Effect of Level Overlap and Color Coding on Attribute Non-Attendance in Discrete Choice Experiments

Comparing heuristic valuation processes between health state valuation from child and adult perspectives

OBJECTIVES

Health state valuation assumes that respondents trade off between all aspects of choice tasks and maximize their utility. Yet, respondents may use heuristic valuation processes, i.e., strategies to simplify or avoid the trade-offs that are core to health state valuation. The objective of this study is to explore if heuristic valuation processes are more prevalent for valuation from a 10-year-old child's perspective compared to the use of an adult perspective.

METHODS

We reused existing data in which EQ-5D health states were valued from adult and child perspectives with composite time trade-off (cTTO) and discrete choice experiment (DCE) tasks. Our analyses focused on comparing completion time and responding patterns across both perspectives. We also explored how reflective of a set of heuristic strategies respondents' choices were in both perspectives.

RESULTS

We found no evidence for systematic differences in completion time across perspectives. Generally, we find different responding patterns in child perspectives, e.g., more speeding, dominance violations, and clustering of utilities at 1.0, 0.8, and 0. Very few heuristic strategies provide a coherent explanation for the observed DCE responses.

CONCLUSION

Our results provide some, albeit indirect, evidence for differences in heuristic valuation processes between perspectives, although not across all data sources. Potential effects of heuristic valuation processes, such as transfer of responsibility, may be identified through studying responding patterns in cTTO and DCE responses.

Stated-Preference Survey Design and Testing in Health Applications

Following the conceptualization of a well-formulated and relevant research question, selection of an appropriate stated-preference method, and related methodological issues, researchers are tasked with developing a survey instrument. A major goal of designing a stated-preference survey for health applications is to elicit high-quality data that reflect thoughtful responses from well-informed respondents. Achieving this goal requires researchers to design engaging surveys that maximize response rates, minimize hypothetical bias, and collect all the necessary information needed to answer the research question. Designing such a survey requires researchers to make numerous interrelated decisions that build upon the decision context, selection of attributes, and experimental design. Such decisions include considering the setting(s) and study population in which the survey will be administered, the format and mode of administration, and types of contextual information to collect. Development of a survey is an interactive process in which feedback from respondents should be collected and documented through qualitative pre-test interviews and pilot testing. This paper describes important issues to consider across all major steps required to design and test a stated-choice survey to elicit patient preferences for health preference research.

Estimating a Preference-Based Value Set for the Mental Health Quality of Life Questionnaire (MHQoL)

BACKGROUND

Health economic evaluations using common health-related quality of life measures may fall short in adequately measuring and valuing the benefits of mental health care interventions. The Mental Health Quality of Life questionnaire (MHQoL) is a standardized, self-administered mental health-related quality of life instrument covering 7 dimensions known to be relevant across and valued highly by people with mental health problems. The aim of this study was to derive a Dutch value set for the MHQoL to facilitate its use in cost-utility analyses.

METHODS

The value set was estimated using a discrete choice experiment (DCE) with duration that accommodated nonlinear time preferences. The DCE was embedded in a web-based self-complete survey and administered to a representative sample (N = 1,308) of the Dutch adult population. The matched pairwise choice tasks were created using a Bayesian heterogeneous D-efficient design. The overall DCE design comprised 10 different subdesigns, with each subdesign containing 15 matched pairwise choice tasks. Each participant was asked to complete 1 of the subdesigns to which they were randomly assigned.

RESULTS

The obtained coefficients indicated that "physical health,""mood," and "relationships" were the most important dimensions. All coefficients were in the expected direction and reflected the monotonic structure of the MHQoL, except for level 2 of the dimension "future." The predicted values for the MHQoL ranged from -0.741 for the worst state to 1 for the best state.

CONCLUSIONS

This study derived a Dutch value set for the recently introduced MHQoL. This value set allows for the generation of an index value for all MHQoL states on a QALY scale and may hence be used in Dutch cost-utility analyses of mental healthcare interventions.

HIGHLIGHTS

A discrete choice experiment was used to derive a Dutch value set for the MHQoL.This allows the use of the MHQoL in Dutch cost-utility analyses.The dimensions physical health, mood, and relationships were the most important.The utility values range from -0.741 for the worst state to 1 for the best state.

Public preferences for policies promoting a healthy diet: a discrete choice experiment

BACKGROUND

Worldwide obesity rates have nearly tripled over the past five decades. So far, policies to promote a healthier diet have been less intrusive than those to reduce tobacco and alcohol consumption. Not much is known about public support for policies that aim to promote a healthy diet. In this study, a discrete choice experiment (DCE) was used to elicit stated preferences for policies varying in intrusiveness among a representative sample of the public of The Netherlands.

METHODS

The choice tasks presented respondents a hypothetical scenario of two policy packages, each comprising a mix of seven potential policies that differed in level of intrusiveness. We estimated mixed logit models (MXL) to estimate respondents' preferences for these policies and performed latent class analyses to identify heterogeneity in preferences.

RESULTS

The MXL model showed that positive financial incentives like subsidies for vegetables and fruit yielded most utility. A tax of 50% on sugary drinks was associated with disutility while a tax of 20% was associated with positive utility compared to no tax at all. We identified three subgroups with distinct preferences for the seven policies to promote a healthy diet, which were characterized as being "against", "mixed" and "pro" policies to promote a healthy diet.

CONCLUSION

Preferences for policies promoting a healthy diet vary considerably in the Dutch population, particularly in relation to more intrusive policies. This makes selection and implementation of a policy package that has wide public support challenging.

An investigation of age dependency in Dutch and Chinese values for EQ-5D-Y

Aims

The primary aim was to explore the age dependency of health state values derived via trade-offs between health-related quality of life (HRQoL) and life years in a discrete choice experiment (DCE). The secondary aim was to explore if people weigh life years and HRQoL differently for children, adolescents, adults, and older adults.

Methods

Participants from the general population of the Netherlands and China first completed a series of choice tasks offering choices between two EQ-5D-Y states with a given lifespan. The choice model captured the value of a year in full health, disutility determined by EQ-5D-Y, and a discount rate. Next, they received a slightly different choice task, offering choices between two lives that differed in HRQoL and life expectancy but produced the same number of quality-adjusted life years (QALYs). Participants were randomly assigned to fill out the survey for three or four age frames: a hypothetical person of 10, 15, 40, and 70 years (the last one only applicable to China) to allow the age dependency of the responses to be explored.

Results

A total of 1,234 Dutch and 1,818 Chinese people administered the survey. Controlling for time preferences, we found that the agreement of health state values for different age frames was generally stronger in the Netherlands than in China. We found no clear pattern of differences in the QALY composition in both samples. The probability distribution over response options varied most when levels for lifespan or severity were at the extremes of the spectrum.

