Predictors of perceived health in hospitalized older persons: A cross-sectional and longitudinal study

OBJECTIVES: To identify the predictors of perceived health and predictors of changes in perceived health in frail hospitalized older persons during the year after hospitalization. DESIGN: Both cross-sectional and longitudinal multivariate analysis of data from a cohort followed for 1 year. SETTING: Six hospitals in a group practice model health maintenance organization (HMO) in Southern California. PARTICIPANTS: A total of 1889 persons aged 65 or older who met at least one of 13 inclusionary criteria for a randomized trial of Comprehensive Geriatric Assessment consultation at admission and completed three Functional and Health Status Questionnaires (FHSQ) during a 12-month period. MEASUREMENTS: Functional and health status measures included basic and intermediate activities of daily living (BADL and IADL) and social activities (SA) scales from the Functional Status Questionnaire as well as the mental health index (MHI) and current health perception (CHP) scales from the Medical Outcomes Study short form. Subject's severity of disease was measured by the Resource Demand Scale (RD Scale). RESULTS: In the cross-sectional analyses, MHI score, IADL score, RD Scale, history of falls during the 3 months before hospitalization, and female gender were significant predictors of perceived health in all models for each time point. BADL score, age, presence of incontinence, and ethnicity were significant in the model for baseline only, and SA score was significant in models for 3 months and 12 months only. In the longitudinal analyses, the baseline CHP score and the changes in MHI, IADL, and BADL score predicted CHP change from 0 to 3 months and from 3 to 12 months. CONCLUSIONS: Functional and psychosocial health indicaters are the most important and most consistent predictors of perceived health. Our study showed that several predictors of perceived health in cross-sectional analysis did not predict changes in perceived health over a 1-year period. Hence, to fully understand the medical and social contributors to perceived health, a comprehensive approach using both cross-sectional and longitudinal analyses is necessary.