TEST-RETEST RELIABILITY IN PSYCHIATRIC-PATIENTS OF THE SF-36 HEALTH SURVEY

The objectives of this study were to examine the test-retest reliabili ty of the SF-36 Health Survey (SF-36) among patients seen in an outpat ient department of psychiatry, to develop a computer-administered vers ion of the SF-36 and test the reliability compared to the paper versio n, and to examine the patient acceptability of the SF-36. At stage one , both the paper and computer version of the SF-36 were administered t o 138 consecutive patients who had previously been seen in the departm ent. At stage two, 61 patients returning within one month for a follow -up appointment completed either the paper or computer version both be fore and after that appointment as well as a study evaluation form. Re liability as measured by intraclass correlation coefficients is high f or simultaneous administration of the paper and computerized versions. Reliability over a short interval (before and after the patient's app ointment) is acceptable. Reliability over a longer interval between th e two consecutive appointments (mean and median 18 days) also appears to be acceptable for both computer and paper versions. Three-quarters of the patients felt that the SF-36 could provide useful information t o their physicians. Among psychiatric outpatients, the test-retest rel iability of the SF-36 appears to be acceptable for use in outcome stud ies. The SF-36 can be administered using a computer without jeopardizi ng reliability. Both versions are acceptable for use with psychiatric patients.