The role of health care providers and significant others in evaluating the quality of life of patients with chronic disease.

Health-related quality of life (HRQL) studies sometimes rely, in part, on proxy information obtained from patients' significant others (spouse or close companion) or health care providers. This review: (1) provides a quantitative analysis of the results that have been reported in recent studies assessing the level of agreement between patient and proxy HRQL ratings, and (2) addresses a number of key methodological issues surrounding the use of proxy raters in HRQL research. This review concentrates on 23 studies, published between 1991-2000, that describe patient-proxy agreement for a number of well-known multidimensional HRQL instruments. In general, moderate to high levels of patient-proxy agreement were reported. Lower levels of agreement were found predominantly in studies employing a small sample size (approximately 50 patient-proxy pairs or less). In larger studies comparing patients and their significant others, median correlations were between 0.60-0.70 for physical HRQL domains and about 0.50 for psychosocial domains. Mixed results were reported in studies comparing patients and their health care providers, but most of these studies employed a relatively small sample size. Proxy raters tended to report more HRQL problems than patients themselves, but the magnitude of observed differences was modest (median standardized differences of about 0.20). Based on the current evidence, we conclude that judgements made by significant others and health care providers about several aspects of patients' HRQL are reasonably accurate. Substantial discrepancies between patient and proxy ratings occur in a minority of cases. We recommend that future studies focus on: (a) the reliability and validity of proxy ratings according to common psychometric methods, and (b) the balance between information bias due to proxy ratings and potential selection bias due to exclusion of important patient subgroups from HRQL studies.