Trust, Cambridge, UK No patient-reported disease-specific outcome measures are available for assessing the impact of pulmonary arterial hypertension (PAH) on patients. Qualitative interviews were conducted with 35 PAH patients to generate items to include in a measure assessing PAH-specific symptoms and functioning. We report on the first stage of testing the new instrument; item reduction. Items were selected for the questionnaire that covered issues that had been consistently raised by interviewees and that were judged by the clinical team to be a direct consequence of the condition. A postal survey was conducted with PAH patients who were currently under the care of a major clinical centre in Southern England. Results from the survey were used to reduce the number of items and to ensure that the scales fitted both principal component factor analysis and Rasch models. Seventy-nine patients responded to the questionnaire (25 males and 54 females; mean age 52.0 (15.3) years). Four symptom scales fitting the Rasch model were identified; breathlessness, fatigue, oedema and mood. These scales had a values ranging from 0.98 to 0.99 and person separation indices (psi) from 0.79 to 0.92. It was also possible to establish an overall symptoms scale consisting of 39 of these items that fitted the Rasch model with an a of 0.99 and w of 0.93. A separate scale for functioning was also identified (a 0.93, w ¼ 0.96). The new measure represents a potential means of determining the impact of PAH on the patient. Scales were shown to fit both factor analysis and Rasch models with good internal consistency. A second, larger postal survey is planned to reduce the number of items further and to determine the test–retest reliability and construct validity of the new scales.

837 # 1491/QUALITY OF LIFE AFTER VENOUS THROMBOEMBOLISM Mirjam Locadia & Mirjam Sprangers, Medical Psychology; Patrick Bossuyt, Clinical Epidemiology & Biostatistics, Academic Medical Centre, Amsterdam; Peep Stalmeier, Medical Technology Assessment, University Medical Centre Nijmegen, Nijmegen; Martin Prins, Clinical Epidemiology, Academic Hospital Maastricht, Maastricht, The Netherlands Patients with deep venous thrombosis or pulmonary embolism (venous thromboembolism) are treated with oral anticoagulants (OAC) for at least 3 months. During this treatment patients have an increased risk of major bleeding. Post-thrombotic syndrome (PTS) is a frequent chronic complication of deep venous thrombosis. We compared the quality of life (QOL) of patients who recently experienced an episode of venous thromboembolism (n ¼ 50), a major bleeding event during treatment with OAC (n ¼ 14), and patients with PTS (n ¼ 47). In addition, we compared the QOL of these patient groups with normative data of the Dutch population. Patients completed questionnaires on generic QOL (SF-36) and demographics. Multivariate analyses of variance indicated significant differences on QOL between the groups, corrected for age and gender. Patients with PTS reported better physical, physical role, and social functioning than patients who experienced an episode of venous thromboembolism or a major bleeding event (p < 0.01). QOL did not differ significantly after venous thromboembolism or major bleeding. As compared to the general population, the QOL of patients with PTS was impaired on the physical aspects (p < 0.01), but not on social functioning, emotional role functioning and mental health. QOL was impaired on all aspects (p < 0.01), except for mental health, after venous thromboembolism, and on all aspects (p < 0.01), except for emotional role functioning and mental health, after major bleeding. In conclusion, QOL was least impaired in patients with PTS and comparable after an episode of venous thromboembolism or a major bleeding event. Each patient group had lower levels of QOL than the general population. # 1211/HEALTH-RELATED QUALITY OF LIFE AND CARDIOVASCULAR RISK Francesca Baiardi, Health Economics & Outcomes Research, CliCon Srl; Ezio Degli Esposti, Clinical Effectiveness Unit, LHU Ravenna; Giorgia Valpiani, Maria Grazia Puglia & Pier Luigi Ceccarelli, Health Economics & Outcomes Research, CliCon Srl; Alberto Gualdrini, General Practice Setting, LHU Ravenna, Faenza, Ravenna, Italy The aim of this study, which is part of a larger project called Faenza, was to assess the Health-related Quality of Life (HrQoL) and cardiovascular risk (CVR) in a cohort of patients of the Ravenna Local Health Service General Practitioners. Individual level of CVR was calculated collecting the following data: age, gender, glycemia, total cholesterolemia, systolic (SBP) and diastolic (DBP) blood pressure, diabetes, smoking habits. QoL assessment was performed using the questionnaire on the state of health SF-36, administered on occassion of the screening visit. Patients were classified according to SF-36 Physical (PCS) and Mental (MCS) Component Summaries quartiles distributions. Differences in baseline characteristics between groups were analysed with one-way analysis of variance in the case of continuous variables (age, SBP, DBP, cholesterolemia, glycemia) and with v2 tests for categorical variables (gender, diabetes, smoking habits, CVR level). The null hypothesis was rejected for p values of less than 0.05.

All analysis was carried out using SPSS for Windows 11.0 software. The mean age of the 1704 subjects enrolled was 48.08 ± 16.22 years and 792 (46.5%) were males. As to PCS, statistically significant differences (p < 0.05) appeared in the comparison among groups relative to all considered indicators, except for that of smoking habits. The prevalence of subjects with high CVR progressively decreases from 1st to 4th quartile (9.4, 8.2, 4.2, 3.3%). As far as MCS is concerned, the only data significantly different are age (p ¼ 0.003) and gender (p < 0.001): the subjects of the 1st quartile are those with the highest mean age and with less proportion of males. Perceptions concerning patients’ own physical conditions seem to reflect the real state of health, since PCS correlates significantly with metabolic values. The psychological wellness does not seem to be influenced by the presence of CVR factors and this aspect could have a heavy implication on the patient’s compliance. # 1093/HEALTH RELATED QUALITY OF LIFE AND SYMPTOMATOLOGY IN CHILDREN DIAGNOSED WITH ATTENTION DEFICIT/ HYPERACTIVITY DISORDER Anne F. Klassen & Anton Miller, Pediatrics, University of British Columbia; Stuart Fine, ADHD Clinic, BC Children’s Hospital; Mary DeVera, Nicola Butler, Pediatrics, University of British Columbia, Vancouver, BC, Canada Attention deficit/hyperactivity disorder (ADHD) is a common psychiatric disorder, affecting approximately 3–5% of children. It is characterized by developmentally inappropriate levels of inattentiveness, distractibility, impulsivity and hyperactivity, and is associated with significant academic, social and emotional problems at home and at school. The aim of treatment for ADHD is to decrease symptoms, enhance functionality and improve wellbeing for the child and his or her close contacts. However, measurement of treatment response is often limited to measuring a reduction in symptoms using behavior rating scales and checklists completed by teachers and parents. Since so much of the focus has been on symptom reduction, less is known about the impact of ADHD on other aspects of health and wellbeing. In this study, we used the Child Health Questionnaire (CHQ-PF50) to assess the health of children with ADHD from the parent perspective. Our specific objectives were to measure the HRQoL of children diagnosed with ADHD, and to investigate the relationship between ADHD symptoms and HRQoL. Subjects were recruited in 2001-2002 from the ADHD clinic at BC Children’s Hospital in Vancouver (Canada), a provincial referral center for children aged 6–17. Parents were sent a questionnaire package in the mail prior to their child’s appointment. The package included the following instruments: the Child or Adolescent Symptom Inventory 4, (measures symptoms for a wide range of DSM-IV diagnoses); the Hyperactivity Index of the Conners Parent and Teacher Rating Scale (measures overall psychopathology in children presenting with ADHD); and the CHQ-PF50. 335 questionnaire packages were mailed. Twenty three percent of patients were deemed ineligible, most commonly because the referral had been cancelled. The overall response rate for eligible subjects was 64.3% (n ¼ 166). In this paper, we will present summary statistics for the CHQ-PF50, for the sample as a whole, and for subgroups (by agegroup, sex, ADHD subtype, presence of co-morbidity). We will also describe the relationship between HRQoL and ADHD symptomatology.

838 # 1611/HEALTH STATUS INDICATORS AND HEALTH CARE NEEDS IN CHILDREN WITH CHRONIC CONDITIONS: A CROSSNATIONAL PERSPECTIVE Silke Schmidt, Corinna Petersen & Monika Bullinger, Institute of Medical Psychology, University Clinic, Hamburg; Angela Gosch, Robert Koch Institute, Berlin, Germany; Ute Thyen, Pediatric Clinic, University of Luebeck, and the DISABKIDS GROUP, EU This study explains heterogeneities in childrensÕs health status by analysing a variety of disease-related and consequence-based health indicators in children with chronic conditions from a cross-national perspective. Quality of life from the childrenÕs perspective was explained by sociodemographic, disease-related variables, health status and health care needs indices in a population of children with seven chronic conditions, pooled and per country. In a sample of 372 children, aged 8–16 and diagnosed with seven chronic conditions in seven different European pediatric clinics, the following instruments were employed: Parental measures included the FS2-R, special health care needs with the 5-item CSHCN screener as well as a health service related questionnaire, disease severity with clinical measures and quality of life assessed with a brief instrument of both KINDL and CHQ items. ANOVAS and regression analyses were employed. Country differences in children’s QoL showed moderate effects when entered separately, however disappeared when including chronic conditions (effect size ¼ 0.01) in the analyses which explained a higher proportion of variance (effect size ¼ 0.07). General health status and disease severity were both correlated with overall Qol and health care needs, however interpersonal functioning negatively. The identification of CSHCN varied both across countries and condition, but was not related to QoL. Within various health service related questions, QoL of children was only a function of receiving specialist care, and this effect was independent of culture. Results indicate that the associations between consequences-based criteria, such as limitations and health care parameters and childrenÔs QoL showed higher cross-cultural variations than biomedical indicators in this particular data set. Despite the limitations of the sample size given the wide variety of factors involved in health status, results suggest a range of universal relationships between health indicators and quality of life from the childrenÕs perspective. # 1132/QUALITY OF LIFE AND HEALTH RISK BEHAVIORS AMONG STUDENTS WITH AND WITHOUT DISABILITIES IN WASHINGTON STATE, USA Todd C. Edwards, Tari D. Topolski & Donald L. Patrick, Health Services, University of Washington, Seattle, WA, USA The Washington State Department of Health conducted the Washington State Healthy Youth Survey in autumn 2002 to assess health risk and protective factors focusing on lifestyle and behaviors and the social environment. Health-related risk and protective factors are commonly grouped into three general categories including lifestyle and behavior; environmental exposure, encompassing both the physical and social environments; and biologic and genetic characteristics. Social environment includes the school, peer, community and home environments, as well as individual assets. The survey included some factors directly related to health, but most of the risk and protective factors were associated with intermediary behaviors, such as drug and

tobacco use, violence and staying in school. The Youth Quality of Life Instrument - Disability Screener (YQOL-DS) and six items of the Youth Quality of Life Instrument-Surveillance Version (YQOL-S) were incorporated into the version used with 8th–12th grade students. The YQOL-DS was used to identify students with self-reported disabilities. Approximately 20,000 students in grades 6, 8, 10, and 12 were sampled, across 125 schools statewide (6% of students at these grade levels). Based on the 2002–2003 enrollment figures, the student response rates were 60.4% in 6th grade, 68.2% in 8th grade, 44.5% in 10th grade, and 43.6% in 12th grade. Results showed that approximately 25% of adolescents reported YQOL scores of 63 or less. These scores are comparable to the average scores obtained from youth with mobility limitations and craniofacial conditions using the YQOL-Research version. Looking at all youth, 50% of 8th graders, 44% of 10th graders, and 35% of 12th graders who reported that they used tobacco reported YQOL scores of 63 or less (which is the lowest quartile). This presentation will discuss the YQOL-DS as a tool for identifying adolescents with disabilities and health risk behaviors, and quality of life among youth with and without disabilities. # 1571/COMPARING QUALITY OF LIFE (QOL) OF YOUTH WITH CRANIOFACIAL DIFFERENCES (CFDS) TO A NO CHRONIC CONDITIONS GROUP Tari D. Topolski, Todd C. Edwards & Donald L. Patrick, School of Public Health and Community Medicine, University of Washington, Seattle, WA, USA QoL is often invoked as the justification for craniofacial surgeries and other interventions for youth with CFDs. Few studies have been conducted to ascertain the quality of these youths lives and research has shown that objective medical indicators do not reliably predict psychological adjustment in individuals with congenital CFDs. Moreover, most often it is the physician or the parents who make the decision as to how the intervention will improve QoL, without considering the youths own desires or expectations. To fill the gap in understanding QoL among these youth, the Youth Quality of Life Instrument-Research Version (YQOL-R) was used to explore differences between a group of adolescents with CFDs and a group without any diagnosed chronic conditions (NCC). Fifty-six adolescents with CFDs were recruited from Childrens Hospital, Seattle to participate in the study. YQOL-R scores for these youth were compared to the scores for 102 NCC youth, who were previously recruited to participate in a study of adolescents with and without chronic conditions. The data were analysed using MANCOVA adjusting for age, gender and depressive symptomatology. The results showed that youth with CFDs reported significantly lower QoL scores. Only one YQOL-R domain, Relationships, did not show a significant difference between groups. Looking at the items that comprise this domain, adolescents with CFDs scored very similar to the adolescents with NCC on most of the family variables and slightly higher on feeling understood by their parents and getting along with their parents. They scored somewhat lower on the items having to do with peers, for instance feeling that they were being treated with less respect by their peers. These results suggest that adolescents with craniofacial conditions confront significant challenges to their self-identity while experiencing higher quality of life from relationships, possibly resulting from their need to negotiate and maintain close family support.

839 # 1601/QUALITY OF LIFE IN CHILDREN WITH OTITIS MEDIA (OMAVAX) C.N.M. Brouwer & A.R. Maille´, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht; R.H. Veenhoven, Department of Pediatrics, Spaarne Hospital Haarlem, Haarlem; E.A.M. Sanders, Department of Pediatric Immunology; D.E. Grobbee & M.M. Rovers, Julius Center for Health Sciences and Primary Care; A.G.M. Schilder, Department of Otolaryngology, University Medical Center Utrecht, Utrecht, The Netherlands Purposes of the study were: (1) Assessment of health-related and otitis media-specific quality of life (QoL) in children with recurrent otitis media (OM), and (2) Comparison of health-related QoL in children with recurrent OM with that in children from a general population and those with mild to moderate asthma. Patients: 307 children aged 1–7 year with recurrent OM (118 with 2–3 episodes per year; 189 with 4 or more episodes per year i.e. otitis prone children, mean age 2.8 year), 117 healthy controls (mean age 3.0 year), 64 children with mild to moderate asthma (mean age 4.2 year). Their parents completed Dutch versions of questionnaires measuring health related QoL (RAND and FSQ Specific) and OM specific QoL (OM-6). Scores were transformed to a 0–100 scale. Results showed that (1) Otitis prone children scored significantly poorer than children with fewer episodes of OM on the RAND (50.4 vs. 56.4, p ¼ 0.003), the FSQ Specific (78.4 vs. 83.9, p ¼ 0.002) and the OM-6 (61.1 vs. 71.7, p < 0.001). (2) Children with OM scored significantly poorer (p < 0.001) than healthy children on the RAND and the FSQ Specific. The scores of the otitis prone group on the RAND and FSQ Specific were significantly lower as compared to the scores of children with mild to moderate asthma (p < 0.001). In conclusion, parents of otitis prone children rated the health related QoL of their children significantly poorer than parents of children with fewer episodes of OM. Otitis-prone children expressed more behavioural problems related to illness (FSQ Specific) than the children with fewer episodes and healthy controls. The parents of otitis prone children noticed more complaints related to OM (OM-6). Parents of otitis prone children also rated the health related QoL of their children poorer than parents of healthy children or asthmatic children. # 1071/SELF-REPORTED RELIGIOSITY AND MENTAL HEALTH: A POPULATION-BASED STUDY Azita Goshtasebi & Mariam Vahdaninia, Iranian Institute for Health Sciences Research; Kazem Mohammad & Abbas Rahimi, Tehran University of Medical Sciences; Ali Montazeri, Iranian Institute for Health Sciences Research, Tehran, Iran The objective of this study was to investigate the relationship between religiosity and mental health among the general population in Tehran, Iran. Using a cluster sampling method in all, 4163 healthy individuals aged 15 years and over were interviewed. Religiosity was measured by a self-reported Likert scale indicating the highest to the lowest degree of religiosity. Mental health was assessed using the Short Form Health Survey Questionnaire (SF-36). Ninety-seven individuals did not indicated their religious commitment and this led to analysis of data for the 4066 remaining individuals. The mean age of the respondents was 35.1 (SD ¼ 16.0) years, 52% were female, mostly married (58%) and the mean years of their formal education was 10.0 (SD ¼ 4.5). The mean mental health score for the whole population was 67.0 (SD ¼ 17.9). Using one-way analysis of variance the results showed that there was no significant difference between religiosity and mental health (p ¼ 0.11). In general, those with normal religious beliefs showed a better mental health status compared to those indicated the highest and the lowest degree of religiosity (mean score 67.1 vs. 66.0 and 63.6, respectively). Indeed, those indicated the lowest degree of religiosity showed the worse mental health status. The study finding suggests that normal religious commitment could be regarded as a source of positive psychological health and may play an important role in improved quality of life. However, in studying the relationship between religiosity and mental health there is need for further investigation to indicate which aspect of psychological health and which definitions of religiosity are correlated.

# 1061/SELF-REPORTED RELIGIOSITY AND GENERAL HEATH PERCEPTIONS: A POPULATION-BASED STUDY Mariam Vahdaninia & Azita Ghoshtasebi, Iranian Institute for Health Sciences Research; Kazem Mohammad & Abbas Rahimi, Tehran University of Medical Sciences; Ali Montazeri, Iranian Institute for Health Sciences Research, Tehran, Iran The objective of this study was to investigate the relationship between religiosity and general health among the general population in Tehran, Iran. Using a cluster sampling method in all, 4163 healthy individuals aged 15 years and over were interviewed. Religiosity was defined by a self-reported Likert scale indicating the highest to the lowest degree of religiosity. General health perceptions was measured using the Short Form Health Survey Questionnaire (SF-36). Ninety-seven individuals did not indicated their religious commitment and this led to analysis of data for the 4066 remaining individuals. The mean age of the respondents was 35.1 (SD ¼ 16.0) years, 52% were female, mostly married (58%) and the mean years of their formal education was 10.0 (SD ¼ 4.5). The mean general health score for the whole population was 67.4 (SD ¼ 20.3). Using one way analysis of variance the results showed that there was significant difference between religiosity and general health (p < 0.0001). Those with normal religious beliefs showed better general health perceptions compared to the highest and the lowest extremes (mean score 69.7 vs. 62.3 and 66.3, respectively). The findings indicate that those with normal degree of religiosity experience a better general health status and this suggests that a balanced religiosity may play an important role in improved quality of life. Since religion is multifaceted in nature, in studying the relationship between religiosity and general health there is need to include measures of institutional religion, ideological religion, and personal devotion to avoid ambiguous findings. # 1603/QUALITY OF LIFE IN A FREE CLINIC Shannon O’Connor, Theodore G. Ganiats, Ellen Beck & Robert M. Kaplan, Family and Preventive Medicine, University of California San Diego, La Jolla, CA, USA Background: Three medical student-run free clinics were started in 1997 with a mission to provide accessible, quality healthcare for the underserved. The clinic serves a large portion of minorities and most patients are females. The clinic has many anecdotal cases of patients reporting improvement, but there has been no objective evaluation. The goal of this project is to implement a quantitative measure of health related quality of life in the free clinic population and to follow the quality of life of this population over time. This abstract presents the results of the initial 2 months of data collection. Methods: The selfadministered version of the Quality of Well-Being Scale (QWB-SA) was incorporated into the three sites. Consenting patients were asked to fill out a QWB-SA form at each visit up to once per month. New patients and returning patients were analysed separately. All of the forms are collected and delivered to a central office for scoring. The data is then returned to the Free Clinic in a database for analysis. Results: In the first 2 months 165 patients completed their first QWBSA, including 60 new and 105 existing patients. The overall mean QWB-SA score was 0.596 (SD ¼ 0.16). The average score of the 65 men was similar to the 100 women (0.602, SD ¼ 0.19 vs. 0.591, SD ¼ 0.15). As expected, the QWB-SA scores varied by age, with the youngest having the highest QOL score. Returning patients had a QWB-SA score of 0.599, SD ¼ 0.17, and new patients had a score of 0.590, SD ¼ 0.15. Conclusions: The early results of this study show that the QWB-SA can be implemented into free clinics and that the data can be collected up to monthly. The preliminary data show trends that are not yet statistically significant. It is difficult to know what results to expect; those who have not received medical care recently may experience an increase or a transient decrease in HRQOL with an initiation of medical therapy. This ongoing study will provide important insights into the health outcomes of high-risk individuals who are reintroduced into a health care system.

840 # 1440/HEALTH PERCEPTIONS OF ADULT POPULATION IN PROVINCE VOJVODINA Vera Grujic, Mirjana Martinov Cvejin, Vesna Mijatovic Jovanovic, Erzebet Ac Nikolic & Olja Niciforovic Surkovic, Department of Social medicine, Institute of Public Health Novi Sad, Novi Sad, Serbia, Serbia & Montenegro, Yugoslavia Introduction: The definition of health-related quality of life covers five broad concepts that combine quantity and quality of life on a value scale or continuum: opportunity, health perception, functional status, impairments, and death or duration of life-domains of these concepts. Perceived health is an indicator of the value of medical care and health status. Self ratings are important predictor of mortality. Methodology: Cross sectional study performed in Province Vojvodina, health interview survey, the sample drawn by electorall registar, the mode of data collection – face to face in a household settings (n ¼ 2239), sample size determined by estimated frequency of observed variables, homogenity and confidence level of 95. Results: In 2000, almost 50% of general adult population rated their health as relatively good and good, while every sixth person assessed their health as fair or poor. Self-assessment of health status differs by gender, age, educational levels, economic status. More women, elderly, people with the lowest educational level and with a low economic status statistically significant assessed their health as fair and poor. Risk factors for noncommunicable diseases (physical inactivity, obesity, hypertension) are more spread among those people. Their self esteem about current health compared with one in the past is lower than in other population groups. Conclusion: Analysis of health perception of adult population indicated strong relation between different demographic and socioeconomic variables and self-assessed health status. # 1179/COMPARING THE QUALITY OF LIFE OF THE PHYSICALLY DISABLED TO THE GENERAL POPULATION: DIFFERENCES AND THE PERSONAL AND SOCIAL FACTORS Cretien van Campen, Health Care, Social and Cultural Planning Office of The Netherlands, The Hague, The Netherlands An object of the health policy in the Netherlands is, like in many other countries, to provide a life as normal as possible to the disabled. The government wants to reduce the differences in quality of life between the disabled and the non-disabled. The purpose of this research is to gain insight into these differences and into the personal and social factors that explain them. Research questions: (1) What are the differences in quality of life between the disabled and the general population in the Netherlands? (2) Which personal and social factors explain these differences? Methods: Survey data of a sample selection of the Dutch population are analysed. The sample consists of ca. 6000 persons, including 700 physically disabled persons. The survey contains measuring instruments for quality of life (SF-12) and physical disability (OECD) and a range of personal and social indicators: e.g. age, sex, education, labor, income, housing, social networks, participation, health, health care use, and leisure activities. The data set is already available and multivariate analyses are conducted this spring and summer. Results: The following results will be presented at the time of the conference: First, a description will be given of the differences between the disabled and the general population on eight health related dimensions of quality of life: physical functioning, role-physical, bodily pain, general health, vitality, social functioning, role-emotional, and mental health. Second, the results of the multivariate analyses of personal and social factors will be presented. What factors explain these differences and what is their relative share in the explanation? Discussion: In the discussion, the results will be considered from a health policy perspective. Which decisive personal and social factors are relevant for policymakers that want to reduce the differences in quality of life between the disabled and nondisabled?

# 1612/RELATIONSHIP BETWEEN PAST 12-MONTH PHYSICAL ACTIVITY AND QUALITY OF LIFE IN THE SU VI MAX STUDY Anne Vuillemin, Ste´phanie Boini, Serge Brianc¸on & Francis Guillemin, School of Public Health – EA3444, University Henri Poincare, Vandoeuvre-les-Nancy; Sandrine Bertrais, U557 Inserm (UMR Inserm/Inra/Cnam), Paris; Jean-Michel Oppert, Nutrition Department, Hoˆtel Dieu Hospital, Paris; Pilar Galan, Serge Hercberg, U557 Inserm (UMR Inserm/Inra/Cnam), Paris, France The purpose of our research was to study the relationship between the level of physical activity and various dimensions of quality of life. Among the 12,741 volunteers involved in the SUVIMAX. study, 5632 healthy subjects (2323 men, mean age: 51.4 ± 4.7; 3309 women, mean age: 46.9 ± 6.5) answered both physical activity and quality of life (SF-36) questionnaires. Past 12-month leisure physical activity (h/ week, Mets-h/week) and time spent watching television (min/d) were measured using the french version of the modifiable activity questionnaire. Subjects were classified in four groups: inactivity, irregular, moderate, intense for physical activity and <1 h/d, 1–2 h/d, 2–3 h/d, >3 h/d for television. Associations between physical activity and quality of life were estimated for men and women separately using age-adjusted Spearman’s correlation and analysis of variance. A higher level of physical activity was associated with a higher level of quality of life scores. Higher correlations were found for physical functioning and vitality scores in men and women (r ¼ 0.18–0.27, p < 0.0001). A higher time spent watching television was associated with a lower level of quality of life (especially physical functioning score: r ¼ )0.08 in men and r ¼ )0.11 in women, p < 0.0001). Regarding the analysis by groups, whatever the indicator considered (physical activity or television), each group was significantly associated with quality of life (p < 0.0001). In general, the mean quality of life scores were significantly different between groups, but more specifically between extreme groups. For instance, physical functioning score for intense physical activity were 95.8 ± 0.5 and 94.8 ± 0.6, there were 87.1 ± 0.8 and 84.6 ± 0.7 for inactivity (p < 0.0001) for men and women, respectively. Although the relationships between physical activity and quality of life are complex, these preliminary results show that a higher level of physical activity may exert a positive effect on quality of life, particularly on physical functioning and vitality dimensions. These results should encourage people to adopt an active behaviour during leisure-time. # 1388/IS SELF-RATED HEATLH ASSOCIATED WITH BIOMARKERS? Carola Bardage, Pia Svedberg & Nancy L. Pedersen, Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden The association between self-rated health and morbidity and mortality is well-known. However, the biological mechanisms behind a good or bad self-rated health have not been studied to any great extent. The purpose of this study was to examine the relationship of frequently used biomarkers and self-rated health. The study used data from the Swedish Twin Registry. Preliminary results showed that higher levels of triglycerides and cholesterol, lower levels of high-density lipoprotein cholesterol (HDL-C), lower albumin levels, and higher creatinine levels were associated with fair or poor self-rated health. The relationships did not remain after adjusting for sex, age, education, life-style factors, blood pressure, coronary heart disease, stroke, diabetes, cancer, and gastrointestinal diseases. We conclude that self-ratings of health have biological explanations. Self-rated health is a sensitive indicator of the biological state of the organism, and reflects differences also at subclinical levels of the biomarkers.

