Proxy reports on the EQ-5D-5L in residential aged care facilities: a validation study

BACKGROUND: The EuroQol 5-Dimension 5-Level questionnaire (EQ-5D-5L) is a common scale to measure health-related quality of life in residential aged care facilities. Many residents are unable to answer the scale independently; therefore, a proxy usually responds on the residents' behalf. The quality of data reported by proxy is not well understood. This study examines the psychometric properties, including convergent validity, known-group validity, interrater reliability, and responsiveness of proxy reports on the EQ-5D-5L. METHODS: The study used data from 281 residents participating in a 12-month deprescribing randomised control trial in Australia. Interview-administered questionnaires were used for data collection. Convergent validity, known-group validity, and responsiveness were evaluated using a hypothesis-testing approach. Interrater reliability was assessed by calculating the intra-class correlation coefficient, weighted Cohen's kappa coefficient, and generating Bland-Altman plots and bar charts. RESULTS: Proxy (staff) reports on EQ-5D-5L were significantly associated with residents' cognitive functioning, physical functioning, neuropsychiatric behaviours and frailty. Proxy (staff) reports on EQ-5D-5L were significantly different across groups of patients with varying levels of cognitive and physical functioning, frailty and neuropsychiatric behaviours. Changes over 12 months in residents' cognitive and physical functioning, frailty and neuropsychiatric behaviours were associated with changes in proxy reports on EQ-5D-5L. Interrater correlation at the total score measured by the intra-class correlation coefficient was 0.744 at baseline and 0.664 at 12 months (moderate). Interrater correlation at the domain level, measured by weighted Cohen's kappa, ranged from 0.206 to 0.432 (minimal to weak). CONCLUSION: Proxy reports on the EQ-5D-5L demonstrated satisfactory convergent validity, known-group validity, and responsiveness. Interrater reliability was moderate at the total utility score but poor at the domain level. Proxy reports on the EQ-5D-5L can serve as an alternative method to measure residents' HRQOL. However, data from proxy reports at the domain level is not recommended for future research and practice evaluation. CLINICAL TRIAL REGISTRATION: This study is part of a large deprescribing trial known as 'Opti-Med', which has been registered with the Australian New Zealand Clinical Trials Registry (Registry number: ACTRN12613001204730, dated 4 November 2013) and the World Health Organisation Clinical Trials Registry Platform (the universal trial number: U1111-1148-6094).