Background: Proxy-reporting has been proposed as an alternative to self-report of health-related quality of life (HRQoL) for patients with poor cognition. There are 2 possible perspectives from which to complete a proxy-report, answer as the patient would (proxy-patient) or from the proxy's own perspective (proxy-proxy). Most research has not differentiated between perspectives. Agreement between patient and proxy-reports from either perspective has not been investigated using the Euroqol-5D (EQ-5D) among elderly hospital patients undergoing rehabilitation.
Objectives: Identify agreement levels between proxy-patient and patient self-report as well as proxy-proxy and patient self-report of the EQ-5D and investigate interaction effects of timing (admission vs. discharge) and basic cognition (intact vs. not intact).
Research design: Repeated measures, inter-rater agreement investigation of clinician proxy-report, and patient self-report incorporating; proxy-patient reports (perspective A) and proxy-proxy reports (perspective B).
Subjects: Geriatric rehabilitation patients (n = 272) and their proxies (treating physiotherapists n = 29).
Measures: EQ-5D for HRQoL and Mini Mental State Examination for cognition.
Results: One hundred fifty (89%) proxy-patient and 130 (98%) proxy-proxy datasets were complete, 51 perspective A and 52 perspective B patients did not have basic cognition intact. Proxy-patient assessments had strong agreement with self-report at discharge across all cognition levels (kappa = 0.76-0.95), but at admission had stronger agreement among patients with better cognition (kappa = 0.70-0.86) than patients with lower cognition (kappa = 0.47-0.76). At admission and discharge proxy-proxy assessments generally had moderate agreement with self-report among patients with poor cognition on most domains with proxies giving lower scores than patients (kappa = 0.23-0.81), this is in contrast to proxy-proxy assessments and patients with better cognition (kappa = 0.55-0.95).
Conclusions: Clinician (physiotherapist) proxy-reports among this population generally had good agreement with patient self-report though this was affected by proxy perspective, patient cognition, and timing.