Head-to-head comparison between the EQ-5D-5L and the EQ-5D-3L in general population health surveys

Background: The EQ-5D has been frequently used in national health surveys. This study is a head-to-head comparison to assess how expanding the number of levels from three (EQ-5D-3L) to five in the new EQ-5D-5L version has improved its distribution, discriminatory power, and validity in the general population. Methods: A representative sample (N = 7554) from the Catalan Health Interview Survey 2011-2012, aged >= 18, answered both EQ-5D versions, and we evaluated the response redistribution and inconsistencies between them. To assess validity of this redistribution, we calculated the mean of the Visual Analogue Scale (VAS), which measures perceived health. The discriminatory power was examined with Shannon Indices, calculated for each dimension separately. Spanish preference value sets were applied to obtain utility indices, examining their distribution with statistics of central tendency and dispersion. We estimated the proportion of individuals reporting the best health state in EQ-5D-5L and EQ-5D-3L within groups of specific chronic conditions and their VAS mean. Results: A very small reduction in the percentage of individuals with the best health state was observed, from 61.8% in EQ-5D-3L to 60.8% in EQ-5D-5L. In contrast, a large proportion of individuals reporting extreme problems in the 3 L version moved to severe problems (level 4) in the 5 L version, particularly for pain/discomfort (75.5%) and anxiety/depression (66.4%). The average proportion of inconsistencies was 0.9%. The pattern of the perceived health VAS mean confirmed the hypothesis established a priori, supporting the validity of the observed redistribution. Shannon index showed that absolute informativity was higher in the 5 L version for all dimensions. The means (SD) of the Spanish EQ5D- 3L and EQ-5D-5L indices were 0.87 (0.25) and 0.89 (0.22). The proportion of individuals with the best health state within each specific chronic condition was very similar, regardless of the EQ-5D version (<= 30% in half of the 28 chronic conditions). Conclusion: Although the proportion of individuals with the best possible health state is still very high, our findings support that the increase of levels provided by the EQ-5D-5L contributed to the validity and discriminatory power of this new version to measure health in general population, as in the national health surveys.