841 # 1392/MAKING SENSE OF QOL RESULTS IN ANEMIA: DESCRIBING FACT-AN RESULTS IN TERMS OF SF-36 PHYSICAL FUNCTIONING Simon Ferrazzi & Pieway Hwang, Health Economics & Outcomes Research, Janssen Ortho Inc., Toronto, Ontario, Canada Purpose of research: Determining the clinical significance of changes in QOL scores can be challenging. To assist clinicians in interpreting QOL, we characterized FACT-An (Functional Assessment of Cancer Therapy-Anemia) scores observed in a randomized, double-blind, placebo-controlled trial (Littlewood et al., 2001), using SF-36 Physical Functioning (PF) items. Subject Sample and Statement of Methods: Patients aged >17 years with solid or hematologic nonmyeloid malignancies receiving nonplatinum chemotherapy were assigned 2:1 to receive 150–300 IU/kg EPO or placebo SC tiw (maximum 28 weeks) for cancer-related anemia. FACT-An (FACT-G and Anemia subscale) and SF-36 scores were obtained at baseline and final assessment. On average, EPO patients had improvements in scores of +2.5 for the FACT-G and +4.0 in the Anemia subscale whereas placebo patients had decrements of minus 3.6 and 2.6, respectively. We analysed the impact of these changes in FACT-An scores by mapping the SF-36 PF items to these strata. Summary of results: A >4 point increase (Anemia subscale) in the EPO group resulted in a higher level of physical functioning vs. placebo patients with a >2.5 point decrement for all activities (p < 0.02). Similar results were obtained for FACT-G scores (data not shown). Comparing changes in the FACT-An to SF-36 PF items provides understanding of the clinical impact of changes in FACT-An scores. # 1405/LINEAR ANALOGUE SCALE ASSESSMENT (LASA) SCORES: WHAT DO THEY MEAN IN TERMS OF PHYSICAL FUNCTIONING FOR ANEMIC PATIENTS? Pieway Hwang & Simon Ferrazzi, Health Economics & Outcomes Research, Janssen-Ortho Inc., Toronto, Ontario, Canada Purpose of research: Interpreting QOL scores can be problematic. In order to provide clinicians with a new perspective in assessing LASA scores as observed in a randomized, double-blind, placebo controlled trial (Littlewood et al., 2001), we characterized LASA scores using the Physical Functioning (PF) items of the SF-36. Subject sample and statement of methods: Patients >17 years of age with solid or hematologic nonmyeloid malignancies receiving nonplatinum chemotherapy were assigned 2:1 to receive 150–300 IU/kg EPO or placebo SC 3x/ week (maximum 28 weeks) for cancer-related anemia. LASA and SF36 scores were obtained at baseline and final assessment. On average, EPO patients had improvements in scores of +8.1 (Energy Level), +7.5 (Ability to Do Daily Activities) and +4.8 (Overall QOL) whereas placebo patients had decrements of minus 5.8, 6.0 and 6.0, respectively. We analysed the impact of changes in LASA scores by mapping the PF items of the SF-36 to these strata. Summary of results: A >8 mm increase in Energy Level rating for EPO patients resulted in a higher level of physical functioning vs. placebo patients with a >5.5 mm drop in rating (p < 0.03). Similar results were obtained for the two other LASA scales (data not shown). Comparing changes in LASA ratings to changes in SF-36 PF items provides easy understanding of the impact an intervention has on QOL.

# 1413/DECIPHERING VITALITY IN ANEMIC PATIENTS USING FACT-AN RESULTS Simon Ferrazzi & Pieway Hwang, Health Economics & Outcomes Research, Janssen-Ortho Inc., Toronto, Ontario, Canada Purpose of research: Results from disease specific instruments may be better interpreted in the context of generic QOL results. The SF-36 Vitality (VT) scale was used to enhance understanding of Functional Assessment of Cancer Therapy-Anemia (FACT-An) scores observed in a randomized, double-blind, placebo-controlled trial (Littlewood et al., 2001). Subject sample and statement of methods: Patients aged >17 years with solid or hematologic nonmyeloid malignancies receiving nonplatinum chemotherapy were assigned 2:1 to receive 150– 300 IU/kg EPO or placebo SC 3x/week (maximum 28 weeks) for cancer-related anemia. FACT-An (FACT-G, Anemia subscale, Fatigue subscale) and SF-36 scores were obtained at baseline and final assessment. On average, EPO patients had improvements in scores of +2.5 (FACT-G), +3.0 (Fatigue) and +4.0 (Anemia) whereas placebo patients had mean decrements of minus 3.6, 2.2 and 2.6, respectively. We analysed the impact of these changes in FACT-An scores on patient Vitality by mapping the SF-36 VT items to these strata. Summary of results: A P3 point increase in the Fatigue subscale for either EPO (achieved by 50.5%) or Placebo (achieved by 36.7%) treated patients resulted in a higher degree of Vitality in the previous 4 weeks vs. declining scores of >2.0 (p < 0.01). Similar results were obtained for the FACT-G and Anemia subscale (data not shown). Generic instruments can assist in interpreting scores from disease specific questionnaires bringing clarity to the impact of interventions on patient QOL. # 1610/EFFECTIVENESS OF FOOT ORTHOSES IN THE TREATMENT OF PLANTAR FASCIITIS: A RANDOMISED CLINICAL TRIAL Karl B. Landorf, School of Exercise and Health Sciences, University of Western Sydney, Penrith South DC, NSW, Australia; Anne-Maree Keenan, Rheumatology and Rehabilitation Unit, University of Leeds, Leeds, United Kingdom; Robert Herbert, School of Physiotherapy, University of Sydney, Lidcombe, NSW, Australia Foot orthoses are commonly used in the treatment of plantar fasciitis, a common foot problem causing pain in the heel. This research project evaluated the effect foot orthoses had on health-related quality of life when used to treat plantar fasciitis. One hundred and thirty five participants were randomly allocated to one of three treatment groups: (1) a sham orthosis, (2) a pre-fabricated orthosis, (3) a customised orthosis. Primary outcome measures were the pain and function domains of the Foot Health Status Questionnaire – a validated, foot specific, health-related quality of life instrument – at 3 and 12 months. (Secondary outcomes included four domains from the SF-36: physical function, general health, vitality and social functioning.) Analysis of the data was by intention-to-treat. A linear regression approach to ANCOVA was used to compare the effects of treatments. At 3 months, estimates of treatment effect on function favoured the pre-fabricated and customised orthoses over the sham orthosis. Compared to the sham orthosis, function improved 8.4 points [0–100 scale] more for the pre-fabricated orthosis (95% CI, 1.0–15.8; p ¼ 0.026) and 7.5 points more for the customised orthosis (95% CI, 0.3–14.7; p ¼ 0.041). These effects are statistically significant, however it is not clear if they are large enough to be clinically worthwhile. There were no other significant differences in the primary analysis. These results suggest there may be a short-term benefit in function if appropriate orthoses are prescribed for the treatment of plantar fasciitis. As there were no differences between the pre-fabricated and customised orthoses over the 12 months, the cheaper pre-fabricated orthosis is recommended. Finally, the lack of significant findings at 12 months may simply indicate that plantar fasciitis has a natural course, which for most people is one of gradual resolution.

842 # 1684/PREVALENCE OF ‘GAP’ BETWEEN QUALITY OF LIFE AND PERCEIVED SEVERITY OF LOW BACK PAIN AND ITS RELATIONSHIP WITH PSYCHOSOCIAL FACTORS: A POPULATION-BASED SURVEY Natsuko Takahashi, Department of Epidemiology and Health Care Research, Kyoto University, Kyoto; Shinichi Kikuchi & Shinichi Konno, Department of Orthopedic Surgery, Fukushima Medical University, Fukushima; Satoshi Morita, Department of Epidemiological & Clinical Research; Yoshimi Suzukamo & Shunichi Fukuhara, Department of Epidemiology and HealthCare Research, Kyoto University, Kyoto, Japan The specific aims of this study were, among those having Low Back Pain (LBP), to estimate prevalence of dissociation (gap) between Quality of Life (QOL) and perceived severity of pain, and to examine the relationship of the psychosocial factors with this dissociation. We used the data from a cross-sectional population-based survey in 2002, in which 4500 residents were randomly sampled. The self-administered questionnaire includes 0–10 visual analogue scale measuring severity of LBP (VAS), 24 items of Roland-Morris Disability Questionnaire (RDQ) as a LBP-related QOL instrument, SF-36 and Perceived Stress Scale (PSS). ANCOVA was used to test association of psychosocial factors with this gap. The response rate was 66% (2966). The overall prevalence of LBP was estimated to be 30% (906). RDQ score increased with age. We identified two subgroups among those with LBP, who showed significant dissociation between VAS and RDQ score category (we set three score categories; ‘the best’, ‘moderate’ and ‘the worst’); Group 1 with the worst RDQ despite the best VAS (n ¼ 25, 3%), Group 2 with the best RDQ despite the worst VAS (n ¼ 80, 9%). As a control group, we set Group 3 with RDQ level and VAS scores in the same score category (n ¼ 441, 49%). ANCOVA showed that age was a factor significantly associated with this ‘gap’ when adjusted for sex, comorbidities, employment status, and physical loading at work; the least square mean (LSM) age of the each groups was 58 for Group 1, 48 for Group 2, 53 for Group 3 respectively (p < 0.01). In addition, PSS and Mental Health of SF-36 (MH) scores were also significant factors; LSM of PSS (high score means more stress) was 24 for Group 1, 20 for Group 2, 22 for Group 3 (p ¼ 0.03), as for MH, 62 for Group 1, 70 for Group 2, 64 for Group 3, respectively (p ¼ 0.04). In conclusion, there were certain number of people (12%) among those with LBP who have a gap between severity of pain and QOL. This study confirmed that psychosocial factors such as mental health and perceived stress are important factors significantly associated with this gap. # 1531/HEALTH RELATED QUALITY OF LIFE AT 6 MONTHS AND TWO YEARS AFTER TOTAL KNEE REPLACEMENT Antonio Escobar, Research Unit, Hospital de Basurto, Bilbao; J. Marı´a Quintana, Research Unit, Hospital de Galdakao, Galdakao; Jesus Azkarate, Orthopedic Surgery, Hospital de Mendaro, Mendaro; J. IGnacio Gu¨enaga, Orthopedic Surgery, Hospital de Santiago, Vitoria; Iratxe Lafuente, Research Unit, Hospital de Galdakao, Galdakao, Spain Health-related quality of life (HRQoL) measures have been administered over the past years as a global assessments of patient’s health status. Total knee replacement (TKR) is one of the procedures that are currently increasing more. The objective of this study was to measure the change in HRQoL of patients who have undergone a TKR in two times, at 6 months and 2 years post-surgical intervention. The study

sample consisted in 366 patients from three tertiary Hospitals in the Basque Country (Spain). Patients with diagnosis of osteoarthritis undergoing TKR were recruited and fullfilled the following HRQoL questionnaires three times, first while the patients were in the waiting list, secondly 6 months after the intervention and finally 2 years postintervention: SF-36 as generic and WOMAC as specific questionnaires. ANOVA models were employed to study HRQoL changes. At 6 months there were a remarkable gain in HRQoL as much in the WOMAC as in the SF-36. The change in the WOMAC questionnaire were: pain from 54.7 prior to the intervention to 22.7 6 months afterwards; stiffness from 56.4 to 30.4 and functional capability from 59.9 to 31.9. In the SF-36 there were as well a gain in all domains from 3.2 points to 27.9, being the domains that more have improved the Role Physical, Physical Functioning and Bodily Pain (+27.9; +23.5 and +18.5 points). From 6 months to 2 years, there were a slightly gain in WOMAC, in all three domains, going from 5.8 in stiffness to 2.1 points in functional capability, while in the SF-36 gain only was observed in the Role Physical domain, 4.7 points and even there were a decrease in the rest of domains. TKR is an effective mean of relieving pain and improving function. The largest gains in HRQoL are seen at 6 months post-intervention so much in the specific questionnaire as in the generic one. At 2 years the gains in quality of life are very small, even in the generic questionnaire decreases are observed in the punctuations, possibly due to other comorbidities. # 1710/QUALITY OF LIFE/LIFE SATISFACTION AMONG WOMEN WITH BREAST CANCER WHO HAVE RECEIVED COMPLEMENTARY CARE AND A MATCHED GROUP OF WOMEN WITHIN CONVENTIONAL MEDICINE Marianne Carlsson, Public Health and Caring Sciences, Uppsala University, Uppsala, Sweden; Maria Arman, Department of Caring Sciences, Abo Akademi University, Vasa, Finland; Marie Backman, Clinical Nursing, The Swedish Red Cross University College of Nursing, Stockholm; Ursula Flatters, Administration; Vidar Clinic, Jarna; Thomas Hatschek, Oncology, Karolinska Hospital, Stockholm, Elisabeth Hamrin, Medicine and Care, Linkoping University, Linkoping, Sweden The perceived quality of life/life satisfaction was studied in two samples of women with breast cancer (60 women in each). The first sample was women who were treated in a hospital with alternative/ complementary care (anthroposophic medicine) during a mean period of 14 days, the second sample consisted of individually matched patients treated only within conventional care in the same geographic area. Repeated measurements of QoL was studied during 1 year with EORTC QLQ-C30 and an instrument developed within the project Life Satisfaction Questionnaire, LSQ. Some medical parameters as well as immunology were also included. The comparisons between the two samples (the matched pairs) showed that although the women who had received the complementary care reported somewhat lower QoL on admission to the study, their QoL on both instrument increased significantly during the 1-year study period. Expressed as effect sizes (ES) in the majority of the subscales in the EORTC QLQ-C30 as well as the LSQ. There was no significant difference in survival 1 year after entrance. In conclusion the QoL seemed to have improved by the stay in the anthroposophic hospital, although we cannot exclude the fact that the design was a matching procedure since randomization was not possible. The study was sponsored by the Swedish Cancer Society and other foundations in Sweden and Finland.

843 # 1714/QUALITY OF LIFE AFTER SURGERY OF CARCINOMA OF THE PANCREASHEAD Thomas Kuechler & Jens-Ulrich Siebmann, Department of General and Thoracic Surgery, Reference Centre on Quality of Life in Oncology; Bodo Schniewind, Department of General and Thoracic Surgery, Christian-Albrechts-University, Kiel, Schleswig-Holstein, Germany; Doris Henne-Bruns, Department of Visceral Surgery, University of Ulm, Ulm, Baden-Wu¨rtemberg; Bernd Kremer, Department of General and Thoracic Surgery, Christian-Albrechts-University, Kiel, Schleswig-Holstein, Germany Carcinoma of the pancreas (PCA) remains a challenging problem not only for gastrointestinal surgeons but as well for QoL researchers. Although significant progress has been made in diagnosis, preoperative staging and safety of surgery, longterm survival is rare. Therefore assessing QoL in patients with PCA may serve as a paradigm for the possibilities (determining the treatment leading to best possible QoL in the remaining short lifespan) as well as for the limitations in QoLresearch (e.g. longterm survivors not representative for the whole sample). Since PCA has to be divided into subtypes (head, tail, papilla, diff. histological types, etc.) the first challenge is to define meaningful subgroups within the whole sample. Additionally variations of the Whipple-procedure (e.g. pylorus-preserving vs. classic Whipple or extend of lymphadenectomy) may have different impact on the QoL of these patients. This study, which aims to contribute to surgical decision-making, included only patients with adenocarcinoma of the head of the pancreas (n ¼ 172). Between 09/1996 and 03/2003 we collected (mostly) prospective QoL-data of 134 of these patients (77.9%) preand postoperatively, after 3, 6, 12 and 24 month (334 compl. questionnaires altogether), using the EORTC QLQ-C30 plus a sufficiently reliable German pancreas adhoc-module. At least 102 patients died, most of them (71.6%) within the first year after surgery. Results: (1) After curative Whipple-procedure Function-scales and Global Health recover within 6–12 month despite some severe persisting symptoms like fatigue, diarrhoea, insomnia and pain. (2) After palliative surgery QoL of surviving patients recovers to a preoperative level within 6 month as well, but patients after curative surgery show better physical and role functioning, less dyspnoea, pain and appetite loss. (3) The detailed subgroup analysis that has to follow will be presented at the meeting. Since there are very few systematically gathered QoLdata (standardized instruments, preoperative assessment, several follow-ups) on patients with PCA in Europe, we regard this ongoing study as a major contribution to the field. # 1620/THE TIME TRAJECTORY TO QUALITY OF LIFE OVER SIX MONTHS IN PATIENTS WITH RECTUM CANCER Heejung H. Yoo, Psychiatry & Jincheon J. Kim, Surgery, Asan Medical Center, Ulsan University College of Med, Seoul, Korea We assessed quality of life (QOL) of patients with rectum cancer at baseline (n ¼ 98), 1 month (n ¼ 52), 6 month (n ¼ 29) after receiving colorectomy. We examined the relationships among various psychological and medical variables, stoma and QOL to verify that FACT-C can be a valid assessment tool for the time trajectory to QOL of the patients with rectum cancer. The participants completed the Functional Assessment of Cancer Therapy-Colorectal scale (FACT-C), the shortened forms of the profile of mood states (brief POMS: BPOMS), the eastern cooperative oncology group (ECOG) performance status rating, neuroticism, and functional living index-cancer (FLIC). The recovery trajectory is showed as follows: First, the trend for overall QOL was parabolic, worsening at 1 month after colorectomy, and then recovering nearly back to the pre-colorectomy at 6 months. Second, at 1 month after colorectomy, stoma group was only lower in general well-being than the nonstoma group. At 6 month stoma group was lower in general QOL including CCS (colorectal symptom). This study showed that the quality of life (QOL) of patients with colorectomy does not begin to recover until 1 month after colorectomy, and that full recovery back to the pre-colorectomy level occurs only 6 months after colorectomy. We verified that FACT-C can be a valid assessment tool for the QOL of rectum cancer patients.

# 1834/EXPLAINING FACTORS OF QUALITY OF LIFE IN POSTSURGERY BREAST CANCER PATIENTS Berta Varela, Bettina Caporale, Mercedes Viera & Ana Galain, Medical Psychology; Lucı´a Delgado, Mastology Unit; Laura Schwartzmann, Medical Psychology, Medical College, Montevideo, Uruguay Quality of life is a multidimensional construct determined by many factors among which are coping strategies and emotional state. Most women with breast cancer must face a long and complex treatment that affects their quality of life, so it is important to study how these other factors may also influence it. The purpose of this study is to assess the influence of coping strategies and emotional state over quality of life in women who underwent breast surgery. It shows preliminary results of a longitudinal study aimed at assessing these aspects along the whole treatment. We interviewed 31 patients who had undergone breast surgery on the previous week and administered HADS (Hospital Anxiety and Depression Scale), MAC (Mental Adjustment to Cancer) and SF-36. We used nonparametric correlations and multiple linear regression to analyze the data. The following negative correlations (r: Spearman) between MAC coping strategies and SF-36 domains were found: Hopelessness with Physical function (r: 0.38) and Social function (r: 0.44); Anxious preoccupation with Physical role (r: 0.36), General health (r: 0.35) and Mental health (r: 0: 062); Avoidance with Physical role (r: 0.37) and Mental health (r: 0.36); and Fatalism with Pain (r: 0.37) and Vitality (r: 0.41).The following negative correlations for SF-36 domains (R2) were found: Physical function with Hopelessness (r2: 0.27); Physical role with Avoidance (r2: 0.20); General health, Vitality and Social function with HADS Depression (r2: 027, r2: 022 and r2: 030), and Mental health with HADS Anxiety and Anxious preoccupation (r2: 076). Almost all SF-36 domains are negatively correlated with those coping strategies which are expected to diminish quality of life. Anxiety and Depression disorders as well as Hopelessness and Anxious preoccupation coping strategies were the factors that most negatively influenced quality of life. These results suggest the need to detect them in order to design interventions at an early stage of treatment. # 1109/HEALTH RELATED QUALITY OF LIFE IN YOUNG DRUG USERS Antonia Domingo-Salvany & MCarmen Rodriguez-Llera, Health Services Research Unit, IMIM, Barcelona, Spain; M. Teresa Brugal & Mireia Ambro´s, Age`ncia de Salut Publica de Barcelona, ASPB, Barcelona; Teresa Silva, Luı´s de la Fuente, Centro Nacional de Epidemiologia, Instituto de Salud Carlos III, Madrid, Spain The study aimed to ascertain what variables are related with HRQL among young opiate users, including previous overdose episodes. Young (18–30 years) current opiate users were recruited in open scenes of three Spanish cities (Barcelona, Madrid, Sevilla) as part of the ITINERE project. Standardised computer assisted interviews about socio-demographic, drug use patterns and health problems data were administered, along with the Severity Dependence Scale (SDS, range 0–15) for heroin and for cocaine, and the Nottingham Health Profile (NHP, range 0–100). Information on 786 subjects (72.7% males) was collected (mean age ¼ 25.7 years). Gender differences were observed in age, residence, current job, ever in jail, years of drug use, ever injected, and being HIV positive. The 67 subjects (8.6%) reporting an overdose in the previous 12 months more frequently were from Barcelona, had achieved secondary school, were homeless or with unstable residence, reported injecting ever and last 12 months and had hepatitis C. The mean global NHP score was 35.7 (SD: 23.3). Worse NHP scores in all dimensions were found among the lower educated and those with higher heroine and cocaine SDS scores. Multiple linear regression showed that women, lower educational level, no current job, either stable residence or homeless, reporting important health problems, and higher score (worse) in heroin or cocaine SDS, but not age, were associated with a higher (worse) NHP global score. Overdoses in the previous 12 months were only associated with higher NHP global score in a model including only ever injector subjects (515). HRQL is impaired in young opiate users (NHP global score in general adult population: 11.0 (SD: 13.6)), severity of dependence being an important determinant (an increase in one point in heroin SDS is associated to an increase of 1.5 points in the NHP global score). Having suffered an overdose in the previous 12 months is also associated with lower HRQL among injectors (increase of 6.2 points in the NHP global score). Supported by FIPSE 3035/99, FIS 00/1017 and CIRIT 22001SGR00405.

844 # 1812/QUALITY OF LIFE IN ALCOHOL DEPENDENT SUBJECTS AND CHANGES WITH ABSTINENCE Maria Coccossis, Ioannis Liappas, Eugenia Triantafillou & George N. Christodoulou, Medical School, University of Athens, Athens, Greece

# 1537/QUALITY OF LIFE IN PANIC DISORDER Sara Martinez-Barrondo, Maria P. Garcia-Portilla, Pilar A. Saiz & Julio Bobes, Department of Psychiatry, University of Oviedo, Oviedo, Asturias, Spain

In the realm of medical care, alcohol dependent individuals are considered to experience considerable restrictions in their perceived quality of life. The purpose of this study is to investigate the changes in the levels of perceived quality of life (QOL) before and after cessation of consumption. Sociodemographic and psychiatric data were collected for a group of 37 alcohol dependant individuals who were admitted in an in-patient psychiatric ward because of alcohol dependence. The sample’s mean age was 46.54 years. In terms of gender, six were women, while in terms of education 30% approximately were primary school graduates and 27% had a higher education. Concerning marital status, 55% approximately were married. The subjects were assessed at two different times (at psychiatric intake and after 3 weeks of hospitalization) with the following inventories: (1) the WHO Quality of Life Inventory, WHOQOL-100, (2) the Goldberg’s General Health Questionnaire, GHQ-28 and (3) the Life Satisfaction Inventory, LSI. The above instruments have been validated in Greece presenting satisfactory psychometric properties. The data were analysed with the use of SPSS program. The analysis indicated a statistically significant change in the level of quality of life after cessation of alcohol consumption. Specifically, the patients’ perceived QOL during the time of alcohol consumption (which was assessed with admission for hospitalization) was less favorable than the QOL reported after 3 weeks of hospitalization. The results indicated that the patients’ somatic and psychological well being were significantly improved after 3 weeks of in-patient psychiatric treatment with medication and psychological support, as well as the subsequent cessation of alcohol consumption. The implications of these results for alcohol treatment are discussed.

To evaluate the level of quality of life of panic disorder patients according to the severity of the disorder. Up to date 62 patients of panic disorder (DSM-IV criteria) were included. Evaluations were made by a unique rater (first author) using the Bandelow Panic and Agoraphobia Scale (PAS) and the Short Form-36 (SF-36). The mean age was 35.5 (SD 12.4); 37.1% were male; and 54.9% were married. The mean age of onset was 31.42 and mean time of illness evolution was 4.1 years. PAS total score was 17 (SD 8.95). The PAS subscales scores were: panic attacks 3.64; agoraphobia 4.66; anticipatory anxiety 3.30; disability 3.24; and worries about health 2.1. The more affected areas in SF-36 were vitality (50), and mental health (52) followed by general health (55), emotional role (58.6), physical role (57.2), social functioning (64.9), bodily pain (61.8), and physical functioning (88.8). In the global scores mental health media was 36.56 (SD 14.5) and in physical health media was 48.83 (SD 6.7). No statistical significant differences were found in quality of life and illness severity in relation to gender. All PAS subscales, except worries about health, showed statistical significant correlation with the mental health summary score; in contrast only disability and worries about health subscales showed statistical significant correlation with the physical health summary score. In conclusion: the most affected areas of quality of life were: vitality and mental health. The greater the severity of illness, the worse the quality of life was.

