TEST-RETEST RELIABILITY OF A QUESTIONNAIRE THAT IDENTIFIES ELDERS AT RISK FOR HOSPITAL ADMISSION

OBJECTIVE: To determine the test-retest reliability of a questionnaire designed to measure elderly persons' probability of repeated admission (P-ra) to a hospital within 4 years. DESIGN: Participants received the test questionnaire by mail; respondents to the test questionnaire received the retest questionnaire 3 weeks later. PARTICIPANTS: Elderly (65+) community-dwelling enrollees in the Medical Assistance (Medicaid) program of Ramsey County, MN (n = 192). MAIN OUTCOME MEASURE: The correlation (r) between the P-ra values computed from the test-retest correlations (k) of the questionnaire's individual items were also measured. RESULTS: The response rate was 63% (121/192) for the test questionnaire and 69% (84/121) for the retest questionnaire. The test and retest values of P-ra computed from responses to the questionnaires were highly correlated (r = 0.78, P < 0.0001). For the individual questionnaire items, the range of k values was 0.50-1.00 (P < 0.0001 for all items). Slightly higher values of r and k were obtained when the respondents were women and when the same person (either proxy or self) completed both questionnaires. Age was not consistently related to reliability. CONCLUSION: the test-retest reliability of individual items and of the computed P-ra was high, suggesting that responses by elderly persons to mailed questionnaires pertaining to health status are stable over brief periods of time. The resulting P-ra values may be useful in identifying elders at high risk for hospital admission.