ORIGINAL RESEARCH ARTICLE

Patient 2012; 5 (1): 33-44 1178-1653/12/0001-0033/$49.95/0

ª 2012 Adis Data Information BV. All rights reserved.

Patient Preferences for First-Line Oral Treatment for Mild-to-Moderate Ulcerative Colitis A Discrete-Choice Experiment Paul Hodgkins,1 Paul Swinburn,2 Dory Solomon,3 Linnette Yen,1 Sarah Dewilde2 and Andrew Lloyd2 1 Global Health Economics & Outcomes Research, Shire Pharmaceuticals, Wayne, PA, USA 2 Patient Reported Outcomes, Oxford Outcomes Ltd, Oxford, UK 3 Clinical Development and Medical Affairs, Shire Pharmaceuticals, Wayne, PA, USA

Abstract

Background: Patients with ulcerative colitis (UC) frequently require long-term therapy to prevent relapse. Treatments such as 5-aminosalicylic acid (5-ASA [mesalazine]) are efficacious and well tolerated, but adherence to treatment is often poor. Objective: This discrete-choice experiment (DCE) was conducted to estimate differences in patient preferences for 5-ASA treatment in mild-to-moderate UC based on levels of self-reported adherence. Inclusion of patients residing in the US, UK, Germany, and Canada allowed for assessment of possible cultural differences in patient preferences. Methods: DCE attributes were determined through literature review, clinician consultation, and patient interviews. Six treatment attributes were identified: ease of swallowing, time of day, quantity, extent of flare resolution, likelihood of flare occurrence, and cost. A total of 400 patients in four countries completed the DCE and adherence (Modified Morisky Scale) surveys. Data were analyzed using generalized estimating equations to estimate patient preference and willingness to pay (WTP) by levels of self-reported adherence and country of residence. Results: All attributes had expected polarity and were significant predictors of patient preference. Self-reported ‘good’ versus ‘poor’ adherers significantly preferred symptom control (p = 0.0108) and mucosal healing (p = 0.0190) attributes. All patients stated preference for symptom control/mucosal healing and flare risk attributes; the latter attribute was significantly preferred across all countries. Country differences in patient preference for convenience versus clinical attributes were found. Overall, patients were willing to pay d29.24 ($US46.27) per month for symptom control and mucosal healing, and an additional d78.81 ($US124.70) per month for reduction in flare risk to 10% per year (WTP costs were equalized between each country using the published

Hodgkins et al.

34

2008 purchasing power parity). Those with flares in the past year significantly preferred avoiding future flares (p < 0.0001) versus other attributes, as well as lower risk of flares (10%, likelihood ratio: 0.64–0.70). Conclusions: Findings indicate that self-reported adherers to UC therapy have a stronger preference for clinical benefits over other treatment attributes, suggesting that positive patient assessment of effectiveness may influence adherence. Ongoing clinician assessment of patient preferences for treatment attributes, as well as education on the importance of adherence, may help improve treatment outcomes in UC.

Key points for decision makers  Evidence suggests that patient adherence is a significant issue in ulcerative colitis that may lead to poor outcomes  This study, conducted in North America and Europe, generated data on patients’ strength of preference for aspects of treatments in ulcerative colitis and their self-reported treatment adherence  Study participants most valued the effectiveness of treatments (symptom control and risk of disease flare up) but geographical differences existed in the strength of preference for the convenience of taking the medication and their willingness to pay  This study suggests adherence behavior of a patient may be partly driven by the value participants place on the perceived benefits of treatments

Background Ulcerative colitis (UC) is a chronic inflammatory disease of the large intestine characterized by episodes of relapse and remission. The recommended first-line treatment for patients with mild-to-moderate UC is 5-aminosalicylic acid (5-ASA [mesalazine]). There are a variety of different 5-ASA delivery systems available that attempt to achieve maximal drug delivery to the colon while minimizing systemic absorption. Because of the chronic nature of UC, longterm drug therapy must continue on an ongoing basis to minimize the risk of symptomatic flares. Nevertheless, an increasing body of evidence suggests that a significant proportion of patients do not adhere to their prescribed therapeutic regimen.[1,2] Non-adherence to 5-ASA therapy is associated with a 5-fold increase in the risk of ª 2012 Adis Data Information BV. All rights reserved.

disease relapse and an increased risk of colorectal cancer.[3,4] Factors that influence adherence behavior in patients are complex and may be related to treatment, symptoms of disease, or physician-patient relationships. Treatment-related factors include side effects or the high cost of medication.[5] The extent and severity of disease, as well as the occurrence and intensity of flares, may also affect adherence to therapeutic regimens.[6,7] In addition, patients’ trust in their physician has been shown to be highly predictive of adherence to maintenance therapies.[7,8] Other findings suggest that reduced pill burden for quiescent disease leads to improved adherence, at least in the short term.[2] Studies of adherence need to consider the impact of sociocultural differences. Cultural preferences, sometimes reflected in lifestyles, can be directly observed to influence behavior, albeit not universally Patient 2012; 5 (1)

