Measuring health-related quality of life of disabled older adults in China: comparison of self-report and proxy-report data

Background and objective Health-related Quality-of-life is a key indicator in aged and long-term care, but disabled-older-adults cannot always self-report, making proxy-assessment necessary. However, evidence on self-proxy agreement from China is limited. We aimed to assess agreement between self-reported and proxy-reported EQ-5D-5L among disabled-older-adults by comparing proxy-proxy (P1) and proxy-person (P2) ratings with self-report, characterise individual-discrepancy patterns, and examine factors associated with self-proxy discrepancies. Methods In a cross-sectional survey in Nanjing, China, disabled-older-adults completed the self-report EQ-5D-5L and proxies completed proxy-proxy and proxy-person versions in fixed order. Perspective differences and agreement were evaluated using Friedman-tests, weighted kappa, intraclass-correlation-coefficients, and Bland–Altman plots. Individual-discrepancy-patterns were examined using three-stage EQ-5D health profile analysis, and multiple linear regression identified correlates of proxy–self discrepancies. Results Data from 641 dyads were analysed. Utility and EQ-VAS differed across perspectives overall and in most subgroups. Agreement was higher for observable dimensions (mobility, self-care and usual activities, k ≥ 0.75) than for subjective dimensions (pain/discomfort and anxiety/depression, k = 0.62–0.65). Agreement was high for utility (ICC 0.846–0.849) but lower for EQ-VAS (ICC 0.637–0.698). P2 showed slightly closer agreement with self-report than P1 among nursing-home, severe-cognitive-impairment, and formal-caregiver subgroups. Education, caregiver burden, depressive symptoms, and cognitive impairment were associated with self-proxy discrepancies. Conclusion Self- and proxy-reported EQ-5D-5L showed high utility agreement, with stronger agreement for observable than subjective dimensions. Proxy-person perspective aligned slightly more closely with self-report than proxy-proxy perspectives in specific care contexts. Proxy perspective should be selected according to care context in long-term care quality assessment and economic evaluation.