# 1449/POSTTRAUMATIC SYMPTOMS WERE WORST AMONG QUAKE VICTIMS WITH INJURIES FOLLOWING THE CHI–CHI QUAKE IN TAIWAN Hsien-Wen Kuo, Institute of Environmental Health, China Medical College, Taichung, Taiwan Objective: To investigate the prevalence of posttraumatic stress disorder (PTSD) and psychological health status among quake victims 1 year after the quake. Method: Two hundred and seventy-two quake victims from temporary housing units were interviewed. PTSD was assessed using the Davidson Trauma Scale, Chinese version (DTSC). Psychological health status was measured using the Chinese Health Questionnaire (CHQ). Results: The prevalence rate of PTSD was 16.5%. Fifty-seven percent of victims were found to have psychological problems using the CHQ. PTSD and psychological problems were more prevalent among women (22.2 and 64%), compared to men (9.2 and 47.9%). DTS-C and CHQ scores were highest among 25–44-year-olds, and lowest in the over 60-years group. Based on linear and logistic regression models, age and injury were the only two factors that significantly affected DTS-C and CHQ scores. Conclusion: Consistent with the previous studies, it is vitally important to continue providing psychological counselling and social support for quake victims, particularly victims who sustained an injury. Keywords: Posttraumatic stress disorder (PTSD); victims; injuries; quake.

# 1621/THE FUNCTIONING AND WELL-BEING OF UNTREATED SUBJECTS WITH PROBABLE MINOR PSYCHIATRIC MORBIDITY SEEKING PHYSICAL CHECK-UPS IN GENERAL MEDICAL SETTINGS Tsai-Chung Li, Institute of Chinese Medical Science, China Medical College, Taichung; Yih-Dar Lee, Lilly Taiwan, Eli Lilly and Company, Taipei; Cheng-Chieh Lin & Chiu-Shong Liu, Department of Family Medicine, China Medical College Hospital; Jim-Shoung Lai, Institute of Environmental Health, China Medical College, Taichung, Taiwan, Republic of China Background: In order to estimate the burden of probable minor psychiatric morbidity (PMPM), the aim of this study is to examine the degree to which untreated subjects with PMPM reduce health-related quality of life (HRQOL) in a general medical setting for physical checkups. Methods: The study sample consists of consecutive subjects admitted to the hospital of China Medical College for a 1- or 2-day health examination program from July 1996 to December 1997. All study subjects were given self-administered questionnaires to collect sociodemographic data, Chinese Health Questionnaire (CHQ)-30, and the Short-Form (SF-36). CHQ, a reliable and valid instrument, was particularly used as a screening test of PMPM within the community and primary care settings for Chinese. Results: Of 462 subjects, 192 (41.6%) subjects with PMPM were identified. Regression models estimated the effects of PMPM on functioning and well-being (SF-36) by comparing subjects with PMPM to subjects screened as being without PMPM. After controlling for age and gender, the estimated effect of PMPM on all subscales for physical, social and emotional functioning was negative, often as much as a 20–30 point reduction on this 100point scale. The three scales that discriminate best between PMPM and non-PMPM are vitality, general health, and mental health scales while the scale with least discriminant ability is physical functioning scale. Poor HRQOL of untreated subjects with PMPM was observed and the impact of PMPM were comparable with or worse than the effects of mental disorder such as anxiety disorders or medical conditions such as arthritis and diabetes.

845 # 1651/SUBJECTIVE AND OBJECTIVE QUALITY OF LIFE AS MEASURED BY THE LANCASHIRE QUALITY OF LIFE PROFILE: HOW RELATED ARE THEY? Georg Kemmler, Bernhard Holzner, Alex Hofer, Salvatore Giacomuzzi & Wolfgang W. Fleischhacker, Department of Psychiatry, Innsbruck University Clinics, Innsbruck, Austria Background: The importance of both subjective and objective aspects of quality of life (QOL) has often been stressed. However, detailed investigations regarding the relationship between both are rare. Aim of the present study was a comparison of subjective and objective QOL in two different groups of patients treated at our department. Method: Eighty schizophrenia patients (mean age 37.3 ± 11.7) and 61 opioid addicts (age 28.9 ± 6.2) completed the Lancashire QOL Profile (LQOLP) in interview mode; moreover, the psychopathological status of the schizophrenia patients was assessed using the PANSS. The LQOLP is a widely used QOL instrument for mental health patients covering both subjective (satisfaction with life domains) and objective aspects of QOL (functioning, standard of living). Data were analysed by relating subjective QOL subscales with objective QOL items/subscales belonging to the same domain, using correlation analysis and linear regression. Results: In schizophrenia patients, correlations between objective and subjective QOL were generally low for the individual domains (Spearman r: 0.1–0.35), except for the financial domain (r ¼ 0.49). Though in the group of opioid addicts these correlations were somewhat higher, fairly strong correlations of r>0.5 were obtained only for two domains of seven (work, financial situation). Results of linear regression were similar. Education, disease duration and psychopathology had only a minor impact on the results. Conclusions: There may be the following explanations for our findings: (1) the patients’ subjective experiences of their QOL may in fact be fairly unrelated to objective or functional aspects; (2) the ‘objective’ items of the LQOLP may not sufficiently reflect the actual life situation of the patients, especially in the schizophrenia group; (3) satisfaction alone may not fully cover the patients’ subjective QOL. Further aspects will be discussed. # 1058/BRACHYTHERAPY FOR PROSTATE CANCER: THE PATIENT’S PERSPECTIVE Peggy A. Ward-Smith, School of Nursing, University of Missouri – Kansas City, Kansas City, MO, USA Prostate cancer is the most common malignancy, and the second most common cause of cancer among men in the United States. The treatment of localized prostate cancer is highly controversial, with many treatment options. These options include radical prostatectomy, hormonal manipulation, radiation therapy, brachytherapy, or any combination of the above, as well as watchful waiting. Brachytherapy is the permanent implantation of radioisotopes directly into the tumor. This has become a popular treatment, for it is both convenient and cost-effective. Impotence, urinary and bowel dysfunction, as well as disease progression have frequently been utilized as disease-specific measurements of quality of life (QoL) among those with localized prostate cancer. The purpose of this qualitative study was to explore the impact brachytherapy has on QoL, from the patient’s perspective. Utilizing a phenomenological approach, themes emerged from the data that surrounded the physical changes attributed to the treatment, treatment choices, knowledge of the disease, receiving the diagnosis and use of medications to control symptoms post-treatment. The findings from this study can be utilized for patient education and to assist patients in making appropriate treatment choices.

# 1329/PATIENTS’ AND CAREERS’ PERSPECTIVES OF A PALLIATIVE CARE PROGRAM: QUALITATIVE RESEARCH IN RURAL AUSTRALIA Kate A. Senior, Centre for Health Service Development, University of Wollongong, Wollongong, NSW, Australia Exploration of quality of life and satisfaction with care by both patients and their carers. Maintaining quality of life is a key component of palliative care, but it is difficult, both ethically and practically to explore how palliative care influences quality of life of patients and their carers. Standard qualitative methods, such as interviews may be too emotionally and physically burdensome for patients. Also, research focusing on carer’s accounts may not reflect important issues for the patient. The aim was to build up non-threatening relationships with patients and carers and observe how the project worked for them. Access to the patients and carers was a key issue. The palliative care program was home based. The initial emphasis focused on building relationships with staff delivering the service. After a period of discussion and feedback, the staff agreed to me accompanying them on visits to clients and observing their interactions. I became immersed in the daily activities of the palliative care project. I also attended oncology clinics, visited patients who were in the hospital and attended weekly case conferences where each patients’ needs were discussed. Observing people’s everyday interactions with a palliative care program, revealed much about the aspects of the program that influenced their quality of life. It also highlighted important tensions between the quality of life of patients, their carers and the staff who delivered their care. Qualitative research is an important tool in assessing the quality of life issues in end of life care. This research needs to proceed within an atmosphere of trust and at a pace that is directed by the patient and their carer. # 1418/HOPE AND QUALITY OF LIFE IN CYSTIC FIBROSIS (CF) PATIENTS COMPARED TO THE NORWEGIAN NORMAL POPULATION Berit R. Hanestad, Department of Public Health and Primary Health Care, University of Bergen, Bergen; Astrid K. Wahl, Faculty of Nursing, Oslo University College, Oslo; Eva Gjengedal, Department of Public Health and Primary Health Care, University of Bergen, Bergen; Tone Rustoen, Faculty of Nursing, Oslo University College, Oslo, Norway The purpose of this study was to analyze how adult patients with CF experience their quality of life and hope. The sample consisted of 86 adults (48% female, mean age 29 (SD 9.0) years, age range 18–54) with CF recruited from the Norwegian Competence Center for Cystic Fibrosis (84% response rate) and 1893 persons (52% female, mean age 45.2 (SD 16) years, range 18–81) from the general Norwegian population randomly drawn from the National Register by Statistics (49% response rate). The patients and the general population filled in a questionnaire consisting of (1) demographic and clinical data, (2) Herth Hope Index (HHI) (12 items), score range from 12–48 with higher score indicating higher hopefulness, (3) Quality of Life Scale (QOLS-N) (16 items) score range from 16 to 112 with higher score indicating better quality of life. Mean score in CF patients on HHI was 36.1 and mean score on QOLS was 84.9. For the normal population the scores were 37.2 and 83.3 respectively. The CF patients score on the HHI was significantly lower and the score on QOLS-N significantly higher compared to the general population (p < 0.05). In the CF group those with poorest lung function (FEV1% <30%) reported significantly lower hope and quality of life (p < 0.05). These results indicate that CF patients, except from those with poorest lung function, are satisfied with their life and hopeful for the future. The reason for this might be a redefinition of values in life.

846 # 1182/ASTHMA CONTROL AND HEALTH STATUS IN PATIENTS WITH INCREASING ASTHMA SEVERITY Ann-Christin Mork, Klas Svensson & Elisabeth Stahl, Clinical Science, AstraZeneca R&D Lund, Lund, Sweden; Eric van Ganse, Unite´ de Pharmacoe´pide´miologie, Centre Hospitalier Lyon-Sud (CHLS), PierreBenite Cedex, France The aim was to assess the relationship between asthma control and disease severity in asthma patients. Forty-one patients (mean age 34 years) from France, Spain and the UK completed the Asthma Control Questionnaire (ACQ, 0 ¼ good control, 6 ¼ poor control) and the SF-12 (0–100, a higher score indicates a better health status). The Physical Component Score (PCS) and Mental Component Score (MCS) of the SF-12 was calculated. All four GINA severity groups were represented in the sample. In addition, patients rated their perceived asthma severity. Thirty percent of the patients perceived their asthma as intermittent (n ¼ 12), 40% as mild (n ¼ 16), 22% as moderate (n ¼ 9) and 8% as severe (n ¼ 3). According to GINA stages 20% of the patients had intermittent asthma (n ¼ 8), 44% mild (n ¼ 18), 12% moderate (n ¼ 5) and 24% had severe asthma (n ¼ 10). The ACQ mean score was 0.7 for intermittent, 1.2 for mild, 2.3 for moderate and 3.0 for severe asthma (GINA severity, p ¼ 0.03). The PCS mean score was 52 for intermittent, 47 for mild, 49 for moderate and 37 for severe asthma (p ¼ 0.01). The MCS mean scores were 50 for intermittent, 51 for mild, 48 for moderate and 42 for severe asthma (ns). The patients underestimated their disease severity. Level of asthma control and PCS decreased significantly with increasing disease severity. The ACQ may be useful in identifying patients with poor asthma control. # 1409/ADEQUACY OF SYMPTOM CONTROL AND HEALTH-RELATED QUALITY OF LIFE IN THAI ASTHMATIC PATIENTS Chulaporn Limwattananon, Supranit Penkam & Pailin Kaepum, Faculty of Pharmaceutical Sciences, Khon Kaen University, Amphoe Muang, Khon Kaen, Thailand; Cynthia Gross, College of Pharmacy, University of Minnesota, Minneapolis, MN, USA Symptom control and health-related quality of life (HRQL) are important outcomes of treatment for chronic asthma. In Thailand, accessibility to steroid inhalers is limited due to unaffordable drug prices and improper use of inhaler devices. This study determined the adequacy of symptom control and evaluated HRQL in Thai asthmatic patients with continuous treatment. An additional aim was to assess if the Thai version of the SF-36 was sensitive to differences in asthma severity. All persons 15 years of age or older who were diagnosed with asthma and had at least two outpatient visits during a 5-month period of 2002 in a tertiary care hospital were identified (n ¼ 170). A set of self-administered questionnaires intended to measure the severity of asthma symptoms and HRQL was mailed to these patients. Severity in four levels (intermittent, mild, moderate, and severe) was measured by an instrument modified from the method of the National Asthma Education and Prevention Program by the Thai Thoracic Physician Association. HRQL was measured using the SF-36 Thai version. The response rate was 70% (n ¼ 119). Forty-nine percent of those prescribed steroid medications (n ¼ 67) and 40% of those prescribed nonsteroidal medications (n ¼ 52) reported moderate to severe symptoms, reflecting inadequate control of asthma. Using multivariate analyses adjusted for age, gender, duration of being diagnosed with asthma, comorbidities, and use of steroid inhalers, the asthma severity was a statistically significant predictor of SF-36 scores for six domains. Magnitude of the increased SF-36 scores due to a unit decrease in the severity level were 11.95 for RE, 9.33 for BP, 7.04 for GH, 6.98 for PF, 5.90 for SF, and 5.23 for MH. Self-reported asthma severity could be differentiated by HRQL using the Thai SF-36. The high proportion of patients reporting moderate to severe symptoms in both steroid and non-steroid recipients indicates the need for a disease management program to improve HRQL of patients. Further nationwide studies on quality of care as related to HRQL outcomes in chronic asthma are warranted.

# 1824/LIFE QUALITY AND PSYCHOLOGICAL WELL BEING AFTER SHORT-TERM PULMONARY REHABILITATION FOR SEVERELY COPD PATIENTS Joan J. Branin, Health Services Management, University of La Verne, Altadena, CA, USA Chronic obstructive pulmonary diseases (COPD) are a major cause of death and chronic disability. The purpose of this research was to evaluate pulmonary function and endurance; psychological well being, quality of life, and physical functioning; and hospital and other health services utilization before and after a 3-week inpatient pulmonary rehabilitation program for older patients with severe COPD. Few studies have examined these outcomes at admission, discharge, and over multiple time periods post-hospital discharge for patients with severe COPD. Eighty-four older patients with severe COPD (mean percent predicted FEV1 ¼ 34%) were interviewed at admission and discharge and at 2, 4, and 6 months after discharge using the Medical Outcomes Study Questionnaire (SF-36). Pulmonary function and walk scores were obtained from patient records, while hospital and health care utilization was based on self-reports at each of the three follow-ups. Physical functioning significantly improved from admission to discharge from the program and was maintained at 6-month follow-up. Despite the severity of their disease, these patients experienced significant improvement in psychological well being from admission to discharge. All three aspects of quality of life (physical role, emotional role, and social functioning) improved significantly and were maintained at 6-month follow-up. Six months after discharge, these patients were performing their physical, emotional, and social roles at the same levels as the normal population. These patients were hospitalized significantly fewer days and less frequently after the program. Significant reductions in emergency room visits were reported. These findings support the use and benefits of short-term inpatient pulmonary rehabilitation for older adults with severe COPD in improving physical and psychological outcomes while reducing health care utilization particularly relevant in today’s managed care environment. # 1607/DOES PHYSICAL FUNCTIONING IMPROVE THE PREDICTION OF LUNG FUNCTION? Pao-Hsuan Lin, Department of Medical Research, China Medical College Hospital, Taichung; Tsai-Chung Li, Institute of Chinese Medical Science; Jim-Shoung Lai, Institute of Environmental Health; Cheng-Chieh Lin, Department of Family Medicine, China Medical College Hospital, Taichung, Taiwan, Republic of China Introduction: Lung function has been used as outcome measures of several pulmonary disorders either in clinical studies or trials. Although the association between physical functioning and lung function has been identified, previous prediction models of lung function do not evaluate its predictive ability. In addition, the prediction model of lung function for Chinese was built for more than one decade ago with limited sample size. Thus, the objective of the study is to assess the predictive ability of physical functioning on lung function after considering age, gender, smoking, height and weight. Methods: A crosssectional study was conducted among subjects admitted for health checkup in China Medical College Hospital during 1996–1999. Four hundreds and fifty-six volunteers completed a questionnaire, including demographic characteristics, smoking and physical functioning (PF) of Short Form 36. Lung function was measured by forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV1), abstracted from health checkup results. Results: After adjusting for age, sex, height, weight and smoking, the partial correlation coefficients for PF with FVC and FEV1 were significant (both partial r ¼ 0.14). The prediction models of FVC and FEV1 were built according to smoking and gender. Smoking females were excluded from analysis due to small sample size. Our results showed PF was an independent predictor of FVC and FEV1 for smoking and nonsmoking males given that age, height, and squared height already in the model, but not for nonsmoking females. About 3% of variation of FVC was further explained by PF for both smoking and nonsmoking males while 8.0% and 2.2% of variation of FEV1 for both smoking and nonsmoking males, respectively. Our study findings indicated that PF improved the precision of the prediction for lung function.

847 # 1059/ARE EFFECT SIZES REFERRING TO THE OVERALL SIZE OF EFFECTS? AN EXAMPLE COMPARING THE MOS SF-36 WITH THE MOS SF-12 Corina Guethlin & Harald Walach, Institute for Hospital Epidemiology, University Hospital, Freiburg, Germany Purpose of research: One of the standardised quality-of-life measurements to assess the effects of medical treatment is the MOS SF12, which results in two summary scales of physical and mental health status. Even more often used is the MOS SF-36, which results in eight subscales. These can also be summarised into two summary scales of physical and mental health status. The size of effects of a given treatment may be expressed as the effect size (f.e. Cohen’s d). We were interested in the difference of the resulting effect sizes when using the MOS SF-12 analysis compared with the SF-36 in the same population. Subject sample and statement of methods: About 100 patients treated with homeopathy were given the MOS SF-36 prior to the treatment and 6 months thereafter. We analysed the eight subscales and two summary scales of the SF-36 pre- and post-treatment. The second analysis persists of the same data set, but was analysed according to the rules of the SF-12. Effect sizes were calculated as mean(post) ) mean(pre)/SD(pre). More sophisticated ways of calculating the effect size will be applied in the poster. Summary of results: The results are as follows: the effect size of PCS(SF36) was 0.35, the effect size of MCS(SF36) was 0.39. The effect size of PCS(SF12) was )0.27, the effect size of MCS(SF12) was 0.17. The effect sizes of the eight subscales ranged from )0.02 (physical functioning) to 0.5 (general health). The corresponding item from the general health scale which went into the SF-12 showed a negative change from pre- to post-treatment. The results show that calculating effect sizes out of the MOS SF-12 may result in considerable loss of effect size compared to either the summary scales of the MOS SF-36 or the subscales. This is at least partly due to the loss of information when condensing 36 items into 12 items. In this sample of patients the items chosen for the SF-12 did not seem to be representative for the matching subscales. As underestimating the size of effects may especially be a problem in small randomised controlled studies due to loss of power we would recommend strongly to use the SF-36. # 1215/NET BENEFIT FOR HRQL MEASURES – AN EFFICIENT APPROACH? Klas Svensson, Clinical Science, AstraZeneca R&D, Lund, Sweden The clinical significance of a unit change in a HRQL score is often uncertain, although much effort has been put into establishing a Minimal Important Difference, MID, for many measures. This lack of knowledge can cause some interpretation problems, especially when HRQL scores are used as effectiveness variables in cost-effectiveness analysis and similar approaches. Consequently, new ways of expressing change in HRQL measures are explored. One such way is the concept of Net Benefit (NB), defined as ‘the percentage of patients with increase in score of at least MID minus the percentage of patients with a decrease in score of at least MID’, provided an increase is compatible with an improvement in health status [1, 2]. We have investigated the efficiency of the NB approach theoretically under certain conditions compared to the standard analysis of change in HRQL measures. For various distributions of change in HRQL measures, assumed to be Gaussian but varying in location for two treatments, the difference DNB in Net Benefit for the two treatments are calculated and tested, and the results are compared to the results from the traditional hypothesis testing of the difference in mean scores. We find, that under realistic assumptions, the efficiency of the NB approach varies between roughly 70 and 80%. Data from four different clinical studies in asthma shows that in certain situations, e.g. with different shapes of the distributions for the two samples, the NB approach can be more efficient than the standard approach. 1. Price D, et al. The cost effectiveness of chlorofluorocarbon-free beclomethasone dipropionate in the treatment of chronic asthma. Pharmacoeconomics 2002:20(10):653–664. 2. Price D, et al. Budesonide/formoterol with an adjustable maintenance plan costs less and is as effective as fixed dosing. ERJ 2002;20(Suppl 38):397s Abs P2452.

# 1821/MCID: A COMPARISON ACROSS METHODOLOGICAL APPROACHES Dorcas E. Beaton, Mobility Program Clinical Research Unit, St Michael’s Hospital; James G. Wright, Hospital for Sick Children, Toronto, ON, Canada; Jeffrey N. Katz, Brigham and Womens Hospital, Boston, MA, USA; Claire Bombardier, Research, Institute for Work and Health, Toronto, ON, Canada Purpose: The minimal clinically important difference (MCID) is an important attribute of scales and used for interpretation of change, and sample size calculations. The purpose of this study was to compare values of MCID if different approaches were used in the same sample. Methods: The study prospectively enrolled 200 patients undergoing treatment of an upper limb musculoskeletal disorder. Questionnaires were completed pre-treatment and 12 weeks after and included the DASH Outcome Measure and markers of overall improvement. Six published methods for determining MCID could be reproduced: Norman’s 0.5 standard deviation; Wyrwich’s SEM using a and test–retest (1999); Jaeschke (modified for a 11 point change scale rather than 15) (1989); Juniper’s modification of Jaeschke (1994); Stratford 1998 (most accurate change to detect a substantive change). MCID values were compared across the different approaches, and to the bounds of error or minimal detectable change (MDC-95). Results: 86% of the sample completed follow-up. Mean DASH scores at baseline were 44.5/100 (SD 22.7). The values for the MCID varied from 3.9 (SEM, a approach) to 16.6 using the Juniper 1994 approach. Other results: 1/2 standard deviation ¼ 11.4; SEM test–retest ¼ 4.6; Jaeschke ¼ 12.8; Stratford ¼ 15. MDC-95 for DASH ¼ 12.75. The proportion of the sample that would considered to be improved more than an MCID using different MCID methods varied from 33 to 63%. Conclusions: The value of the MCID varies depending on the method used. Until a preferred method is determined, Juniper’s approach yielded the most conservative estimate of MCID (which also surpassed the MDC-95) and is recommended as a lower threshold of change. # 1811/RESPONSIVENESS OF THE MOTIVATION AND ENERGY INVENTORY (MEI) Sheri E. Fehnel & Lori D. McLeod, Patient-Reported Outcomes, RTI Health Solutions; Susan L. Hogue, Global Health Outcomes, GlaxoSmithKline, RTP, NC, USA The Motivation and Energy Inventory (MEI) is a multi-dimensional patient-reported questionnaire developed to measure changes in activation (e.g., mental and physical energy, social motivation) among depressed individuals. While, results of previous psychometric evaluations provided strong evidence to support the internal consistency, test–retest reliability, and construct validity of the MEI. The initial results pertaining to responsiveness, were more equivocal, because the analysis of data from two clinical trials failed to demonstrate the ability of the MEI to show clear treatment advantages for two antidepressants when compared with placebo. As a result, a second evaluation of responsiveness, investigating the ability of the MEI to distinguish those who responded during the course of the trials (regardless of treatment arm), was undertaken in the current study. First, patients in the two trials were classified as responders if their scores on the HAM-D decreased by 50% or more during the treatment period; all others were classified as non-responders. A series of pairwise t-tests was then conducted and effect sizes computed to examine the differences in MEI change scores between responders and non-responders. Finally, the proportions of responders and non-responders in each treatment arm were computed to investigate the possibility of a placebo effect within the trials. The results of these analyses indicated that each MEI subscale was consistently able to distinguish responders from nonresponders (p < 0.001 for all t-tests; effect sizes ranged between 1.19 and 1.78). Additionally, the proportion of responders in the placebo arm was similar to those in both of the active treatment arms, indicating a large placebo effect. The results of this study demonstrate that limiting evaluations of responsiveness to the detection of treatment effects may be insufficient and unjustly cast doubt on the utility of a new questionnaire. Certainly, there is now strong evidence that the MEI is responsive to changes in the severity of depressive symptoms across time.