Patient Treatment Preferences for UC

reported.[9] Cultural factors can influence the doctor-patient relationship, which may influence adherence to therapeutic regimens.[10] Indeed, it is plausible that the nature of the healthcare delivery system itself in a given country may result in a disparity in patient attitudes and preferences when compared with other regions. Discrete-choice experiment (DCE) – sometimes referred to as conjoint analysis – is an attribute-based survey method for assessing preferences. DCEs involve presenting respondents with a sequence of hypothetical scenarios (choice sets) composed of two or more competing alternatives that vary along several attributes.[11] In recent years, DCEs have been increasingly utilized to help understand preferences in the field of health and healthcare.[12-17] By including cost as an attribute, willingness to pay (WTP) can be calculated to demonstrate value and strength of preference for other attributes assessed. Moreover, when adherence behavior is measured, differences in patient preferences can be assessed based on adherence or non-adherence. The key role of patient preferences in relation to treatment adherence has been recognized by both WHO[18] and the UK National Institute for Health and Clinical Excellence.[19] The present study was conducted to understand differences in patient preferences for oral 5-ASA therapies in mild-to-moderate UC based on self-reported adherence. Inclusion of patients residing in the US, UK, Germany, and Canada allowed for assessment of possible cultural differences between four countries across two continents. Methods Subject Recruitment

A total of 400 patients with mild-to-moderate UC were recruited to participate in this DCE from four study countries (n = 100 per country). Eligibility requirements included UC diagnosis (self-reported); age ‡18 years; residency in the US, UK, Germany, or Canada; and informed consent. Participants were identified by local independent third-party patient recruitment services who maintain databases of patients willing ª 2012 Adis Data Information BV. All rights reserved.

35

to contribute to research studies. The protocol was approved by an Independent Institutional Review Board (#IRB3563) prior to study initiation; all participants were required to provide informed consent to participate in this study. Survey Development

The selection of attributes, and levels per attribute, was based on the outcome of literature reviews and in-depth, semi-structured interviews with patients with UC. Initial interviews were conducted in the UK (n = 10). Additional interviews were conducted in Germany (n = 2) and the US (n = 2) to confirm findings and account for cultural differences in attitudes toward health and medication. Interviews were conducted in the local language by native-speaking trained interviewers, and audiorecorded for transcription and off-line analysis. Full interview transcripts were analyzed using the qualitative analysis software tool, ATLAS.ti (Scientific Software Limited, Berlin, Germany). Open-ended coding was used to establish thematic boundaries based on qualitative evidence. Findings were examined in a cultural context. Six attributes were identified for inclusion in the study (figure 1): ease of swallowing, time of day when the medication must be taken, quantity needed per administration, extent of flare resolution, likelihood of flare occurrence within 12 months, and cost of therapy. The number of levels per attribute was determined based on design of prior DCEs, as described in the literature.[15,16] A brief description or scenario was provided for each level by attribute to help participants select the appropriate preference level. A pilot test of the DCE survey was conducted, consisting of cognitive debriefing to confirm comparable understanding of survey questions and responses in geographies intended for this study (Canada, n = 8; Germany, n = 5; UK, n = 7; and US, n = 7). Based on comparable understanding of the survey instrument across geographies tested, we commenced with the currently reported study. Web-Based Data Collection

Cross-sectional anonymized survey data were collected from patients via a secure study website. Patient 2012; 5 (1)

Hodgkins et al.

36

Attribute 1.

Ease of swallowing.

Some people can find certain medications difficult to swallow whether because of their texture or their size. Rough or gritty treatments or medication which is large in size may cause problems with swallowing. This can make taking medication everyday unpleasant. Consider treatments that vary in how easy it is to take the medication. a. b. c. Attribute 2.

No problems swallowing Slight problems swallowing Some problems swallowing Number of doses/Time of day.

The time period(s) at which the medication must be taken regularly. a. b. c.

Attribute 3.

Only in the morning or only in the evening In the morning and in the evening In the morning, afternoon and evening

Quantity of medication needed.

The amount of medication that is taken at one time. a. b. c. Attribute 4.

One Two Three Treating your symptom flare.

Please imagine that you are currently experiencing a sudden worsening of symptoms (sometimes called a flare) and you need to start taking a new treatment. We will present you with a series of different treatments which vary in terms of the following features. Symptom flares can be associated with bleeding from the rectum (when you pass a stool) and an increase in the number of stools each day. The pain and discomfort of colitis is often worse during these periods. Treatments can stop this symptom flare. According to the treatment you choose you can expect to experience: a. b. c.

Attribute 5.

Improvement in symptoms (Fewer stools and/or less bleeding) Return to normal stool frequency and no bleeding Return to normal stool frequency, no bleeding and some healing of the inner lining of your bowel.

