Limitations of the SF-36 in a sample of nursing home residents

Objective: to assess test characteristics of the Medical Outcomes Study SF-36 (Short-Form 36) with residents of nursing homes. Research design: nursing home residents with 17 or more points on the Mini-Mental State Examination (MMSE) and greater than or equal to 3 months residence (128 of 552 screened) were selected randomly. Interviewers administered the SF-36 (repeated after 1 week), Geriatric Depression Scale and MMSE. We recorded activities of daily living and medication data from medical records. Data analysis included test-retest: intraclass correlations, item completion, score distributions and SF-36 correlations with measures of physical and mental functioning. Results: 97 nursing home residents (75.8%) consented. Test-retest intraclass correlation coefficients were good to excellent (range = 0.55 to 0.82). Convergent validity between SF-36 physical health scales and the activities of daily living index Mras modest (r range = -0.37 to -0.43). About 25% of residents scored zero (lowest score) on at least one SF-36 physical function measure. SF-36 mental health scales correlated strongly with the Geriatric Depression Scale (r range = -0.63 to -0.71) and modestly with bodily pain (r = -0.35). No SF-36 scales correlated strongly with the MMSE. Conclusion: only one in five nursing home residents met minimal participation criteria, suggesting limited utility of the SF-36 in nursing homes. Reliability and validity characteristics were fairly good. Skewed scores were noted for some SF-36 scales. The utility of thee SF-36 may be limited to assessments of subjects with higher cognitive and physical functioning than typical nursing home residents. The SF-36 might benefit from modification for this setting, or by tests of proxy ratings.