Conclusion/discussion

The magnitude and direction of age effects on values seemed dimension- and country specific. In the Netherlands, we found a few differences in dimension-specific weights elicited for 10- and 15-year-olds compared to 40-year-olds, but the overall age dependency of values was limited. A stronger age dependency of values was observed in China, where values for 70-year-olds differed strongly from the values for other ages. The appropriateness of using existing values beyond the age range for which they were measured needs to be evaluated in the local context.

Comparison of Preferences and Data Quality between Discrete Choice Experiments Conducted in Online and Face-to-Face Respondents

INTRODUCTION

Discrete choice experiments (DCE) are increasingly being conducted using online panels. However, the comparability of such DCE-based preferences to traditional modes of data collection (e.g., in-person) is not well established. In this study, supervised, face-to-face DCE was compared with its unsupervised, online facsimile on face validity, respondent behavior, and modeled preferences.

METHODS

Data from face-to-face and online EQ-5D-5L health state valuation studies were compared, in which each used the same experimental design and quota sampling procedure. Respondents completed 7 binary DCE tasks comparing 2 EQ-5D-5L health states presented side by side (health states A and B). Data face validity was assessed by comparing preference patterns as a function of the severity difference between 2 health states within a task. The prevalence of potentially suspicious choice patterns (i.e., all As, all Bs, and alternating As/Bs) was compared between studies. Preference data were modeled using multinomial logit regression and compared based on dimensional contribution to overall scale and importance ranking of dimension-levels.

RESULTS

One thousand five Online respondents and 1,099 face-to-face screened (F2FS) respondents were included in the main comparison of DCE tasks. Online respondents reported more problems on all EQ-5D dimensions except for Mobility. The face validity of the data was similar between comparators. Online respondents had a greater prevalence of potentially suspicious DCE choice patterns ([Online]: 5.3% [F2FS] 2.9%, P = 0.005). When modeled, the relative contribution of each EQ-5D dimension differed between modes of administration. Online respondents weighed Mobility more importantly and Anxiety/Depression less importantly.

DISCUSSION

Although assessments of face validity were similar between Online and F2FS, modeled preferences differed. Future analyses are needed to clarify whether differences are attributable to preference or data quality variation between modes of data collection.

Valuing the EQ Health and Wellbeing Short Using Time Trade-Off and a Discrete Choice Experiment: A Feasibility Study

OBJECTIVES

The EQ Health and Wellbeing Short (EQ-HWB-S) is a new generic measure that covers health and wellbeing developed for use in economic evaluation in health and social care. The aim was to test the feasibility of using composite time trade-off (cTTO) and a discrete choice experiment (DCE) based on an international protocol to derive utilities for the EQ-HWB-S and to generate a pilot value set.

METHODS

A representative UK general population was recruited. Online videoconference interviews were undertaken where cTTO and DCE tasks were administered using EuroQol Portable Valuation Technology. Quality control (QC) was used to assess interviewers' performance. Data were modeled using Tobit, probit, and hybrid models. Feasibility was assessed based on the distribution of data, participants, and reports of understanding from the interviewer, QC and modeling results.

RESULTS

cTTO and DCE data were available for 520 participants. Demographic characteristics were broadly representative of the UK general population. Interviewers met QC requirements. cTTO values ranged between -1 to 1 with increasing disutility associated with more severe states. Participants understood the tasks and the EQ-HWB-S states; and the interviewers reported high levels of understanding and engagement. The hybrid Tobit heteroscedastic model was selected for the pilot value set with values ranging from -0.384 to 1. Pain, mobility, daily activities, and sad/depressed had the largest disutilities, followed by loneliness, anxiety, exhaustion, control, and cognition in the selected model.

CONCLUSIONS

EQ-HWB-S can be valued using cTTO and DCE. Further methodological work is recommended to develop a valuation protocol specific to the EQ-HWB-S.

Taking the Shortcut: Simplifying Heuristics in Discrete Choice Experiments

Health-related discrete choice experiments (DCEs) information can be used to inform decision-making on the development, authorisation, reimbursement and marketing of drugs and devices as well as treatments in clinical practice. Discrete choice experiment is a stated preference method based on random utility theory (RUT), which imposes strong assumptions on respondent choice behaviour. However, respondents may use choice processes that do not adhere to the normative rationality assumptions implied by RUT, applying simplifying decision rules that are more selective in the amount and type of processed information (i.e., simplifying heuristics). An overview of commonly detected simplifying heuristics in health-related DCEs is lacking, making it unclear how to identify and deal with these heuristics; more specifically, how researchers might alter DCE design and modelling strategies to accommodate for the effects of heuristics. Therefore, the aim of this paper is three-fold: (1) provide an overview of common simplifying heuristics in health-related DCEs, (2) describe how choice task design and context as well as target population selection might impact the use of heuristics, (3) outline DCE design strategies that recognise the use of simplifying heuristics and develop modelling strategies to demonstrate the detection and impact of simplifying heuristics in DCE study outcomes.

Efficient Designs for Valuation Studies That Use Time Tradeoff (TTO) Tasks to Map Latent Utilities from Discrete Choice Experiments to the Interval Scale: Selection of Health States for TTO Tasks

BACKGROUND

In eliciting utilities to value multiattribute utility instruments, discrete choice experiments (DCEs) administered online are less costly than interviewer-facilitated time tradeoff (TTO) tasks. DCEs capture utilities on a latent scale and are often coupled with a small number of TTO tasks to anchor utilities to the interval scale. Given the costly nature of TTO data, design strategies that maximize value set precision per TTO response are critical.

METHODS

Under simplifying assumptions, we expressed the mean square prediction error (MSE) of the final value set as a function of the number J of TTO-valued health states and the variance VJ of the states' latent utilities. We hypothesized that even when these assumptions do not hold, the MSE 1) decreases as VJ increases while holding J fixed and 2) decreases as J increases while holding VJ fixed. We used simulation to examine whether there was empirical support for our hypotheses a) assuming an underlying linear relationship between TTO and DCE utilities and b) using published results from the Dutch, US, and Indonesian EQ-5D-5L valuation studies.

RESULTS

Simulation set (a) supported the hypotheses, as did simulations parameterized using valuation data from Indonesia, which showed a linear relationship between TTO and DCE utilities. The US and Dutch valuation data showed nonlinear relationships between TTO and DCE utilities and did not support the hypotheses. Specifically, for fixed J, smaller values of VJ reduced rather than increased the MSE.

