Search

Background: Accounting for preference heterogeneity is a growing analytical practice in health-related discrete choice experiments (DCEs). As heterogeneity may be examined from different stakeholder perspectives with different methods, identifying the breadth of these methodological approaches and understanding the differences are major steps to provide guidance on good research practices. Objectives: Our objective was to systematically summarize current practices that account for preference heterogeneity based on the published DCEs related to healthcare. Methods: This systematic review is part of the project led by the Professional Society for Health Economics and Outcomes Research (ISPOR) health preference research special interest group. The systematic review conducted systematic searches on the PubMed, OVID, and Web of Science databases, as well as on two recently published reviews, to identify articles. The review included health-related DCE articles published between 1 January 2000 and 30 March 2020. All the included articles also presented evidence on preference heterogeneity analysis based on either explained or unexplained factors or both. Results: Overall, 342 of the 2202 (16%) articles met the inclusion/exclusion criteria for extraction. The trend showed that analyses of preference heterogeneity increased substantially after 2010 and that such analyses mainly examined heterogeneity due to observable or unobservable factors in individual characteristics. Heterogeneity through observable differences (i.e., explained heterogeneity) is identified among 131 (40%) of the 342 articles and included one or more interactions between an attribute variable and an observable characteristic of the respondent. To capture unobserved heterogeneity (i.e., unexplained heterogeneity), the studies largely estimated either a mixed logit (n = 205, 60%) or a latent-class logit (n = 112, 32.7%) model. Few studies (n = 38, 11%) explored scale heterogeneity or heteroskedasticity. Conclusions: Providing preference heterogeneity evidence in health-related DCEs has been found as an increasingly used practice among researchers. In recent studies, controlling for unexplained preference heterogeneity has been seen as a common practice rather than explained ones (e.g., interactions), yet a lack of providing methodological details has been observed in many studies that might impact the quality of analysis. As heterogeneity can be assessed from different stakeholder perspectives with different methods, researchers should become more technically pronounced to increase confidence in the results and improve the ability of decision makers to act on the preference evidence.

Background: New technologies have enabled the potential for stratified medicine in psoriasis. It is important to understand patients’ preferences to enable the informed introduction of stratified medicine, which is likely to involve a number of individual tests that could be collated into a prescribing algorithm for biological drug selection to be used in clinical practice. Objectives: To quantify patient preferences for an algorithm-based approach to prescribing biologics (‘biologic calculator’) in psoriasis. Methods: An online survey comprising a discrete choice experiment (DCE) was conducted to elicit the preferences of two purposive samples of adults living with psoriasis in the UK, identified from a psoriasis patient organization (Psoriasis Association) and an online panel provider (Dynata). Respondents chose between two biologic calculators and conventional prescribing described using five attributes: treatment delay; positive predictive value; negative predictive value; risk of infection; and cost saving to the National Health Service. Each participant selected their preferred alternative from six hypothetical choice sets. Additional data, including sociodemographic characteristics, were collected. Choice data were analysed using conditional logit and fully correlated random parameters logit models. Results: Data from 212 respondents (67 from the Psoriasis Association and 145 from Dynata) were analysed. The signs of all estimated coefficients were consistent with a priori expectations. Respondents had a strong preference for a high predictive accuracy and avoiding serious infection, but there was evidence of systematic differences in preferences between the samples. Conclusions: This study indicates that individuals with psoriasis would value a biologic calculator and suggested that such a biologic calculator should have sufficient accuracy to predict future response and risk of serious infection from the biologic.

Introduction There have been promising developments in technologies and associated algorithm-based prescribing (‘stratified approach’) to target biologics to sub-groups of people with rheumatoid arthritis (RA). The acceptability of using an algorithm-guided approach in practice is likely to depend on various factors. Objective This study quantified preferences for an algorithm-guided approach to prescribing biologics (termed ‘biologic calculator’). Methods An online discrete choice experiment (DCE) was designed to elicit preferences from patients and the public for using a ‘biologic calculator’ compared with conventional prescribing. Treatment approaches were described by five attributes: delay to starting treatment; positive and negative predictive value (PPV/NPV); risk of infection; and cost saving to the UK national health service. Each survey contained six choice sets asking respondents to select their preferred option from two hypothetical biologic calculators or conventional prescribing. Background questions included sociodemographics, health status and healthcare experiences. DCE data were analysed using mixed logit models. Results Completed choice data were collected from 292 respondents (151 patients with RA and 142 members of the public). PPV, NPV and risk of infection were the most highly valued attributes to respondents deciding between prescribing strategies. Conclusion Respondents were generally receptive to personalised medicine in RA, but researchers developing personalised approaches should pay close attention to generating evidence on both the PPV and the NPV of their technologies. Supplementary Information The online version contains supplementary material available at 10.1007/s40271-021-00533-z.