848 # 1190/FEASIBILITY, RELIABILITY AND VALIDITY OF THE TNOAZL PRE-SCHOOL CHILDREN QUALITY OF LIFE (TAPQOL) QUESTIONNAIRE Eveline M. Bunge & Marie-Louise Essink-Bot, Public Health, Erasmus Medical Center, Rotterdam, ZH; Margot Buskop, Thuiszorg Salland, Ommen, OV; Lisette W. van Suijlekom-Smit & Henriette A. Moll, Hein Raat, Public Health, Erasmus Medical Center, Rotterdam, ZH, The Netherlands Health related quality of life (HRQoL) is an important outcome measure in addition to clinical measures in evaluations of paediatric interventions. In pre-school children the use of a proxy, mostly a parent, is necessary to measure HRQoL. The TAPQOL, an originally Dutch questionnaire, was developed to measure HRQoL in pre-school children. The TAPQOL is a generic instrument that consists of 12 scales that cover physical, social, cognitive and emotional functioning. We investigated the feasibility, internal consistency and test–retest reliability and discriminative validity of the TAPQOL in a large sample of pre-school children, aged 2 months–4 years old. We especially focussed on the children of 2–12 months old, because HRQoL was not previously studied in this age group. A mailed questionnaire was sent to a random sample of 500 parents of Dutch pre-school children and a random subgroup of 116 parents was sent a retest after 2 weeks. The response rate was 82% at the test and 74% at the retest. Nine scales had Cronbach’s as >0.70. Six scales showed ceiling effects (>50% of the respondents had the maximum score). In general, score distributions and Cronbach’s as were comparable for the different age groups. Test–retest showed no significant differences in mean scale scores, but only four scales had Intra Class Correlations >0.70. Five scales showed significant differences between children with 0 conditions vs. children with two or more self-reported chronic conditions; Cohen’s effect sizes were large for skin and lung problems. In conclusion, our study showed that the TAPQOL is a feasible instrument and supports the reliability and discriminative validity of the majority of its scales for 2–12 months olds as well as 1–4 years olds. Further research needs to be done for the TAPQOL in clinical populations. # 1512/FURTHER VALIDATION OF THE WRINKLING AND SAGGING MODULE: PSYCHOMETRIC PROPERTIES AND CLINICAL VALIDITY Linda Abetz, Mapi Values, Bollington, Macclesfield, Cheshire, UK; Marci Clark, Outcomes Research, Pfizer, Ann Arbor, MI, USA Menopausal women are often bothered by changes in their skin due to the aging process, yet few measures are available to assess the signs of skin aging in women. This study assessed the psychometric properties of the Wrinkling Sagging Module (WSM), a 23-item patient reported questionnaire examining facial skin wrinkling and sagging. One hundred and twenty-one women, ages 45–74, were evaluated by two or more dermatologists (once and again 10–21 days later) using the Photonumeric Scoring Manual (PSM). The WSM and Skin Assessment Module (SAM) (54 items assessing aging symptoms, quality of life, ultra-violet exposure, and perceived changes in skin aging symptoms) were administered to the women at baseline. The WSM’s summary scales’ reliability and construct, concurrent (with SAM), and clinical validity (with PSM) were assessed. The Wrinkling Summary and Sagging Summary scales met the tests of item-convergent and item-discriminant validity and internal consistency reliability (Cron-

bach’s a: 0.95; 0.91, respectively). Correlations between WSM Wrinkling Summary and Sagging Summary scales and the SAM facial skin quality item, the SAM facial aging scale, and the SAM concern about aging scale were moderate (>0.40) indicating the WSM summary scales’ concurrent validity. Significant differences (p < 0.0001) in Wrinkling Summary and Sagging Summary scale scores were observed by patient reports of skin quality groups using SAM Item 1 (i.e. excellent, very good, good, fair and poor), indicating the known groups validity of the WSM. The WSM Wrinkling Summary and Sagging Summary scales were also able to discriminate between clinician’s ratings of wrinkling and sagging using the PSM (F ¼ 10.62, p < 0.0001), suggesting that as clinician reports of wrinkling or sagging severity increased, patients reports of wrinkling or sagging severity also increased, indicating clinical validity. Meaningful differences on the WSM summary scales will be discussed. These data confirm the psychometric integrity and clinical validity of the WSM. Further research in a longitudinal study is necessary to adequately assess the WSM’s sensitivity to change over time. # 1142/THE PEDSQLTM: RELIABILITY AND VALIDITY OF THE SHORT-FORM GENERIC CORE SCALES AND ASTHMA MODULE IN CHILDREN AND ADOLESCENTS WITH ASTHMA Kitty S. Chan, Health Program, RAND Corporation, Santa Monica; Rita Mangione-Smith, Department of Pediatrics, UCLA, Los Angeles; Mayde Rosen, Health Program, RAND Corporation, Santa Monica; Tasha M. Burwinkle, Center for Child Health Outcomes, Children’s Hospital and Health Center, San Diego; James W. Varni, Colleges of Architecture and Medicine, Texas A&M University, College Station, TX, USA This study evaluated the reliability and validity of the short-forms of the PedsQL 4.0 Generic Core Scales, a validated pediatric HRQoL instrument, and the PedsQL 3.0 Asthma Module in children and adolescents with asthma. Adolescents with asthma (aged 12–18, n ¼ 145) provided self-report and parents of children with asthma (aged 2–11, n ¼ 366) provided proxy-report to the short forms of the PedsQL 4.0 Generic Core Scales (15 items) and the Asthma Module Symptoms Scale (11 items) and Treatment Problems Scale (11 items) by telephone between March 1, 2000 and March 1, 2001. Internal consistency reliability was assessed using Cronbach’s a. Validity was evaluated using the known groups method. Reliability of the Generic Core Scales Total Score and Asthma Module Scales was good (Generic Core Total: 0.83–0.89; Symptoms: 0.87–0.91; Treatment Problems: 0.72–0.87, across age groups and adolescent self-report and parent proxy-report). Children with mild intermittent asthma had higher HRQoL than those with mild persistent or moderate/severe persistent asthma (Generic Core Total: +8.0, +17.3; Symptoms: +17.0, +32.5; Treatment Problems: +7.7, +12.7, difference from mild persistent and moderate/severe persistent asthma, respectively, all p < 0.05). Children had higher HRQoL scores when receiving appropriate asthma care (Generic Core Total: +7.6; Symptoms: +13.3; Treatment Problems: +5.1, all p < 0.05) and had no co-morbid conditions (Generic Core Total: +7.3; Symptoms: +9.8; Treatment Problems: +4.4, all p < 0.05). (Note: higher scores indicate fewer problems or symptoms). The PedsQL Generic Core-Short Form and Asthma Module Symptoms and Treatment Scales were demonstrated to be reliable and valid and would be useful for measuring the generic and asthmaspecific HRQoL of children and adolescents with asthma.

849 # 1460/VALIDATION OF THE SPANISH McNEW QLMI QUESTIONNAIRE FOR PATIENTS WITH MYOCARDIAL INFARCTION A. Ribera, G. Permanyer-Miralda, P. Cascant, I. Ferreira & C. Brotons, Department of Cardiology, Hosp. Vall d’Hebron, Barcelona, Spain The McNew QLMI assesses HRQL in pts with myocardial infarction (AMI) and it has recently been adapted into Spanish. In the context of a longitudinal study of HRQL of pts admitted to a tertiary hospital for a first AMI, the McNew QLMI questionnaire was administered and its validity studied using short term results. Generic HRQL was measured with the SF36 health profile. Clinical status was described as presence of angina and functional class (FC). Tendency to report negative feelings was measured with a Negative Affectivity (NA) scale. Of the 153 pts asked to participate during 1 year, 112 accepted and information on HRQL was available for 90 at 2 weeks and for 70 at 2 months after discharge. To assess face validity previous hypotheses were formulated about the correlations expected between the scores of the McNew QLMI and the eight dimensions of the SF36. Discriminative properties were evaluated by correlating the scores with other health indicators: the Physical Component Summary (PCS) of the SF36 and the FC as indicators of physical health and the Mental Component Summary (MCS) and the NA as indicators of mental health. To assess the evaluative properties we defined two criteria of change: (1) perceived change: pts declaring to have changed in a transitional question and (2) clinical change: pts improving in functional class. The emotional domain correlated as expected with all but two dimensions of the SF36, and physical and social domains showed the expected correlation with four of eight dimensions. Mental health indicators were highly correlated with the emotional domain, while physical health indicators were only moderately correlated with the physical and social domains. The improved groups showed significant mean change and effect sizes between 0.44 and 0.76. Pts classified as unchanged showed significant but lower mean improvement in the physical and social domains but not in the emotional domain. In conclusion, the Spanish McNew QLMI shows good face validity and construct validity in its emotional domain. Lower validity in physical and social domains may reflect cross-cultural differences. # 1599/KOREAN VALIDATION OF THE LIVER DISEASE QUALITY OF LIFE INSTRUMENT (LDQOL-1.0): A PILOT STUDY Sun Y. Kwak, Internal Medicine; Sehyun Kim, Preventive Medicine; Seong G. Hwang & Gyu S. Lim, Internal Medicine, College of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do, Korea Assessment of quality of life of patients with chronic liver disease will provide information to evaluate the care delivered. The objective of this study was to validate a Korean translation of the Liver Disease Quality of Life instrument (LDQOL-1.0). We have followed rigorous translation steps guided by the LDQOL Working Group. Two translators who are native Korean speakers and fluent in English translated the LDQOL instructions, items, and response choices. After completing a forward translation step, we have conducted a pilot study for cognitive testing. The Korean translated LDQOL instrument was administered to 20 outpatients who were diagnosed as chronic liver disease. The mean age (±SD) was 44.9 (±7.2). Mean (±SD) scale scores were 62.98 (±16.35) for symptoms, 65.02 (±17.83) for effects of liver disease, 67.71 (±24.99) for concentration, 75.31 (±22.71) for memory, 61.00 (±10.21) for quality of social interaction, 56.56 (±27.47) for health distress, 54.38 (±11.97) for sleep, 78.25 (±19.08) for loneliness, 63.33 (±28.92) for hopelessness, 74.27 (±26.84) for stigma, 83.33 (±11.65) for sexual problems (male), and 62.70 (±34.95) for sexual problems (female). All scales had high internal consistency reliabilities (Cronbach’s as ranged 0.63–0.95) except for three scales (symptoms, hopelessness, male sexual problems). Focus group meeting was organized followed by a pilot study with patients who agreed to participate. Re-wording of instructions and response choices reflecting psychosocial and cultural viewpoint of Korean was done after focus group meeting. We are currently planning to convene the second pilot study in order to finalize forward translation prior to the field test.

# 1359/CROSS-VALIDATION AND NORMS ESTABLISHMENT OF THE WHOQOL-BREF TAIWAN VERSION Kaiping G. Yao, Department of Psychology; Jung-Der Wang, Institute of Occupation, Medicine and Industrial Hygiene, College of Public Health; Chen-Miao Chen, Department of Psychology, National Taiwan University, Taipei; Yih-Jian Tsai, Population and Health Research Center, Bureau of Health Promotion, Department of Health, TaiChun, Taiwan, ROC By stratified randomly sampling, 1068 subjects from 17 hospitals throughout Taiwan, the WHOQOL-BREF Taiwan version was developed in 1999. This questionnaire contains 26 cross-cultural items plus two additional national items with good reliability and validity. The two national items represent two new facets, i.e., ‘being respected/accepted’ and ‘eating/food’. The purpose of this study is to cross-validate the WHOQOL-BREF Taiwan version by using the data from the 2001 National Health Interview Survey (NHIS) in Taiwan. The NHIS conducted a stratified multistage systematic sampling which collected 13,064 subjects aged 20 and above and each completed the questionnaire of WHOQOL-BREF Taiwan version. Psychometric properties (factor structures and various types of reliability and validity) were assessed on this questionnaire by using the new sample. The crossvalidation results show that the psychometric properties were comparable to the previous study. In addition, norms for general population, and for subjects with diverse diseases (such as cardiac disease, pulmonary disease, hypertension, diabetic mellitus, hyperlipidemia, gout, stroke, peptic ulcer, sinusitis, liver disease, renal disease, prostate disease, uterine and ovary disease) were established. Norms by gender and age were also calculated. Extended norm comparisons between the general population and the diverse disease groups and across gender and age were discussed. # 1277/THE PACEMAKER PATIENTS QUALITY OF LIFE STUDY (THE PAPQOL STUDY) Stefan Ho¨fer, Psychology, Royal College of Surgeons in Ireland, Dublin, Ireland; Werner Benzer, Interventional Cardiology, Academic Hospital Feldkirch, Feldkirch; Michael Anelli-Monti, Cardiac Surgery, University Hospital Graz, Graz; Thomas Berger & Florian Hintringer, Cardiology, University Hospital Innsbruck, Innsbruck, Austria Health-related quality of life (HRQL) is an increasingly important and acceptable outcome to clinicians and patients when considering the effectiveness of clinical interventions. In cardiac pacing clinical practice permits the use of different pacemaker systems. Little is known about the HRQL of pacemaker patients, due to limited availability of disease specific instruments and the short history of pacemaker treatment. This study compared three different HRQL instruments to assess pacemaker patients quality of life. The well-known and psychometrically established SF-36 was chosen as a generic instrument. The Karolinska Quality of Life Instrument (Karolinska) was chosen as a disease specific HRQL instrument showing evidence of validity for pacemaker patients. Finally the MacNew Heart Disease Questionnaire (MacNew) was included as a disease specific HRQL instrument. The MacNew was originally developed in English for patients with myocardial infarction (MI), meanwhile translated into more than 15 languages, and proofed to be a reliable and valid instrument in German speaking MI-patients and angina patients. In the PAPQOL study patients (n ¼ 70) with indication for pacemaker implantation received a self-administered battery including the SF-36, MacNew and Karolinska before implantation, 1, 3 and 6 months after initial treatment. Disease severity was assessed with New York Heart Association Scale (NYHA) and the 6-min walk test was performed. In addition relevant clinical data (e.g. pacing mode, left ventricular function, ejection fraction) were collected. Test–retest reliability, intra-class correlation, Cronbach’s a, validity coefficients, sensitivity analyses (effect sizes) and confirmatory factor analysis were carried out. Results will be presented at the conference.

850 # 1054/GENDER AND QOL AMONG AGING COUPLES Gail Low, Nursing, University of Victoria, Victoria, British Columbia, Canada The purpose of this cross-sectional study was to elucidate the role of gender in shaping the QOL perceptions of older adult couples. Sixtyseven community-dwelling older adults coping with mild, moderate, and severe chronic lung disease (COPD) and their non-COPD spouses took part. All participants completed the generic Sickness Impact Profile (SIP) and the COPD-specific St George’s Respiratory Questionnaire (SGRQ). Non-COPD spouses provided proxy reports of their COPD spouse’s QOL. For both groups, significant differences by gender were found for specific QOL domains at specific stages of COPD severity. On the SIP, females with mild COPD rated their physical and psychosocial QOL significantly lower (Tau C ¼ )0.646, p < 0.001; Tau C ¼ )0.646, p < 0.001). On the SGRQ, males with moderate COPD perceived the impairment from COPD symptoms to be significantly higher (Tau C ¼ 0.609, p < 0.01). Stronger patterns were noted in non-COPD spouse proxy reports using the SIP. Male spouses perceived their COPD spouse as having greater physical (Tau C ¼ 0.938, p < 0.001) and psychosocial (Tau C ¼ 0.813, p < 0.001) impairment. On the SGRQ, female spouses felt the level of impairment from COPD symptoms (Tau C ¼ )0.642, p < 0.001) was far greater. Most QOL research focuses on males with severe lung disease and the perceptions of non-COPD spouses receive little empirical attention. From these findings, future research that can be of benefit to both practitioners and scholars includes an exploration of gender differences at varying stages of COPD severity. Gender differences in certain domains of QOL also suggest the need for further empirical attention to specific and significant types of felt impairment. Community-dwelling older adults with COPD are most often cared for by a spouse. Spouse reports offer a chance to compare within-couple differences and identify which aspects of living are most impacted from a caregiver’s perspective. Further research on non-COPD groups can help to determine whether the findings in this study are unique to chronic lung disease alone. Gail Low PhD Student, UVIC School of Nursing Victoria, British Columbia, Canada. # 1578/CORRELATES OF HEALTH-RELATED QUALITY OF LIFE IN THE COMMUNITY-BASED ELDERLY Jennifer D. Walker & Colleen J. Maxwell, Community Health Sciences, University of Calgary, Calgary, AB; Walter P. Wodchis & Adriana Venturini, Toronto Rehabilitation Institute, Toronto, ON; David F. Feeny, Institute of Health Economics, Edmonton; David B. Hogan, Jenny X. Zhang, University of Calgary, Calgary, AB, Canada A new measure of health-related quality of life (HRQL), the Minimum Data Set-Health Status Index (MDS-HSI), has recently been developed for continuing care settings based on items analogous to those in the Health Utilities Index-Mark 2 (HUI2). The MDS-HSI measures observed HRQL while the HUI2 provides a self-reported score. We examined the correlates of both the MDS-HSI and HUI2 to explore the utility of such measures in a home care sample. This cross-sectional study included 308 older (65+) community care clients from Calgary, Canada and Detroit, USA. All subjects received MDS-HC and HUI assessments to calculate the HRQL measures. Correlates examined in multivariate linear regression included age, sex, informal care, caregiver status, comorbidity and polypharmacy. Multivariate analyses

showed that high levels of caregiver stress were associated with lower HRQL for both the HUI2 and MDS-HSI measures (0.08 units lower and 0.09 units lower, respectively). Both measures showed a significant decrease in HRQL with increasing numbers of chronic conditions. No associations were found between HRQL and sex or type of informal support for either measure. The results for HUI2 and MDS-HSI differed in their association with polypharmacy and age. The MDS-HSI was significantly higher (0.06 units) for subjects using fewer than nine drugs compared with those using nine or more. With increasing age, there was a significant decrease in HRQL as measured by the MDSHSI. Baseline results suggest that the MDS-HSI and the HUI2 measures are able to distinguish older subjects who may be at high risk for poor outcomes. Caregiver stress was identified as the strongest correlate of HRQL on both measures. However, the MDS-HSI appears to be more sensitive to polypharmacy and age. These preliminary findings are limited by the cross-sectional nature of our data. Ongoing longitudinal assessments will permit further analyses of the determinants and consequences of HRQL decline and clarification of the relative strengths and limitations of the HRQL measures. # 1563/QUALITY OF LIFE ISSUES AMONG OLDER ADULTS IN ISRAEL Tal Narkiss-Guez, Behavioral Sciences, Ben Gurion University of The Negev; Marianne Amir, Behavioral Sciences and Social Work; Yoram Ben Ya’acov, Behavioral Sciences, Ben Gurion University of The Negev, Beer Sheva, Israel The present report is part of the WHOQOL-OLD project. The main objective of this current study is to develop a module for the assessment of quality of life (QoL) in older adults. This development will occur through the adaptation of the recently published measure of QoL for younger adults which has been produced by the World Health Organization and which is known as the WHOQOL. Another objective of the program is to use the WHOQOL-OLD in an innovative cross-cultural study of healthy vs. unhealthy ageing. More specifically, groups of older adults from a wide variety of cultures will be assessed on a range of psychological, socio-economic, and health-related measures in order to assess factors that contribute to healthy vs. unhealthy ageing. In the first stage of the study, focus groups took place in 23 countries. Following are the results from data collected in Israel: We ran focus groups with older adults with relevant professionals and, where appropriate, with carers of older adults in order to identify the areas to be included in the older adults WHOQOL module. Four focus group discussions took place in residential care housing, 2-day care centres and one nursing home. Professionals and carers groups took place in the nursing home. There were a total of 34 participants. Within the older adults groups two were with 60–80 year olds and two with 80+ year olds (M ¼ 79.3). There were both healthy and unhealthy people. The results showed some important domains and factors related to QoL for old people in Israel: the most important domain was health, especially physical functioning and exercise; attitudes toward the older adults from society and the professional staff (respect); leisure time activities; relationship with children and grandchildren; general security (in Israel, terror threat); personal safety and loneliness. A central issue was the celebration of Shabbat and other Jewish holidays where the family gathers together. Additional issues were Holocaust memories and the preservation of the value of Zionism.

851 # 1724/DO CZECH ELDERLY WOMEN HAVE LOWER QUALITY OF LIFE THAN MEN? RESULTS OF A PILOT STUDY Eva Dragomirecka & Pavla Selepova, Department of Social Psychiatry, Prague Psychiatric Center, Praha 8 – Bohnice, Czech Republic

# 1403/SPIRITUALITY, MORALE, AND QUALITY OF LIFE IN THE AGED Jose L. Pais-Ribeiro & Tania G. Pombeiro, Psicologia, Universidade do Porto, Porto, Portugal

The term aging often connotes decline of abilities, interests, satisfaction and generally quality of life. Especially elderly women have been often mentioned as a risky group due to their loneliness, reduced financial resources and worse health condition. The aim of this presentation was to investigate the differences in priorities, needs and other quality of life aspects of elderly women and men and to understand the changes of these characteristics over the time. As a part of the EU project the measurement of quality of life in older adults and its relationship to healthy ageing, 350 older adults aged 58–95 living in Prague completed the set of questionnaires comprised two quality of life assessments (WHOQOL-100 and WHOQOL-OLD), Importance questions, Co-morbid conditions list and Sociodemographics form. The sample consisted of equal number of men and women, with mean age 74. The majority of male participants were married or lived with a partner (71%) while only one-third of women (32%) lived with a partner and half of them (49%) were widowed. Most prevalent reported medical conditions were mobility, cardiovascular and sensory problems; 64% of the men and 60% of the women reported themselves as healthy. All differences indicated higher satisfaction with life domain for men than for women. Men reported significantly higher satisfaction in a number of items such as personal relationship, family support, financial situation, social care services, health conditions and life as a whole while women scored significantly higher in items of loneliness, lack of money and worries about future. The gender differences increased with age of respondents; men aged 80 and over were significantly more satisfied with their life and reported higher quality of life than women of the same age group. The results that supported the initial hypotheses could contribute to the broad public discussion how to improve quality of life in retirement community residents. Supported by QLRT-2000-00320.

Spirituality becomes an important dimension for personal life in general and QOL, especially for aged people. The aim of the present study is to inspect the relationships between Morale, Spirituality and QOL in older people. Participants include 57 individuals, aged between 64 and 94 years (M ¼ 75.5) 31.6% males, 52.6% married, 57.9% living in their own home. Instruments include the Portuguese version of the Lawton Philadelphia Geriatric Center Morale Scale (PGC Morale Scale) (14 items in three dimensions, Agitation, Attitude Toward Own Aging, and Loneliness; 17 items in the original version). To Lawton, morale consists of positive self-regard, a struggle for mastery and acceptance of reality. The domain of spirituality of the QOL-WHO questionnaire includes four questions referring to religion, spirituality and any other beliefs one may hold. And the EuroQoL-5D which is a six-item questionnaire encompassing five aspects of current health including mobility, self-care, and limitations to main activities, pain, and mood, followed by a single overall measure of health. Results show a statistically significant correlation between spiritual beliefs and the PGC Morale Scale: r ¼ 0.30 for Lonely dimension, r ¼ 0.41 for Attitude Toward Own Aging, r ¼ 0.38 for the total scale. Results also show a statistically significant correlation between spiritual beliefs and EuroQoL-5D results: r ¼ )0.30 for Mental Health (better mental health more appropriate beliefs) and r ¼ 0.26 for actual health perception. We also found a statistically significant correlation between the PGC Morale Scale and the Mental Health dimension of EuroQoL-5D (between r ¼ )0.35 and r ¼ )0.42, with an r ¼ )0.52 for the total PGC Morale Scale); also between Attitude Toward Own Aging from the PGC Morale Scale and Physical Domain (r ¼ )0.27), Mental Health (r ¼ )0.35), and Health Today, (r ¼ 0.44) from EuroQoL-5D scale, as well as Spirituality (r ¼ 0.41). An overall inspection of the results suggests an important relationship between Mental Health, Attitude Toward Own Aging and Spirituality which can play an important role in the well-being of the elderly.

# 1461/INFLUENCE OF SOME SOCIO-DEMOGRAPHIC CHARACTERISTICS ON THE HEALTH RELATED QUALITY OF LIFE AMONG ELDERLY IN VOJVODINA Erzebet Ac Nikolic, Olja Niciforovic Surkovic & Svetlana Kvrgic, Department of Social Medicine, Institute of Public Health, Novi Sad, Serbia, Serbia and Montenegro, Yugoslavia Aim of this study was to determine influence of some socio-demographic characteristics on the health related quality of life (HRQOL) among elderly in Province Vojvodina. Measurement of health related quality of life was performed using original instrument made by modifying of McMaster Health Index Questionnaire. Technique of defining of unique numeric value of health related quality of life as well as numeric values for scales entering instrument is especially determined for this modified instrument. The final result was QOL-OLD Index, on which value influence values of the six scales: physical function, selfcare, interpersonal relations, social usefulness, social functioning and self-assessment scale. Each scale as well as QOL-OLD Index have numeric value, from 0 to 100 (higher value mens better quality of life) and is comparable with other instruments for measurement of HRQOL among elderly. Cross-sectional study was performed on the random sample of 373 elderly people in the population aged 60 and above throughout poll. Results of the study showed that elderly in elderly homes have QOL-OLD mean value of 67.9 which is significantly lower than value for elderly who live in their own homes (p ¼ 0.04). Significant influence on the QOL-OLD Index have gender, in favour of males (p ¼ 0.009). Population of elderly was stratified in two groups: first from 60 to 74 years old and second 75 years and above. Younger elderly have significantly better quality of life (p ¼ 0.02). Higher level of education have positive influence on HRQOL (p < 0.01). Elderly people living together with partners have also better HRQOL (p ¼ 0.04). It may be concluded that instrument originally created is suitable for measuring HRQOL among elderly population and its’ practical value is getting unique numeric value which is comparable with other instruments for measuring HRQOL. Selected socio-demographic characteristics have significant influence on HRQOL.

