Estimating health state utility values from discrete choice experiments--a QALY space model approach

The EORTC QLU-C10D: the Hong Kong valuation study

OBJECTIVE

The EORTC QLU-C10D is a new preference-based measure derived from the EORTC QLQ-C30. Country-specific value sets are required to support the cost-utility analysis of cancer-related interventions. This study aimed to generate an EORTC QLU-C10 value set for Hong Kong (HK).

METHODS

A HK online panel was quota-sampled to achieve an adult general population sample representative by sex and age. Participants were invited to complete an online discrete choice experiment survey. Each participant was asked to complete 16 choice-pairs, randomly assigned from a total of 960 choice-pairs, each comprising two QLU-C10D health states and a duration attribute. Conditional and mixed logistic regression analyses were used to analyse the data.

RESULTS

The analysis included data from 1041 respondents who had successfully completed the online survey. The distribution of sex did not differ from that of the general population, but a significant difference was found among age groups. A weighting analysis for non-representative variable (age) was used. Utility decrements were generally monotonic, with the largest decrements for physical functioning (- 0.308), role functioning (- 0.165), and pain (- 0.161). The mean QLU-C10D utility score of the participants was 0.804 (median = 0.838, worst to best = - 0.169 to 1). The value of the worst health state was - 0.223, which was sufficiently lower than 0 (being dead).

CONCLUSIONS

This study established HK utility weights for the QLU-C10D, which can facilitate cost-utility analyses across cancer-related health programmes and technologies.

Estimating the EQ-5D-5L value set for the Philippines

Background

The Philippines has recommended the use of Quality-Adjusted Life Years (QALYs) in government health technology assessments (HTA). We aimed to develop a value set for the EQ-5D-5L based on health preferences of the healthy general adult population in the Philippines.

Methods

Healthy, literate adults were recruited from the Philippine general population with quota targets based on age, sex, administrative region, type of residence, education, income, and ethnolinguistic groups. Each participant’s preference was elicited by completing Composite Time Trade-Off (C-TTO) and Discrete Choice Experiment (DCE) tasks. Tasks were computer-assisted using the EuroQol Valuation Technology 2.0. To estimate the value set, we explored 20- and 8-parameter models that either use c-TTO-only data or both c-TTO and DCE (also called hybrid models). Final model choice was guided by principles of monotonicity, out-of-sample likelihood, model fit, and parsimony.

Results

We recruited 1000 respondents with demographic characteristics that approximate the general population such as 49.6% Female, 82% Roman Catholic, 40% in urban areas, and 55% finished high school. None of the 20-parameter models demonstrated monotonicity (logical worsening of coefficients with increasing severity). From the 8-parameter models, the homoscedastic TTO-only model exhibited the best fit. From this model, mobility and pain/ discomfort had the highest effect on utilities.

Conclusion

The selected model for representing the Philippine general population preferences for EQ-5D-5L health states was an 8-parameter homoscedastic TTO-only model. This value set is recommended for use in QALY calculations in support of HTA-informed coverage decisions in the Philippines.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11136-022-03143-w.

Willingness to pay for a quality-adjusted life year: a systematic review with meta-regression

The use of a threshold for cost-utility studies is of major importance to health authorities for making the best allocation decisions for limited resources. Regarding the increasing number of studies worldwide that seek to establish a value for a quality-adjusted life year (QALY), it is necessary to review these studies to provide a global insight into the literature. A systematic review on willingness to pay (WTP) studies focusing on QALY was conducted in eight databases up to June 26, 2020. From a total of 9991 entries, 39 studies were selected, and 511 observations were extracted for the meta-analysis using the ordinary least squares method. The results showed a predicted mean empirical value of $52,619.39 (95% CI 49,952.59; 55,286.19) per QALY in US dollars for 2018. A 1% increase in income led to an increase of 0.6% in the WTP value, while a 1-year increase in respondent age led to a decrease of 3.3% in the WTP value. Sex, education level and employment status had significant effects on WTP. Compared to face-to-face interviews, surveys conducted by the internet or telephone were more likely to have a significantly higher value of WTP per QALY, while out-of-pocket payment tended to lower the value. The prediction made for the province of Quebec, Canada, provided a QALY value of approximately USD $98,450 (CAD $127,985), which is about 2.3 times its gross domestic product (GDP) per capita in 2018. This study is consistent with the extant literature and will be useful for countries that do not yet have a preference-based survey for the value of a QALY.