Background Antibiotics have led to considerable increases in life expectancy. However, over time, antimicrobial resistance has accelerated and is now a significant global public health concern. Understanding societal preferences for the use of antibiotics as well as eliciting the willingness to pay for future research is crucial. Objective To investigate preferences for different strategies to optimize antibiotic use and to understand the willingness to pay for future research in antimicrobial resistance and antimicrobial drug development. Methods A discrete-choice experiment was administered to a sample of the UK general population. Respondents were asked to make nine choices, each offering three options—two hypothetical “doctor and antibiotics” and one “no doctor—no antibiotics”—defined by five attributes: treatment, days needed to recover, risk of bacterial infection that needs antibiotics, risk of common side effects, and risk of antimicrobial resistance by 2050. Data were analyzed using random parameters logit models. A double-bounded contingent valuation was also included in the survey to explore the willingness to pay for policies to contain antimicrobial resistance. Results Among the 2579 respondents who completed the survey, 1151 always selected “no doctor—no antibiotics” and 57 never varied their choices; therefore, 1371 responses were used in the analysis. Risk of antimicrobial resistance by 2050 was the most important attribute and the “treatment” was the least important attribute, although this was sensitive to a higher risk of bacterial infection. The aggregate annual willingness to pay for containing antimicrobial resistance was approximately £8.35 billion (~£5–£10 billion). Conclusions The antimicrobial resistance risk is relevant and important to the general public. The high willingness to pay suggests that large investments in policies or interventions to combat antimicrobial resistance are justified. Supplementary Information The online version contains supplementary material available at 10.1007/s40273-021-01076-9.

635 Background: Although a handful of studies have elicited treatment preferences in renal cell carcinoma (RCC), most focused on advanced disease. This study elicited United States patients’ and physicians’ preferences for adjuvant treatment characteristics. Methods: Patients with physician-confirmed RCC and (physician-defined) intermediate high or high risk of recurrence and physicians who treat such patients completed online surveys in Q1-Q2 2022 with a discrete-choice experiment. Hypothetical treatments were described by median disease-free survival (DFS); 5-year overall survival (OS) rate; mode and frequency of administration; need for concomitant daily pill; treatment duration; and the risks of severe diarrhea, fatigue, and dizziness. Preference weight estimates from random parameter logit analysis were used to calculate the conditional relative importance of attributes and risk tolerance measures. Results: 250 patients (50% post-nephrectomy) and 250 physicians (64% oncologists; 36% urologists) completed the survey. OS was the most important attribute to both patients and physicians, but DFS was also important (Table). OS had a greater influence on physicians’ choices than on patients’ choices. On average, OS was 3.2 and 2.5 times as important as DFS and 5.8-9.1 and 2.4-3 times more important than the evaluated risks for physicians and patients, respectively. Further, DFS was 1.8-2.9 times more important to physicians than the evaluated risks, while the importance of DFS and risks were nearly equivalent for patients. The need for concomitant oral medication was the least important attribute to patients and physicians. Both groups were willing to accept more than a 25-percentage-point increase in the risks of severe diarrhea, fatigue, and dizziness for improvements (from 45% to 60% or 85%) in OS. Conclusions: While both patients and physicians weighted OS improvements more than the other treatment attributes, including risks, physicians tended to place lower importance on changes in risk and administration than patients. Physicians and patients should discuss potential benefits and harms when considering adjuvant RCC therapies. [Table: see text]

661 Background: To inform shared decision-making, it is important to learn how patients or physicians trade off the features of adjuvant treatments and whether there is heterogeneity within preferences. Methods: An online discrete-choice experiment survey was administered to patients with physician-confirmed renal cell carcinoma (RCC) and physician-defined intermediate high/high risk of recurrence and physicians who treat such patients. Hypothetical treatment choices were defined by median disease-free survival (DFS); 5-year overall survival (OS) rate; mode and frequency of administration; need for concomitant daily pill; treatment duration; and the risks of severe diarrhea, fatigue, and dizziness. After making an adjuvant treatment choice, respondents were presented with the opportunity to opt out of treatment. Patient and physician choice data were analyzed separately using latent class (LC) models, which identify clusters within patients and physicians making similar choices. Each class’ preference weights were used to calculate the conditional relative attribute importance. Results: LC analysis identified three classes among the 250 patients (% respondents) that placed greater relative importance on: 1) 5-year OS and opting into treatment (37.5%), 2) median DFS and opting into treatment (26.9%), and 3) treatment duration and opting out of treatment (35.5%) (Table). Among the 250 physicians, the LC analysis identified three classes that placed greater relative importance on: 1) 5-year OS and recommending treatment (37.5%), 2) median DFS and recommending treatment (37.8%), and 3) not recommending treatment (24.7%). Additionally, each LC analysis showed that the other treatment attributes evaluated were less important, but the importance varied by LC (Table). Conclusions: Heterogeneity in physician and patient preferences for RCC adjuvant therapy was found, highlighting a need for shared decision-making. Discordance within patients and physicians in the propensity to opt out of adjuvant treatment suggests patient-physician dialogue is important. [Table: see text]

