<p>Quality adjusted life years (QALYs) are used to measure health gains from interventions. A classical method to elicit QALY utility-weights is time trade-off (TTO). TTO is often applied assuming that linear utility of life duration holds, which has been shown to be frequently invalid, leading to biased elicitation outcomes. The corrective approach introduced by Lipman et al. (J Health Econ 28(7):843–854, 2019b. <a href="https://doi.org/10.1002/hec.3895">https://doi.org/10.1002/hec.3895</a>) corrects QALY-weights based on prospect theory (PT). Published studies have yielded results in line with theoretical predictions, but arguably low external validity. The present study investigates if corrected TTO-weights reflect preferences more accurately than uncorrected TTO-weights. The following method was applied: First, QALY-weights were elicited using TTO. Second, participant´s loss aversion was quantified based on PT, and QALY-weights were corrected accordingly. Third, corrected QALY-weights were validated by presenting corrected and uncorrected QALY-weights and asking participants to choose which more accurately reflects their preferences and enabling adjustments using a visual analogue scale (VAS). QALY-weights for three health states <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\({\beta}_{1-3}\)</EquationSource> </InlineEquation> (EQ-5D-5L health states 21211, 31231, 33342) were elicited from 66 German university students, using online interviews. The corrected QALY-weights and those which were adjusted by participants using VAS were lower than uncorrected QALY-weights. Most participants indicated that uncorrected QALY-weights reflect their preferences more accurately than corrected QALY-weights (Binomial test, <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\:p<0.05\)</EquationSource> </InlineEquation> for <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\({\beta}_{1}\)</EquationSource> </InlineEquation>, <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\(\:p>0.05\)</EquationSource> </InlineEquation> for <InlineEquation ID="IEq5"> <EquationSource Format="TEX">\({\beta}_{2-3}\)</EquationSource> </InlineEquation>). This suggests a need for correction, but that the corrective approach does not yield QALY-weights more consistent with direct elicitation of subject’s preferences. Future work should replicate this study with a representative sample and individual-level measures of utility curvature.</p>

The corrective approach for health utility measurement using time trade-off: a direct test of validity

期刊论文
发表日期：

2025年12月19日

查看全文

Jacob A. Richelshagen,
Katja C. Senn,
Stefan A. Lipman

摘要

Quality adjusted life years (QALYs) are used to measure health gains from interventions. A classical method to elicit QALY utility-weights is time trade-off (TTO). TTO is often applied assuming that linear utility of life duration holds, which has been shown to be frequently invalid, leading to biased elicitation outcomes. The corrective approach introduced by Lipman et al. (J Health Econ 28(7):843–854, 2019b. https://doi.org/10.1002/hec.3895) corrects QALY-weights based on prospect theory (PT). Published studies have yielded results in line with theoretical predictions, but arguably low external validity. The present study investigates if corrected TTO-weights reflect preferences more accurately than uncorrected TTO-weights. The following method was applied: First, QALY-weights were elicited using TTO. Second, participant´s loss aversion was quantified based on PT, and QALY-weights were corrected accordingly. Third, corrected QALY-weights were validated by presenting corrected and uncorrected QALY-weights and asking participants to choose which more accurately reflects their preferences and enabling adjustments using a visual analogue scale (VAS). QALY-weights for three health states \({\beta}_{1-3}\) (EQ-5D-5L health states 21211, 31231, 33342) were elicited from 66 German university students, using online interviews. The corrected QALY-weights and those which were adjusted by participants using VAS were lower than uncorrected QALY-weights. Most participants indicated that uncorrected QALY-weights reflect their preferences more accurately than corrected QALY-weights (Binomial test, \(\:p<0.05\) for \({\beta}_{1}\) , \(\:p>0.05\) for \({\beta}_{2-3}\) ). This suggests a need for correction, but that the corrective approach does not yield QALY-weights more consistent with direct elicitation of subject’s preferences. Future work should replicate this study with a representative sample and individual-level measures of utility curvature.