# 1771/QUALITY OF LIFE AND SOCIO-DEMOGRAPHIC CHARACTERISTICS OF OLDER ADULTS (THE WHOLQOL-OLD PROJECT) Ramona Lucas, Catalan Institute on Ageing; Rosa Monteserı´n, EAP Sardenya, CAP; Salvi Junca, Departamento de Sanidad, Generalitat de Catalunya, Barcelona, Spain We present preliminary results on quality of life (QoL) on a sample of older people (as a part of the WHOQOL-OLD Project) using the WHOQOL-100 in relation to sociodemographic variables. The WHOQOL-OLD is a project funded by the European Commission involving 23 countries worldwide. In Barcelona, Spain 302 adults 60 and older participated on the pilot stage. All participants signed the Informed Consent and completed the questionnaires: the WHOQOL-100, a specific module for Older Adults, Importance questions related to facets, a list of Comorbid Conditions and Sociodemographic information including age, gender, marital status, education, living arrangement, work and voluntary work 56.6% of the sample were women; mean age was 74.5; 30% were between 60–69, 42% between 70–79 and 28% were 80 and older. 62.6% were married, 60% were living at home unsupported, 69.5% were retired and 27% reported doing voluntary work. Gender differences were found on three domains (Physical, S. Relationships and Spirituality) and five facets (Pain, Sleep, Body Image, N. Feelings and Spirituality) and Age groups differences in one domain (Independence) and nine facets (Energy, P. Feelings, Mobility, ADL, Medication, Working Capacity, Health and Social Care, Environment and Transport). Educational level difference among categories was found on the Psychological domain. Married people scored higher on Physical and Independence domains. People with higher level of education scored higher, and people doing voluntary work scored higher in all domains of the WHOQOL-100 compared to those who did not voluntary work. Women, people 60–79, married people, people with higher level of education, those living unsupported, and those doing voluntary work had higher QoL scores. Knowledge of these specific aspects will contribute to improve the QoL of the Elderly.

852 # 1246/GENERAL US POPULATION NORMS FOR THE FUNCTIONAL ASSESSMENT OF CANCER THERAPY-GENERAL (FACTG) David Cella, Center on Outcomes, Research and Education, Evanston Northwestern Healthcare and Northwestern University; Kathleen J. Yost & Jin-shei Lai, Center on Outcomes, Research and Education, Evanston Northwestern Healthcare, Evanston, IL; Martin J. Zagari, Ortho Biotech, Raritan, NJ, USA The Functional Assessment of Cancer Therapy-General (FACT-G) is a multidimensional measure of quality of life (QoL), which has been widely used to assess the QoL in patients with chronic illnesses. Scores can be computed for the total FACT-G and for four subscales of well-being: physical (PWB, 7 items), social/family (SFWB, 7 items), emotional (EWB, 6 items), and functional (FWB, 7 items). The interpretation of total FACT-G and subscale scores can be greatly enhanced by comparing these scores to normative data; thus, the purpose of this study was to establish general US population norms for the FACT-G. Twenty-one of the 27 FACT-G items were administered to a sample of 1400 individuals randomly selected from more than 100,000 individuals enrolled in an internet-based survey panel, which was representative of the general US adult population. Six of the 27 FACT-G items address concerns specific to patients with a chronic illness (e.g., ‘I have accepted my illness’) and were not included. Scores were computed such that the possible range of scores for the total FACT-G and subscales in the 21-item instrument were comparable to those in the 27-item instrument. Of the 1400 panel members selected, 1078 (77%) responded and FACT-G scores were available for 1075. Ages ranged from 18 to 91 years with a mean (SD) of 45.9 (16.6), 50.6% were female, 75.9% were white, and 87.8% had at least a high school education. General US population norms (mean and standard deviation) for the full sample of 1075 were: total FACT-G 80.1 (18.1); PWB 22.7 (5.4); SFWB 19.1 (6.8); EWB 19.9 (4.8); FWB 18.5 (6.8). Norms were also established for subgroups by gender and 10 year age-group. Figures showing all possible raw scores relative to general US population T-scores were produced for the total FACT-G and subscales. The figures allow clinicians and investigators to immediately discern the QoL of their patients relative to that of the general US population. These FACT-G norms can facilitate meaningful interpretation of QoL in patient populations. # 1070/FACTORS PREDICTING PHYSICAL AND MENTAL COMPONENT SUMMARIES OF THE SHORT FORM HEALTH SURVEY (SF-36) Ali Montazeri, Azita Goshtasebi & Mariam Vahdaninia, Iranian Institute for Health Sciences Research, Tehran, Iran A population-based study was conducted to measure health and collect information on normative data from the general population in Tehran, Iran. Using stratified multi-stage sampling, the Iranian version of the SF-36 was administered to a random sample of healthy individuals aged 15 years and over living in Tehran. This paper presents data on the physical (PCS) and mental (MCP) component summaries of the SF-36. In all 4163 individuals were interviewed. The mean age of the respondents was 35.1 (SD ¼ 16.0) years, 52% were female, mostly married (58%) and the mean years of their formal education was 10.0 (SD ¼ 4.5). The mean score (SD) for the PCS and the MCS

of the SF-36 was 78.2 (23.6) and 69.5 (23.2) respectively. Logistic regression analysis was performed to determine which characteristics of the respondents predicted the PCS and the MCS. For the analysis relative to individuals’ mean scores on the PCS and the MCS, they were divided into two groups: those with a low level of physical and mental health and those with a higher level of the PCS and the MCS. Then these were used as outcome variables and the other variables including age, sex, martial status, years of education, and employment status were entered into the model as independent variables. The results showed that physical and mental component summaries of the SF-36 were best predicted by age, gender and employment status (p < 0.001), while education and marital status were not significant predictors. The findings suggest that socio-demographic characteristics of the individuals have significant effect on people’s physical and mental health. The findings lend support to the fact that health related quality of life should be measured in the context of each individual’s culture, and socio-economic status. # 1735/QOL AND ECONOMIC EVALUATIONS OF TREATMENT ALTERNATIVES FOR PATIENTS WITH ADVANCED PROSTATE CANCER Nobuo Koinuma & Shunichi Namiki, Health Administration and Policy, Tohoku University Graduate School of Medicine, Sendai, Miyagi, Japan Purpose: The majority of patients with advanced prostate cancer will respond to testicular androgen blockades such as castration or LH-RH analogue with or without anti-androgens. Once the disease becomes hormone-refractory, however, it is difficult to cure the patient because no effective therapy has been established. We have two treatment alternatives, low-dose cisplatin (CDDP) and 5-FU or UFT and estramustine phophate (EMP), in metastatic prostate cancer offering almost the same expectations of survival. This study documents the QOL and cost-effectiveness of these alternatives. Materials and methods: 14 consecutive patients with clinically hormone-refractory prostate cancer at the University Hospital between January 1995 and December 2002 were treated. The patients comprised three treatment groups; CDDP, EMP, and the other hormonal treatment (OT). We measured EORTC prostate Cancer QOL and EuroQol-5D. Considering cost-effective was compared in three groups retrospectively. Results: Surveys were returned from 85.7%. The median patient age was 71.7 years old and the median follow-up period was 32.5 months. Clinical stage (n): stage C (4), stage D1 (2), stage D2 (6); Histological grade (n): moderately diff. (8), well diff. (4); the median treatment time in the CDDP group was longer than the EMP group and the OT group (22 vs. 8.5 and 14.7 months respectively). The OT group scored higher than the other two groups in every QOL domain score except for sexual function. There was no difference between the CDDP group and the EMP group. The cost increases in the CDDP group appeared to be associated with greater drug charge. Conclusion: (1) Quality of life showed almost he same scores in three groups and assessment and treatment of pain is essential for a good quality of life in men with prostate cancer. (2) The first-year screening and treatment costs were highest and dependent on injection therapy and drug charge including hormonal therapy. The costs of year therapy decreased as time went by and after four years treatment costs were almost the same. We are now collecting prospective multi-institutional outcome data.

853 # 1690/QUALITY OF LIFE: STUDY OF AFFECTIVE MEANINGS FOR A GENERAL POPULATION Suely S. Viski Zanei, Department of Nursing, Sa˜o Paulo Federal University; Ieda M. Reis, COSEAS, Sa˜o Paulo University, SP; Maria Ame´lia D. Moura, Unidade Ba´sica de Sau´de Salvador de Leoni, Secretaria Mun de Sau´de de Itapecerica da Serra, Embu´guac¸u; Miako Kimura, Vera Lu´cia C. Santos, School of Nursing, Sa˜o Paulo University, SP, Sa˜o Paulo, Brasil This study aimed at identifying the affective meanings underlying the expression quality of life (QOL) for a population sample in Sa˜o Paulo and analyzing the possible relations with its socio-demographic characteristics. The sample was figured out based on governmental data (n ¼ 620) as well as on data collected in public places through interviews with people over 15 years of age. The tool applied 23 pairs of adjectives randomly ordered making up the Bipolar Scales from Osgood’s Semantic Differential (DS). Seven-interval scales contain opposing adjectives in each extremity, through which individuals evaluate the word or expression, indicating the positive or negative direction of each judgement (from +3 to )3). Regression patterns with socio-demographic variables were devised and the averages for each adjective and factor, as well as Internal Contradiction Index (ICI) were analysed. In Brazil the following factors were validated: value, power, stability, simplicity, usefulness, dimension and agility. ICI indicates which concept is unknown, may carry two or more meanings or be controversial. Researched subjects applied a positive connotation to the QOL expression, evidenced through positive averages in all pairs of adjectives. Highly positive adjectives were: great, useful, necessary, clean, good, human, happy, crucial, hopeful and pretty. Pairs of adjectives presenting high ICI were simple/complicated, long lasting/ephemeral, easy/difficult, thick/thin, quick/slow and light/heavy. Schooling, income and age were variables influencing connotation applied to the QOL concept. Individuals with lower schooling levels considered less happy than those with higher ones. Individuals with lower income rates considered QOL worse, less happy, less great and less comfortable than those with higher income ones. The oldest ones considered it more perfect and long lasting than the youngest ones. # 1791/PREDICTORS OF CHANGE IN CAREGIVER HRQL AFTER STROKE A. Simon Pickard & Mehul Dalal, College of Pharmacy, University of Illinois, Chicago, IL, USA Studies examining the relationship between stroke patient/caregiver characteristics and caregiver burden or quality of life have been primarily cross-sectional in nature. The purpose of this study was to identify significant factors associated with changes in overall caregiver health-related quality of life (HRQL) in the 6 month interval after stroke. Stroke patients and their caregivers were recruited prior to patient discharge from two large teaching hospitals to self-assess the EQ-5D and Health Utilities Index (HUI). Stepwise multiple regression (F to enter p < 0.10, F to remove p < 0.20) was used to explore factors explaining change in overall HRQL. Models were developed using the EQ-5D VAS, EQ-5D Index-based score (EQ-Index), and HUI overall utility score (HUI3) as dependent variables. Independent variables included: patient and caregiver age and gender, caregiver relationship to patient, co-habitation prior to stroke, discharge location, and patient HRQL. Ninety-one patients and caregivers completed the HRQL measures at 6 months. Regression analysis revealed that after adjusting for baseline caregiver HRQL, lower caregiver change scores were associated with more elderly, female caregivers, and caregiver change scores were systematically higher for patients with better HRQL scores at 6 months. These findings were significant for the EQVAS, EQ-Index, and HUI3. In addition, for the HUI3, discharge to patient’s own home contributed to significantly lower caregiver change scores (p < 0.01), while for the EQ-Index, significantly higher change scores were associated with discharge to a family member’s home where the patient did not previously reside (p < 0.05). Few previous studies have examined predictors of change in caregiver HRQL. Similar explanatory factors were identified for each of the HRQL measures. Although exploratory and findings require replication, this study may help to inform research and policy in the design of support programs for caregiver subgroups whose HRQL is impacted to a greater extent by stroke.

# 1845/USING MOBILE PATIENT-REPORTED OUTCOMES TO ENHANCE PATIENT CARE Chih-Hung Chang, Feinberg School of Medicine, Northwestern University, Chicago; DerShung Yang, Caracal, Inc., Buffalo Grove, IL, USA Purpose: The purpose of the study was to develop a Mobile PatientReported Outcomes (mPROs) system to administer and collect PROs and clinical information utilizing available wireless mobile networks and cell phone technology to inform health care providers priority areas needing timely attention and to facilitate patient–physician informed treatment decision making. Methods: The potential information flow between the physician and patient was established and then implemented. Each patient used a cell phone to access a secure network and to report pre-designed PROs questions displayed on the touchscreen panel. All data transmitted between the phone and the central server were encrypted to ensure security. Patients could call into the system virtually from anywhere and anytime to self-administer the questionnaires prior to their scheduled clinic visits. The physician could also utilize the same technology to retrieve information being entered and summarized. Results: This technical approach was to implement a solution to improve interaction between patients and physicians to focus on priority issues needing timely attention. A refined program was made to accommodate comments and suggestions to improve system stability and increase patients’ willingness to participate. Physicians could then utilize the concise information summarized from the data to better understand patient’s needs and to make better treatment plans. Discussions: This study envisioned eliminating labor intensive interviewer administration of PROs questionnaires and establishing mechanisms for effectively capturing data using mobile/wireless technology and then presenting the data to both treating physicians and patients in a manner that was sensitive and responsive to clinical needs. To date, however, the use of mobile technology in a busy clinic has not been fully tested. This study provides an efficient way of collecting PROs information and has great potential to facilitate clinicianpatient shared decision-making and to enhance patient care. # 1452/AN AUSTRALIAN ON-LINE KNOWLEDGE MANAGEMENT FRAMEWORK TO FACILITATE OPTIMAL CHOICE AND USE OF HRQOL MEASURES FOR BOTH CLINICAL RESEARCH AND PATIENT MANAGEMENT Bruce Shadbolt, Canberra Clinical School, The Canberra Hospital, Garran, ACT; Jan Sansoni, Centre for Health Service Development, University of Wollongong, Wollongong; Madeleine King, Centre for Health Economics, Research and Evaluation, University of Technology, Sydney, NSW; Paul S. Craft, Medical Oncology Unit, The Canberra Hospital, Garran, ACT, Australia The large number and variety of health-related quality of life (HRQOL) instruments reflects the breadth of conceptual and operational definitions of HRQOL. It also makes it hard to find the best instrument for a particular context. Specialist databases, such and Mapi’s QOLID, aim to identify and describe HRQOL instruments, help choose appropriate questionnaires and facilitate access to them. This paper describes a new technology, called Protocol Hypothesis Testing (PHT), which has the potential to facilitate both the use and usefulness of such databases, and to make the best use of the HRQOL data that is subsequently generated. PHT is an on-line research framework that uniquely blends scientific methodology (including the conduct of randomized controlled trials) with decision support (including evidence-based medicine and quality improvement). The system has specific interfaces for ‘authors’ (such as Mapi’s QOLID and Cochrane review databases), clinicians and consumers. PHT can be used in individual patient management by allowing real-time evidence-based decision support and rapid feedback of HRQOL assessments. The system can also be used to generate group-based evidence. Driven by clinical relevance and cross-institution collaboration, many study types can be conducted in the context of incremental improvement on established decision rules, treatment protocols and research evidence. PHT is a technology that has developed in parallel with other technologies (such as computerised adaptive testing and web-based surveys) and knowledge repositories (such as QOLID and Cochrane). In the near future, it may be possible to combine these technologies and knowledge repositories to support better decision making in clinical practice and more relevant clinical research.

854 # 1666/GP PRIORITIES FOR PATIENT-REPORTED OUTCOMES – A DELPHI SURVEY Munira C. Blacking & Elaine McColl, Centre for Health Services Research, School of Population and Health Sciences, Newcastle, UK We have used the Delphi survey technique to elicit consensus of general practitioners (GPs) regarding the use of patient-reported outcomes (PROs) in their routine clinical practice. The use of standardized PRO scales is widely encouraged in the field of evaluative research, but their use in routine clinical practice is less well-documented. A prior Delphi survey amongst GPs identified asthma, hypertension and diabetes as priority conditions for the development of measures of patient outcome. The purpose of the current survey is to measure consensus on the main barriers and facilitating factors to PRO use; priorities for clinical conditions for PRO use and suggestions of what PRO measures would be useful in achieving this. We will estimate GP willingness to use PROs and will provide suggestions for change, with the purpose of facilitating the introduction of PROs into daily work. Our sample of 400 eligible GP principals in Northern England were selected from 63 primary care trusts. The round 1 Delphi questionnaire consisted of free-text boxes for the panel to enter their priorities for clinical conditions, PROs, barriers and facilitating factors. Round 2 is a multiple-choice style questionnaire; results will be available in July 2003. In round 2, respondents will be invited to rank the topics identified in round 1 to gain consensus. Median values and inter-quartile ranges will represent the degree of consensus. In this way, we hope to show changes in perceptions towards PROs over the last 10 years, and identify ways in which groups that develop PROs could make them more ‘physician-friendly’. Content analysis of round 1 responses has identified a range of 47 conditions, including depression and angina, for measuring PROs in routine practice. Most doctors identified PROs but some chose measures of process, or biomedical markers. The main barriers included time and money; IT systems are viewed as facilitators. # 1515/THE MUSIQOL PROJECT: AN INTERNATIONAL PATIENTCENTERED QUALITY OF LIFE SCALE FOR MULTIPLE SCLEROSIS S. Robitail, K. Barrau & A. Beresniak, Perceived Health Research Unit, Public Health Department, School of Medecine, Marseille, France; O. Fernandez, Hospital Carlos Haya, MUSIQoL Group, Marbella, Spain; P. Flachenecker, Neurologische Klinik, MUSIQoL Group, Wurzburg, Germany; J. Pelletier, Hospital Timone, School of Medecine, Marseille, France; S. Stecchi, Villa Mazzacorati, MUSIQoL Group, Bologna, Italy; C. Constantanescu, Haya Queen’s Medical Center, MUSIQoL Group, Nottingham, UK; M.C. Simeoni, Perceived Health Research Unit, Public Health Depart, School of Medecine, Marseille, France; Donald Patrick, Washington University, MUSIQoL Group, Seattle, USA; P. Auquier, Perceived Health Research Unit, Public Health Department, School of Medecine, Marseille, France The Multiple Sclerosis International Quality of Life (MuSIQoL) Questionnaire is a multidimensional self-administered specific health-related quality of life scale for multiple sclerosis (MS). The questionnaire focuses on concerns identified by patients with MS and has been developed simultaneously in several languages: English, French, German, Spanish, Italian. We interviewed 94 in- and out-patients with varying levels of disease, having MS according to Poser criteria. After analyzing the content of the audio-taped interviews, we selected 73

items describing the impact of MS in everyday life. At this step 11 countries have joined this project: the USA, Canada, Turkey, Brazil, Norway, Sweden, Lebanon, Argentina, Israel, Russia and South Africa, for which we established transcultural equivalence using forward– backward translation. The acceptability and comprehensibility of this preliminary questionnaire were tested among patients in all these countries. After a reduction step (n ¼ 178), we conducted a validation study to evaluate the psychometric properties of this questionnaire including 2500 patients (150–200 by country) with relapsing–remitting, primary progressive and secondary progressive MS and with isolated clinical syndromes. Patients fulfilled MuSIQol, SF36, symptoms questionnaire at day 0 and day 21. In the meantime, physicians reported socio-demographic data, clinical evaluation (history, current health status, EDSS, ambulation index, cognitive problems, CGI, treatments). Preliminary analyses, issued from three countries, isolated 31-items describing six major domains: Physical Well-Being, Psychological Well-Being, Social Relationships (Family, Friends), Relations with caregivers, Self-Esteem and Body Image (MAP analysis, Factor Analysis, Rasch). Cronbach’s a are over 0.7 and ICC satisfactory. The final results will be presented. An implementation survey, following patients during 1 year, is ongoing. # 1010/CHANGES IN QUALITY OF LIFE OVER 6 MONTHS AFTER A FIRST ISCHEMIC STROKE Krystyna D. Jaracz & Krystyna Gorna, Department of Nursing, University of Medical Sciences, Poznan, Poland The majority of studies on quality of life (QOL) after stroke are crosssectional. They congruently show that post-stroke QOL is significantly impaired. However, little is known about how QOL changes over time. There are only few longitudinal researches and their results are inconsistent. The purpose of this study was to assess changes in poststroke QOL in relation to neurological and functional recovery. A sample of 72 consecutive patients (26 men, 46 women; mean age 65.1 year) who were hospitalised for a first-ever ischemic stroke and fulfilled defined criteria was followed up at 3 and 6 months since stroke onset and compared with 110 stroke-free community controls. All patients were interviewed at their homes. They completed a questionnaire that included sociodemographic and clinical variables, Sickness Impact Profile (SIP), Ferrans and Powers Quality of Life Index (QLI), Barthel Index (BI), Scandinavian Stroke Scale (SSS) and Zung Depression Scale. Dependent and independent t tests and ANOVA were used for statistical analyses. The mean SSS and BI scores showed the most recovery in the first 3 months, but even between 3 and 6 months there were still significant improvements (p < 0.02). Between 3 and 6 months after stroke, there were significant improvements in the total, Physical and Psychosocial dimensions of the SIP (p < 0.001). Also, there were significant improvements in the overall, Health and functioning and Psychological subscales of the QLI (p < 0.01), but not in the Socio-economic and Family subscales. Changes in QOL (SIP only) were associated with the level of neurological impairment (p < 0.02). The Psychosocial dimension scores of the SIP deteriorated for patients with severe stroke. Six months after stroke, both SIP and QLI scores of the patient group were significantly worse than in the control group (p < 0.001). In summary, this study suggests that during the initial 6 months after stroke, QOL of stroke survivors improves, although still maintains highly abnormal in comparison with the stroke-free subjects. More attention should be paid to severe stroke victims to protect their QOL from further deterioration.

855 # 1259/HEALTH-RELATED QUALITY OF LIFE IN PATIENTS WITH AN ACOUSTIC NEUROMA BEFORE AND AFTER SURGERY Dominique Gall-Kleebach & Josef Ilmberger, Physical Medicine and Rehabilitation, Jo¨rg-Christian Tonn, Neurosurgery, University Clinics of Munich, Munich, Germany Objective: Outcome of surgery in patients with acoustic neuromas in terms of health-related quality of life (HRQOL) is controversly discussed; there are no studies with a preoperative assessment. We conducted a prospective study in which HRQOL as well as neuropsychological and depressive symptoms were assessed. Methods: Two standardised HRQOL questionnaires, a generic (SF-36) and a specific instrument (EORTC-QLQ-C30), assessments for neuropsychological functioning and depressive disorders (BDI) and a questionnaire concerning pre- and post-operative symptoms were used with 20 patients before as well as 3 and 6 month after surgery. Results: In the SF-36, HRQOL was partly reduced already preoperatively; after 6 months, values on the following scales were significantly reduced with respect to preoperative measures: physical functioning, general health perception, vitality and social functioning. In the EORTC-QLQ-C30 values of the scales role functioning and global health status were reduced; measures of fatigue and pain were increased. There were no correlations with neuropsychological performance. There were no depressive symptoms preoperatively, but 15% of the patients showed a mild depressive disorder and 10% a severe depression after 6 months. One important factor for this outcome seems to be the occurrence of a facial paresis, as patients with a paresis showed lower scores in several SF-36 dimensions, significant reductions in EORTC-QLQ-C30 scales and significant higher scores in BDI compared to patients without paresis. Conclusions: The present prospective study supports the demand for the standardised use of generic and specific HRQOL measures in combination with an instrument for depressive disorders in order to identify patients at risk for depression and significant loss of quality of life. Psychological support starting from diagnosis may then be provided. # 1297/QUALITY OF LIFE AND COMPLEMENTARY CARE SEEKING BEHAVIOR IN PATIENTS WITH MULTIPLE SCLEROSIS Jo¨rg Klewer, Karen Vogt, Ilse Worm & Joachim Kugler, Public Health, Dresden Medical School, Dresden, Germany Multiple sclerosis (MS) is a common neurological disease. Due to modern therapies, the prognosis is better than some years before. Although, the chronic condition of MS motivates several patients to make use of complementary and alternative medicine (CAM). In collaboration with the German Multiple Sclerosis Association, 827 MS patients in Saxony with confirmed MS were asked to work on a mailed anonymous questionnaire. The questions were related to demographical data, impairments, therapeutical course, use of CAM, illness related costs and quality of life (SF-36). Altogether 757 patients returned their questionnaires (response rate: 84%; average age: 49 years, 74% females). Over 75% of the patients reported disturbed coordination and sensibility. Around 50% complained about muscle weakness, spasticity and visual disturbances. Almost 5% reported regularly treatment by non-medical practitioners. Fewer than 10% admitted spending money for CAM (5–1000 monthly), but 50% reported use of CAM, especially homeopathy, vitamins, diets and acupuncture. Compared with the MS patients not using CAM, these patients experienced decreased quality of life in the SF-36-dimensions physical role and emotional role. The results indicate, that despite modern therapies, a considerable number of MS patients still uses CAM to cope with illness-related problems. Therefore, better care for those patients by their physicians is required. By this, MS patients would be protected from unproved CAM and their quality of life will be improved. Additionally, the results point to the methodological problem, that more patients report to use CAM than those who admit to spend money for CAM.