Discrete choice experiments (DCEs) are increasingly used to elicit preferences for health and healthcare. Although many applications assume preferences are homogenous, there is a growing portfolio of methods to understand both explained (because of observed factors) and unexplained (latent) heterogeneity. Nevertheless, the selection of analytical methods can be challenging and little guidance is available. This study aimed to determine the state of practice in accounting for preference heterogeneity in the analysis of health-related DCEs, including the views and experiences of health preference researchers and an overview of the tools that are commonly used to elicit preferences.An online survey was developed and distributed among health preference researchers and nonhealth method experts, and a systematic review of the DCE literature in health was undertaken to explore the analytical methods used and summarize trends.Most respondents (n = 59 of 70, 84%) agreed that accounting for preference heterogeneity provides a richer understanding of the data. Nevertheless, there was disagreement on how to account for heterogeneity; most (n = 60, 85%) stated that more guidance was needed. Notably, the majority (n = 41, 58%) raised concern about the increasing complexity of analytical methods. Of the 342 studies included in the review, half (n = 175, 51%) used a mixed logit with continuous distributions for the parameters, and a third (n = 110, 32%) used a latent class model.Although there is agreement about the importance of accounting for preference heterogeneity, there are noticeable disagreements and concerns about best practices, resulting in a clear need for further analytical guidance.

Background: Formative qualitative research is foundational to the methodological development process of quantitative health preference research (HPR). Despite its ability to improve the validity of the quantitative evidence, formative qualitative research is underreported. Objective: To improve the frequency and quality of reporting, we developed guidelines for reporting this type of research. The guidelines focus on formative qualitative research used to develop robust and acceptable quantitative study protocols and corresponding survey instruments in HPR.Methods: In December 2018, a steering committee was formed as a means to accumulate the expertise of the HPR community on the reporting guidelines (21 members, seven countries, multiple settings, and disciplines). Using existing guidelines and examples, the committee constructed, revised, and refined the guidelines. The guidelines underwent beta-testing by three researchers and further revision to the guidelines were made based on their feedback as well as from comments from members of the International Academy of Health Preference Research (IAHPR) and the editorial board of The Patient.Results: The guidelines have five components: introductory material (4 domains); methods (12); results/findings (2); discussion (2); and other (2). They are concordant with existing guidelines, published examples, beta testing results, and expert comments.Conclusions: Publishing formative qualitative research is a necessary step towards strengthening the foundation of any quantitative study, enhancing the relevance of its preference evidence. The guidelines should aid researchers, reviewers and regulatory agencies as well as promote transparency within HPR more broadly.

Background Technological progress has led to changes in the antenatal screening programmes, most significantly the introduction of non-invasive prenatal testing (NIPT). The availability of a new type of testing changes the type of information that the parent(s) require before, during and after screening to mitigate anxiety about the testing process and results. Objectives To identify the extent to which economic evaluations of NIPT have accounted for the need to provide information alongside testing and the associated costs and health outcomes of information provision. Methods A systematic review of economic evaluations of NIPTs (up to February 2018) was conducted. Medline, Embase, CINAHL and PsychINFO were searched using an electronic search strategy combining a published economic search filter (from NHS economic evaluations database) with terms related to NIPT and screening-related technologies. Data were extracted using the Consolidated Health Economic Evaluation Reporting Standards framework and the results were summarised as part of a narrative synthesis. Results A total of 12 economic evaluations were identified. The majority of evaluations (n = 10; 83.3%) involved cost effectiveness analysis. Only four studies (33.3%) included the cost of providing information about NIPT in their economic evaluation. Two studies considered the impact of test results on parents’ quality of life by allowing utility decrements for different outcomes. Some studies suggested that the challenges of valuing information prohibited their inclusion in an economic evaluation. Conclusion Economic evaluations of NIPTs need to account for the costs and outcomes associated with information provision, otherwise estimates of cost effectiveness may prove inaccurate.