Risk of symptom flares.

Different treatments can help to prevent your next symptom flare if taken as prescribed. In the next twelve months, depending on the treatment you take, the risk of having a symptomatic flare may be different. According to the treatment you choose, your risk will be: a. b. c.

Attribute 6.

10% (1 in 10 patients will experience a flare within the next 12 months) 40% (4 in 10 patients will experience a flare within the next 12 months) 70% (7 in 10 patients will experience a flare within the next 12 months)

Cost. *

Different treatments may cost you more to receive. Please imagine that you are asked to pay the full cost each month in order to receive these new treatments. Think about how you could afford to pay this each month and what you may be willing to do without. a. b. c.

You pay £20 per month You pay £40 per month You pay £60 per month

*Presented in the appropriate regional currency and adjusted using the OECD purchase power parity figures for 2008.

Fig. 1. Attribute descriptions with their associated levels.

ª 2012 Adis Data Information BV. All rights reserved.

Patient 2012; 5 (1)

Patient Treatment Preferences for UC

37

Patients were instructed to self-administer three surveys, including the DCE survey (see figure 2 and the Appendix 1 in the Supplemental Digital Content [SDC], http://links.adisonline.com/PBZ/ A31), a background demographic and diseaserelated questionnaire, and the Modified Morisky Scale (MMS) – a self-reported measure of medication adherence.[20,21] The MMS consists of six questions on motivation and knowledge; scoring is based on each question answered with ‘No’ (score = 1). Total scores ranged from 0 to 6. Higher total scores predicted greater adherence to prescribed medication. Total scores from 0 to 3 indicated poor adherence. Total scores from 4 to 6 were considered good adherence.[20,21] Statistical Analysis

The statistical analysis was conducted using SAS 9.1 (SAS Software Ltd, proc GENMOD, SAS Institute Inc., Cary, NC, USA). A generalized estimating equations (GEE) model[22] was fitted with maximum likelihood using a binomial distribution with a logit link. This regression approach was selected given its appropriateness to DCEs where modeling of random and fixed effects is needed, as well as an array of distributions and repeated measures.[23] The model specification took the repeated nature of the data into account. The analysis started with a fully fitted model including all patient characteristics, disease characteristics, good and poor adherence

based on MMS total scores, scoring on the DCE survey by attribute, and interaction terms of MMS total score and DCE survey attribute scores. The model complexity was then reduced based on a likelihood ratio test to achieve parsimony while retaining the significant predictors of choice. For continuous variables, such as cost, linearity was assumed. Pooled analysis of data from all patients was conducted. Patient data from individual countries were analyzed separately using the same methodology. Odds ratios (OR) and WTP were derived for each model comparison. All tests of statistical significance were conducted with an alpha level of 0.05. Based on GEE model results, the following results on patient preference were tabulated: Importance of attributes. Attributes assessed from the DCE survey were considered important in determining treatment preferences based on likelihood statistics. Attributes were considered important to patients if they were statistically significant predictors of treatment choice based on OR and WTP. Strength of preferences. WTP was used as a proxy for strength of preferences and was calculated based on the ‘cost’ attribute assessed in the DCE survey. Attributes for which patients were willing to expend greater amounts of money were deemed more desirable. WTP was estimated by calculating the extent to which people were willing to trade cost of the drug against the other treatment attributes. This was achieved by taking the ratio of the parameter estimates for each

Treatment A

Treatment B

No problems swallowing In the morning and in the evening

Slight problems swallowing In the morning, afternoon and evening

Quantity needed each time medication is taken

Three

One

Treating your symptom flare

Return to normal stool frequency and no bleeding

Return to normal stool frequency, no bleeding and some healing of the inner lining of the bowel

Ease of swallowing Time of day

Risk of symptom flares

10% (1 in 10 chance)

40% (4 in 10 chance)

Cost

You pay £60 per month

You pay £20 per month

Which treatment do you prefer? Fig. 2. An example choice question from the survey.

ª 2012 Adis Data Information BV. All rights reserved.

Patient 2012; 5 (1)

Hodgkins et al.

38

Table I. Patient characteristics by study region Characteristic

Country

Significance level

Canada

Germany

UK

US 100

n

100

100

100

Mean age [y (SD)]a

47.2 (12.9)

32.1 (9.9)

42.8 (13.2)

47.0 (13.4)

F = 32.4, p < 0.001

Female (%)

73

58

65

72

w2 = 6.6, NS

Employed (%)

56

67

64

52

w2 = 6.0, NS

Education after age 18 y (%)

61

57

62

81

w2 = 15.2, p = 0.002

Time (mean) since diagnosis [y (SD)]

14.6 (11.5)

4.3 (5.9)

10.0 (11.1)

10.1 (9.4)

F = 18.9, p < 0.001 w2 = 20.9, NS

No. of flares in the past year (%)b 0

26

11

21

18

1

19

29

21

14

2

19

13

25

22

3

16

31

18

19

‡4

20

16

15

27

54

23

65

49

Concomitant tablet taking (%)b,c MMS-reported adherence (%)

a

w2 = 38.1, p < 0.001 w2 = 1.68, NS

b

good

57

63

65

59

poor

43

37

35

41

Post hoc comparisons of age indicated that each country was significantly different to every other country in terms of age profile, except for the US and Canada, who were not significantly different.

b

These values are self-reported.

c

Using medication for non-UC-related conditions or taking supplements.