# 1298/QUALITY OF LIFE IN SINGLE LIVING PATIENTS WITH MYASTHENIA GRAVIS Joachim Kugler & Jo¨rg Klewer, Public Health, Dresden Medical School, Dresden; Rosemarie Amman, Head Office, German Myasthenia Association, Bremen; Dieter Po¨hlau, Deparment of Neurology, Kamillus Hospital, Asbach, Germany Myasthenia gravis (MG) is a common neuromuscular disease. Progress in medical therapy has continuously increased life expectancy in MG patients. Despite this and due to the social problems of single living persons, until now no comprehensive study investigated the situation of MG patients living alone. In collaboration with the German Myasthenia gravis Association, the self-help organization for patients, 2150 patients were asked to work on a mailed questionnaire on health care situation. The questions were related to demographical data, impairments, therapeutical course, use of complementary therapies, illness related costs and quality of life (SF-36). Around 71% responded and altogether 313 of these MG patients reported living alone (proportion females: 82.7%, average age: 61.8 years). In average, the single living MG patients were significantly 6 years older than those not living alone, they lived significantly more in bigger towns and this sample included more females. These MG patients suffered longer from MG, showed increased impairments, especially in mobility, and used significantly more often complementary medicine. Additionally, quality of life was reduced in six dimension of the SF-36. It becomes obvious, that impaired quality of life results from the constellation of living alone, higher age, female gender and longer lasting course of MG. Therefore these MG patients living alone need more attention by health care providers and measures for improving living situation of these patients are still required. # 1299/EXPERIENCES WITH HEALTHCARE SERVICES AND QUALITY OF LIFE AMONG PEOPLE WITH MULTIPLE SCLEROSIS IN THE UNITED STATES AND GERMANY Jo¨rg Klewer, Joachim Kugler, Karen Vogt & Ilse Worm, Public Health, Dresden Medical School, Dresden; Thilo Kroll & Phillip W. Beatty, MedStarResearch Institute, NRH Center for Health and Disability Research, Washington, Washington DC, USA Multiple sclerosis (MS) is a debilitating neurological condition that affects approximately 100,000 people in Germany and 300,000 in North America. In this paper, we will present findings from research conducted with American and German nationals with MS about their healthcare experience and quality of life. In the United States and Germany data were collected using structured mail surveys. In the US data were collected through national disability, community organizations and listservs as part of a national longitudinal study about the healthcare experience of people with physical disabilities. 161 people with MS (mean age ¼ 48.5 years; 72.7% women) participated. In Germany 757 individuals with MS (mean age ¼ 49 years; 74% women), members of the German MS Association in Saxony completed the survey, including the SF-36. Response rates exceeded 80%. All respondents experienced frequent symptoms of spasticity (US: 60%; Germany: 50%) and visual problems (US: 31%; Germany: 50%). Most respondents in both countries had health insurance. Fewer than onethird of the US respondents were covered by public programs (Medicare, Medicaid), 96% of the German participants had mandatory coverage. Around 60% of the German respondents reported to spent more than 50 EURO monthly to cope with MS related problems. Additionally, they presented reduced quality of life in all dimensions of the SF-36. The results indicate, that despite improved therapies and high quality health care systems, patients quality of life is still reduced. Therefore, reduction of illness related problems and financial burden and measures for improving quality of life as well are still required in US and German MS patients.

856 # 1442/THE RELATIONSHIP BETWEEN PSYCHOLOGICAL ADJUSTMENT AND QOL IN PARKINSON’S DISEASE Yoshimi Suzukamo, Deparment of Epidemiology and Health Care Research, Kyoto University School of Public Health, Kyoto; Sadayoshi Ohbu, Department of Neurology, Yokohama Municipal Citizen’s Hospital, Yokohama; Tomoyoshi Kondo & Junko Kohmoto, Department of Neurology, Wakayama Medical University, Wakayama; Shunichi Fukuhara, Department of Epidemiology and Health Care Research, Kyoto University School of Public Health, Kyoto, Japan The objective of this study was to investigate what effect psychological adjustment to the disease had on the Quality of Life (QOL) of patients with Parkinson’s disease. Two hundred and two patients (88 males and 114 females with a mean age of 65.4) were evaluated using the validated Japanese version of Parkinson’s Disease Questionnaire (PDQ-39) and the Nottingham Adjustment Scale (NAS-J). Multiple regression analysis was performed using age and gender as the explanatory variables, and comparisons were made of the changes in the R2 when clinical severity (Yahr) and psychological adjustment subscales were added. In addition, comparisons were made of the QOL score between groups with dierent levels of psychological adjustment. The components of psychological adjustment were confirmed based on an analysis of variance. The R2 increase was higher in the model including severity in comparison with the psychological adjustment subscale model in the areas of mobility and ADL. However the R2 was higher when the psychological adjustment subscale was included in comparison with severity in the other subscales (emotional well being, stigma, social support, cognition, communication and bodily discomfort). The QOL scores adjusted sex, age and severity was significantly lower in the low adjustment group in comparison with other groups. On covariance structural analysis, psychological adjustment also influenced QOL more strongly than did clinical severity. Psychological adjustment had a greater eect than severity of disease on the QOL of patients with Parkinson’s disease. This finding suggests that not only is suppression of the procession of the disease symptoms important but that psychological intervention may also be eective in the enhancement of the QOL of patients with Parkinson’s disease. # 1177/QUALITY OF LIFE OF ELDERLY EPILEPTICS Fabio Arpinelli, Giovanni Visona & Maurizio Roncolato, Medical Department, GSK S.p.A., Verona, Italy It is well known that prevalence of epilepsy is greater in aged population. Cognitive capability is impaired frequently, and antiepileptic drugs can affect mood. We aimed to measure Quality of Life (QoL) and prevalence of cognitive disorders and depression of aged epileptic patients (pts). We carried out an observational, cross-sectional, multicentre study. Eligible pts were aged 60 or more, treated at least for 4 weeks, able to fill in SF-36 questionnaire. Neurologists collected demographic, clinical and therapeutic data on a specific form. Pts were administered Beck Scale, Mini Mental State (MMSE) and SF-36 questionnaires. Forty-eight neurologists recruited 434 evaluable pts. Males were 51.8%, mean age was 69 y.o., 68% had low or no education. About 49% of pts had not driving license. Seizures appear only during daytime in 47% of pts. Epilepsy was cryptogenic (37.3%), symptomatic (50.2%), idiopathic (9.7%). Seizures were partial (49.3%) or generalized tonic-clonic (29%). More than 73% of pts suffered from

at least another concurrent disease (especially cardiovascular). Monotherapy was given to 64% of pts, 30% received two drugs. Most used drugs were barbiturates (45%) and carbamazepin (44.5%). Adverse events are reported by 39% of patients (especially drowsiness 29%). About 8% of pts did not comply with therapy (usually they reduced prescribed dosage). About 26% of pts showed an impaired cognitive status. Significative depressive symptoms were collected in 16% of pts, while 30.4% suffered from a mild depression. SF-36 mean scores showed that our sample had lower mean scores than Italian norm (adjusted for sex and age) in General Health, Role Physical, Vitality, Social Functioning and Emotional Role domains. A good correlation was detected between Beck Scale and some SF-36 domains (Mental Health, Vitality, Social Functioning and General Health). Correlation was poor between SF-36 and MMSE. Our results show that depression is able to significantly impair QoL of aged epileptics, especially mental components. # 1089/THE BIG LUNG TRIAL (BLT) – QUALITY OF LIFE (QOL) STUDY: DETERMINING THE EFFECT OF CHEMOTHERAPY FOR SUPPORTIVE CARE (SC) PATIENTS WITH NON-SMALL CELL LUNG CANCER (NSCLC) Julia M. Brown, Helen Thorpe & Vicky Napp, Northern & Yorkshire Clinical Trials & Research Unit, University of Leeds, Leeds; Mick D. Peake, Glenfield Hospital, Leicester, UK The BLT detected a survival benefit in the cisplatin-based chemotherapy arm for SC patients with NSCLC. Here we report fully on the results of the QoL sub-study. Patients randomised to receive SC alone or SC with chemotherapy were compared using EORTC QLQ-C30 and LC17, collected at 0, 6–8, 12, 18 and 24 weeks, and daily diary cards completed over 12 weeks. The primary endpoint was global QoL score at 12 weeks. 273 patients were randomised, 138 to SC alone, 135 to SC with chemotherapy. Analyses of covariance of the main endpoints adjusted for (i) baseline QoL scores, (ii) minimisation factors, potential QoL prognostic factors, were performed. To ease clinical interpretation, estimates from a logistic regression model, investigating the proportion of patients improving from baseline, were also obtained (Osoba 1999). However in these analyses missing data were a problem: 34% had baseline or 12-week assessments missing. To investigate the impact of missing data, various analytic models using different mechanisms of missing data and data at all time-points were employed. Standardised area under the curve analysis and multilevel modelling for repeated measures data were performed to further investigate trends in QoL over time. No statistically significant differences in global QoL were detected at any time-point; however, there was no evidence of a large detrimental effect. Secondary measurements of physical, emotional functioning, dyspnea and fatigue, were not statistically significant different at any time-point. Considering all time-points there was less pain in the SC with chemotherapy group (p ¼ 0.02). This study represents the most detailed investigation into QoL in advanced NSCLC. It uses standard questionnaires, adequate pre-defined sample size to determine ‘large’ effects and a variety of analyses to check the sensitivity of the conclusions. The study shows that there are no large detrimental QoL effects of using chemotherapy in these patients. The trial was supported by a research grant from Pierre Fabre Oncology.

857 # 1770/QUALITY OF LIFE (QOL) OF NON-HODGKIN’S LYMPHOMA SURVIVORS James C. Lynch & Mary E. Morris, Preventive and Societal Medicine; Gregory Bociek, Philip J. Bierman, Julie M. Vose & James O. Armitage, Internal Medicine-Oncology/Hematology, University of Nebraska Medical Center, Omaha, NE, USA Non-Hodgkin’s lymphoma (NHL) is an increasingly common malignancy with improving clinical outcome, resulting in a large, but understudied, number of survivors. To evaluate the QOL of this survivor population, we conducted a cross-sectional mail survey. Potential participants surviving at least 2 years after start of initial treatment with a standard chemotherapy regimen were identified from the Nebraska Lymphoma Study Group Registry and, after obtaining physician approval, asked to complete a demographic form and three quality of life instruments: SF-36, City of Hope-Quality of Life Cancer Patient (COH), and the Functional Assessment of Cancer Therapy-General (FACT-G.) Four hundred twenty-six eligible participants were contacted and 53% (n ¼ 218) responded. Participants were a median of 85 months (range 34–218) post-treatment at the time of survey. The median age at survey was 65 years (range 29–92), 56% are female, and 99% are white, non-Hispanic. After stratification on age and gender, participants did not report statistically significant deficits in any SF-36 domain compared to published norms. Females reported a significantly lower (p ¼ 0.013) physical functioning SF-36 subscale score than males, but none of the other SF-36 domains differed between males and females. Using five categories (<45, 45–54, 55–64, 65–74, P75), age was significantly related to the physical functioning (p < 0.001) and rolephysical functioning (p < 0.0001) subscales of the SF-36, but not to other domains. Females reported higher scores on the COH social concerns subscale (p < 0.01), with no differences between males and females on other COH subscales or total score. Only the COH psychological well-being and social concerns subscales, and total score were statistically significantly different (all p < 0.05) across the age categories. Males and females did not differ on any of the FACT-G subscales or the total scale. Only the FACT-G social concerns score differed across age categories (p < 0.01). While NHL survivors report QOL that is comparable to the general population, differences by sex and by age in selected domains merit further study. # 1020/LONG-TERM QUALITY OF LIFE (QOL) OF PATIENTS WITH HEMATOLOGICAL MALIGNANCIES WHO UNDERWENT HEMATOPOIETIC STEM CELL TRANSPLANTATION: COMPARISON BETWEEN THE UTILITY AND THE PROFILE APPROACHES Chiun Hsu, Oncology, National Taiwan University Hospital; Jung-Der Wang, College of Public Health, National Taiwan University; JingShiang Hwang, Institute of Statistical Science, Academia Sinica; Grace Yao, Psychology, National Taiwan University; Jih-Luh Tang, Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan The relative merits of bone marrow (BMT) vs. peripheral blood stem cell transplantation (PBSCT) in the treatment of hematological malignancies remain undetermined. Long-term QOL for BMT and PBSCT patients was compared in this study. The cohort consisted of 140 patients with hematological malignancies (77 BMT, 63 PBSCT). Seventy-one of the 86 surviving patients agreed to be interviewed. The patients’ utility was obtained by the standard gamble (SG) method and the patients’ QOL by using the European Organization for the Research and Treatment of Cancer (EORTC)-QLQ-C30 questionnaire and the brief form of the World Health Organization quality of life (WHOQOL-BREF) questionnaire. Both the utility and the QOL measurements were integrated with the survival function of the cohort to obtain the quality-adjusted life months (utility approach) and survivaladjusted score-months (profile approach). The 5-year overall survival was 67.5% for BMT and 39.2% for PBSCT (p ¼ 0.02). The mean SG scores were similar between the two groups (0.695 ± 0.359 vs. 0.607 ± 0.326, p ¼ 0.31). BMT patients had significantly better QOL scores in the physical domain of WHOQOL-BREF and physical functioning of EORTC-QLQ-C30. The SG scores had significant (p < 0.05) correlation with the following QOL scores: physical domain of WHOQOL-BREF; cognitive and social functioning, nausea/vomiting, pain, appetite loss and financial difficulty of EORTC-QLQ-C30. The mean quality-adjusted survival time was 44.6 ± 4.8 months for BMT and 27.6 ± 4.5 months for PBSCT (p < 0.01). The survival-adjusted

score-months was also significantly better for BMT patients in all the functioning domains and symptom items. Our results suggest that for patients with hematological malignancies, BMT patients may have better long-term QOL than PBSCT patients, as indicated by both the utility and the profile approaches. Utility of our patients may be influenced by individual physical functions. (Supported by grants NTUH91S053 and NHRI-EX92-9204PP). # 1860/PSYCHOLOGICAL ADJUSTMENT OF LUNG CANCER SURVIVORS Geraldine V. Padilla, School of Nursing, University of California San Francisco, South San Francisco; Linda Sarna, School of Nursing; Mary-Lynn Brecht, Integrated Substance Abuse Program, University of California Los Angeles, Los Angeles, CA, USA This report describes the findings of a prospective cross sectional study of 142 disease-free survivors of non-small cell lung cancer. We hypothesized that fear-uncertainty, purpose, and control in life were important predictors of psychological adjustment of lung cancer survivors. Fear-uncertainty was measured with a 5-item Likert scale, and purpose and control in life were each measured with a 1-item scale. Psychological adjustment was measured with the SF36 mental health scale. The average age of the participants was 71 years, 91% were over 60, and 72% had more than a high school education. Linear regression analysis partially supported the hypothesis (R2 ¼ 0.438). Using the <0.01 level of significance, fear-uncertainty, and control in life, but not purpose in life, contributed significantly to the variance of the mental health component of the SF36. Neither physical distress from comorbid conditions, side eects, and illness distress contributed significantly to the variance in mental health. The results suggest the importance of assuaging patient fears of cancer recurrence, metastasis, second cancers and diagnostic tests and oering them a sense of control over events in their life to promote psychological adjustment to survivorship. # 1817/QUALITY OF LIFE OF PEDIATRIC PATIENTS RECEIVING ALLOGENEIC STEM CELL TRANSPLANTATION: INTERIM RESULTS OF A MULTI-CENTER LONGITUDINAL STUDY Rosemarie Felder-Puig, St. Anna Children’s Hospital, Vienna, Austria; Alain di Gallo, University Clinic of Child and Adolescent Psychiatry, Basel, Switzerland; Marion Waldenmair, St. Anna Children’s Hospital, Vienna; Peter Norden, Department of Pediatrics, University of Innsbruck, Innsbruck; Helmut Gadner, Reinhard Topf, St. Anna Children’s Hospital, Vienna, Austria This presentation describes first results of an ongoing effort to assess the quality of life (QoL) and adjustment of pediatric patients receiving allogeneic stem cell transplantation (SCT). Eligible subjects include patients who receive SCT between January 2000 and December 2003. The patients are assessed seven times: 4–6 weeks (T1) and 7 days (T2) prior to SCT, and 10 (T3), 28 (T4), 100 (T5), 180 (T6) and 360 (T7) days post SCT. For the measurement of various domains of QOL, standardized instruments (Health Utilities Index and Pediatric Quality of Life Inventory) have to be completed by patients, parents, and physicians. Further information gathered include treatment-related coping and compliance of child and parent, the child’s and the family’s history and medical data. So far, we have collected data about 40 SCT recipients aged 3–17 years at the time of SCT. Scores of QoL show a large variability. Generally, global QoL is worst at T3 and T5, but is better at T7 than at baseline (T1). T3 is the most critical time point with regard to emotion, pain, and nausea. Worries about the side-effects of treatment and fear of a relapse remain relatively stable between T1 and T5, and slightly increase after discharge from the SCT unit. Physical functioning of patients is significantly reduced at baseline, is even worse at T6 but improves until T7. Discrepancies between physicians’ and parents’ or patients’ assessments are most evident during T2–T4. At all other measurement points, correlations are in the medium range without any clear pattern as to whether physicians tend to over- or underestimate problems. Further, we will present individual cases that show the relative importance of transplant type, treatmentrelated complications and coping of patient and/or mother on specific QoL domains at the various stages of treatment. Altogether, the instruments’ scores reflect well the psychological and physical stressors patients and families are confronted with during SCT therapy.

858 # 1225/THE IMPACT OF PROSTATE CANCER DIAGNOSIS ON MENTAL HEALTH Ida J. Korfage & Marie-Louise Essink-Bot, Public Health; Monique J. Roobol & Fritz H. Schroeder, Urology; Harry J. de Koning, Public Health, Erasmus University Medical Center, Rotterdam, The Netherlands A longitudinal study on the health-related quality of life (HRQoL) effects of primary therapy in localized prostate cancer (PC) has been ongoing in Rotterdam, the Netherlands, since 1996. Men with recently diagnosed PC (n ¼ 314) were followed from before treatment (radical prostatectomy RP or external radiotherapy ER) till four to five years afterwards. Long term results showed that, contrary to other scales, SF-36 mental health scores in both therapy groups improved after treatment and remained at that high level till end of follow-up. The average improvement amounted to 10 points on a scale of 100 in RPpatients and 12 in ER-patients. We hypothesised that low mental health scores between diagnosis and treatment were caused by the diagnosis; that men had experienced better mental health before they were diagnosed with prostate cancer; and that treatment had a favourable effect on mental health (Sprangers, Psycho-Oncology, 2002). The European Randomised Study on Screening for Prostate Cancer in Rotterdam enabled us to test this hypothesis. We aimed at recruiting 2000 screening participants to complete a HRQoL questionnaire before screening, and 50 screening participants in whom PC was subsequently diagnosed for two follow-up telephone interviews (before treatment and 6 months afterwards). Questionnaires consisted of EQ-VAS and SF-36 Mental Health and Vitality. Until now, 14 men with screen-detected PC were recruited for interviews. First results showed a decrease of 8 points on a scale of 100 after diagnosis in mental health. Vitality scores were 85 (SD 14) before and 81 (SD 11) after diagnosis. EQ-VAS values dropped from 86 (SD 10) before to 77 (SD 20) after diagnosis. Results of more men and of the second followup interview will be available at the time of the conference. Although the study is on-going, first results suggest a large effect of PC diagnosis on mental health. # 1133/EXAMINATION OF HEALTH-RELATED QUALITY OF LIFE (HRQL) IN PATIENTS WITH PERITONEAL SURFACE MALIGNANCIES Tatiana Beresnev & Sharon Mavroukakis, Surgery Branch, NCI/NIH; Susan Marden, Nursing Department, NIH; Donald White, H. Richard Alexander, James Pingpank & Steven Libutti, Surgery Branch, NIH/ NCI, Bethesda, MD, USA Peritoneal surface malignancies (PSM) of non-gynecologic origin including malignant mesothelioma (MM), low-grade (LG), and highgrade (HG) gastrointestinal adenocarcinoma (Ca) are associated with poor prognoses. Patients experience chronic abdominal pain, weight loss, ascites, and die of complications related to intraperitoneal tumor progression. Little is known about the impact of PSM on HRQL. This study examined HRQL in patients with PSM and ECOG status 0 prior to participation in Institutional Review Board approved clinical studies. HRQL was measured by physical (PCS) and mental (MCS) health scores from the SF-36 (norm ¼ 50). Cancer-specific HRQL was measured by physical (P), emotional (E), social (S), and functional (F) subscales (range 0–100), colon subscale (C; range 0–28), and total score (Total-C; range 0–136; higher score ¼ better HRQL) of the Functional Assessment of Cancer Therapy-Colon (FACT-C). Scores

were compared to published norms. Patients (n ¼ 81; Female 43, Male 38; mean age 48 year; ascites 31%) with PSM (MM 30%, LGCa 52%, HGCa 18%) were evaluated. The median disease duration at presentation was 6 months (range 1–75), 80% of patients had prior surgical debulking, and 32% had systemic chemotherapy. FACT-C mean (SEM) scores were P ¼ 71(2.1); S ¼ 76(1.8); E ¼ 62(1.6); F ¼ 64(1.8); C ¼ 20(0.5) and Total-C ¼ 104(2.1). Pain, role-physical, social functioning, role-emotional, and mental health SF-36 scores were significantly lower than 1998 values for cancer patients from the US general population (GP), but vitality, physical functioning, and general health were significantly higher (Table). No significant differences in SF-36 or FACT-C scores were identified between MM, LGCa, and HGCa groups. The results suggest patients with PSM and ECOG 0 perceive impaired physical and mental health particularly in areas of pain and social functioning compared to published norms. PSM of different histologies appears to affect HRQL similarly. Better general health and vitality may be related to ECOG 0 status in this selected sample. Treatment outcome analysis for patients with PSM should include HRQL endpoints. # 1433/EVALUATION OF SPIRITUAL WELL-BEING QOL MEASURED BY THE KOREAN-VALIDATED FACIT-Sp IN KOREAN CANCER PATIENTS Sehyun Kim, Preventive Medicine; Hye J. Park, Nursing, College of Medicine, Pochon Cha University, Sungnam; Kyong J. Jeong, Pharmacy, Bundang Cha General Hospital, Sungnam; Jong E. Lee, Nursing, College of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do; Sung S. Han, Nursing, Catholic University, Seoul; So Y. Chong & Do Y. Oh, Internal Medicine, College of Medicine, Pochon Cha University, Sungnam, Kyonggi-Do, Korea Shamanism is the oldest traditional method of healing and mind body spirit relationship as the path to good health and harmony in Korea. When people have contacted life-threatening disease (such as cancer), they seek spiritual help from the God or the higher being regardless of religious background in addition to medical treatments. Previous research suggested that spirituality was an important aspect of QOL for cancer patients. The impact of religion on spirituality wellbeing never been evaluated in Korean cancer patients. The objective of the study was to evaluate the association between religious background and spirituality QOL in cancer patients in Korea. A consecutive sample of 58 patients with cancer was included in the study. Mean age of the subjects is 51.6 years (SD ¼ 11.04; range 27–82). The most subjects had clinical characteristics of stage III and fair Karnofsky scale (mean 71.09), and most of them were married, under college education, and had religion (75% of sample). The Korean translation FACIT-Sp had high reliability in FACT-G, FACIT-Sp-12, and FACIT-Sp total (Cronbach’s a 0.66–0.89). The various domains of QOL including spirituality did not show significant associations with the type of cancer, stage, current status, and the method of treatments. The faith factor and FACIT-Sp-12 total scores were significantly higher among those who have a religion than those who do not have any religion (p < 0.02). Faith subscale of spiritual QOL, which measures comfort and strength derived from one’s own faith, was related with religious background. Therefore, a further research on a broader concept of spiritual palliative care program is needed to provide benefit to patients diagnosed with cancer.