BACKGROUND: Literature reviews show stated-preference studies, used to understand the values individuals place on health and health care, are increasingly administered online, potentially maximising respondent access and allowing for enhanced response quality. Online respondents may often choose whether to use a desktop or laptop personal computer (PC), tablet or smartphone, all with different screen sizes and modes of data entry, to complete the survey. To avoid differences in measurement errors, frequently respondents are asked to complete the surveys on a PC despite evidence that handheld devices are increasingly used for internet browsing. As yet, it is unknown if or how the device used to access the survey affects responses and/or the subsequent valuations derived.METHOD: This study uses data from a discrete choice experiment (DCE) administered online to elicit preferences of a general population sample of females for a national breast screening programme. The analysis explores differences in key outcomes such as completion rates, engagement with the survey materials, respondent characteristics, response time, failure of an internal validity test and health care preferences for (1) handheld devices and (2) PC users. Preferences were analysed using a fully correlated random parameter logit (RPL) model to allow for unexplained scale and preference heterogeneity.RESULTS: One thousand respondents completed the survey in its entirety. The most popular access devices were PCs (n = 785), including Windows (n = 705) and Macbooks (n = 69). Two-hundred and fifteen respondents accessed the survey on a handheld device. Most outcomes related to survey behaviour, including failure of a dominance check, 'flat lining', self-reported attribute non-attendance (ANA) or respondent-rated task difficulty, did not differ by device type (p > 0.100). Respondents accessing the survey using a PC were generally quicker (median time to completion 14.5 min compared with 16 min for those using handheld devices) and were significantly less likely to speed through a webpage. Although there was evidence of preference intensity (taste) or variability (scale) heterogeneity across respondents in the sample, it was not driven by the access device.CONCLUSION: Overall, we find that neither preferences nor choice behaviour is associated with the type of access device, as long as respondents are presented with question formats that are easy to use on small touchscreens. Health preference researchers should optimise preference instruments for a range of devices and encourage respondents to complete the surveys using their preferred device. However, we suggest that access device characteristics should be gathered and included when reporting results.

Publications reporting discrete choice experiments of healthcare interventions rarely discuss whether patient and public involvement (PPI) activities have been conducted. This paper presents examples from the existing literature and a detailed case study from the National Institute for Health Research-funded PATHWAY programme that comprehensively included PPI activities at multiple stages of preference research. Reflecting on these examples, as well as the wider PPI literature, we describe the different stages at which it is possible to effectively incorporate PPI across preference research, including the design, recruitment and dissemination of projects. Benefits of PPI activities include gaining practical insights from a wider perspective, which can positively impact experiment design as well as survey materials. Further benefits included advice around recruitment and reaching a greater audience with dissemination activities, amongst others. There are challenges associated with PPI activities; examples include time, cost and outlining expectations. Overall, although we acknowledge practical difficulties associated with PPI, this work highlights that it is possible for preference researchers to implement PPI across preference research. Further research systematically comparing methods related to PPI in preference research and their associated impact on the methods and results of studies would strengthen the literature. Electronic supplementary material The online version of this article (10.1007/s40271-020-00439-2) contains supplementary material, which is available to authorized users.

Background There are increasing examples of linking data on healthcare resource use and patient outcomes from different sectors of health and social care systems. Linked data are generally anonymised, meaning in most jurisdictions there are no legal restrictions to their use in research conducted by public or private organisations. Secondary use of anonymised linked data is contentious in some jurisdictions but other jurisdictions are known for their use of linked data. The publics’ perceptions of the acceptability of using linked data is likely to depend on a number of factors. This study aimed to quantify the preferences of the public to understand the factors that affected views about types of linked data and its use in two jurisdictions. Method An online discrete choice experiment (DCE) previously conducted in Scotland was adapted and replicated in Sweden. The DCE was designed, comprising five attributes, to elicit the preferences from a representative sample of the public in both jurisdictions. The five attributes (number of levels) were: type of researcher using linked data (four); type of data being linked (four); purpose of the research (three); use of profit from using linked data (four); who oversees the research (four). Each DCE contained 6 choice-sets asking respondents to select their preferred option from two scenarios or state neither were acceptable. Background questions included socio-demographics. DCE data were analysed using conditional and heteroskedastic conditional logit models to create forecasts of acceptability. Results The study sample comprised members of the public living in Scotland (n = 1004) and Sweden (n = 974). All five attributes were important in driving respondents’ choices. Swedish and Scottish preferences were mostly homogenous with the exception of ‘who oversees the research using linked data’, which had relatively less impact on the choices observed from Scotland. For a defined ‘typical’ linked data scenario, the probability (on average) of acceptance was 85.7% in Sweden and 82.4% in Scotland. Conclusion This study suggests that the public living in Scotland and Sweden are open to using anonymised linked data in certain scenarios for research purposes but some caution is advisable if the anonymised linked data joins health to non-health data.

There is emerging interest in the use of discrete choice experiments as a means of quantifying the perceived balance between benefits and risks (quantitative benefit-risk assessment) of new healthcare interventions, such as medicines, under assessment by regulatory agencies. For stated preference data on benefit-risk assessment to be used in regulatory decision making, the methods to generate these data must be valid, reliable and capable of producing meaningful estimates understood by decision makers. Some reporting guidelines exist for discrete choice experiments, and for related methods such as conjoint analysis. However, existing guidelines focus on reporting standards, are general in focus and do not consider the requirements for using discrete choice experiments specifically for quantifying benefit-risk assessments in the context of regulatory decision making. This opinion piece outlines the current state of play in using discrete choice experiments for benefit-risk assessment and proposes key areas needing to be addressed to demonstrate that discrete choice experiments are an appropriate and valid stated preference elicitation method in this context. Methodological research is required to establish: how robust the results of discrete choice experiments are to formats and methods of risk communication; how information in the discrete choice experiment can be presented effectually to respondents; whose preferences should be elicited; the correct underlying utility function and analytical model; the impact of heterogeneity in preferences; and the generalisability of the results. We believe these methodological issues should be addressed, alongside developing a 'reference case', before agencies can safely and confidently use discrete choice experiments for quantitative benefit-risk assessment in the context of regulatory decision making for new medicines and healthcare products.