CONCLUSIONS

Given that, in practice, the underlying relationship between TTO and DCE utilities may be nonlinear, health states for TTO valuation should be placed evenly across the latent utility scale to avoid systematic bias in some regions of the scale.

HIGHLIGHTS

Valuation studies may feature a large number of respondents completing discrete choice tasks online, with a smaller number of respondents completing time tradeoff (TTO) tasks to anchor the discrete choice utilities to an interval scale.We show that having each TTO respondent complete multiple tasks rather than a single task improves value set precision.Keeping the total number of TTO respondents and the number of tasks per respondent fixed, having 20 health states directly valued through TTO leads to better predictive precision than valuing 10 health states directly.If DCE latent utilities and TTO utilities follow a perfect linear relationship, choosing the TTO states to be valued by weighting on the 2 ends of the latent utility scale leads to better predictive precision than choosing states evenly across the latent utility scale.Conversely, if DCE latent utilities and TTO utilities do not follow a linear relationship, choosing the states to be valued using TTO evenly across the latent utility scale leads to better predictive precision than weighted selection does.In the context of valuation of the EQ-5D-Y-3L, we recommend valuing 20 or more health states using TTO and placing them evenly across the latent utility scale.

Estimating an EQ-5D-Y-3L Value Set for Indonesia by Mapping the DCE onto TTO Values

BACKGROUND AND OBJECTIVES

Methods for estimating health values in adult populations are well developed, but lag behind in children. The EuroQol standard protocol to arrive at value sets for the youth version of the EQ-5D-Y-3L combines discrete choice experiments with ten composite time trade-off values. Whether ten composite time trade-off values are sufficient remains to be seen and this is one of the reasons the protocol allows for experimental expansion. In this study, 23 health states were administered for the composite time trade-off. This methodological research is embedded in a study aimed at generating a representative value set for EQ-5D-Y-3L in Indonesia.

METHODS

A representative sample of 1072 Indonesian adults each completed 15 discrete choice experiment choice pairs via face-to-face interviews. The discrete choice experiment responses were analysed using a mixed-logit model. To anchor the discrete choice experiment values onto the full health-dead quality-adjusted life-year scale, composite time trade-off values were separately obtained from 222 adults living in Java for 23 EQ-5D-Y-3L states. The derived latent discrete choice experiment values were mapped onto the mean observed composite time trade-off values to create a value set for the EQ-5D-Y-3L. Linear and non-linear mapping models were explored to estimate the most efficient and valid model for the value set.

RESULTS

Coefficients obtained from the choice model were consistent with the monotonic structure of the EQ-5D-Y-3L instrument. The composite time trade-off data showed non-linearity, as the values for the two worst states being evaluated were much lower than predicted by a standard linear model estimated over all composite time trade-off data. Thus, the non-linear mapping strategies with a power term outperformed the linear mapping in terms of mean absolute error. The final model gave a value range from 1.000 for full health (11111) to - 0.086 for the worst health state (33333). Values were most affected by pain/discomfort and least by self-care.

CONCLUSIONS

This article presents the first EQ-5D-Y-3L value set for Indonesia based on the stated preferences of adults asked to consider their views about a 10-year-old child. Mapping the mixed-logit discrete choice experiment model with the inclusion of a power term (without a constant) allowed us to generate a consistent value set for Indonesian youth. Our findings support the expansion of the composite time trade-off part of the EQ-5D-Y valuation study design and show that it would be wise to account for possible non-linearities in updates of the design.

The Garbage Class Mixed Logit Model: Accounting for Low-Quality Response Patterns in Discrete Choice Experiments

OBJECTIVES

To introduce the garbage class mixed logit (MIXL) model as a convenient alternative to manually screening and accounting for respondents with low data quality in discrete choice experiments.

METHODS

Garbage classes are typically used in latent class logit analyses to designate or identify group(s) of respondents with low data quality. Yet, the same concept can be applied to MIXL models as well.

RESULTS

Based on a reanalysis of 4 discrete choice experiments that were originally analyzed using a standard MIXL model, it is shown that garbage class MIXL models can achieve the same effect as manually screening for (and excluding) respondents with low data quality based on the more commonly used root likelihood test, but with less effort and ambiguity.

CONCLUSIONS

Including a garbage class in MIXL models removes the influence of respondents with a random choice pattern from the MIXL model estimates, provides an estimate of the number of low-quality respondents in the dataset, and avoids having to manually screen for respondents with low data quality based on internal or statistical validity tests. Although less versatile than the combination of standard MIXL estimates with separate assessments of data quality and sensitivity analyses, the proposed garbage class MIXL model provides an attractive alternative.

Patient and Public Preferences for Coordinated Care in Switzerland: Development of a Discrete Choice Experiment

Objective

Our objective was to develop and test a discrete choice experiment (DCE) eliciting public and patient preferences for better-coordinated care in Switzerland.

Methods

We applied a multistage mixed-methods procedure using qualitative and quantitative approaches. First, to identify attributes, we performed a review of the DCE literature in healthcare with a focus on chronic care. Next, attribute selection involved stakeholders (N = 7) from various healthcare sectors to select the most relevant and actionable attributes, followed by three organized focus groups involving the general public and patients (N = 21) to verify the selection and the clarity of the DCE tasks and explanations. Finally, we conducted an online pilot in the target population to test the survey and obtain priors for a final six tested attributes to refine the final design of the experiment.

Results

After identifying an initial 33 attributes, a final list of six attributes was selected following stakeholder involvement and the three focus groups involving the target population. At the online pilot-testing stage with 301 participants, the majority of respondents found the DCE choice tasks socially relevant for Switzerland but challenging. The quality of the answers was relatively high. Most attributes had signs matching those in the literature and focus group discussions.

Conclusion

This article will be useful to researchers designing DCEs from a broad health policy perspective. The multistage approach involving a range of stakeholders was essential for the development of a DCE that is relevant for policy makers and well-accepted by the general public and patients.

Supplementary Information

The online version contains supplementary material available at 10.1007/s40271-021-00568-2.

Estimating an anchored utility tariff for the well-being of older people measure (WOOP) for the Netherlands

OBJECTIVE

Health economic evaluations using common health-related quality of life measures may fall short in adequately incorporating all relevant benefits of health and social care interventions targeted at older people. The Well-being of Older People measure (WOOP) is a broader well-being measure that comprises nine well-being domains. The objective of this study was to estimate a utility tariff for the WOOP, to facilitate its application in cost-utility analyses.

METHODS

A discrete choice experiment (DCE) with duration approach was set up and fielded among 2,012 individuals from the Netherlands aged 65 years and above. Matched pairwise choice tasks, colour-coding and level overlap were used to reduce the cognitive burden of the DCE. The choice tasks were created using a Bayesian heterogeneous D-efficient design. The estimation procedure accommodated for nonlinear time preferences via an exponential discounting function.