859 # 1375/PARTNERSHIP IN TREATMENT DECISION-MAKING AND HEALTH-RELATED QUALITY OF LIFE IN PATIENTS WITH CANCER M.S. Salek & A. Al-Mohammadi, WSP Centre for Socioeconomic Research, Cardiff University; T. Maughan & M. Mason, UWCM; P.J. Nicholls, Welsh School of Pharmacy, Cardiff University, Cardiff, UK Communication between clinicians and patients may influence patients’ subjective assessment of their QoL. Affective quality of communication appears to be the most important factor in determining a patient’s HRQoL. However, previous knowledge and subjective meaning of facts and events about diseases and related processes were found to affect the QoL of patients with cancer. Therefore, the aim of this study was to assess the impact of cancer patients information seeking behaviours and willingness to participate in treatment decision-making on HRQoL. A prospective, cross-sectional study examining patients self reports (n ¼ 105) on the Patient Communication Profile (PCP), assessing information needs, understanding, satisfaction and mental distress, and the revised McGill QoL Questionnaire (rMQOL) was carried out. Higher rMQOL scores represent better QoL. Mean age was 57.96 years (range ¼ 18–88) and 57% were females. Patients who preferred an active role in their treatment decisionmaking showed lower total mean scores in their QoL than those who preferred the passive role (p ¼ 0.049). While patients who expressed higher needs for information showed lower total mean scores in their HRQoL (p ¼ 0.03) especially in the psychological domain (p ¼ 0.015) than those with lower needs. Patients who were more satisfied with the amount and the quality of information being received showed higher total mean scores of their QoL than those less satisfied (p ¼ 0.028 and 0.047 respectively). A high total mean score of QoL was also recorded for patients who reported better understanding of their medical information (p ¼ 0.027). Patients who expressed higher mental distress showed lower QoL scores on all the domains (p < 0.01). The findings indicate that an individualised intervention that improves the communicated medical information for patients with cancer will decrease their mental distress and hence improve their QoL. # 1558/QUALITY OF LIFE AMONG YOUNG MINORITY WOMEN WITH BREAST CARCINOMA Ann F. Chou, School of Public & Environmental Affairs, Indiana University, Indianapolis, IN; Joan R. Bloom, School of Public Health, University of California, Berkeley, CA, USA Albeit limited, preliminary evidence suggests that minority women with breast cancer experience more disruptions in and concerns about quality of life (QoL). More research is needed to assess how ethnocultural factors influence cancer adjustment and its implications. This study examines the variation in QoL outcomes in a tri-ethnic sample of breast cancer patients. The study employed data from the California Tumor Registry and Young Women with Breast Cancer Survey. The sample included 391 women, of which 31% were minority (16% Asian, 7.5% Latina, and 8% African American). Women in this cohort were mostly married (83%), college educated (80%), employed (77%) and diagnosed at age 50 or younger. The distribution of disease stage was 14% in situ, 46% local, 37% regional, and 2% remote. About 45% had a lumpectomy and 55% had a mastectomy. Close to 47% received radiation therapy, 57% had chemotherapy, and 22% used Tamoxifen. QoL outcomes were collected at baseline, 4, 8, and 12 months post baseline and repeated measures regression analyses were conducted to examine ethnic differences in eight categories and two composite SF-36 QoL measures. Controlling for demographic and clinical factors, findings showed greater improvement in pain and physical functioning among Asian women and in vitality, physical functioning and overall mental health among African Americans in comparison to their White counterparts. Both groups reported less improvement in physical and social role functioning. With the exception of general health, Latina women as a group suffered worse outcomes in several QoL categories: pain, mental health, physical functioning, emotional role functioning, overall mental and physical health. Ethnocultural difference can impact decision-making regarding screening and treatment among minority women, which in turn affect QoL outcomes. Variations observed in QoL outcomes in this sample may be results of ethnic differences in psychosocial adjustment, social resources and appraisal of illness. Understanding ethnic differences in QoL outcomes lends in-

sights into the development of strategies aiding minority women in reducing the burden of cancer. # 1003/SHORT AND LONG-TERM EFFECTS OF ACUTE MYELOID LEUKEMIA ON PATIENT’S QUALITY OF LIFE Alberto Redaelli, Global Outcomes Research-Oncology, Pharmacia Corporation, Milan, Italy; Jennifer M. Stephens, Health Economics Research and QOL Evaluation Service, HERQuLES, Cambridge, MA, USA Objectives: The treatment of Acute Myeloid Leukemia (AML) includes a number of aggressive therapies, namely chemotherapy (CT) and bone marrow transplantation (BMT) that are known to significantly affect the quality of life (QOL) of the leukemic patient. To address this issue, an extensive literature survey to assess what has been published from the international scientific community was performed. Materials and methods: Systematic electronic review of the quality of life literature for AML focused on English-language articles published in 1990–2002, integrated with additional articles published before 1990, and manual review as appropriate for the completeness of the survey. Results: The survey has identified 28 articles which were imported into a master reference database, reviewed by two experienced staff members belonging to the authors’ companies and retrieved for full analysis. The results of this analysis constitute the main body of the poster. Conclusion: The disease and the intensive therapy (re. chemotherapy, BMT) required to cure a patient affected by AML has substantial negative impact on his/her QOL as measured by a number of different instruments (re. generic, cancer-specific and leukemia specific). However, the heavy burden on the patient’s QOL is seen after the diagnosis of the disease itself and during the course of the therapy. The long-term survivors appear to have recovered almost completely as shown by the values reached by quality of life domains such as physical, psychological and emotional distress that are not significantly different from the ones measured in the normal agematched population. # 1608/QUALITY OF LIFE IN ADULT BONE MARROW TRANSPLANT PATIENTS FOLLOWED AT LEAST FIVE YEARS AFTER TREATMENT: A COMPARISON WITH HEALTHY CONTROLS Martin Kopp, Verena Meraner, Bernhard Holzner, Barbara SpernerUnterweger & Georg Kemmler, Department of Psychiatry; David M. Nachbaur, Department of Hematology, Innsbruck University Hospital, Innsbruck, Austria Background: Bone marrow transplantation (BMT) is an established treatment modality for patients with hematological malignancies. As long-term survivors become more numerous, studies addressing the issue of long-term follow-up are necessary. Quality of life (QOL) is an important subjective outcome measure of BMT. In this study, we report on the QOL of BMT-patients, who were alive at least at 5 years after transplantation or longer in comparison to an age and sex-matched sample of healthy controls assessed in the same time-period and the same geographical region. Patients and methods: The EORTCQuality of Life Questionnaire (EORTC-QLQ C30) were mailed to patients who had there BMT at least 5 years before the beginning of the study. Thirty four BMT survivors answered the questionnaire and were analysed in comparison with healthy controls drawn from a database of 977 cases with the criterias age and sex. For each patient two healthy controls (68) were included in the analysis. Results: MannWhitney-U tests identified significant lower QOL on the dimensions physical and social functioning and the symptom scale financial impact. Discussion: In summary, patients having survived their BMT for more than 5 years did generally well in terms of global QOL. As expected and shown in an earlier study of Kiss et al. (J Clin Oncol, 2002; 20: 9) CML-patients, who were alive at least 10 years after BMT report lower physical functioning in comparison to healthy controls. Deficits regarding social functioning and financial aspects should be readdressed in future studies. Problems in these fields should be manageable during an intensive rehabilitation process integrating patients, family members and significant others. Keeping these limitations of QOL in mind, interdisciplinary (medical, psychological and social) treatment of patients should be continued after the acute phase during the medical controls and check-ups following transplantation.

860 # 1499/EQUIPRECISE MEASUREMENT OF CANCER FATIGUE: AN ITEM BANK FOR CLINICAL USE Kelly Dineen, Center on Outcomes, Research and Education (CORE), Evanston Northwestern Healthcare & Northwestern University, Evanston; Jin-shei Lai, Center on Outcomes, Research and Education (CORE); David Cella, CORE, ENH & NU, Evanston; Rita K. Bode, Rehabilitation Institute of Chicago & NU; Jamie Von Roenn, NU, Chicago; Richard C. Gershon, CORE, ENH & NU; Michael Bass, CORE, ENH, Evanston, IL, USA Fatigue is the most common unrelieved symptom experienced by cancer patients. It is under-recognized and under-treated, due in part to a lack of clinically sensitive instruments that integrate easily into the clinic setting. Modern computerized adaptive testing (CAT) can overcome some of the obstacles by enabling precise individual assessments of fatigue without requiring the administration of a large number of questions. A calibrated item bank is essential for a working CAT platform. This study describes the building of a calibrated item bank for use in clinical settings with the ultimate goal of improving fatigue identification and treatment. The sample included 301 cancer patients from the metropolitan Chicago area. Psychometric properties of items were examined by using an Item Response Theory (IRT) model. The final bank included 72 items. These 72 items satisfied IRT (Rasch) measurement requirements, demonstrated high Cronbach’s a (0.99) and reasonable item-total correlation (0.51–0.85). This bank covered the range of fatigue that clinicians working with patients would consider important. The resulting bank was amenable to the development of a CAT platform. Our simulation studies have suggested that the accurate assessment of fatigue in individual patients to within 0.2 standard error of measurement could be obtained with as few as five items from this bank. A short-form has been developed which allows one to reference fatigue scores to the continuum defined by the entire bank of 72 items. It can be used in clinical trials and other situations where grouped data are the unit of analysis. # 1728/USE OF STRUCTURAL EQUATION MODELING IN QUALITY OF LIFE MEASUREMENT WITH MECHANICALLY VENTILATED PATIENTS Yookyung Kim, Department of Health and Community Systems; Leslie A. Hoffman, Fred Tasota & Carmella Scharfenberg, Department of Acute and Tertiary Care; Michael P. Donahoe, Medicine, University of Pittsburgh, Pittsburgh, PA, USA The purpose of this study was to use structural equation modeling (SEM) to provide support for the Health Assessment Questionnaire (HAQ) as a quality of life measure in patients who require mechanical ventilation (MV). Seventy patients completed the HAQ based on functional ability prior to intensive care unit (ICU) admission. Using structural equation modeling, reliability of the scale was based on the squared multiple correlations (R2) values, and validity was based on a confirmatory factor analysis (CFA) and path coecients. R2 estimated the variance in the observed score that can be explained by the model. CFA examined relationships among the HAQ items. SEM results showed evidence of reliability and validity of the HAQ based on R2 values ranging from 0.66 to 0.91, and highly statistically significant path coecients ranging from 0.81 to 0.95, a reflection of the small standard errors. The measures of model-data fit (Normed Fit Index ¼ 0.90; Comparative Fit Index ¼ 0.92; Incremental Fit Index ¼ 0.92; Root Mean Squared Residual ¼ 0.05; Standardized Root Mean Squared Residual ¼ 0.04; Average O-Diagonal Absolute Standardized Residuals ¼ 0.03) favored the CFA model of a single latent factor. The HAQ shows excellent measurement properties in the MV population. The SEM results support the quality and the hypothesis of a single latent construct, indicating the unidimensionality of the HAQ in the MV population. This study provides evidence that the HAQ is a reliable and valid measure of physical quality of life with the MV population and as such can be a valuable tool to use in outcomes research in this population.

# 1587/PSYCHOMETRIC PROPERTIES OF THE WORLD-HEALTH ORGANIZATION QUALITY OF LIFE SHORT VERSION INSTRUMENT (WHOQOL-BREF) IN MALE ALCOHOLICS Ana Fla´via B. Silva Lima, Marcelo P. Fleck & Flavio Pechansky, Department of Psychiatry and Legal Medicine, Federal University of Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brasil Objectives: To evaluate the content and discriminant validity of the World Health Organization Quality of life (WHOQOL-BREF) instrument and verify concurrent validity against the Medical Outcomes Study Short Form (Sf-36) and Symptom Check List (Scl-90) questionnaire. Methods: 36 male alcohol dependents (13-mild/moderate dependence, 23-severe dependence) was evaluated to assess quality of life by the WHOQOL-BREF and SF-36 instruments. Results: The WHOQOL-BREF adequately discriminated the quality of life in this sample, showing a large effect size for the psychological domain and a moderate effect size for the other domains. As for the content validity, the internal consistency for all variables showed satisfactory rates. The concurrent validity between SF-36 and WHOQOL-BREF showed larger correlation between the physical and psychological domains of the WHOQOL, with all domains of the SF-36. An inverse correlation between the domains of the WHOQOL-BREF and total scores of the SCL-90, showing that the larger the number of symptoms, worse were the escores of quality of life. Conclusions: Both WHOQOL-BREF and SF-36 instruments discriminate quality of life in accordance with the severity of the dependence in an appropriate form. However, since the concurrent validity was moderate for the most of the domains of these instruments, it seems that they measure quality of life based on different constructs. # 1595/RELIABILITY AND VALIDITY OF THE WHOQOL BREF IN A SAMPLE OF BRAZILIAN OUTPATIENTS WITH DEPRESSIVE DISORDERS Marcelo T. Berlim, Daniele P. Pavanello, Marco A. Caldieraro & Marcelo P. Fleck, Psychiatry and Forensic Medicine, Federal University of Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brazil Objective: To assess the psychometric properties of the Portuguese version of the WHOQOL BREF and to validate it in sample of adult depressed outpatients. The WHOQOL BREF is a 26-items self-report instrument that examines four quality of life (QOL) domains (physical health, psychological, social relationships, and environmental), and two facets for assessing overall QOL and general health. Along with the WHOQOL BREF, subjects have also been administered the Beck Depression Inventory (BDI), the Quality of Life in Depression Scale (QLDS) and the Hamilton Depression Rating Scale (HAM-D). Methods: Data have been collected in a mood disorders clinic of a university hospital located in Southern Brazil. In addition, a subsample has been followed prospectively for a mean of 12 weeks in order to assess WHOQOL BREF’s sensitivity to change. Results: The instrument was cross-sectionally administered to 120 depressed subjects, and 66 have been reassessed after a mean of 12 weeks following the start of treatment. No differences in the mean scores for the four domains were found on the basis of patients’ social and demographic characteristics. The WHOQOL BREF has acceptable internal consistency (a range: 0.82–0.95 across domains), and it is responsive to change in clinical condition, as evidenced by predicted score improvement after initial treatment of depression. Convergent validity between the WHOQOL BREF and the BDI, the HAM-D and the QLDS was statistically significant, as well as its ability to discriminate between outpatients on the basis of their level of depressive symptoms. Conclusions: The present study shows that the WHOQOL BREF is a psychometrically valid and reliable instrument, and that it is suitable for evaluating the quality of life of adult depressed outpatients.

861 # 1718/UNDERSTANDING THE COMPONENTS OF HRQL AMONGST THE KAMBA OF KENYA: PUTTING A QUALITATIVELY DERIVED CONCEPTUAL MODEL TO THE PSYCHOMETRIC TEST Julia Fox-Rushby, Health Policy Unit, London School of Hygiene & Tropical Medicine, London; Annabel Bowden, The Lewin Group, Quintiles UK Ltd, Bracknell; Sarah C. Smith & Donna L. Lamping, Health Services Research Unit, London School of Hygiene & Tropical Medicine, London, UK The KENQOL structured survey is a new measure of HRQL developed for use amongst Kamba adults (P16 years) in Makueni district, Kenya. It represents five concepts of HRQL developed through extensive qualitative work: contentment, corporeal capacity, cleanliness, co-operation and completeness. Following our previous report of the psychometric evaluation of the full 112-item KENQOL, we now report on the psychometric testing of the underlying conceptual model that guided questionnaire development. In field testing with 447 community respondents, we produced an item-reduced version (112-items reduced to 69-items) based on missing data, endorsement rates, item redundancy and internal consistency of subscales. Results of psychometric testing of the conceptually derived subscales are mixed. Factor analysis revealed only limited empirical support for the a priori conceptual model. Psychometric properties are similar to those for the full 112-item version, indicating that despite the substantial reduction in respondent burden, reliability and validity are not compromised. We discuss (i) the methodological challenges in combining qualitative and quantitative findings to understand conceptions of HRQL in a crosscultural context and (ii) our attempts to refine the conceptual model based on a critical evaluation of discrepancies between qualitative and quantitative findings. # 1886/DEVELOPMENT OF A SHORT HIV QUALITY OF LIFE QUESTIONNAIRE BY EQUATING ITEMS FROM TWO EXISTING INSTRUMENTS Nuria Perulero, Xavier Badia & Roset Montserrat, Health Outcomes Research Europe Group, Barcelona, Spain MOS-HIV and MQOL-HIV, HRQOL questionnaires in patients with HIV, were validated population and showed good psychometric properties, but both questionnaires had too many items for the clinical practice. The development of a new short Quality of Life Questionnaire for HIV patients will allow to assess the quality of life in a clinical practice on the actual disease context. MOS-HIV and MQOL-HIV were administered by random assignment to a different group of VIH patients (MOS-HIV, n ¼ 314; MQOL-HIV, n ¼ 323). The new questionnaire based on the assemblage (equating) of the items of two existing questionnaires (MOS-HIV and MQOL-HIV) using a common anchor test (EuroQol-5D) was also given to both groups. The high reliability showed the unidimensionality of both questionnaires. Within each set, equating consisted in connecting MOS-HIV and MQOL-HIV through their link with the EuroQol-5D. Equating was carried out through the Rasch mathematical model. Quantitative (item statistics) and qualitative reductions (expert opinion) of the equated sets resulted in a 17item questionnaire. Although the new instrument requires further empirical validation, it provides a promising alternative for use in clinical practice to currently existing longer questionnaires.

# 1344/DEVELOPMENT AND PRELIMINARY VALIDATION OF DEEP VENOUS THROMBOSIS QUALITY OF LIFE INSTRUMENT (DVTQOL) FOR PATIENTS WITH ACUTE DEEP VENOUS THROMBOSIS (DVT) Ewa C. Hedner, Ka´roly R. Kulich, Ingela Wiklund & Jonas Carlsson, Clinical Science, AstraZeneca R&D, Mo¨lndal, Sweden DVT, both its sequels and therapy, adversely affects patients’ healthrelated quality of life (HRQL). The objective was to develop a diseasespecific HRQL measure, DVTQOL, for patients treated with proximal DVT. Items for the new DVTQOL questionnaire were generated from individual interviews with DVT patients and medical specialists. A total of 121 consecutive outpatients, (50% males; mean age 61.2, SD ¼ 14 years) treated with warfarin for symptomatic proximal DVT were recruited in connection to a routine monitoring visit in an open study. Patients were treated with warfarin for at least 4 weeks and up to 6 months. Patients completed the DVTQOL, SF-36 and EQ-5D using electronic data capture. The initial 65 items were reduced to 29 by using exploratory factor analysis and Rasch analysis. Items with high ceiling and floor effects were deleted. The final DVTQOL consists of 29 items divided into six dimensions depicting problems with: Symptoms (e.g. pain, swollen ankles, sense of fatigue), Emotional distress, problems with Physical functioning, Sleep disturbances, Food/Drink problems and hassles with Monitoring. The internal consistency reliability was high. Pearson correlation was acceptable between the relevant dimensions of the DVTQOL, SF-36 and EQ-5D thereby confirming its construct validity (Table 1). # 1376/THE DEVELOPMENT AND PSYCHOMETRIC VALIDATION OF A VENOUS LEG ULCER QUALITY OF LIFE (VLU-QOL) QUESTIONNAIRE Diane J. Wild & Helen Doll, Cassington, Oxford, Oxfordshire; Asha Hareendran, Outcomes Research, Pfizer Ltd, Sandwich, Kent; Peter J. Franks & Christine J. Moffatt, Centre for Research and Implementation of Clinical P, Thames Valley University, London, UK Venous Leg Ulcers (VLU) affects various dimensions of a patient’s life. In evaluating treatment options for VLU, treatment outcomes are not viewed only in terms of clinical measures like ulcer size, and rate of healing, but also in terms of level of comfort, symptom reduction and improvement in daily activities. A study was conducted to develop and validate a Quality of Life questionnaire (the VLU-QoL) to evaluate outcomes of treatment. The instrument was based on the Skindex-29 (for dermatological conditions) with adaptation of the language to make it appropriate for VLU patients. VLU-specific items were generated from in-depth interviews and focus groups with VLU patients. 124 VLU patients were recruited from four centres for instrument validation. The VLU-QoL, the SF-36 and clinical information forms were completed at recruitment, with the VLU-QoL being completed 48–72 and 8 weeks later. An item-response model was also applied. The final questionnaire included 17 items adapted from the Skindex and 17 VLU-specific items. On factor analysis the items fell into three domains. Two domains had particularly high internal consistency (Cronbach’s a >0.9). High test–retest reliability was also shown (intra class correlation coefficients >0.8, p < 0.001). Convergent and divergent construct validity was shown. Instrument sensitivity was demonstrated in relation to scores on clinical measures, with responsiveness shown by relationship to symptom severity improvement and change in ulcer diameter. An item response model showed that most questions functioned well in relation to the underlying construct. The VLU-QoL is disease specific QoL instrument which has good psychometric properties. The data supports the use of the VLU-QoL for the evaluation of outcomes for patients with venous leg ulcers.

862 # 1012/THE CONSTRUCTION AND PRELIMINARY VALIDATION OF A SCALE TO MEASURE QUALITY OF LIFE IN PATIENTS WITH FAECAL INCONTINENCE IN INFLAMMATORY BOWEL DISEASE Chris Speed & Elaine MCColl, Centre For Health Services Research, University of Newcastle Upon Tyne, Newcastle Upon Tyne; Mark Welfare, Regional School of Medicine, North Tyneside General Hospital, North Shields, Tyneside, UK We report on the development and preliminary validation of a new scale – Quality of Life in Faecal Incontinence (QoLiFI) to measure the impact of Faecal Incontinence (FI) in patients with Inflammatory Bowel Disease (IBD). FI has been defined as the involuntary or inappropriate passage of faeces. Here we have used a broader definition to include the involuntary passage of liquid matter or gas (anal incontinence). There is a misconception that FI results only from neurological disorders or obstetric trauma. However, FI is a significant feature of IBD, highlighted in studies of rectal pathophysiology, yet it is under-reported by patients and rarely asked about by clinicians. We adopted a patientcentred, multi-method approach. 553 IBD patients responded to a postal survey to assess the prevalence and severity of FI. Results indicate 87% of respondents had experienced FI in adulthood and 71% had experienced FI in the 4 weeks up to completion of the questionnaire. A purposive sample took part in in-depth and semi-structured interviews designed to inform the understanding of the fears and concerns of living with FI and the impact of the condition upon life and lifestyle. Emergent themes were drafted into questionnaire items which were rated by participants (n ¼ 79) for applicability and importance. Following initial item reduction, factor analysis yielded a unitary 36 item scale. Due to poor face validity one item was removed from the final scale. Cronbach’s a ¼ 0.97 suggested redundancy and a shorter 18 item scale (face validity being the selection criterion) was constructed (Cronbach’s a ¼ 0.96). The test–retest reliability coefficient for the 35 and 18 item scales was 0.93 and 0.91 respectively suggesting a high level of stability over time. Further analyses show a strong positive correlation between the QoLiFI scale, severity and frequency of FI, lifestyle alteration due to FI and worry about FI. Existing scales e.g., Hospital Anxiety and Depression Scale produce similar moderate to good correlations with the QoLiFI items. # 1833/QUALITY OF WORKING LIFE: DEVELOPMENT AND VALIDATION OF A MEASURING INSTRUMENT FOR NURSES Dirley M. Carandina & Miako Kimura, Department of Medical Surgical Nursing, School of Nursing, University of Sa˜o Paulo, Sa˜o Paulo, Brazil The objective of this study was to develop and validate an instrument for measuring the quality of working life of nurses. This is a methodological study using the theoretical, empirical and analytical procedures proposed by Pasquali (1998), for psychological scale development. Theoretical procedures included the literature review on the subject and, as a starting point, the adoption of an existing model proposed by Eda Conte Fernandes (1996). In this initial phase, a qualitative research with a sample of 70 nurses was also performed to identify specific elements of the construct for nurses. Bardin’s method of content analysis were used. A pilot instrument was obtained by adding the specific elements to the Fernandes model. This instrument was submitted to a panel of eight judges for analysis of semantic and conceptual adequacy of the items (content validity). The instrument was then pre-tested in a sample of 50 nurses. Empirical procedures included a quantitative research for gathering empirical evidence of the psychometric properties of the instrument, which were explored in a sample of 348 nurses. Both studies (qualitative and quantitative) were performed in four hospitals in Sa˜o Paulo city – 2 public and 2 private. Analytical procedures included the statistical analysis to test the reliability and validity of the instrument. Internal consistency reliability was estimated through Cronbach’s a. Construct, convergent, discriminant and concurrent validity were analysed. This study allowed the construction of an instrument composed by 65 items and 13 domains, achieving highly satisfactory validity and reliabiliy (a ¼ 0.93) for the total items and above 0.92 for the domains.

# 1744/DEVELOPMENT OF THE HERPES OUTBREAK IMPACT QUESTIONNAIRE (HOIQ) Lynda C. Doward, Stephen P. McKenna & David M. Meads, Galen Research, Manchester, UK Purpose: The RGHQoL questionnaire has been employed in the assessment of quality of life (QoL) in patients with recurrent genital herpes (RGH); it is concerned with the impact of the virus on the everyday life of patients. No measure has been available for assessing impact of a herpes outbreak on patients with the condition. Some interventions for the condition attempt to abort an outbreak or to reduce its length and/or severity. An instrument was required to determine the benefit of such interventions from the patients’ perspective, in the context of a clinical trial. Methods: Content of the HOIQ was derived from 40 qualitative interviews with RGH patients who were asked to describe symptoms and functional limitations imposed by a herpes outbreak. Nineteen cognitive debriefing interviews were conducted with RGH patients, employing the draft HOIQ. Finally, the measure was employed in a web-based survey involving 158 patients in order to identify the final set of items and to establish unidimensionality, reproducibility, internal consistency and construct validity. Results: Interviews identified a number of areas of functioning that were affected during an outbreak; social, work and other activities, relationships, personal hygiene, choice of clothing, need to avoid becoming tired, worry and anxiety and fear of discovery. De-briefing interviews indicated that the content was appropriate for this patient group and that the measure was easy to understand and complete. Following Rasch analysis of the survey data a 12-item unidimensional measure was identified with good reproducibility, internal consistency and construct validity. Conclusions: The HOIQ represents an effective method for determining the impact of a herpes outbreak. The measure is scheduled to be used on a daily basis during such outbreaks in a trial designed to determine the effectiveness of a new treatment for RGH. # 1469/DEVELOPMENT AND VALIDATION OF QUALITY OF LIFE MEASUREMENT IN PRECARIOUSNESS: QoL-P18 T. Iordanova, Perceived Health Research Unit, Public Health Department; P. Jean, A. Chevalier, M. Declemy, R. Dubuisson, G. Pedrant, A. Krajevitch, H. Liauthaud, J.M. Minguet, P. Olivier, R. Rouvier, D. Debensason, S. Robitail & P. Auquier, Public Health Department, School of Medecine, Marseille, France Precariousness is a major public health problem. Precarious people in France present a low health status albeit they are recurrent users of health services and notably of emergency departments (ED). Probably this population is not well identified by the health system. The assessment of perceived health could allow a better identification and by the mean an improvement of health care specific to this population. None tool is available for this purpose in the international literature. We report the development and the validation of the first specific instrument: the Quality of Life questionnaire in Precariousness (QoL-P18). We have conducted 80 audio-taped interviews with patients in 4 ED (different subgroups of ICD, acute and chronic illness, migrant...). Content analysis (classical and computerised process) isolated 141 items. Item selection (300 patients) led to 62 items. Validation study was conducted with 720 in- and out-patients, in 10 randomised ED in south east of France. Patients fulfilled QoL-P, Duke Health Profile (DHP), demographic and socio economic features, visual analog scales (VAS) assessing major domains of life issued from literature. Physicians reported a short clinical, social and psychological evaluation. Parallel analyses and exploratory factor analysis with varimax rotation were performed. Six factors explained 72% of the variance: Social relationships (S), Psychological well-being (P), Self-esteem (SE), Autonomy (A), Perception of the future (F), Access to care (AC) and 18 items were retained. Reliability was satisfactory with Cronbach’s a coefficients ranging from 0.70 to 0.85. Major differences were found between precarious subjects vs. well-know nonprecarious subjects for: S, P, AC (p < 0.001). Missing data were low (<9%). Criterion validity of QOL-P18 (correlations with 9 VAS and DHP) is satisfactory.