Background Public health ethics and law (PHEL) is a core professional competency for the public health workforce. However, few data are available describing the extent to which UK public health workforce members experience ethical and legal issues or have sufficient educational and/or training background to adequately deal with such issues. Methods An anonymous online survey was developed for dissemination via member mailing lists of the: Faculty of Public Health, Royal Society of Public Health, and UK Public Health Register. Public Health England also included a link to the survey in their newsletter. The survey included questions about education, training, and experience in relation to PHEL. The survey was deployed from October 2017 to January 2018. Results The survey was completed by a diverse sample of five hundred and sixty-two individuals. The majority of respondents reported: (i) regularly encountering ethical issues, (ii) resolving ethical issues through personal reflection, (iii) having little or no education and training in PHEL, and (iv) questioning whether they have dealt with ethical issues encountered in practice in the best way. Conclusions The results suggest that there is a need to develop and support wider PHEL capacity within the UK public health workforce through the provision of PHEL education, training, guidance, and mentoring.

Background Online survey-based methods are increasingly used to elicit preferences for healthcare. This digitization creates an opportunity for interactive survey elements, potentially improving respondents’ understanding and/or engagement. Objective Our objective was to understand whether, and how, training materials in a survey influenced stated preferences. Methods An online discrete-choice experiment (DCE) was designed to elicit public preferences for a new targeted approach to prescribing biologics (“biologic calculator”) for rheumatoid arthritis (RA) compared with conventional prescribing. The DCE presented three alternatives, two biologic calculators and a conventional approach (opt out), described by five attributes: delay to treatment, positive predictive value, negative predictive value, infection risk, and cost saving to the national health service. Respondents were randomized to receive training materials as plain text or an animated storyline. Training materials contained information about RA and approaches to treatment and described the biologic calculator. Background questions included sociodemographics and self-reported measures of task difficulty and attribute non-attendance. DCE data were analyzed using conditional and heteroskedastic conditional logit (HCL) models. Results In total, 300 respondents completed the DCE, receiving either plain text (n = 158) or the animated storyline (n = 142). The HCL showed the estimated coefficients for all attributes aligned with a priori expectations and were statistically significant. The scale term was statistically significant, indicating that respondents who received plain-text materials had more random choices. Further tests suggested preference homogeneity after accounting for differences in scale. Conclusions Using animated training materials did not change the preferences of respondents, but they appeared to improve choice consistency, potentially allowing researchers to include more complex designs with increased numbers of attributes, levels, alternatives or choice sets. Electronic supplementary material The online version of this article (10.1007/s40271-019-00391-w) contains supplementary material, which is available to authorized users.

The desire to understand the preferences of patients, healthcare professionals and the public continues to grow. Health valuation studies, often in the form of discrete choice experiments, a choice based survey approach, proliferate as a result. A variety of methods of pre-choice process analysis have been developed to investigate how and why people make their decisions in such experiments and surveys. These techniques have been developed to investigate how people acquire and process information and make choices. These techniques offer the potential to test and improve theories of choice and/or associated empirical models. This paper provides an overview of such methods, with the focus on their use in stated choice-based healthcare studies. The methods reviewed are eye tracking, mouse tracing, brain imaging, deliberation time analysis and think aloud. For each method, we summarise the rationale, implementation, type of results generated and associated challenges, along with a discussion of possible future developments.

This article reports and reflects on an element of a recent survey of UK public health professionals, specifically in relation to the Public Health Knowledge and Skills Framework (PHSKF) and the ethical requirements that underpin public health practice. Only 38.4% of respondents reported accessing the PHKSF and a mere 13.7% reported accessing the accompanying background paper on ethical public health practice. Given that ethical practice underpins the PHSKF, it is concerning that so few respondents are familiar with the PHSKF and one of the source documents. While issuing frameworks and guidance is one way to support public health practice, there is a further need for greater integration of skills and knowledge around ethical public health practice within education and training initiatives.