RESULTS

The estimation results showed that 'physical health', 'mental health', and 'making ends meet' were the most important well-being domains for older people, followed by 'independence' and 'living situation'. Of somewhat lesser importance were domains like 'social life', 'receiving support' and 'feeling useful'. The generated utility tariffs can be used to translate well-being states described with the WOOP to a utility score between -0.616 and 1.

CONCLUSIONS

This study established a tariff for the WOOP, which will facilitate its use in economic evaluations of health and social care interventions targeted at older people, first of all in the Netherlands.

Not all respondents use a multiplicative utility function in choice experiments for health state valuations, which should be reflected in the elicitation format (or statistical analysis)

Discrete choice experiments (DCEs) that include health states and duration are becoming a common method for estimating quality-adjusted life year (QALY) tariffs. These DCEs need to be analyzed under the assumption that respondents treat health and duration multiplicatively. However, in the most commonly used DCE duration format there is no guarantee that respondents actually do so; in fact, respondents can easily simplify the choice tasks by considering health and duration separately. This would result in valid DCE responses but preclude subsequent QALY tariff calculations. Using a Bayesian latent class model and data from two existing valuation studies, our analyses confirm that in both datasets the majority of respondents do not appear to have used a multiplicative utility function. Moreover, a statistical correction for respondents who used an incorrect function changes the range of the QALY weights. Hence our results imply that one can neither assume that respondents use the theoretically required multiplicative utility function nor assume that the type of utility function that respondents use does not affect the estimated QALY weights. As a solution, we advise researchers to use an alternative, more constrained DCE elicitation format that avoids these behavioral problems.

EQ-VT protocol: one-size-fits-all? Challenges and innovative adaptations used in Egypt: a cross-sectional study

OBJECTIVES

To present the challenges and adaptations done to the EuroQol Valuation Technology (EQ-VT) protocol to fit the Egyptian culture during the extensive pilot phase of the Egyptian EuroQol 5 Dimension five level (EQ-5D-5L) valuation study DESIGN: This study was a cross-sectional, interviewer-administered face-to-face survey of representative Egyptians using the Arabic version of the EuroQol Group Valuation Technology (EQ-VT-2.1) and a country specific questionnaire pertaining to participants' demographics and opinions about health, life and death SETTING: Participants were recruited from workplaces, university campuses, sporting clubs, shopping malls and other public areas from different Egyptian governorates representing all geographical areas of the country.

PARTICIPANTS

A total of 1378 participants were interviewed from July 2019 to March 2020 by 12 interviewers to select a representative sample in terms of: geographical distribution, age and gender, of which 75 participants did not complete the interview, 298 interviews were pilot and 1005 interviews were real of which 974 interviews were used for the valuation study. Two participants did not complete the country-specific questionnaire but completed the valuation protocol; therefore, 1301 interviews were included in the final analysis of country specific questions.

RESULTS

Some modifications were applied to the protocol. The 'wheelchair example' was modified to 'migraine example' since most of the participants in the pilot interviews considered being in a wheelchair 'worse than dead'. There was some ambiguity in the Egyptian translated version for the EQ-5D-5L between levels 4 and 5 of the pain and depression dimensions. This was overcome by using colour coding to express the different levels of severity. A pictorial representation for the EQ-5D-5L health states was used to interview illiterate and less educated participants.

CONCLUSION

In the Egyptian valuation study, the modifications made to the EQ-VT protocol made it feasible and culturally acceptable to the Egyptian participants.

Exploring Decisional Conflict With Measures of Numeracy and Optimism in a Stated Preference Survey

Objectives

Low optimism and low numeracy are associated with difficulty or lack of participation in making treatment-related health care decisions. We investigated whether low optimism and low self-reported numeracy scores could help uncover evidence of decisional conflict in a discrete-choice experiment (DCE).

Methods

Preferences for a treatment to delay type 1 diabetes were elicited using a DCE among 1501 parents in the United States. Respondents chose between two hypothetical treatments or they could choose no treatment (opt out) in a series of choice questions. The survey included a measure of optimism and a measure of subjective numeracy. We used latent class analyses where membership probability was predicted by optimism and numeracy scores.

Results

Respondents with lower optimism scores had a higher probability of membership in a class with disordered preferences (P value for optimism coefficient = 0.032). Those with lower self-reported numeracy scores were more likely to be in a class with a strong preference for opting out and disordered preferences (P = 0.000) or a class with a preference for opting out and avoiding serious treatment-related risks (P = 0.015).

Conclusions

If respondents with lower optimism and numeracy scores are more likely to choose to opt out or have disordered preferences in a DCE, it may indicate that they have difficulty completing choice tasks.

A Systematic Review of Discrete Choice Experiments in Oncology Treatments

BACKGROUND

As the number and type of cancer treatments available rises and patients live with the consequences of their disease and treatments for longer, understanding preferences for cancer care can help inform decisions about optimal treatment development, access, and care provision. Discrete choice experiments (DCEs) are commonly used as a tool to elicit stakeholder preferences; however, their implementation in oncology may be challenging if burdensome trade-offs (e.g. length of life versus quality of life) are involved and/or target populations are small.

OBJECTIVES

The aim of this review was to characterise DCEs relating to cancer treatments that were conducted between 1990 and March 2020.

DATA SOURCES

EMBASE, MEDLINE, and the Cochrane Database of Systematic Reviews were searched for relevant studies.

STUDY ELIGIBILITY CRITERIA

Studies were included if they implemented a DCE and reported outcomes of interest (i.e. quantitative outputs on participants' preferences for cancer treatments), but were excluded if they were not focused on pharmacological, radiological or surgical treatments (e.g. cancer screening or counselling services), were non-English, or were a secondary analysis of an included study.

ANALYSIS METHODS

Analysis followed a narrative synthesis, and quantitative data were summarised using descriptive statistics, including rankings of attribute importance.

RESULT

Seventy-nine studies were included in the review. The number of published DCEs relating to oncology grew over the review period. Studies were conducted in a range of indications (n = 19), most commonly breast (n =10, 13%) and prostate (n = 9, 11%) cancer, and most studies elicited preferences of patients (n = 59, 75%). Across reviewed studies, survival attributes were commonly ranked as most important, with overall survival (OS) and progression-free survival (PFS) ranked most important in 58% and 28% of models, respectively. Preferences varied between stakeholder groups, with patients and clinicians placing greater importance on survival outcomes, and general population samples valuing health-related quality of life (HRQoL). Despite the emphasis of guidelines on the importance of using qualitative research to inform attribute selection and DCE designs, reporting on instrument development was mixed.