Dutch utility weights for the EORTC cancer-specific utility instrument: the Dutch EORTC QLU-C10D

Purpose

To measure utilities among cancer patients, a cancer-specific utility instrument called the European Organization for Research and Treatment of Cancer (EORTC) QLU-C10D has been developed based on EORTC quality of life core module (QLQ-C30). This study aimed to provide Dutch utility weights for the QLU-C10D.

Methods

A cross-sectional valuation study was performed in 1017 participants representative in age and gender of the Dutch general population. The valuation method was a discrete choice experiment containing 960 choice sets, i.e. pairs of QLU-C10D health states, each health state described in terms of the 10 QLU-C10D domains and the duration of that health state. Each participant considered 16 choice sets, choosing their preferred health state from each pair. Utility scores were derived using generalized estimation equation models. Non-monotonic levels were combined.

Results

Utility decrements were generated for all 10 QLU-C10D domains, with largest decrements for pain (− 0.242), physical functioning (− 0.228), and role functioning (− 0.149). Non-monotonic levels of emotional functioning, pain, fatigue, sleep problems, and appetite loss were combined. No decrement in utility was seen in case of a little or quite a bit impairment in emotional functioning or a little pain. The mean QLU-C10D utility score of the participants was 0.85 (median = 0.91, interquartile range = 0.82 to 0.96).

Conclusion

Dutch utility decrements were generated for the QLU-C10D. These are important for evaluating the cost-utility of new cancer treatments and supportive care interventions. Further insight is warranted into the added value of the QLU-C10D alongside other utility instruments.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11136-021-02767-8.

Consumer health informatics approach for personalized cancer screening decisions using utility functions

A consumer health informatics approach is used to investigate the development of a patient-centered decision support system (DSS) with individualized utility functions. It supports medical decisions that have uncertain benefits and potential harms. Its use for accepting or declining cancer screening is illustrated. The system's underlying optimization model incorporates two user-specific utility functions-one that quantifies life-saving benefits and one that quantifies harms, such as unnecessary follow-up tests, surgeries, or treatments. The system requires sound decision making. Therefore, the decision making process was studied using a decision aid in the form of a color-coded matrix with the potential outcomes randomly placed in proportion to their likelihoods. Data were collected from 48 study participants, based on a central composite experimental design. The results show that the DSS can be effective, but health consumers may not be rational decision makers.

The cost-effectiveness of adjunctive corticosteroids for patients with septic shock

OBJECTIVE

To determine whether hydrocortisone is a cost-effective treatment for patients with septic shock.

DESIGN

Data linkage-based cost-effectiveness analysis.

SETTING

New South Wales and Queensland intensive care units.

PARTICIPANTS AND INTERVENTION

Patients with septic shock randomly assigned to treatment with hydrocortisone or placebo in the Adjunctive Glucocorticoid Therapy in Patients with Septic Shock (ADRENAL) trial.

MAIN OUTCOME MEASURES

Health-related quality of life at 6 months using the EuroQoL 5-dimension 5-level questionnaire. Data on hospital resource use and costs were obtained by linking the ADRENAL dataset to government administrative health databases. Clinical outcomes included mortality, health-related quality of life, and quality-adjusted life-years gained; economic outcomes included hospital resource use, costs and cost-effectiveness from the health care payer perspective. We also assessed cost-effectiveness by sex. To increase the precision of cost-effectiveness estimates, we conducted unrestricted bootstrapping.