MMS = Modified Morisky Scale; NS = not significant.

attribute to the parameter estimate of the cost attribute. The level of the cost attribute and income brackets was equalized between each country through the use of the published 2008 purchasing power parity (PPP) figures for each of the countries.[24] For consistency and ease of comparison, the WTP amounts for the combined analyses (all countries) are presented in d and $US. Influence of immediate prior clinical status. Additional interaction terms were included in the GEE model to assess the likelihood for immediate prior clinical status to predict patient preferences for treatment attributes. Results Patient Characteristics

Variations in patient characteristics were observed across the four countries (table I). Each country (apart from Canada and the US) was significantly different from every other country in mean age; the youngest patients were from Germany (mean age 32.1 years). Females and ª 2012 Adis Data Information BV. All rights reserved.

those employed represented the largest proportion of each patient sample by country (females: range 58–73%; employed: range 52–67%), but no significant differences were observed across countries for either group. More than 50% of patients across countries had completed some post-high school education, with the highest educational levels observed in the US (81%). Time since UC diagnosis ranged from 4.3 years in Germany to 14.6 years in Canada. The number of recent symptom flares was greatest in Germany and the US (47% and 46%, respectively, had experienced three or more in the past 12 months). Concomitant medication utilization was common in all countries except Germany, which had a reported 23% utilization. The majority of patients self-reported ‘good’ adherence, and these findings were similar across the four countries. Overall Population Analysis

All parameter estimates for attributes were in the expected/logical direction such that patients Patient 2012; 5 (1)

Patient Treatment Preferences for UC

39

were less likely to choose a treatment with a logically worse attribute level (table II). All attributes apart from the degree of difficulty experienced in swallowing were statistically significant predictors of patient choices. While ‘no problems’ in swallowing was a highly significant predictor of choice in the survey, no difference was found between ‘slight’ or ‘some’ problems. The attribute with the largest likelihood for patient preference was reduced flare risk of

10% compared with the reference case of 70% flare risk per year (OR 9.60; 95% CI 7.91, 11.66). Patients were willing to pay an additional d78.81 ($US124.70) per month for this attribute (table II). Symptom control and mucosal healing were also highly preferred. Patients were willing to pay d9.85 ($US15.59) for complete resolution of symptoms (compared with only partial relief) and an additional d19.39 ($US30.68) to a total of

Table II. Generalized estimating equations analysis of discrete-choice experiment attributes and willingness to pay (WTP) – overall study population (n = 400) Attribute and level

Estimate (SE)

p-Value

OR (95% CI)

Intercept

-0.7059 (0.1068)

<0.0001

no problems

0.1959 (0.0552)

0.0004

1.22 (1.09, 1.36)

slight problems

0.0486 (0.0556)

0.3815

1.05 (0.94, 1.17)

WTP (95% CI) £

$US

Swallowing

some problems

b

6.83 (3.05, 10.60)

10.81 (4.83, 16.77)

a

b

No. of doses/time of day morning

0.2141 (0.0553)

0.0001

1.24 (1.11, 1.38)

7.46 (3.69, 11.23)

11.80 (5.84, 17.77)

morning and evening

0.1293 (0.0536)

0.0158

1.14 (1.02, 1.26)

4.51 (0.85, 8.16)

7.14 (1.34, 12.91)

morning, afternoon, and evening

b

b

Quantity one tablet

0.2438 (0.0549)

<0.0001

1.28 (1.15, 1.42)

8.49 (4.75, 12.24)

13.43 (7.52, 19.37)

two tablets

0.1852 (0.0537)

0.0006

1.20 (1.08, 1.34)

6.45 (2.78, 10.12)

10.21 (4.40, 16.01)

three tablets

b

b

Symptoms back to normal and healing

0.8391 (0.0890)

<0.0001

2.31 (1.94, 2.76)

29.24 (23.16, 35.31)

46.27 (36.65, 55.87)

back to normal

0.2828 (0.0862)

0.0010

1.33 (1.12, 1.57)

9.85 (3.97, 15.74)

15.59 (6.28, 27.91)

some improvement

b

b

Flare risk (%) 10

2.2619 (0.0991)

<0.0001

9.60 (7.91, 11.66)

78.81 (72.04, 85.58)

124.70 (113.99, 135.41)

40

1.2089 (0.0936)

<0.0001

3.35 (2.79, 4.02)

42.12 (35.73, 48.51)

66.65 (56.53, 76.76)

70 Cost

b

b

-0.0287 (0.0014)

<0.0001

-0.1835 (0.0661)

0.0055

0.2812 (0.1104)

0.0108

0.2639 (0.1126)

0.0190

Adherence good Symptomsc good adherence back to normal Symptomsc good adherence back to normal and healing a

WTP cannot be calculated if the attribute is not a significant predictor of choice.

b

Values not calculated for reference level.

c

Denotes an interaction.

