Modelling sociodemographic determinants of healthrelated quality of life (HRQoL) loss among older adults in Spain using discretised beta regression

Abstract

Purpose: This study aims to identify the main factors associated with health-related quality of life (HRQoL) loss among older adults in Spain by modelling EQ-5D-5L health utility outcomes using alternative regression frameworks. Methods: Data comes from the population-based health survey conducted in Catalonia (Spain) for 2024. The survey includes sociodemographic, health, and lifestyle information from 1,227 individuals aged 60 years and older. To account for the discrete and bounded nature of HRQoL scores, we use the discretised beta regression (DBR) to analyse factors associated with HRQoL loss, a modelling approach not previously applied to health utility outcomes. The performance of the DBR is compared with traditional zero-one inflated regression models. Results: The DBR model showed higher predictive accuracy. Greater perceived HRQoL loss was associated with economic hardship, weak social support, sex, obesity, disability and limited mobility. Ageing initially appears to be associated with a greater loss (mean=0.16, q5=0.12, q95=0.25), but this relationship was no longer supported once disability was accounted for (mean=0.03, q5=-0.02, q95=0.07), indicating that the observed deterioration with age is largely mediated by functional decline rather than ageing itself. Conclusion: DBR provides a theoretically grounded alternative to conventional regression methods commonly applied in HRQoL analysis and is particularly well suited to the distributional properties of the EQ-5D utility decrements. Perceived HRQoL loss in later life is not primarily driven by chronological age, as differences in HRQoL loss among older adults are linked to persistent social and economic inequalities rather than ageing itself.