Reporting heterogeneity in EQ-5D-5L and their impact for population health measurement by sex

Background

Evidence on reporting heterogeneity in health-related quality of life (HRQoL) measures suggest that using the response categories varies systematically for respondents with same underlying health state. This study addresses reporting heterogeneity in HRQoL by analysing sex differentials in the EQ-5D-5L.

Methods

Population-based survey from 2025 in Estonia provided nationally representative data on 4221 respondents aged 16–84 years of age. Hierarchical Ordered Probit (HOPIT) model was used to study reporting heterogeneity in 5 dimensions of EQ-5D (mobility, self-care, usual activities, pain/discomfort, anxiety/depression) by: a) constructing a latent health indicator based on self-reported data on diagnosis or treatment of 10 health conditions within past 12 months, b) modelling the threshold effect by sex for 5-level EQ-5D scales, where larger values indicate reporting more severe problems.

Results

In total 2777 (1081 males and 1696 females) responses were included to the analytical sample limited to non-missing data. At least some problems were reported by 30% in mobility, 12% in self-care, 29% in usual activities, 71% in pain/discomfort, and 61% in anxiety/depression domains. Based on preliminary analysis using HOPIT models, statistically significant threshold effects by sex were found for self-care (3 vs 4 level), usual activities (4 vs. 5), pain/discomfort (3 vs. 4), anxiety/depression (1 vs. 2). The regression coefficients indicated that respective cut-off values for females were higher for self-care and usual activities domains and lower for pain/discomfort and anxiety/depression domains.

Conclusions

Study found sex differentials in the use of 5-level response scale of EQ-5D-5L for a given latent health which provides evidence on reporting heterogeneity in HRQoL. While these findings are in accordance with literature suggesting socio-cultural impact on reporting behaviours, interpretation of HRQoL data warrants caution.

Key messages

• Men and women systematically use EQ-5D-5L response categories differently, even when their underlying health is similar.

• These reporting differences can bias comparisons in health outcomes and should be accounted for in health outcome analysis and policy.