OR = odds ratio.

ª 2012 Adis Data Information BV. All rights reserved.

Patient 2012; 5 (1)

Hodgkins et al.

42.12 (35.73, 48.54) [66.65] a Importance is based on statistically significant WTPs in generalized estimating equation. b n = 100 for all countries. c Not statistically significant predictor of choice. WTP = willingness to pay.

78.81 (72.04, 85.58) [124.70] 59.82 (51.13, 68.51) 53.23 (46.01, 60.46)

27.09 (20.17, 34.00) 46.69 (36.41, 56.96) 122.56 (100.76, 144.37) 40

86.75 (75.82, 97.67) 214.27 (191.34, 237.19) 10

Flare risk (%)

ª 2012 Adis Data Information BV. All rights reserved.

d29.24 ($US46.27) per month for a treatment that led to mucosal healing (table II). In addition to clinical benefits, patients stated a preference for convenience attributes. Patients were willing to pay d6.45 ($US10.21) to reduce pill burden at each administration from three pills to two and d8.49 ($US13.43) for a reduction from three pills to one. Additionally, they were willing to pay d7.46 ($US11.80) to move from three administrations per day to just once in the morning. Patients also were willing to pay for treatments that did not cause difficulties in swallowing (d6.83 [$US10.81] per month). Patients who self-reported good adherence significantly preferred treatments that returned the stool frequency to normal (p = 0.0108) and promoted mucosal healing of the bowel (p = 0.019) versus other attributes tested. See the worked example in Appendix 2 of the SDC. A final model that illustrates the calculation of the odds of choosing a treatment with the aforementioned characteristics is also available as SDC 2 at http://links.adisonline.com/ PBZ/A33.

33.41 (25.29, 41.53)

29.24 (23.16, 35.35) [46.27]

9.85 (3.97, 15.74) [15.59] 15.07 (6.87, 23.28) 16.39 (9.40, 23.38)

32.75 (24.32, 41.17) 28.91 (21.70, 36.11) 53.60 (43.21, 63.99)

8.28 (0.19, 16.37)

26.60 (16.15, 37.06)

121.49 (99.40, 143.59)

38.69 (16.84, 60.55) back to normal

Symptoms

11.37 (2.97, 19.79)

9.22 (2.29, 16.15) c c

13.46 (6.37, 20.54) c c

two tablets

one tablet

Quantity

c

c

8.07 (0.93, 15.21)

c c

morning and evening

c

39.85 (17.37, 62.34)

31.81 (10.14, 53.46)

morning

No. of doses/time of day

back to normal and healing

8.49 (4.75, 12.24) [13.43]

6.45 (2.78, 10.12) [10.21]

4.51 (0.85, 8.16) [7.14]

7.46 (3.69, 11.25) [11.80]

6.83 (3.05, 10.60) [10.81] c c

12.98 (4.54, 21.42)

7.06 (0.00, 14.12) c

13.18 (5.97, 20.40) c

c

no problems

slight problems

Swallowing

c

US [$US]b UK [£]b Germany [h]b Canada [$Can]b

WTP (95% CI) Attribute and level

Table III. Regional differences in the importancea of different attributes of 5-aminosalicylic acid (mesalazine) treatment

Total sample £ (95% CI) [$US]

40

Individual Country Analyses

Differences in patients by country of residence were found: patients living in Germany significantly preferred clinical attributes (flare risk and symptom control). In contrast, patients living in the UK stated preferences for all attributes. Patients living in the US demonstrated statistically significant preferences for reductions in quantity of pills taken at each dose, while patients living in Canada significantly preferred reduction in the number of doses per day (table III). The relative strength of the preference for a given attribute across countries was demonstrated using WTP as a proxy (table IV) [see the final model in the SDC]. Flare risk was the primary predictor of choice in all countries. For this attribute, patients from Canada had the strongest preference and, therefore, the highest WTP ($Can446, h79, $US81, and d27 for Canada, Germany, the US, and the UK, respectively, for reducing a 70% flare risk to 40%, using PPPs). Patient 2012; 5 (1)