RESULTS

Of 3800 patients in the ADRENAL trial, 1772 (46.6%) were eligible and 1513 (85.4% of those eligible) were included. There was no difference between hydrocortisone or placebo groups in regards to mortality (218/742 [29.4%] v 227/759 [29.9%]; HR, 0.93; 95% CI, 0.78-1.12; P = 0.47), mean number of QALYs gained (0.10 ± 0.09 v 0.10 ± 0.09; P = 0.52), or total hospital costs (A$73 515 ± 61 376 v A$69 748 ± 61 793; mean difference, A$3767; 95% CI, -A$2891 to A$10 425; P = 0.27). The incremental cost of hydrocortisone was A$1 254 078 per quality-adjusted life-year gained. In females, hydrocortisone was cost-effective in 46.2% of bootstrapped replications and in males it was cost-effective in 2.7% of bootstrapped replications.

CONCLUSIONS

Adjunctive hydrocortisone did not significantly affect longer term mortality, health-related quality of life, health care resource use or costs, and is unlikely to be cost-effective.

Discrete choice experiments to generate utility values for multi-attribute utility instruments: a systematic review of methods

OBJECTIVES

In recent years, discrete choice experiments (DCEs) have become frequently used to generate utility values, but there are a diverse range of approaches to do this. The primary focus of this systematic review is to summarise the methods used for the design and analysis of DCEs when estimating utility values in both generic and condition-specific preference-based measures.

METHODS

Published literature using DCEs to estimate utility values from preference-based instruments were identified from MEDLINE, Embase, Cochrane Library and CINAHL using PRISMA guidelines. To assess the different DCE methods, standardised information was extracted from the articles including the DCE design method, the number of choice sets, the number of DCE pairs per person, randomisation of questions, analysis method, logical consistency tests and techniques for anchoring utilities. The CREATE checklist was used to assess the quality of the studies.

RESULTS

A total of 38 studies with samples from the general population, students and patients were included. Values for health states described using generic multi attribute instruments (MAUIs) (especially the EQ-5D) were the most commonly explored using DCEs. The studies showed considerable methodology and design diversity (number of alternatives, attributes, sample size, choice task presentation and analysis). Despite these differences, the quality of articles reporting the methods used for the DCE was generally high.

CONCLUSION

DCEs are an important approach to measure utility values for both generic and condition-specific instruments. However, a gold standard method cannot yet be recommended.

German value sets for the EORTC QLU-C10D, a cancer-specific utility instrument based on the EORTC QLQ-C30

PURPOSE

The EORTC QLU-C10D is a new multi-attribute utility instrument derived from the EORTC QLQ-C30, a widely used cancer-specific quality of life questionnaire. It covers ten dimensions: physical, role, social, emotional functioning, pain, fatigue, sleep, appetite, nausea, and bowel problems. To allow national health attitudes to be reflected, country-specific valuations are being performed by collaboration of the Multi-Attribute Utility Cancer (MAUCa) Consortium and the EORTC. The purpose of this paper is to provide German value sets (utility weights) for the QLU-C10D.

METHODS

Valuations were run in a web-based setting in two general population samples of approximately 2000 adults in total. As the German version of the QLQ-C30 is presently undergoing a revision of the wording of one response category, valuations for both the current and the new version were performed (Germany 1 and 2). Utilities were elicited using a discrete choice experiment (DCE). Data were analyzed by conditional logistic regression and mixed logits.

RESULTS

Completion rates were 88.3% (1002/1135) and 90.4% (1016/1124) for Germany 1 and Germany 2 valuations, respectively. Dimensions with the largest impact on utility weights were, in this order: physical functioning, pain, role functioning, social functioning and nausea (same ordering for both German versions). Several violations of the logical ordering of levels were observed for Germany 1; this was largely improved for Germany 2.

CONCLUSION

This study established German utility weights for the cancer-specific utility instrument QLU-C10D.

