Discordance between physician-rated health and an objective health measure among institutionalized older people

Background Although physician-rated health is emerging as a potentially useful variable in research, with implications in practice, it has not been analyzed. Moreover, one of its most important aspects, namely, concordance with patients’ objective health state, has not been investigated. This study sought to measure concordance between physician-rated health and an objective health measure, and assess both measures’ validity in predicting death. Methods The data for the study were drawn from a 1998–1999 survey and subsequent mortality follow-up of residential and nursing homes in Madrid (Spain). Study subjects were 630 residents aged ≥65 years, and their respective facility physicians. Measures included agreement between physicians’ rating of residents’ overall health (good, intermediate or poor) and an objective measure of residents’ health (good, intermediate or poor), based on functional capacity, cognitive status, and number of chronic conditions. Overrating was defined as any case where health, rated as good by a physician, was objectively rated as poor. Results Whereas 45 % of physicians and 55 % of residents rated their health as good, only 4 % of such residents had good objective health. Of those who received a physician rating of good/very good health, 39.0 % had poor objective health. There was evidence of clear overrating in 18 % of the population, and clear to moderate overrating in 73 % of the population. In terms of power to predict mortality, the pattern of behavior shown by the objective health measure was good, graded and congruent, and better than that shown by physician-rated health. Conclusion Physician overrating of the overall health of older persons in residential and nursing homes, would appear to be very high. Although some degree of contextualization by physicians in this setting might be considered reasonable, the degree of overrating in our population seems nevertheless excessive.