Patient Treatment Preferences for UC

41

Influence of Immediate Prior Clinical Status

Additional analyses explored the relationship between past flare experience and preferences regarding risk of future flares. Results demonstrated that those who experienced more flares in the past 12 months placed significantly greater importance on avoiding flares in the future (p < 0.001) versus other attributes. Compared with other patients, those with recent flare experience had a greater likelihood of preferring treatments with a 10% risk of flares (likelihood ratio: 0.64–0.70) and a lower likelihood of preferring treatments with a 70% risk of flares (likelihood ratio: 0.24–0.29). These data indicate that, while the effects are relatively small, people who experienced more recent flares had a greater preference for treatments that reduced flare risk in the future. Discussion This study was designed to meet two broad objectives. First, the DCE approach was used to estimate differences in patient preference based on self-assessed adherence, and, second, based on

country of residence as a proxy for cultural differences. Regardless of geography, data from the patient preference survey revealed that participants significantly preferred avoiding acute flares versus other attributes: the OR of choosing a treatment with a 10% annual risk of flares (compared with a treatment with a 70% risk) was 9.60 (95% CI 7.91, 11.66). Patients were willing to pay d78.81 ($US124.70) per month for treatment with a 10% annual risk versus a 70% annual risk of flare. Symptom control was the second most important attribute in the survey. Participants preferred treatments that could promote a return to normal bowel function during a flare (WTP = d9.85 [$US15.59] per month) versus treatment that would leave the participant with some residual symptoms. In addition, patients were willing to pay almost d30.00 ($US46.00) more per month for a return to normal bowel function and mucosal healing. Interesting differences emerged from comparison of country-specific preferences. Patients living in the UK significantly preferred convenience attributes, such as the number of pills taken, and difficulty swallowing medication. In contrast,

Table IV. Strength of patient preference for attributes of 5-aminosalicylic acid (mesalazine) treatment using willingness to pay (WTP) as a proxy Attribute and level

WTP Canada ($Can)

Germany (h)

UK (£)

US ($US)

Swallowing no problems

30.15

2.74

13.18*

31.41*

slight problems

17.22

16.13

7.06*

13.26

once a day

145.01*

8.45

8.07*

14.30

twice a day

115.76*

1.98

1.78

12.90

51.85

11.25

13.46*

27.53*

56.30

7.99

9.22

*

20.05*

back to normal

140.83*

44.76*

16.39*

36.49*

back to normal and healing

442.16*

90.19*

28.91*

79.27*

10

779.80*

145.97*

53.23*

144.81*

40

*

*

*

80.89*

No. of doses/times a day

Quantity one tablet two tablets Symptoms

Flare risk (%)

*

446.05

78.55

27.09

p < 0.05.

ª 2012 Adis Data Information BV. All rights reserved.

Patient 2012; 5 (1)

42

patients living in Germany significantly preferred clinical attributes such as flare risk and symptom control, rather than convenience attributes. This may be a reflection of significantly younger patients residing in Germany versus other countries included in this study or an additional cultural difference irrespective of age. Larsen et al.[25] examined cross-cultural differences in medication adherence in 45 678 participants across Europe. In this study, differences in adherence could not be explained by sociodemographic factors, suggesting that other factors, such as cultural differences, may be important. The purpose of the study by Larsen et al.[25] was to assess nonmodifiable risk factors, rather than clinical and convenience attributes as was done in this study. The literature includes several studies that explored cultural differences in the views of patients with inflammatory bowel disease (IBD). Levenstein et al.[26] explored the views of 2002 patients in eight countries regarding the importance of different aspects of IBD. Several factors associated with IBD, such as the need for an ostomy, medication side effects, and surgery, were patient preferences in all countries studied. However, other factors, such as concern regarding pain and the risk of bowel cancer, varied by country. In contrast, Richardson et al.[27] reported that the views of children with IBD in the UK and Canada were very closely correlated. While these studies are not consistent with our findings, they offer some evidence that cultural differences may affect patient preferences on treatment attributes. The present study found geographic differences in patient WTP. Canadians were willing to pay the most, while Americans were willing to pay the least for clinical benefits. These findings may be due to what patients in each country are accustomed to paying for medication or their ability to pay. It is important to note that this study includes countries with different healthcare delivery and drug reimbursement systems. Drug costs in the US are met through a combination of health insurance and variable co-pays. Most Canadians of working age have private prescription drug coverage, and older Canadians may receive drug coverage through the government. In the UK and Germany, drug costs are fixed and ª 2012 Adis Data Information BV. All rights reserved.

Hodgkins et al.