Overview, Update, and Lessons Learned From the International EQ-5D-5L Valuation Work: Version 2 of the EQ-5D-5L Valuation Protocol

A standardized 5-level EuroQol 5-dimensional questionnaire (EQ-5D-5L) valuation protocol was first used in national studies in the period 2012 to 2013. A set of problems encountered in this initial wave of valuation studies led to the subsequent refinement of the valuation protocol. To clarify lessons learned and how the protocol was updated when moving from version 1.0 to the current version 2.1 and 2.0, this article will (1) present the challenges faced in EQ-5D-5L valuation since 2012 and how these were resolved and (2) describe in depth a set of new challenges that have become central in currently ongoing research on how EQ-5D-5L health states should be valued and modeled.

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.

Assessing utility values for treatment-related health states of acute myeloid leukemia in the United States

Background

Preference valuations of health status are essential in health technology and economic appraisal. This study estimated utilities for treatment-related health states of acute myeloid leukemia (AML) and disutilities of severe adverse events (SAEs) using a representative sample of adults from the general population in the United States (US).

Methods

Treatment-related AML health states, defined based on literature and interviews with clinicians, included complete remission (CR), no CR, relapse, stem cell transplant (SCT), and post SCT short-term recovery. Six attributes with varying levels, including fever, lack of energy, problems with daily function, anxiety/depression, blood transfusions, and hospitalization, were used to define health states. An online survey using discrete choice experiment methodology was designed to capture preferences for health status scenarios including the identified attributes and key grade 3/4 chemotherapy-related SAEs. Health state utilities and SAE disutilities were generated from a conditional logistic regression with generalized estimating equations.

Results

Of the 300 survey participants, the demographic distributions were within a 3% margin of those in the 2010 US Census. CR had the highest utility value (0.875), followed by post-SCT short-term recovery (0.398), relapse (0.355), no CR (0.262), and SCT (0.158). Of the SAEs, serious infection had the highest decline in utility (0.218), followed by severe diarrhea (0.176), abnormally low blood cell counts (0.100), and severe redness/skin peeling (0.060).

Conclusions

AML and treatments can result in reduced quality of life and impaired ability to perform daily activities. Findings of this study underline the value that society places on treatment-related AML health states and SAEs.

Comparison of Value Set Based on DCE and/or TTO Data: Scoring for EQ-5D-5L Health States in Japan

BACKGROUND

The valuation study of the five-level version of the EuroQol five-dimensional questionnaire (EQ-5D-5L) involved composite time trade-off (cTTO) and a discrete choice experiment (DCE). The DCE scores must be anchored to the quality-of-life scale from 0 (death) to 1 (full health). Nevertheless, the characteristics of the statistical methods used for converting the EQ-5D-5L DCE results by using TTO information are not yet clearly known.

OBJECTIVES

To present the Japanese DCE value set of the EQ-5D-5L and compare three methods for converting latent DCE values.

METHODS

The survey sampled the general population at five locations in Japan. 1098 respondents were stratified by age and sex. To obtain and compare the value sets of the EQ-5D-5L, the cTTO and DCE data were analyzed by a linear mixed model and conditional logit, respectively. The DCE scores were converted to the quality-of-life scale by anchoring to the worst state using cTTO, mapping DCE onto cTTO, and a hybrid model.

RESULTS

The data from 1026 respondents were analyzed. All the coefficients in the cTTO and DCE value sets were consistent throughout all the analyses. Compared with the cTTO algorithm, the mapping and hybrid methods yielded very similar scoring coefficients. The hybrid model results, however, produced a lower root mean square error and fewer health states with errors exceeding 0.05 than did the other models. The DCE anchored to the worst state overestimated the cTTO scores of almost all the health states.

CONCLUSIONS

Japanese value sets based on DCE were demonstrated. On comparing the observed cTTO scores, we found that the hybrid model was slightly superior to the simpler methods, including the TTO model.