Self-Rating of Health in Community-Dwelling Elders: A Complex Construct

Kathryn Adams
University of Maryland
13000 Cleveland Drive
Rockville MD 20850
301-216-9665
FAX: 301-216-1506
MDAKBA@aol.com
Abstract: Purpose of Study:  Self-Rating of Health has been established as an independent predictor of mortality (see Idler & Benyamini, 1997, for a review of 27 studies), and of long-term medication and service use (Bath, 1999).  Because this single variable is so important to health and social research, this study has identified characteristics and experiences of elders that contribute to how they rate their health.

Method:  Surveys completed by 327 members over age 65 of a Managed Care Organization included information on demographic characteristics, social and health-related variables.  For this paper, a hierarchical multiple regression was performed on a five-anchor Self-Rating of Health.

Results: The hierarchical regression model explained 47.8% of the variance in Self-Rating of Health, using 15 variables from the survey.  Age, gender, income and relationship status, entered as covariates, explained approximately 7% of the variance.  Other significant predictors in order of entry included number of health conditions, functional impairment, experience of pain, number of doctor visits in the past year, and "feeling of slowing down."  Next, the four most common chronic illnesses suffered by the sample were entered in order of their frequency.  Arthritis and hypertension each proved non-significant when entered, whereas heart disease and diabetes were significant predictors.  Finally, a depressive affect subscale of the Geriatric Depresssion Scale explained an additional 4.4% of the variance in Self-Rating of Health.

Implications:  Self-Rating of Health appears to be a complex construct relating both to objective health and functioning and to subjective experience of physical pain, slowing down, and depressive affect.  Specific illness did not contribute much to the rating after controlling for the number of conditions and functional impairments.  Still, heart disease and diabetes explained unique variance not seen with arthritis (more likely to share variance with functioning and pain variables) or hypertension.