Measuring health status in older patients. The SF-36 in practice

OBJECTIVE: To examine the use of the Medical Outcomes Study short form 36-item (SF-36) health status measure in older patients receiving health care; to explore the influence of age and physical and cognitive status on response to and completion of the SF-36 questionnaire. DESIGN: Prospective observational study. SETTINGS: Hospital and ambulatory patients aged 65 years and over in the medical and surgical wards and outpatient department of a teaching hospital and a local general practitioner's surgery. SUBJECTS: 1014 hospital inpatients, 80 hospital outpatients and 40 patients attending their general practitioner's surgery. MAIN OUTCOME MEASURES: Response rates, overall rates of completion (sufficient to calculate a valid SF-36 score) and completion of individual questions. RESULTS: 37 out of 40 ambulatory patients in general practice (93%) and 71 out of 80 outpatients (89%) returned a self-completed questionnaire. In hospital inpatients the overall response rate was only 46% (369 of 802). This was improved by interview administration to 77.4% (164 of 212). Logistic regression analysis revealed that self-completion, cognitive dysfunction, disability and age were all independently associated with poor overall response rates. Among those patients who returned a completed questionnaire, completion of individual questions sufficient to calculate a valid score was variable. Only 62.5% of inpatients who self-completed a questionnaire gave sufficient response to calculate a score on the mental health subscale, compared with 93.7% of general practice patients. CONCLUSION: The self-administered questionnaire is unacceptable for older hospital inpatients. Use of an interviewer improves response, but factors which influence health status, such as physical and cognitive dysfunction, have a significant effect on response rates. Therefore the utility of the SF-36 in its present form as a routine health status measure for use in older hospital inpatients is questioned.