LIMITATIONS

No formal assessment of bias was conducted, with the scope of the paper instead providing a descriptive characterisation. The review only included DCEs relating to cancer treatments, and no insight is provided into other health technologies such as cancer screening. Only DCEs were included.

CONCLUSIONS AND IMPLICATIONS

Although there was variation in attribute importance between responder types, survival attributes were consistently ranked as important by both patients and clinicians. Observed challenges included the risk of attribute dominance for survival outcomes, limited sample sizes in some indications, and a lack of reporting about instrument development processes.

PROTOCOL REGISTRATION

PROSPERO 2020 CRD42020184232.

Deriving a Preference-Based Measure for People With Duchenne Muscular Dystrophy From the DMD-QoL

OBJECTIVES

This study generates a preference-based measure for capturing the quality of life of people with Duchenne muscular dystrophy (DMD) from a new measure of quality of life, DMD-QoL.

METHODS

A health state classification system was derived from the DMD-QoL based on psychometric performance of items, factor analysis, and item response theory analysis. Preferences for health states described by the classification system were elicited using an online discrete choice experiment survey with life years as an additional attribute, from members of the UK general population (n = 1043). Discrete choice experiment data was modeled using a conditional fixed-effects logit model and utility estimates were directly anchored on the 1 to 0 full health-dead scale.

RESULTS

The health state classification system has 8 dimensions: mobility, difficulty using hands, difficulty breathing, pain, tiredness, worry, participation, and feeling good about yourself. The standard model had mostly statistically significant coefficients and reflected the instrument's monotonic structure. However, 2 dimensions had inconsistent coefficients (where utility increased as health worsened) and a consistent model was estimated that merged adjacent inconsistent severity levels. The best state defined by the classification system has a value of 1 and the worst state has a value of -0.559.

CONCLUSION

The modeled results enable DMD-QoL-8D utility values to be generated using DMD-QoL or DMD-QoL-8D data to generate QALYs for people with DMD. QALYs can then be used to inform economic models of the cost-effectiveness of interventions in DMD. Future research comparing the psychometric performance of DMD-QoL-8D to existing generic preference-based measures, including EQ-5D-5L, is recommended.

Patients Prioritize a Low-volume Bowel Preparation in Colitis-associated Colorectal Cancer Surveillance: A Discrete Choice Experiment

BACKGROUND

Patients with inflammatory bowel disease (IBD) undergo surveillance colonoscopies at fixed intervals to reduce the risk of colorectal cancer (CRC). Taking patients' preferences for determining surveillance strategies into account could improve adherence and patient satisfaction. This study aimed to determine patient preferences for CRC surveillance in IBD.

METHODS

We conducted a web-based, multicenter, discrete choice experiment among adult IBD patients with an indication for surveillance. Individuals were repeatedly asked to choose between 3 hypothetical surveillance scenarios. The choice tasks were based on bowel preparation (0.3-4 L), CRC risk reduction (8% to 1%-6%), and interval (1-10 years). Attribute importance scores, trade-offs, and willingness to participate were calculated using a multinomial logit model. Latent class analysis was used to identify subgroups with similar preferences.

RESULTS

In total, 310 of 386 sent out questionnaires were completed and included in the study. Bowel preparation was prioritized (attribute importance score 40.5%) over surveillance interval and CRC risk reduction (31.1% and 28.4%, respectively). Maximal CRC risk reduction, low-volume bowel preparation (0.3 L laxative with 2 L clear liquid) with 2-year surveillance was the most preferred combination. Three subgroups were identified: a "surveillance avoidant," "CRC risk avoidant," and "surveillance preferring" groups. Membership was correlated with age, educational level, perceived CRC risk, the burden of bowel preparation, and colonoscopies.

CONCLUSIONS

Inflammatory bowel disease patients consider bowel preparation as the most important element in acceptance of CRC surveillance. Heterogeneity in preferences was explained by 3 latent subgroups. These findings may help to develop an individualized endoscopic surveillance strategy in IBD patients.

Public Preferences and Predicted Uptake for Esophageal Cancer Screening Strategies: A Labeled Discrete Choice Experiment

INTRODUCTION:

As novel, less invasive (non)endoscopic techniques for detection of Barrett's esophagus (BE) have been developed, there is now renewed interest in screening for BE and related neoplasia. We aimed to determine public preferences for esophageal adenocarcinoma screening to understand the potential of minimally invasive screening modalities.

METHODS:

A discrete choice experiment was conducted in 1,500 individuals, aged 50–75 years, from the general population. Individuals were repeatedly asked to choose between screening scenarios based on conventional upper endoscopy, transnasal endoscopy, nonendoscopic cell collection devices, breath analysis, and a blood test, combined with various levels of test sensitivity and specificity, and no screening. A multinomial logit model was used to estimate individuals' preferences and to calculate expected participation rates.

RESULTS:

In total, 554 respondents (36.9%) completed the survey. The average predicted uptake was 70.5% (95% confidence interval: 69.1%–71.8%). Test sensitivity (47.7%), screening technique (32.6%), and specificity (19.7%) affected screening participation (all P < 0.05). A low test sensitivity had the highest impact on screening participation, resulting in a 25.0% (95% confidence interval: 22.6%–27.7%) decrease. Respondents preferred noninvasive screening tests over endoscopic and capsule-based techniques, but only if sensitivity and specificity were above 80%.

DISCUSSION:

Our study suggests that individuals generally prefer noninvasive BE screening tests. However, these tests would unlikely improve screening uptake when associated with a much lower accuracy for detecting BE and esophageal adenocarcinoma compared with conventional upper endoscopy. Improving accuracy of minimally invasive screening strategies and informing the target population about these accuracies is therefore essential to maximally stimulate screening participation.

An Exploration on Attribute Non-attendance Using Discrete Choice Experiment Data from the Irish EQ-5D-5L National Valuation Study

BACKGROUND

Generic measures of health-related quality of life (HRQoL) permit comparisons of competing demands for healthcare resources using outcomes that reflect the preferences of tax payers. EQ-5D instruments are the most commonly used generic, preference-based measures of HRQoL. The EQ-5D-5L enables respondents to describe their health state using five dimensions of health, each with five response levels. The standardised protocol for the valuation of EQ-5D-5L health states comprises use of the composite time trade-off valuation technique, supplemented by a discrete choice experiment (DCE).

OBJECTIVE

This paper presents the first exploration on attribute non-attendance (ANA) to the dimensions of the EQ-5D-5L using DCE data collected following the standardised protocol.

