Revisiting health state preferences after 20 years: A new EQ-5D-3L value set for the Netherlands

INTRODUCTION: Value sets for the EQ-5D-3L have been developed since the 1990's, using methods that are now considered outdated. The Dutch EQ-5D-3L value set was developed using data collected in 2003, and population's preferences may have shifted over time. Consequently, the existing value set may no longer accurately represent the preferences of the Dutch general population. This study aims to develop a new EQ-5D-3L value set using state-of-the-art methods and compare it with the existing value set. METHODS: A nationally representative sample of 417 Dutch adults, stratified by age and sex, completed 12 composite time trade-off (cTTO) tasks via online EuroQol Valuation Technology interviews. The data were modelled using a Tobit model that accounts for heteroskedasticity and the censored nature of cTTO data. Agreement between the new and existing value set was assessed using Bland-Altman and scatter plots. RESULTS: Pain/discomfort and mobility were the most important dimensions in the new value set, whereas pain/discomfort and anxiety/depression were most important in the existing value set. The lowest health state value was - 0.723, considerably lower than the - 0.329 of the existing value set. The Bland-Altman and scatter plots indicated limited agreement between the two value sets. CONCLUSION: The new Dutch EQ-5D-3L value set differs substantially from the existing value set. We recommend its adoption and replacement of the previous value set. Our findings suggest that other countries with older value sets should consider similar updates to ensure accurate representation of contemporary societal preferences.