many patients pay nothing at all, meaning that patients in North America likely have greater knowledge of drug costs than patients from the UK and Germany. All of these factors may have influenced patient WTP. Additionally, patient knowledge of co-pays or prescription costs may act to frame costs assessed in the survey, thereby influencing patient preferences. The study was also designed to understand how self-reported adherence influenced patient preferences. Patients who reported good adherence stated preferences for clinical attributes such as resolution of flares and mucosal healing. This finding suggests that patients may adhere to therapy because they believe it is effectively treating their UC. Conversely, patients with poor adherence may place less value on the benefits of their current therapy. Therefore, clinical assessment of patient preference may be an indicator of good or poor adherence, the latter indicating the need for education on the importance of treatment adherence. Improving medication adherence offers the prospect of improving patient outcomes (e.g. symptom control and health-related quality of life for patients) and reducing healthcare costs.[28] The data from the DCE demonstrate how recent experience with the disease can affect the value patients place on treatments. More specifically, the analyses identified a significant interaction effect between the experience of recent flares and patients’ likelihood of preferring treatment that could reduce their risk of flares in the future. Patients who experienced three or more flares in the past 12 months were significantly more likely to choose a treatment with a low flare risk (10%). This study does have limitations. A decision was made to pool the data from different countries to better understand the overall pattern of preferences and to improve the power of the study for examining interaction effects. To equate the value of different currencies, we used recent estimates of PPP. These are estimates from the Organisation for Economic Co-operation and Development (OECD), which determine the value of one currency in a country compared with a different currency in another country. As discussed above, we found differences in the Patient 2012; 5 (1)

Patient Treatment Preferences for UC

importance of attributes when data were analyzed by country. Therefore, the results from the combined analyses across the four countries should be considered alongside the individual country data to understand possible cultural differences. Some national differences may also reflect differences for drug formulations commercially available and reimbursed. Formulations vary in their physical properties, and the availability of particular medications and the prescribing practices of physicians in a region will ultimately influence the actual medication taken by patients. The use of the Internet for data collection may have excluded some potential participants. Finally, the study included a relatively simple measure of selfreported adherence, the MMS. While this instrument has been validated in the general population, it has not been validated in patients with UC. The outcome of this instrument is typically presented as a single summary score; however, we used a median split of scores to classify people as good or poor adherers to simplify interpretation of the findings. This simple dichotomy does not have established validity and may be subject to a degree of reporting bias. Results should therefore be interpreted with caution. Self-reported adherence may not be as accurate as other methods for estimating adherence, such as pill counting. Conclusions The objective of this study was to estimate patient preferences for 5-ASA treatment in mildto-moderate UC based on self-reported adherence in four different geographical regions. The results demonstrated that, while clinical attributes were significantly preferred by the majority of patients, convenience attributes such as medication dosing were also important for study participants in some countries. Self-reported adherers significantly preferred clinical benefits, suggesting that positive patient assessment of effectiveness may influence adherence. Further work should explore whether clinician solicitation of patient preference may be an indicator for treatment adherence. This, coupled with education on the importance of adherence, may help to improve treatment outcomes in UC. ª 2012 Adis Data Information BV. All rights reserved.

43

Acknowledgments PH conceived the need for the study, contributed to the study throughout and co-wrote the manuscript. PS codeveloped all of the study materials, project managed the survey phase, and co-wrote the manuscript. LY assisted with the interpretation of analyses, and reviewed and contributed to the manuscript. SD undertook all analyses of the choice experiment data, helped to interpret the findings and wrote that section of the manuscript. DS contributed to the study throughout, providing expert insight into how colitis affects patients; he also reviewed and contributed to the manuscript. AJL directed the study and designed the choice experiment survey. He also oversaw data collection, assisted with the interpretation of analyses, and co-wrote the manuscript. Editorial support was provided by Lisa E. Melilli and Bari Samson at PAREXEL, and the study was funded by Shire Pharmaceuticals. DS, LY and PH are employees and stock holders of Shire Pharmaceuticals. Shire develops and markets drugs for UC. PS, AL, and SD are employees of Oxford Outcomes, which was paid a set fee to collaborate with Shire Pharmaceuticals in the design, conduct, and reporting of the study.

References 1. Rubin G, Hungin AP, Chinn D, et al. Long-term aminosalicylate therapy is under-used in patients with ulcerative colitis: a cross-sectional survey. Aliment Pharmacol Ther 2002 Nov; 16 (11): 1889-93 2. Kane S, Huo D, Magnanti K. A pilot feasibility study of once daily versus conventional dosing mesalamine for maintenance of ulcerative colitis. Clin Gastroenterol Hepatol 2003 May; 1 (3): 170-3 3. Moody GA, Jayanthi V, Probert CS, et al. Long-term therapy with sulphasalazine protects against colorectal cancer in ulcerative colitis: a retrospective study of colorectal cancer risk and compliance with treatment in Leicestershire. Eur J Gastroenterol Hepatol 1996 Dec; 8 (12): 1179-83 4. Kane SV. The complexity of compliance and persistence in ulcerative colitis. Gastroenterol Hepatol 2007; 3 (9 Suppl. 28): 3-10 5. Ediger JP, Walker JR, Graff L, et al. Predictors of medication adherence in inflammatory bowel disease. Am J Gastroenterol 2007 Jul; 102 (7): 1417-26 6. Kane SV, Brixner D, Rubin DT, et al. The challenge of compliance and persistence: focus on ulcerative colitis. J Manag Care Pharm 2008 Jan; 14 (1 Suppl. A): s2-12; quiz s3-5 7. Sewitch MJ, Abrahamowicz M, Barkun A, et al. Patient nonadherence to medication in inflammatory bowel disease. Am J Gastroenterol 2003 Jul; 98 (7): 1535-44 8. Nguyen GC, LaVeist TA, Harris ML, et al. Patient trust-inphysician and race are predictors of adherence to medical management in inflammatory bowel disease. Inflamm Bowel Dis 2009 Aug; 15 (8): 1233-9 9. Thomas RK. Health services planning. New York: Kluwer Academic/Plenum Publishers, 2003 10. Gao G, Burke N, Somkin CP, et al. Considering culture in physician-patient communication during colorectal cancer screening. Qual Health Res 2009 Jun; 19 (6): 778-89