METHOD

This paper uses the equality constrained latent class model and the endogenous attribute attendance model to examine ANA to the dimensions of the EQ-5D-5L.

RESULTS

The results suggest that respondents are less likely to consider the physical dimensions of the EQ-5D-5L (such as self-care and usual activities) when evaluating the health states. The effects of ANA on utility scores depends on the interpretation of the underlying reasons for ANA.

CONCLUSIONS

We recommend that future value sets based in whole or in part on DCE data examine the impact of and reasons for non-attendance in national valuation studies.

United States Utility Algorithm for the EORTC QLU-C10D, a Multiattribute Utility Instrument Based on a Cancer-Specific Quality-of-Life Instrument

BACKGROUND

The EORTC QLU-C10D is a multiattribute utility measure derived from the cancer-specific quality-of-life questionnaire, the EORTC QLQ-C30. The QLU-C10D contains 10 dimensions (physical, role, social and emotional functioning, pain, fatigue, sleep, appetite, nausea, bowel problems). The objective of this study was to develop a United States value set for the QLU-C10D.

METHODS

A US online panel was quota recruited to achieve a representative sample for sex, age (≥18 y), race, and ethnicity. Respondents undertook a discrete choice experiment, each completing 16 choice-pairs, randomly assigned from a total of 960 choice-pairs. Each pair included 2 QLU-C10D health states and duration. Data were analyzed using conditional logistic regression, parameterized to fit the quality-adjusted life-year framework. Utility weights were calculated as the ratio of each dimension-level coefficient to the coefficient for life expectancy.

RESULTS

A total of 2480 panel members opted in, 2333 (94%) completed at least 1 choice-pair, and 2273 (92%) completed all choice-pairs. Within dimensions, weights were generally monotonic. Physical functioning, role functioning, and pain were associated with the largest utility weights. Cancer-specific dimensions, such as nausea and bowel problems, were associated with moderate utility decrements, as were general issues such as problems with emotional functioning and social functioning. Sleep problems and fatigue were associated with smaller utility decrements. The value of the worst health state was 0.032, which was slightly greater than 0 (equivalent to being dead).

CONCLUSIONS

This study provides the US-specific value set for the QLU-C10D. These estimated health state scores, based on responses to the EORTC QLQ-C30 questionnaire, can be used to evaluate the cost-utility of oncology treatments.

Does Device or Connection Type Affect Health Preferences in Online Surveys?

BACKGROUND AND OBJECTIVE

Recent evidence has shown that online surveys can reliably collect preference data, which markedly decrease the cost of health preference studies and expand their representativeness. As the use of mobile technology continues to grow, we wanted to examine its potential impact on health preferences.

METHODS

Two recently completed discrete choice experiments using members of the US general population (n = 15,292) included information on respondent device (cell phone, tablet, Mac, PC) and internet connection (business, cellular, college, government, residential). In this analysis, we tested for differences in respondent characteristics, participation, response quality, and utility values for the 5-level EQ-5D (EQ-5D-5L) by device and connection.

RESULTS

Compared to Mac and PC users, respondents using a cell phone or tablet had longer completion times and were significantly more likely to drop out during the surveys (p < 0.001). Tablet users also demonstrated more logical inconsistencies (p = 0.05). Likewise, respondents using a cellular internet connection exhibit significantly less consistency in their health preferences. However, matched samples for tablets and cell phones produced similar EQ-5D-5L utility values (mean differences < 0.06 on a quality-adjusted life-year [QALY] scale for all potential health states).

CONCLUSION

Allowing respondents to complete online surveys using a cell phone or tablet or over a cellular connection substantially increases the diversity of respondents and the likelihood of obtaining a representative sample, as many individuals have cell phones but not a computer. While the results showed systematic variability in participation and response quality by device and connection type, this study did not show any meaningful changes in utility values.

Health-related quality of life after traumatic brain injury: deriving value sets for the QOLIBRI-OS for Italy, The Netherlands and The United Kingdom

Purpose

The Quality of Life after Brain Injury overall scale (QOLIBRI-OS) measures health-related quality of life (HRQoL) after traumatic brain injury (TBI). The aim of this study was to derive value sets for the QOLIBRI-OS in three European countries, which will allow calculation of utility scores for TBI health states.

Methods

A QOLIBRI-OS value set was derived by using discrete choice experiments (DCEs) and visual analogue scales (VAS) in general population samples from the Netherlands, United Kingdom and Italy. A three-stage procedure was used: (1) A selection of health states, covering the entire spectrum of severity, was defined; (2) General population samples performed the health state valuation task using a web-based survey with three VAS questions and an at random selection of sixteen DCEs; (3) DCEs were analysed using a conditional logistic regression and were then anchored on the VAS data. Utility scores for QOLIBRI-OS health states were generated resulting in estimates for all potential health states.

Results

The questionnaire was completed by 13,623 respondents. The biggest weight increase for all attributes is seen from “slightly” to “not at all satisfied”, resulting in the largest impact on HRQoL. “Not at all satisfied with how brain is working” should receive the greatest weight in utility calculations in all three countries.

Conclusion

By transforming the QOLIBRI-OS into utility scores, we enabled the application in economic evaluations and in summary measures of population health, which may be used to inform decision-makers on the best interventions and strategies for TBI patients.

Electronic supplementary material

The online version of this article (10.1007/s11136-020-02583-6) contains supplementary material, which is available to authorized users.

Do Patient Preferences Align With Value Frameworks? A Discrete-Choice Experiment of Patients With Breast Cancer

Purpose. Assess patient preferences for aspects of breast cancer treatments to evaluate and inform the usual assumptions in scoring rubrics for value frameworks. Methods. A discrete-choice experiment (DCE) was designed and implemented to collect quantitative evidence on preferences from 100 adult female patients with a self-reported physician diagnosis of stage 3 or stage 4 breast cancer. Respondents were asked to evaluate some of the treatment aspects currently considered in value frameworks. Respondents' choices were analyzed using logit-based regression models that produced preference weights for each treatment aspect considered. Aggregate- and individual-level preferences were used to assess the relative importance of treatment aspects and their variability across respondents. Results. As expected, better clinical outcomes were associated with higher preference weights. While life extensions with treatment were considered to be most important, respondents assigned great value to out-of-pocket cost of treatment, treatment route of administration, and the availability of reliable tests to help gauge treatment efficacy. Two respondent classes were identified in the sample. Differences in class-specific preferences were primarily associated with route of administration, out-of-pocket treatment cost, and the availability of a test to gauge treatment efficacy. Only patient cancer stage was found to be correlated with class assignment (P = 0.035). Given the distribution of individual-level preference estimates, preference for survival benefits are unlikely to be adequately described with two sets of preference weights. Conclusions. Although value frameworks are an important step in the systematic evaluation of medications in the context of a complex treatment landscape, the frameworks are still largely driven by expert judgment. Our results illustrate issues with this approach as patient preferences can be heterogeneous and different from the scoring weights currently provided by the frameworks.