Background Many countries offer screening programmes to unborn and newborn babies (antenatal and newborn screening) to identify those at risk of certain conditions to aid earlier diagnosis and treatment. Technological advances have stimulated the development of screening programmes to include more conditions, subsequently changing the information required and potential benefit-risk trade-offs driving participation. Quantifying preferences for screening programmes can provide programme commissioners with data to understand potential demand, the drivers of this demand, information provision required to support the programmes and the extent to which preferences differ in a population. This study aimed to identify published studies eliciting preferences for antenatal and newborn screening programmes and provide an overview of key methods and findings. Methods A systematic search of electronic databases for key terms identified eligible studies (discrete choice experiments (DCEs) or best-worst scaling (BWS) studies related to antenatal/newborn testing/screening published between 1990 and October 2018). Data were systematically extracted, tabulated and summarised in a narrative review. Results A total of 19 studies using a DCE or BWS to elicit preferences for antenatal (n = 15; 79%) and newborn screening (n = 4; 21%) programmes were identified. Most of the studies were conducted in Europe (n = 12; 63%) but there were some examples from North America (n = 2; 11%) and Australia (n = 2; 11%). Attributes most commonly included were accuracy of screening (n = 15; 79%) and when screening occurred (n = 13; 68%). Other commonly occurring attributes included information content (n = 11; 58%) and risk of miscarriage (n = 10; 53%). Pregnant women (n = 11; 58%) and healthcare professionals (n = 11; 58%) were the most common study samples. Ten studies (53%) compared preferences across different respondents. Two studies (11%) made comparisons between countries. The most popular analytical model was a standard conditional logit model (n = 11; 58%) and one study investigated preference heterogeneity with latent class analysis. Conclusion There is an existing literature identifying stated preferences for antenatal and newborn screening but the incorporation of more sophisticated design and analytical methods to investigate preference heterogeneity could extend the relevance of the findings to inform commissioning of new screening programmes.

Background: Risk is increasingly used as an attribute in discrete choice experiments (DCEs). However, risk and probabilities are complex concepts that can be open to misinterpretation, potentially undermining the robustness of DCEs as a valuation method. This study aimed to understand how respondents made benefit-risk trade-offs in a DCE and if these were affected by the communication of the risk attributes.Methods: Female members of the public were recruited via local advertisements to participate in think-aloud interviews when completing a DCE eliciting their preferences for a hypothetical breast screening programme described by three attributes: probability of detecting a cancer; risk of unnecessary follow-up; and cost of screening. Women were randomised to receive risk information as either: 1) percentages or 2) percentages and icon arrays. Interviews were digitally recorded then transcribed to generate qualitative data for thematic analysis. Results: Nineteen women completed the interviews (icon arrays n=9; percentages n=10). Analysis revealed four key themes where women made references to: 1) the nature of the task; 2) their feelings; 3) their experiences, for instance making analogies to similar risks; and 4) economic phenomena such as opportunity costs and discounting. Conclusion: Most women completed the DCE in line with economic theory; however, violations were identified. Women appeared to visualise risk whether they received icon arrays or percentages only. Providing clear instructions and graphics to aid interpretation of risk and qualitative piloting to verify understanding is recommended. Further investigation is required to determine if the process of verbalising thoughts changes the behaviour of respondents.

Objectives: Discrete choice experiments (DCEs) are increasingly advocated as a wayto quantify preferences for health. However, increasing support does not necessarilyresult in increasing quality. Although specific reviews have been conducted in certaincontexts, there exists no recent description of the general state of the science ofhealth-related DCEs. The aim of this paper was to update prior reviews (1990-2012), toidentify all health-related DCEs and to provide a description of trends, current practiceand future challenges.Methods: A systematic literature review was conducted to identify health-relatedempirical DCEs published between 2013-2017. The search strategy and dataextraction replicated prior reviews to allow the reporting of trends, although additionalextraction fields were incorporated.Results: Of the 7877 abstracts generated, 301 studies met the inclusion criteria andunderwent data extraction. In general, the total number of DCEs per year continued toincrease, with broader areas of application and increased geographic scope. Studiesreported using more sophisticated designs (e.g. D-efficient) with associated software(e.g. Ngene). The trend towards using more sophisticated econometric models alsocontinued. However, many studies presented sophisticated methods with insufficientdetail. Qualitative research methods continued to be a popular approach for identifyingattributes and levels.Conclusions: The use of empirical DCEs in health economics continues to grow.However, inadequate reporting of methodological details inhibits quality assessment.This may reduce decision-makers’ confidence in results and their ability to act on thefindings. How and when to integrate health-related DCE outcomes into decision-makingremains an important area for future research.

