Geographic variations in hospital discharge rates and discretionary hospital use.

Abstract

While past research has shown that there is greater geographic variability in hospital discharge rates for medical conditions where there is less agreement about proper treatment, there is little empirical evidence to support the corollary that high hospital use in a community is primarily the result of a greater volume of such discretionary hospitalizations. This study assesses the contribution of discretionary hospitalizations to higher overall rates of hospital use in communities. Hospital discharge files and Medicare eligibility files were used to estimate adjusted rates of hospital discharge, days of care and adjusted mortality for a sample of 761 geographic communities in four states in the USA. Diagnostic information was used to classify hospitalizations into low, moderate and high discretion categories. Correlation and multiple regression analysis methods were used to test for systematic relationships between a community's overall rate of hospital use and discretion-level mix of hospitalizations. Although about half of the variance in overall rates of hospital use among communities was found to be related to the discretion-level mix of hospital discharges, only a small portion of the explained variance could be attributed specifically to community differences in the prevalence of high discretion hospitalizations. High overall rates of hospital use in communities were not found to be largely the result of high discretion hospital use.