Summarizing Patient Preferences for the Competitive Landscape of Multiple Sclerosis Treatment Options

Objective. Quantitatively summarize patient preferences for European licensed relapsing-remitting multiple sclerosis (RRMS) disease-modifying treatment (DMT) options. Methods. To identify and summarize the most important RRMS DMT characteristics, a literature review, exploratory physician interviews, patient focus groups, and confirmatory physician interviews were conducted in Germany, the United Kingdom, and the Netherlands. A discrete choice experiment (DCE) was developed and executed to measure patient preferences for the most important DMT characteristics. The resulting DCE data (n=799 and n=363 respondents in the United Kingdom and Germany, respectively) were analyzed using Bayesian mixed logit models. The estimated individual-level patient preferences were subsequently summarized using 3 additional analyses: the quality of the choice data was assessed using individual-level R² estimates, individual-level preferences for the available DMTs were aggregated into DMT-specific preference shares, and a principal component analysis was performed to explain the patients' choice process. Results. DMT usage differed between RRMS patients in Germany and the United Kingdom but aggregate patient preferences were similar. Across countries, 42% of all patients preferred oral medications, 38% infusions, 16% injections, and 4% no DMT. The most often preferred DMT was natalizumab (26%) and oral DMT cladribine tablets (22%). The least often preferred were mitoxantrone and the beta-interferon injections (1%-3%). Patient preferences were strongly correlated with patients' MS disease duration and DMT experience, and differences in patient preferences could be summarized using 8 principle components that together explain 99% of the variation in patients' DMT preferences. Conclusion. This study summarizes patient preferences for the included DMTs, facilitates shared decision making along the dimensions that are relevant to RRMS patients, and introduces methods in the medical DCE literature that are ideally suited to summarize the impact of DMT introductions in preexisting treatment landscapes.

Analysis of Patient Preferences in Lung Cancer - Estimating Acceptable Tradeoffs Between Treatment Benefit and Side Effects

OBJECTIVE

Increased treatment options and longer survival for lung cancer have generated increased interest in patient preferences. Previous studies of patient preferences in lung cancer have not fully explored preference heterogeneity. We demonstrate a method to explore preference heterogeneity in the willingness of patients with lung cancer and caregivers to trade progression-free survival (PFS) with side effects.

PATIENTS AND METHODS

Patients and caregivers attending a national lung cancer meeting completed a discrete-choice experiment (DCE) designed through a collaboration with patients. Participants answered 13 choice tasks described across PFS, short-term side effects, and four long-term side effects. Side effects were coded as a one-level change in severity (none-mild, mild-moderate, or moderate-severe). A mixed logit model in willingness-to-pay space estimated preference heterogeneity in acceptable tradeoffs (time equivalents) between PFS and side effects. The study was reported following quality indicators from the United States Food and Drug Administration's patient preference guidance.

RESULTS

A total of 87 patients and 24 caregivers participated in the DCE. Participants would trade 3.7 month PFS (95% CI (CI): 3.3-4.1) for less severe functional long-term treatment side effects, 2.3 months for less severe physical long-term effects (CI: 1.9-2.8) and cognitive long-term effects (CI: 1.8-2.8), 0.9 months (CI: 0.4-1.4) for less severe emotional long-term effects, and 1.8 months (CI: 1.4-2.3) for less severe short-term side effects. Most participants (90%) would accept treatment with more severe functional long-term effects for 8.4 additional month PFS.

CONCLUSION

Participants would trade PFS for changes in short-term side effects and long-term side effects, although preference heterogeneity existed. Lung cancer treatments that offer less PFS but also less severe side effects might be acceptable to some patients.

Population preferences for breast cancer screening policies: Discrete choice experiment in Belarus

BACKGROUND

Reaching an acceptable participation rate in screening programs is challenging. With the objective of supporting the Belarus government to implement mammography screening as a single intervention, we analyse the main determinants of breast cancer screening participation.

METHODS

We developed a discrete choice experiment using a mixed research approach, comprising a literature review, in-depth interviews with key informants (n = 23), "think aloud" pilots (n = 10) and quantitative measurement of stated preferences for a representative sample of Belarus women (n = 428, 89% response rate). The choice data were analysed using a latent class logit model with four classes selected based on statistical (consistent Akaike information criterion) and interpretational considerations.

RESULTS

Women in the sample were representative of all six geographic regions, mainly urban (81%), and high-education (31%) characteristics. Preferences of women in all four classes were primarily influenced by the perceived reliability of the test (sensitivity and screening method) and costs. Travel and waiting time were important components in the decision for 34% of women. Most women in Belarus preferred mammography screening to the existing clinical breast examination (90%). However, if the national screening program is restricted in capacity, this proportion of women will drop to 55%. Women in all four classes preferred combined screening (mammography with clinical breast examination) to single mammography. While this preference was stronger if lower test sensitivity was assumed, 28% of women consistently gave more importance to combined screening than to test sensitivity.

CONCLUSION

Women in Belarus were favourable to mammography screening. Population should be informed that there are no benefits of combined screening compared to single mammography. The results of this study are directly relevant to policy makers and help them targeting the screening population.

On the Optimization of Bayesian D-Efficient Discrete Choice Experiment Designs for the Estimation of QALY Tariffs That Are Corrected for Nonlinear Time Preferences

OBJECTIVES

This article explains how to optimize Bayesian D-efficient discrete choice experiment (DCE) designs for the estimation of quality-adjusted life year (QALY) tariffs that are unconfounded by respondents' time preferences.

METHODS

The calculation of Bayesian D-errors is explained for DCE designs that allow for the disentanglement of respondents' time and health-state preferences. Time preferences are modelled via an exponential, hyperbolic, or power discount function and the performance of the proposed DCE designs is compared with that of several conventional DCE designs that do not take nonlinear time preferences into account.

RESULTS

Based on the achieved D-error, asymptotic standard error, and estimated sample size to obtain statistically significant estimates of the discount rate parameters, the proposed designs outperform the conventional DCE designs.

CONCLUSIONS

We recommend that applied researchers use appropriately optimized DCE designs for the estimation of QALY tariffs that are corrected for time preferences. The TPC-QALY software package that accompanies this article makes the recommended designs easily accessible for health-state valuation researchers.