Background. Discrete choice experiments (DCEs) are increasingly used to elicit preferences for benefit-risk tradeoffs. The primary aim of this study was to explore how eye-tracking methods can be used to understand DCE respondents’ decision-making strategies. A secondary aim was to explore if the presentation and communication of risk affected respondents’ choices. Method. Two versions of a DCE were designed to understand the preferences of female members of the public for breast screening that varied in how risk attributes were presented. Risk was communicated as either 1) percentages or 2) icon arrays and percentages. Eye-tracking equipment recorded eye movements 1000 times a second. A debriefing survey collected sociodemographics and self-reported attribute nonattendance (ANA) data. A heteroskedastic conditional logit model analyzed DCE data. Eye-tracking data on pupil size, direction of motion, and total visual attention (dwell time) to predefined areas of interest were analyzed using ordinary least squares regressions. Results. Forty women completed the DCE with eye-tracking. There was no statistically significant difference in attention (fixations) to attributes between the risk communication formats. Respondents completing either version of the DCE with the alternatives presented in columns made more horizontal (left-right) saccades than vertical (up-down). Eye-tracking data confirmed self-reported ANA to the risk attributes with a 40% reduction in mean dwell time to the “probability of detecting a cancer” ( P = 0.001) and a 25% reduction to the “risk of unnecessary follow-up” ( P = 0.008). Conclusion. This study is one of the first to show how eye-tracking can be used to understand responses to a health care DCE and highlighted the potential impact of risk communication on respondents’ decision-making strategies. The results suggested self-reported ANA to cost attributes may not be reliable.

OBJECTIVES: Scale heterogeneity, or differences in the error variance of choices, may account for a significant amount of the observed variation in the results of discrete choice experiments (DCEs) when comparing preferences between different groups of respondents. The aim of this study was to identify if, and how, scale heterogeneity has been addressed in healthcare DCEs which compare the preferences of different groups.METHODS: A systematic review identified all healthcare DCEs published between 1990 and February 2016. The full-text of each DCE was then screened to identify studies which compared preferences using data generated from multiple groups. Data were extracted and tabulated on: year of publication; samples compared; tests for scale heterogeneity; analytical methods to account for scale heterogeneity. Narrative analysis was used to describe if, and how, scale heterogeneity was accounted for when preferences were compared.RESULTS: A total of 626 healthcare DCEs were identified. Of these 199 (32%) aimed to compare the preferences of different groups specified at the design stage while 79 (13%) compared the preferences of groups identified at the analysis stage. Of the 278 included papers, 49 (18%) discussed potential scale issues, 18 (7%) used a formal method of analysis to account for scale between groups and 2 (1%) accounted for scale differences between preference groups at the analysis stage. Scale heterogeneity was present in 65% (n=13) of studies which tested for it. Analytical methods to test for scale heterogeneity included: coefficient plots (n=5, 2%); heteroscedastic conditional logit models (n=6, 2%); Swait and Louviere tests (n=4, 1%); generalised multinomial logit models (n=5, 2%); and scale-adjusted latent class analysis (n=2, 1%).CONCLUSIONS: Scale heterogeneity is a prevalent issue in healthcare DCEs. Despite this, few published DCEs have discussed such issues and fewer still have used formal methods to identify and account for the impact of scale heterogeneity. The use of formal methods to test for scale heterogeneity should be used otherwise the results of DCEs potentially risk producing biased and potentially misleading conclusions regarding preferences for aspects of healthcare.

Background. The use of qualitative research (QR) methods is recommended as good practice in discrete choice experiments (DCEs). This study investigated the use and reporting of QR to inform the design and/or interpretation of healthcare-related DCEs and explored the perceived usefulness of such methods. Methods. DCEs were identified from a systematic search of the MEDLINE database. Studies were classified by the quantity of QR reported (none, basic, or extensive). Authors (n = 91) of papers reporting the use of QR were invited to complete an online survey eliciting their views about using the methods. Results. A total of 254 healthcare DCEs were included in the review; of these, 111 (44%) did not report using any qualitative methods; 114 (45%) reported “basic” information; and 29 (11%) reported or cited “extensive” use of qualitative methods. Studies reporting the use of qualitative methods used them to select attributes and/or levels (n = 95; 66%) and/or pilot the DCE survey (n = 26; 18%). Popular qualitative methods included focus groups (n = 63; 44%) and interviews (n = 109; 76%). Forty-four studies (31%) reported the analytical approach, with content (n = 10; 7%) and framework analysis (n = 5; 4%) most commonly reported. The survey identified that all responding authors (n = 50; 100%) found that qualitative methods added value to their DCE study, but many (n = 22; 44%) reported that journals were uninterested in the reporting of QR results. Conclusions. Despite recommendations that QR methods be used alongside DCEs, the use of QR methods is not consistently reported. The lack of reporting risks the inference that QR methods are of little use in DCE research, contradicting practitioners’ assessments. Explicit guidelines would enable more clarity and consistency in reporting, and journals should facilitate such reporting via online supplementary materials.

