Background: The economic evaluation of health interventions for older people is complicated by the difficulty in obtaining self-reports of quality of life from persons with cognitive impairments, physical impairments, or both.
Objectives: Using the EQ-5D (EuroQoL) measures, to assess: (1) agreement between subjects and proxies on subject's quality of life ratings at different points in time; (2) agreement between subjects and proxies on change of subject's quality of life ratings over time; and (3) subject and proxy characteristics related to agreement.
Research design: Prospective study of subjects visiting hospital emergency departments (ED). Data were collected at enrollment in the ED and at follow-up, 1 and 4 months after the ED visit.
Subjects: The study comprised 231 pairs of cognitively intact patients aged 65 years or older and their primary caregivers.
Measures: Quality of life was measured using both components of the EQ-5D scale, the index score and the Visual Analogue Scale (VAS). Demographic characteristics and health status (physical and mental) were measured for both subjects and proxies. Subjects and proxies were interviewed either in English or French.
Results: There was low to moderate agreement between subjects and proxies at different points in time (intraclass correlation coefficient [ICC] = 0.22 to 0.59), and between subject and proxy change scores over time (ICC = 0-0.50), on both the index score and the VAS. Better agreement between subjects and proxies was found at the 4 months follow-up, when the subject was less depressed, and when the proxy's native language was English.
Conclusions: Proxy EQ-5D responses, either for a specific point in time or for assessing change over time, may not be valid measures of self-reported quality of life among older medically-ill patients.