The Fold-in, Fold-out Design for DCE Choice Tasks: Application to Burden of Disease

Background In discrete-choice experiments (DCEs), choice alternatives are described by attributes. The importance of each attribute can be quantified by analyzing respondents’ choices. Estimates are valid only if alternatives are defined comprehensively, but choice tasks can become too difficult for respondents if too many attributes are included. Several solutions for this dilemma have been proposed, but these have practical or theoretical drawbacks and cannot be applied in all settings. The objective of the current article is to demonstrate an alternative solution, the fold-in, fold-out approach (FiFo). We use a motivating example, the ABC Index for burden of disease in chronic obstructive pulmonary disease (COPD). Methods Under FiFo, all attributes are part of all choice sets, but they are grouped into domains. These are either folded in (all attributes have the same level) or folded out (levels may differ). FiFo was applied to the valuation of the ABC Index, which included 15 attributes. The data were analyzed in Bayesian mixed logit regression, with additional parameters to account for increased complexity in folded-out questionnaires and potential differences in weight due to the folding status of domains. As a comparison, a model without the additional parameters was estimated. Results Folding out domains led to increased choice complexity for respondents. It also gave domains more weight than when it was folded in. The more complex regression model had a better fit to the data than the simpler model. Not accounting for choice complexity in the models resulted in a substantially different ABC Index. Conclusion Using a combination of folded-in and folded-out attributes is a feasible approach for conducting DCEs with many attributes.

Investigating the relative value of health and social care related quality of life using a discrete choice experiment

A key outcome in the evaluation of health technologies is the quality adjusted life year (QALY) which is often estimated using health measures such as the EuroQol instruments (EQ-5D-3L and EQ-5D-5L). The impacts of many interventions extend beyond a narrow definition of health to include non-health impacts such as social care related dimensions of quality of life (QoL). This means that there are circumstances where the QALY does not capture the full value of an intervention. In response to this, instruments with a wider measurement framework, such as the Adult Social Care Outcomes Toolkit (ASCOT), which measures social care related QoL, have been developed. Given the range of instruments available, it is important that decision-makers have tools to assess value for money comprehensively and consistently. To date, preference elicitation of different aspects of QoL combined within the same valuation procedure has not been tested. We investigate the relationship between health and social care aspects of QoL when assessed jointly by combining EQ-5D-5L and ASCOT in an online discrete choice experiment (DCE). In July 2016, 975 respondents recruited from internet panels completed 15 choice sets from an underlying design of 300. Conditional logit regression was used to estimate coefficient decrements for each attribute and examine their relative magnitude. Latent class and mixed logit modelling were used to understand preference heterogeneity. The results suggest trading across health and social care aspects indicated by coefficient estimates of differing magnitude. Dimensions with the largest disutility include four from EQ-5D-5L and one from ASCOT. There is evidence of preference heterogeneity at more severe dimension levels. We have used an established method to test the joint valuation of concepts measuring different aspects of QoL. The results have implications for the aspects of QoL that are included in QALY estimation and used in resource allocation decision-making.

Attribute level overlap (and color coding) can reduce task complexity, improve choice consistency, and decrease the dropout rate in discrete choice experiments

A randomized controlled discrete choice experiment (DCE) with 3,320 participating respondents was used to investigate the individual and combined impact of level overlap and color coding on task complexity, choice consistency, survey satisfaction scores, and dropout rates. The systematic differences between the study arms allowed for a direct comparison of dropout rates and cognitive debriefing scores and accommodated the quantitative comparison of respondents' choice consistency using a heteroskedastic mixed logit model. Our results indicate that the introduction of level overlap made it significantly easier for respondents to identify the differences and choose between the choice options. As a stand-alone design strategy, attribute level overlap reduced the dropout rate by 30%, increased the level of choice consistency by 30%, and avoided learning effects in the initial choice tasks of the DCE. The combination of level overlap and color coding was even more effective: It reduced the dropout rate by 40% to 50% and increased the level of choice consistency by more than 60%. Hence, we can recommend attribute level overlap, with color coding to amplify its impact, as a standard design strategy in DCEs.

One Method, Many Methodological Choices: A Structured Review of Discrete-Choice Experiments for Health State Valuation

BACKGROUND

Discrete-choice experiments (DCEs) are used in the development of preference-based measure (PBM) value sets. There is considerable variation in the methodological approaches used to elicit preferences.

OBJECTIVE

Our objective was to carry out a structured review of DCE methods used for health state valuation.

METHODS

PubMed was searched until 31 May 2018 for published literature using DCEs for health state valuation. Search terms to describe DCEs, the process of valuation and preference-based instruments were developed. English language papers with any study population were included if they used DCEs to develop or directly inform the production of value sets for generic or condition-specific PBMs. Assessment of paper quality was guided by the recently developed Checklist for Reporting Valuation Studies. Data were extracted under six categories: general study information, choice task and study design, type of designed experiment, modelling and analysis methods, results and discussion.

RESULTS

The literature search identified 1132 published papers, and 63 papers were included in the review. Paper quality was generally high. The study design and choice task formats varied considerably, and a wide range of modelling methods were employed to estimate value sets.

CONCLUSIONS

This review of DCE methods used for developing value sets suggests some recurring limitations, areas of consensus and areas where further research is required. Methodological diversity means that the values should be seen as experimental, and users should understand the features of the value sets produced before applying them in decision making.

Severity-Stratified Discrete Choice Experiment Designs for Health State Evaluations

BACKGROUND

Discrete choice experiments (DCEs) are increasingly used for health state valuations. However, the values derived from initial DCE studies vary widely. We hypothesize that these findings indicate the presence of unknown sources of bias that must be recognized and minimized. Against this background, we studied whether values derived from a DCE are sensitive to how well the DCE design spans the severity range.

METHODS

We constructed an experiment involving three variants of DCE tasks for health state valuation: standard DCE, DCE-death, and DCE-duration. For each type of DCE, an experimental design was generated under two different conditions, enabling a comparison of health state values derived from current best practice Bayesian efficient DCE designs with values derived from 'severity-stratified' designs that control for coverage of the severity range in health state selection. About 3000 respondents participated in the study and were randomly assigned to one of the six study arms.

RESULTS

Imposing the severity-stratified restriction had a large effect on health states sampled for the DCE-duration approach. The unstratified efficient design returned a skewed distribution of selected health states, and this introduced bias. The choice probability of bad health states was underestimated, and time trade-offs to avoid bad states were overestimated, resulting in too low values. Imposing the same restriction had limited effect in the DCE-death approach and standard DCE.

CONCLUSION

Variation in DCE-derived values can be partially explained by differences in how well selected health states spanned the severity range. Imposing a 'severity stratification' on DCE-duration designs is a validity requirement.