863 # 1509/DESIGN OF AN INDIVIDUALISED MEASURE OF THE IMPACT OF DIABETES ON THE QUALITY OF LIFE OF ELDERLY PEOPLE: THE ADDQoL SENIOR Jane Speight, Alison Woodcock, Rosalind Plowright & Clare Bradley, Department of Psychology, Royal Holloway, University of London, Egham, Surrey, UK

# 1712/NATURAL HISTORY OF PSYCHOSOCIAL ADJUSTMENT TO CHRONIC KIDNEY DISEASE: FROM PREDIALYSIS TO ENDSTAGE RENAL DISEASE Gerald M. Devins, Psychiatry, University of Toronto, Toronto, Ontario, Canada; Yitzchak M. Binik, Psychology, McGill University, Montreal, Quebec, Canada

Diabetes affects about 5% of people over the age of 65 (20% over 85) in the UK. The ADDQoL measure of the impact of diabetes on the quality of life (QoL) of adults has been shown to have good evidence for validity, reliability and sensitivity to change. Development of similar questionnaires to measure the impact of macular disease (MacDQoL) and diabetic retinopathy (RetDQoL) have suggested that the ADDQoL may be improved to facilitate completion by elderly people. Modifications to the pilot ADDQoL Senior layout included: enlarged font; vertical format response options; one domain per page. Positive impact options (i.e. QoL improved by diabetes) are rarely used compared with negative impact options, so only one level of ‘positive’ impact was included. Leader questions (e.g. ‘do you have family/relatives?’) replaced ADDQoL ‘not applicable’ options. Changes to content included: the ‘holidays or leisure activities’ domain was split into two items; ‘spiritual/religious life’ was added; ‘working life’ was removed. The ADDQoL Senior was pilot-tested during tape-recorded interviews with one man and three women with Type 2 diabetes (age: 69–91 years) in a residential care home in England. Three interviewees could not read the ADDQoL Senior, due to impaired vision, but completed it easily when read aloud. Their comments led to several important wording changes to aid comprehension and relevance e.g.: ‘leisure activities’ became ‘interests and pastimes’; ‘close personal relationship’ was removed; ‘relationships with the people I see most days’ was added. The resulting ADDQoL Senior now includes two overview items (‘present QoL’ and ‘diabetes-specific QoL’) and 17 domain-specific items relevant to older people with diabetes. Its psychometric properties will be evaluated in a study of care home residents currently underway.

Chronic kidney disease (CKD) often progresses to end-stage renal disease (ESRD) necessitating renal replacement therapies (RRTs) (e.g., hemodialysis). Progression of CKD varies widely. Patients are often unaware that maintenance RRT will be required until it must be initiated on an emergency basis, compounding the stresses associated with the onset of a chronic life-threatening disease. We studied a cohort of 406 CKD patients with (n ¼ 206) and without (n ¼ 200) forewarning of the impending onset of ESRD. Patients were sampled from 5 Nephrology centers in Montreal and Calgary, Canada. Measures of psychosocial adjustment included: (a) psychological wellbeing (Affect Balance Scale); (b) quality of life (Cantril Ladder); (c) life happiness (Atkinson Life Happiness Rating); (d) depressive symptoms (Center for Epidemiologic Studies Depression Scale); (e) marital happiness (Locke-Wallace Marital Adjustment Test); and (f) positive and negative effects of CKD on diverse domains of life experience (Positive and Negative Effects Scale). Data were gathered in structured interviews by trained interviewers at six occasions: (a) predialysis (for those with forewarning) or after initiation of dialysis (no forewarning); (b) 18 months following the onset of ESRD; (c) 30 months post-ESRD onset; (d) 42 months post-ESRD onset; (e) 54 months post-ESRD onset; and (f) 66 months post-ESRD onset. Analyses were undertaken both cross-sectionally (including all data available at each wave) and longitudinally (including only data available across all waves). Overall, forewarning provided a statistically significant but modest advantage in facilitating psychosocial adjustment to the onset of ESRD and RRT. Benefits were most notable early in the course of disease and treatment and attenuated with increasing duration of survival. Results based on longitudinal analyses corroborated but were less striking than those indicated by cross-sectional analyses.

# 1078/RELIABILITY AND VALIDITY OF THE TREATMENT FLEXIBILITY SCALE Clarice P. Hayes & Lee Bowman, Global Health Outcomes Research, Eli Lilly & Company, Indianapolis, IN, USA The Treatment Flexibility Scale (TFS) consists of 10 items designed to assess the extent to which respondents with diabetes perceive they have choice in various aspects of their life (e.g., planning social activities) that potentially could be impacted by their diabetes regimen. This study examined the reliability and validity of the TFS in patients with type 2 diabetes (n ¼ 1640, mean age ¼ 59.7 years) in Germany, Japan, United Kingdom, or United States. Patient treatments were 5% diet/exercise alone, 25% oral monotherapy, 37% oral combination therapy (2 or 3 orals), 17% oral and insulin, 16% insulin monotherapy. Cronbach’s a for the TFS across countries ranged from 0.89 to 0.91. Exploratory factor analysis, using varimax rotation for TFS items and six items of a diabetes treatment satisfaction questionnaire, resulted in three factors with eigenvalues >1, accounting for 64–68% of the variance in country-specific analyses. Item loadings on factors indicated treatment flexibility, as measured by the TFS, is a separate construct from treatment satisfaction. The TFS consists of two factors, choice in food intake and choice in daily activities. Analysis of variance showed significant differences (p < 0.05) in treatment flexibility across treatment groups. Patients treated with insulin monotherapy perceived significantly less treatment flexibility, particularly in daily activity, than patients treated with diet/exercise alone or oral therapy. Patients treated with both insulin and orals perceived significantly less treatment flexibility than patients treated with diet/exercise alone. Patientreported outcomes have increasingly become acknowledged as important to include in the evaluation of diabetes treatment. Treatment flexibility is a dimension that is distinctive from treatment satisfaction. As a measure of treatment flexibility, the TFS demonstrates internal consistency and discriminant validity.

# 1633/QUALITY OF LIFE FOR PATIENTS WITH CHRONICLE RENAL FAILURE: DIFFERENCES FOR TREATMENT Angela Magaz, Research Department, CALVIDA Foundation, Baracaldo, Basque Country; Carmen Bragado, Clinical Psychology Department, Universidad Complutense, Madrid, Spain Research about people’s quality of life with Chronicle Renal Failure (CRF) has showed the existence of different psychological effects related to the type of treatment received. For it, part of the psychological research was focused on people who had received Renal Replacement Therapy (RRT) and centred on two factors: (a) to analyse the psychological consequences depending on the treatment and (b) to identify the psychosocial factors which could influence in a better adaptation of the patients to the illness and, their treatments. In this study, three psychological factors have been analysed (anxiety, depression and, health’s perception) in three groups of patients: transplant patients (TS), patients in haemodialysis (HD) and, patients in peritoneal dialysis (PD). The final purpose of it was to obtain information that allows us to develop intervention protocols to improve the well being of patients and the effectiveness of the employed treatments. The sample was comprised by 392 people who had received dialysis (177 in HD; 47 in PD) or who had received transplants (164 TS). All of them from dialysis centres in the Basque Country. The results obtained showed the existence of psychological differences in function of the treatment received. The patients in dialysis (HD and PD) showed bigger levels of anxiety than the transplanted patients, and among the patients in dialysis, those that were in HD showed bigger levels than those in PD. In the case of the depression variable, transplanted groups showed smaller levels than groups in dialysis. These same results were obtained for health’s perception, manifesting transplant patients a better perception of health. Confirming the existence of psychological differences in function of the treatments received is a prior step to identify risk factors and, being based on them, to develop preventive programs and intervention oriented to facilitate the adaptation to the illness and each specific treatment.

864 # 1769/THE ASSOCIATION BETWEEN OVERACTIVE BLADDER (OAB) SYMPTOMS AND HEALTH RELATED QUALITY OF LIFE (HRQoL) Louise E. Allen, Andreas M. Pleil, Pfizer Global Research & Development, High Wycombe; Con J. Kelleher, Guy’s & St. Thomas’ Hospital Trust, London, UK; Pat R. Reese & Terra L. Slaton, CareScience Inc., RTP, North Carolina Overactive bladder (OAB) is defined by symptoms of frequency, urgency and nocturia, with or without urge incontinence. Other urinary symptoms may also be present. The degree to which these symptoms impact on patient reported HRQoL is an important consideration in evaluating treatment outcomes. This work describes the association between OAB symptoms and HRQoL, as measured by the King’s Health Questionnaire (KHQ). Data from a 12-week double-blind safety and efficacy trial and a 12-month open-label extension were used for these analyses. All patients participating in the extension trial were included (n ¼ 838). The KHQ, a multidimensional disease-specific measure for patients with OAB, is scored on a 0 (best) to 100 (worst), scale. The impact of urinary symptoms was based on the self-reported symptom severity checklist scored from 0 (no reported effect) to 3 (a lot of effect). Each of the 9 KHQ domains were regressed onto three sets of independent variables (1) nocturia only (2) OAB symptoms (nocturia, frequency, urgency, and urge incontinence) and (3) all OAB symptoms plus the six remaining items from the symptom checklist. These equations were estimated for both baseline and end-of-treatment scores. Symptoms explained between 7 and 47% of the variance in KHQ domain scores. OAB specific symptoms contributed much of the explanatory power and were consistently predictive for the physical domains. Nocturia was particularly associated with the Sleep/Energy domain (adj. R2 0.26 and 0.34) and adding other symptoms did little to increase the explained variance. Nocturia was the most predictive of the 4 OAB symptoms for the general health perceptions domain. OAB symptoms impact on patients’ HRQoL as measured by the KHQ. Whereas the OAB symptoms of frequency, urge incontinence and urgency tend to explain dierences in the physical domains of the KHQ, nocturia is unique in that it is explanatory of the Sleep/Energy domain score variance. # 1151/QUALITY OF LIFE IN PATIENTS WITH RENAL TRANSPLANTATION Kuniyasu Ikeoka, Chiaki Nakauchi, Hiroshi Ishii, Masamichi Yano, Youichi Kakuta, Shirou Takahara, Haruyasu Fujita, Kozaburo Adachi, Kanehisa Morimoto & Takashi Mandai, Japanese Society of Quality of Life Research, Kobe, Hyogo, Japan The purpose of this study was to evaluate the quality of life (QOL) in patients with renal transplantation. One hundred and six patients with renal transplantation participated in this study. Self administered questionnaire including 40 questions divided into 15 categories with the Life Satisfaction Index (LSI) were used. Cronbach’s a coefficients of our questionnaire were high enough to accept for clinical use: 0.93 in mental function, 0.87 in social participation, 0.86 in medical service and environmental problem, and 0.82 in dietary problems, respectively. Our questionnaire contained 11 main factors and cumulative contribution was 0.77. Pearson’s correlation coefficients between our questionnaire and the LSI were r ¼ 0.66 (p < 0.01) before renal transplantation, and r ¼ 0.69 (p < 0.01) after renal transplantation. After renal transplantation, 89 patients indicated the improved total

QOLs, but 17 patients showed the deteriolated total QOLs. Significant improvements were shown in strata of well-being, dietary problems, sleep, mental function, physical function, medical service, social participation, environmental problem, and work performance (p < 0.01) after renal transplantation. Renal transplantation showed the significant total QOL improvement (p < 0.01). Furthermore the patients with living renal transplantation demonstrated the significant excellent QOLs in strata of well-being, sleep, psychological problem, sexual life, environmental problem (p < 0.05) and physical function (p < 0.01) compared with the ones with cadaveric renal transplantation. Compared with the patients with short hemodialysis period, the patients with long one demonstrated more desirable effects on QOL after renal transplantation. These findings indicate that our questionnaire has high enough reliability and potency of validity to use for the patients with renal transplantation. Although the recent progress of renal transplantation plays very important roles to improve the QOL’s of patients with renal failure, we must pay more attention to the mental burden or anxiety due to renal transplantation as the mental or social support system besides medical service, too. # 1624/EXPLICATIVE MODELS OF QUALITY OF LIFE IN CHRONICLE RENAL FAILURE PATIENTS Angela Magaz, Research Department, CALVIDA Foundation, Baracaldo, Basque Country; Carmen Bragado, Clinical Psychology Department, Universidad Complutense, Madrid, Spain The study of psychosocial factors that participate in the quality of live of patients with Chronicle Renal Failure is one of the researches that have attracted more attention now. The knowledge of these factors is very useful in the process of making protocols of intervention that allow the patients to achieve a better acceptance to the treatment. Low levels of anxiety and depression, and a good perception of health show the patients good acceptance to the treatment. The knowledge of those variables that could create a negative effect on this acceptance would allow us to identify the risk factors and, based on them, to develop preventive programs orientated to facilitate the acceptance of the illness and the treatment. The main aim of this study was to analyse which variables could be affecting these indicators of acceptance of the illness and each of the treatments regularly used as Renal Replacement Therapy (RRT). The sample comprised 392 people who had received RRT (177 in Haemodialysis; 47 in Peritoneal Dialysis and 164 Transplanted) residents in the Basque country (CAPV). The analysed variables can be grouped in two main categories: medical variables (e.g.; creatinina, urea, haemoglobin, symptoms…) and psychosocial variables (e.g.; social support, control perception, stressing factors…). The results obtained show that factors such as perception of physical symptoms, high creatinine levels, stressing factors and negative health impact, affect negatively on anxiety levels; low haemoglobin, and old age affect on depression levels; and the evidence of physical symptoms, years of treatment, stressing factors and old age affect negatively on health’s perception. The most relevant contribution of the results obtained in this study is that the physical symptoms the patients perceive are those constant factors in the explained samples of anxiety, depression and negative health perception. Therefore a preventive strategy of these problems affecting the patients’ quality of live, is to choose the treatment based on the forecasted symptoms and give the patients accurate advise on self care and the way to deal with symptoms.

865 # 1043/BASELINE FINDINGS FROM THE SUBMACULAR SURGERY TRIALS VISION PREFERENCE VALUE SCALE (SST-VPVS) Marta J. Marsh & Eric B. Bass, Baltimore, MD, USA We sought to measure preference values (PVs) that patients assign to their health and vision status to improve our understanding of the impact of subfoveal choroidal neovascularization (CNV) on HealthRelated Quality of Life (HRQoL). Patients enrolled in the Submacular Surgery Trials (SST) from April 1997 to September 2001 and completed centrally administered HRQoL telephone interviews prior to randomization. Interviews consisted of the National Eye Institute-Visual Functioning Questionnaire (NEI-VFQ), SF-36, and Hospital Anxiety and Depression Scale (HADS). In October 1997 the 3-question SST vision preference value scale (SST-VPVS) was added which asked patients to rate (1) their vision from 0 (blind) to 100 (perfect), (2) their current health from 0 (dead) to 100 (perfect) assuming they were blind, and (3) their current health from 0 (dead) to 100 (perfect) assuming they had perfect vision. A single preference value was calculated and converted to a 0 to 1 scale. Differences between patients with and without usable PVs were assessed. Spearman correlation (S) and Kruskal Wallis (KW) tests were used to find associations between PVs and covariates. Baseline HRQoL interviews including PV questions are available for 996 patients. Eighty percent have usable PVs. Patients without usable values were older than those with usable values (mean 76 vs. 71 years) and had worse visual acuity in their better seeing eye (median 20/50 vs. 20/32) but did not differ in gender or race. The mean PV was 0.64 (median ¼ 0.68, inter-quartile range ¼ 0.50–0.80). PVs correlated highly with self rating of overall health (S ¼ 0.36) and overall vision (S ¼ 0.47). PVs were associated (p < 0.01) with visual acuity, CNV lesion size, anxiety, depression, SF-36 Mental and Physical Component Summary Scores, hypertension, and smoking status, but not age, gender, or race (p > 0.05). Patients with subfoveal CNV have low PVs, indicating a profound impact on how they feel about their HRQoL. The SST-VPVS is useful in measuring PVs in this visually impaired population. # 1188/MEASURING QUALITY OF LIFE IN MACULAR DISEASE: WHAT USE ARE UTILITIES? Jan Mitchell & Clare Bradley, Department of Psychology, Royal Holloway, University of London, Egham, Surrey, UK Macular disease (MD) is a degenerative eye condition and the leading cause of blindness in the Western world. The study compared visionspecific and health time trade-off (TTO) indicators with a single-item measure of perceived health (from SF-36) and a present quality of life (QoL) item and MD-specific QoL item (from the MacDQoL individualised measure of impact of MD on QoL). One hundred and fifty members of the UK MD Society (mean age 79 years, 69% women) completed a questionnaire either by telephone interview or self-completion. Findings did not differ for the two administration groups. The MD-specific item was sensitive to severity of MD (measured by registration as blind, partially sighted or not registered) (Kruskal Wallis: v2 ¼ 25.94, df ¼ 2, p < 0.001). The present QoL item correlated with the perceived health item (r ¼ 0.583, p < 0.001), but not with severity of MD. Only 114 (76%) completed the TTO questions compared with 148 (99%) for other items. Reasons for non-completion included finding the questions disturbing, unrealistic, laughable, and violating religious or other principles. Of those who answered, 43 (39%) said they would not give up any years for perfect vision. In this group, proportions of blind, partially sighted and not registered did not differ from the entire sample. Respondents considered factors other than their QoL in calculating responses. Some cited the importance of their lives to others (e.g. spouse, grandchildren) or vice versa (e.g. would not give up any years while a spouse was alive), or their role as carers. Others felt unmotivated to regain perfect vision because their QoL was otherwise poor (e.g. due to loneliness or poor health). Unlike the MacDQoL MD-specific item, the vision-specific TTO question was not sensitive to severity of MD. Unlike the MacDQoL present QoL item, the health TTO question did not correlate with perceived health (r ¼ 0.174, n.s.). TTO utility values for the sample were: vision-specific ¼ 0.74; health ¼ 0.77. Neither TTO item correlated with the present QoL item. The data suggest that utility values provide no indication of QoL or health in people with MD.

# 1883/A COMPARISON OF UTILITY-BASED MEASURES OF HEALTH RELATED QUALITY OF LIFE COMPARED TO CONDITION-SPECIFIC MEASURES FOR PEOPLE WITH DEMENTIA Gillian R. Currie, Economics and Community Health Sciences; Marlene Reimer, Faculty of Nursing, University of Calgary, Calgary, AB, Canada; Cam Donaldson, School of Population & Health Sciences and Business, University of Newcastle upon Tyne, Newcastle upon Tyne, UK; Susan Slaughter, Faculty of Nursing; Michael Eliasziw, Departments of Clinical Neurosciences and Community, University of Calgary, Calgary, AB, Canada Purpose of research: The purpose of this study was to compare different tools for the measurement of quality of life (QoL) for people with dementia residing in continuing care environments. Both generic measures of QoL that allow the calculation of QALYs and conditionspecific measures were used in the study. The QoL tools included both some proxy measures as well as others that relied on direct observation of residents. Subject sample and statement of methods: We collected data on quality of life (QoL) for people with middle to late stage dementia residing in continuing care facilities as part of an controlled study evaluating costs and QoL for residents. The study compared those residing in a purpose-built special care facility (SCF) with traditional care units in Calgary, Alberta. The research design was a prospective cohort study with category matching on dementia severity and age-adjusted comorbities. QoL instruments were administered at baseline and repeated every 3 months over a 1 year period. There were 62 residents in the intervention group residing in the special care facility and two comparison groups representing usual care that consisted of 64 and 59 residents respectively. Summary of results: All residents had Global Deterioration Scale scores greater than 5, ranged in age from 61 to 94 years and 75% were female. A variety of QoL measures were used in this study including Euroquol and the Health Utilities Index. In addition, condition specific measures of different domains of QoL including the Functional Assessment Staging, Apparent Affect scale, Cohen-Mansfield Agitation Inventory, Brief Cognitive Rating Scale, Pleasant Events Schedule and the Multidimensional Obervation Scale of Elderly Subjects. The QoL outcomes across groups and over time will be presented and the measures compared. Special challenges related to the measurement of QoL for this population will be discussed. # 1354/THE MISCONCEIVED NATURE OF THE QUALITY-ADJUSTED LIFE YEAR (QALY) CALCULATION Luis Prieto & Jose´ A. Sacrista´n, Health Outcomes Research Unite, Eli Lilly & Co., Alcobendas, Madrid, Spain Objectives: The quality-adjusted life-year (QALY) is a measure of the value of health outcomes. Since health is a function of length of life and quality of life (QoL), the QALY was developed as an attempt to combine the value of these attributes into a single index number. Nevertheless, QALYs have been criticised on technical and ethical grounds. A salient problem relies on the numerical nature of its constituent parts. The appropriateness of the QALY arithmetical operation is compromised by the essence of the utility scale: while life-years are expressed in a ratio scale with a true zero, the utility is an interval scale where 0 is an arbitrary value for death. In order to be able to obtain coherent results, both scales would have to be expressed in the same units of measurement. The different nature of these two factors jeopardises the meaning and interpretation of QALYs. Methods: A simple general linear transformation of the utility scale suffices to demonstrate that the results of the multiplication are not invariant. Results: Mathematically, the solution to these limitations happens through an alternative calculation of QALYs by means of operations with complex numbers rooted in the well known Pythagorean theorem. Conclusions: Through a series of examples, the new calculation arithmetic is introduced and discussed.

866 # 1083/HEALTH UTILITY VALUES FOR THE AUSTRIAN VERSION OF THE EQ-5D: RESULTS AND CHALLENGES Stefan Ho¨fer, Psychology, Royal College of Surgeons in Ireland, Dublin, Ireland; Bianca E. Noisternig, Gerhard Rumpold & Gerhard Schu¨ßler, Medical Psychology and Psychotherapy, Univeristy of Innsbruck, Innsbruck, Austria There are several cultural sets of health utility (HU) values for the EQ5D available. It is recommended to carefully assess HU indices for each country as no overall European EQ-5D index has yet been identified. The purpose of this study was to evaluate the 243-health states represented by the EQ-5D in an Austrian sample. Randomly eleven out of the possible 243-health states were chosen to cover the whole range of states. A sample of 161 healthy persons (40.6% male, mean age 24.3 ± 5 SD) received the self-administered EQ-5D. Also a Visual Analog Scale (VAS) and the Standard Gamble (SG) method were used to assess the HU values of the 11 health states. A trained interviewer guided the subjects in groups of max. Ten people through the valuation process. OLS-Regression was used to generate the Austrian EQ-5D index. The mean of the five EQ-5D scales ranged from 1.0 (self-care) to 1.24 (anxiety/depression), and with a mean score of 87.3 (±9.6 SD) for the EQ-5D VAS. Both the additional VAS and SG assessment showed a high consistency regarding the relative ranking of the different health states. However there was a discrepancy for the absolute values depending on the method used. HU values were consistently lower when assessed with VAS compared to SG (e.g.: 21323VAS ¼ 35.7 vs. 21323SG ¼ 74.4). Compared to the set of HU values of the German version, there were minor dierences for the HU values generated by VAS. As there were no German SG values available, the Austrian SG values were compared with the German Time Trade O values showing major dierences. In conclusion the order of the HU values generated either by VAS or SG were similar. However using SG participants were less likely to take the risk of treatment and therefore assigned the health status higher values. This study has contributed to the availability of HU values for Austria by using a healthy sample. Ongoing research of EQ-5D valuation in chronic disease will provide further data on the Austrian EQ-5D values. # 1711/HEALTH STATE VALUES FOR THE HUI2: RESULTS FROM A UK SURVEY Christopher J. McCabe, Katherine Stevens, Jennifer Roberts & John E. Brazier, School of Health and Related Research, University of Sheffield, Sheffield, England, UK This paper reports the results of a study to estimate a statistical health state valuation model for a revised version of the Health Utilities Index Mark 2, using Standard Gamble health state preference data. A sample of 51 health states were valued by a sample of the 198 members of the UK general population. Models are estimated for predicting health state valuations for all 8000 states defined by the

revised HUI2. The recommended model produces logical and significant coefficients for all levels of all dimensions in the HUI2. These coefficients appear to be robust across model specifications. This model performs well in predicting the observed health state values within the valuation sample and for a separate validation sample of health states. However, there are concerns over prediction errors for two health states in the valuation sample. These problems must be balanced against concerns over the validity of using the VAS based health state valuation data of the original HUI2 valuation model. # 1268/PARENT RATINGS OF CHILDREN WITH ADHD: COMPARING STANDARD GAMBLE UTILITY RATINGS WITH A MULTI-DOMAIN QUALITY OF LIFE MEASURE AND THE EQ-5D VISUAL ANALOGUE SCALE Kristina Secnik, Global Health Outcomes, Eli Lilly and Company, Indianapolis, IN; Louis Matza, Medtap International, Bethesda, MD, USA; Anne Rentz, Sindelfingen, Germany; Floyd R. Sallee, Department of Psychiatry, University of Cincinnati, Cincinnati, OH, USA Purpose of research: Attention deficit/hyperactivity disorder (ADHD) is a behavioural disorder originating in early childhood, with a high risk for continued symptoms into adolescence and adulthood. ADHD has been shown to impair quality of life (QOL) in studies using multidimensional questionnaires. However parent-based utility measurement has not been used to assess children with ADHD. Health state utilities are needed to calculate quality-adjusted life years (QALYs), a critical measure of outcomes in cost-effectiveness studies. The purpose of this study was to assess parents’ utility ratings of their children’s current health and to validate this utility methodology for ADHD by comparing these scores to two previously validated QOL measures. Sample and methods: Parents with a child with ADHD were recruited from a US child psychiatric clinic. All parents completed the standard gamble utility assessment, the proxy version of the EQ-5D including the Visual Analogue Scale (VAS), and the 50-item Child Health Questionnaire (CHQ-PF50). Summary of results: 43 parents of children with ADHD between the ages of 5–17 participated in the study. Mean parent ratings of their child’s current health state was estimated to be 0.74 using standard gamble (range from 0 to 1) and 77 when measured using the EQ-5D VAS (range from 1 to 100). The standard gamble utility score was significantly correlated with psychosocial domains of the CHQ (range: 0.32–0.44), the EQ-5D VAS (0.40), and the EQ-5D index score (0.38). The utility score was not significantly correlated with CHQ physical domains. In sum, parent-reported measures were used to assess the QOL of children with ADHD. Standard gamble utility ratings of current health demonstrated construct validity through significant correlations with the EQ-5D VAS and the psychosocial domains of the CHQ. Parent-proxy standard gamble ratings appear to be a valid method for obtaining utility scores that can be used in cost-effectiveness models of drug treatments for ADHD.