Predictors of overnight hospital admission in older African American and Caucasian Medicare beneficiaries.

Abstract

Potential disparities in health care utilization were examined using overnight hospitalization data from the University of Alabama at Birmingham Study of Aging, a longitudinal investigation of a stratified sample of Medicare beneficiaries. Racial differences in self-reported surgical and nonsurgical overnight hospital admissions were examined using Cox proportional hazards models. Andersen's Behavioral Model provided the conceptual framework to identify other potential predictors of admission. Nine hundred and forty-two participants, 50.1% African American, provided data at baseline and at least one follow-up assessment (mean age = 75.3 years, range: 65-106). African Americans were less likely to utilize surgical admissions compared with Caucasians in a bivariate model (hazard ratio = 0.63, 95% confidence interval = 0.41-0.98). This effect was not significant after controlling for demographics and self-reported physical health. Additional bivariate predictors of surgical admission were intact mental status, having private insurance, and higher education. African Americans were less likely to utilize nonsurgical admissions in both bivariate (hazard ratio = 0.74, 95% confidence interval = 0.59-0.93) and covariate-adjusted models (hazard ratio = 0.64, 95% confidence interval = 0.50-0.84). This effect was significantly stronger for men than for women, with African American men only 0.50 times as likely as Caucasian men to report a nonsurgical admission. Other bivariate predictors of nonsurgical admission were increased age, poor physical health, negative psychological characteristics, higher levels of social support, and low perceived discrimination. Underutilization of services has been linked to increased mortality in African Americans. Modifications in mutable domains associated with service utilization such as perceived discrimination, social support, and having private insurance may be beneficial.