A head-to-head comparison of the EQ-5D-5L and 15D descriptive systems and index values in a general population sample.

ABSTRACT

BACKGROUND: The EQ-5D-5L and 15D are generic preference-accompanied health status measures with similar dimensions. In this study, we aim to compare the measurement properties of the EQ-5D-5L and 15D descriptive systems and index values in a general population sample. METHODS: In August 2021, an online cross-sectional survey was conducted in a representative adult general population sample (n = 1887). The EQ-5D-5L and 15D descriptive systems and index values were compared in terms of ceiling and floor, informativity (Shannon's Evenness index), agreement, convergent and known-groups validity for 41 chronic physical and mental health conditions. Danish value sets were used to compute index values for both instruments. As a sensitivity analysis, index values were also estimated using the Hungarian EQ-5D-5L and Norwegian 15D value sets. RESULTS: Overall, 270 (8.6%) and 1030 (3.4*10-6%) unique profiles occurred on the EQ-5D-5L and 15D. The EQ-5D-5L dimensions (0.51-0.70) demonstrated better informativity than those of 15D (0.44-0.69). EQ-5D-5L and 15D dimensions capturing similar areas of health showed moderate or strong correlations (0.558-0.690). The vision, hearing, eating, speech, excretion and mental function 15D dimensions demonstrated very weak or weak correlations with all EQ-5D-5L dimensions, which may indicate potential room for EQ-5D-5L bolt-ons. The 15D index values showed lower ceiling than the EQ-5D-5L (21% vs. 36%). The mean index values were 0.86 for the Danish EQ-5D-5L, 0.87 for the Hungarian EQ-5D-5L, 0.91 for the Danish 15D and 0.81 for the Norwegian 15D. Strong correlations were found between the index values (Danish EQ-5D-5L vs. Danish 15D 0.671, Hungarian EQ-5D-5L vs. Norwegian 15D 0.638). Both instruments were able to discriminate between all chronic condition groups with moderate or large effect sizes (Danish EQ-5D-5L 0.688-3.810, Hungarian EQ-5D-5L 1.233-4.360, Danish 15D 0.623-3.018 and Norwegian 15D 1.064-3.816). Compared to the 15D, effect sizes were larger for the EQ-5D-5L in 88-93% of chronic condition groups. CONCLUSIONS: This is the first study to compare the measurement properties of the EQ-5D-5L and 15D in a general population sample. Despite having 10 fewer dimensions, the EQ-5D-5L performed better than the 15D in many aspects. Our findings help to understand the differences between generic preference-accompanied measures and support resource allocation decisions.