Background The relative benefits and risks of screening programs for breast cancer have been extensively debated. Objectives To quantify and investigate heterogeneity in women's preferences for the benefits and risks of a national breast screening program (NBSP) and to understand the effect of risk communication format on these preferences. Methods An online discrete choice experiment survey was designed to elicit preferences from female members of the public for an NBSP described by three attributes (probability of detecting a cancer, risk of unnecessary follow-up, and out-of-pocket screening costs). Survey respondents were randomized to one of two surveys, presenting risk either as percentages only or as icon arrays and percentages. Respondents were required to choose between two hypothetical NBSPs or no screening in 11 choice sets generated using a Bayesian D-efficient design. The trade-offs women made were analyzed using heteroskedastic conditional logit and scale-adjusted latent class models. Results A total of 1018 women completed the discrete choice experiment (percentages-only version=507; icon arrays and percentages version=511). The results of the heteroskedastic conditional logit model suggested that, on average, women were willing-to-accept 1.72 (confidence interval 1.47–1.97) additional unnecessary follow-ups and willing-to-pay £79.17 (confidence interval £66.98–£91.35) for an additional cancer detected per 100 women screened. Latent class analysis indicated substantial heterogeneity in preferences with six latent classes and three scale classes providing the best fit. The risk communication format received was not a predictor of scale class or preference class membership. Conclusions Most women were willing to trade-off the benefits and risks of screening, but decision makers seeking to improve uptake should consider the disparate needs of women when configuring services.

Discrete choice experiments (DCEs) are used to quantify the preferences of specified sample populations for different aspects of a good or service and are increasingly used to value interventions and services related to healthcare. Systematic reviews of healthcare DCEs have focussed on the trends over time of specific design issues and changes in the approach to analysis, with a more recent move towards consideration of a specific type of variation in preferences within the sample population, called taste heterogeneity, noting rises in the popularity of mixed logit and latent class models. Another type of variation, called scale heterogeneity, which relates to differences in the randomness of choice behaviour, may also account for some of the observed 'differences' in preference weights. The issue of scale heterogeneity becomes particularly important when comparing preferences across subgroups of the sample population as apparent differences in preferences could be due to taste and/or choice consistency. This primer aims to define and describe the relevance of scale heterogeneity in a healthcare context, and illustrate key points, with a simulated data set provided to readers in the Online appendix.

Background: Understanding preferences for the process and outcomes of clinical genetics services (CGS) is a first step to developing these services appropriately. Aim: The aim of this study was to quantify the relative importance of attributes defining the process of service delivery and the patient outcomes of CGS. Methods: An online hybrid conjoint analysis discrete choice experiment (CA-DCE) was piloted in a purposive sample (n = 37) of CGS patients and non-patients to identify (i) service attributes (n = 13) perceived to facilitate informed decision making; (ii) relative preferences for six attributes (5 process, 1 outcome: ability to make an informed decision). A three-step approach was taken to link the data from the CA-DCE using hierarchical information integration and ordered logit and multinomial logit models. Marginal willingness-to-pay (WTP) values were calculated. Results: Services that facilitate informed decision making, with shorter waiting times and involving pre-consultation contact were preferred. Estimated WTP values were: service location (£3170; 95 % CI −391 to 15,098); waiting time (−£1080; 95 % CI −3659 to −603); pre-consultation contact (£7765; 95 % CI 2542–33,937); improved informed decision making (£2254; 95 % CI 775–9866). Conclusion: This study suggests that hybrid stated preference experiments offer a practical solution to understanding preferences for how CGS services are delivered. Refereed/Peer-reviewed

Discrete choice experiments (DCEs) are used to elicit preferences of current and future patients and healthcare professionals about how they value different aspects of healthcare. Risk is an integral part of most healthcare decisions. Despite the use of risk attributes in DCEs consistently being highlighted as an area for further research, current methods of incorporating risk attributes in DCEs have not been reviewed explicitly. This study aimed to systematically identify published healthcare DCEs that incorporated a risk attribute, summarise and appraise methods used to present and analyse risk attributes, and recommend best practice regarding including, analysing and transparently reporting the methodology supporting risk attributes in future DCEs. The Web of Science, MEDLINE, EMBASE, PsycINFO and Econlit databases were searched on 18 April 2013 for DCEs that included a risk attribute published since 1995, and on 23 April 2013 to identify studies assessing risk communication in the general (non-DCE) health literature. Healthcare-related DCEs with a risk attribute mentioned or suggested in the title/abstract were obtained and retained in the final review if a risk attribute meeting our definition was included. Extracted data were tabulated and critically appraised to summarise the quality of reporting, and the format, presentation and interpretation of the risk attribute were summarised. This review identified 117 healthcare DCEs that incorporated at least one risk attribute. Whilst there was some evidence of good practice incorporated into the presentation of risk attributes, little evidence was found that developing methods and recommendations from other disciplines about effective methods and validation of risk communication were systematically applied to DCEs. In general, the reviewed DCE studies did not thoroughly report the methodology supporting the explanation of risk in training materials, the impact of framing risk, or exploring the validity of risk communication. The primary limitation of this review was that the methods underlying presentation, format and analysis of risk attributes could only be appraised to the extent that they were reported. Improvements in reporting and transparency of risk presentation from conception to the analysis of DCEs are needed. To define best practice, further research is needed to test how the process of communicating risk affects the way in which people value risk attributes in DCEs.