Patient 2012; 5 (1)

44

11. Gerard K, Ryan M, Amaya-Amaya M. Introduction: benefit valuation in health economics. In: Ryan M, Gerard K, Amaya-Amaya M, editors. Using discrete choice experiments to value health and health care. Dordrecht: Springer, 2008: 1-2 12. Bridges JFP, Kinter ET, Kidane L, et al. Things are looking up since we started listening to patients: trends in the application of conjoint analysis in health 1982-2007. Patient 2008; 1 (4): 273-82 13. de Bekker-Grob EW, Ryan M, Gerard K. Discrete choice experiments in health economics: a review of the literature. Health Econ. Epub 2010 Dec 19 14. Marshall D, Bridges JFP, Hauber AB, et al. Conjoint analysis applications in health: how are studies being designed and reported? An update on current practice in the published literature between 2005 and 2008. Patient 2010; 3 (4): 249-56 15. Hall J, Fiebig DG, King MT, et al. What influences participation in genetic carrier testing? Results from a discrete choice experiment. J Health Econ 2006 May; 25 (3): 520-37 16. Scott A, Watson MS, Ross S. Eliciting preferences of the community for out of hours care provided by general practitioners: a stated preference discrete choice experiment. Soc Sci Med 2003 Feb; 56 (4): 803-14 17. Watson V, Ryan M, Brown CT, et al. Eliciting preferences for drug treatment of lower urinary tract symptoms associated with benign prostatic hyperplasia. J Urol 2004 Dec; 172 (6 Pt 1): 2321-5 18. Sabate´ E. Adherence to long-term therapies: evidence for action. Geneva: WHO, 2003 19. Nunes V, Neilson J, O’Flynn N, et al. Clinical guidelines and evidence review for medicines adherence: involving patients in decisions about prescribed medicines and supporting adherence. London: National Collaborating Centre for Primary Care and Royal College of General Practitioners, 2009 20. Morisky DE, Ang A, Krousel-Wood M, et al. Predictive validity of a medication adherence measure in an outpatient setting. J Clin Hypertens (Greenwich) 2008 May; 10 (5): 348-54

ª 2012 Adis Data Information BV. All rights reserved.

Hodgkins et al.

21. Case Management Society of America. Modified Morisky Scale. In: Case management adherence guidelines. Version 2.0. 2006 [online]. Available from URL: http://www.cmsa. org/Individual/Education/CaseManagementAdherenceGui delines/tabid/253/Default.aspx [Accessed 2011 Mar 22] 22. Agresti A. An introduction to categorical data analysis. Hoboken (NJ): John Wiley & Sons, Inc., 1996 23. McIntosh E, Clarke P, Frew E, et al. Applied methods of cost-benefit analysis in health care. Huntington Beach (CA): Oxford University Press, 2010 24. Organisation for Economic Co-operation and Development (OECD). Purchase power parities 2008 [online]. Available from URL: http://stats.oecd.org/Index.aspx?datasetcode= SNA_TABLE4 [Accessed 2010 Jan 12] 25. Larsen J, Stovring H, Kragstrup J, et al. Can differences in medical drug compliance between European countries be explained by social factors: analyses based on data from the European Social Survey, round 2. BMC Public Health 2009; 9: 145 26. Levenstein S, Li Z, Almer S, et al. Cross-cultural variation in disease-related concerns among patients with inflammatory bowel disease. Am J Gastroenterol 2001 Jun; 96 (6): 1822-30 27. Richardson G, Griffiths AM, Miller V, et al. Quality of life in inflammatory bowel disease: a cross-cultural comparison of English and Canadian children. J Pediatr Gastroenterol Nutr 2001 May; 32 (5): 573-8 28. Mitra D, Davis K, Hodgkins P, et al. Assessing the relationship between adherence and healthcare costs: evidence from a managed care population with ulcerative colitis. Society for Medical Decision Making Annual Meeting; 2009 Oct 18-21; Hollywood (CA)

Correspondence: Dr Paul Hodgkins, Global Health Economics & Outcomes Research, Shire Pharmaceuticals, 725 Chesterbrook Boulevard, Wayne, PA 19087, USA. E-mail: [email protected]

Patient 2012; 5 (1)