Objective: Racial/ethnic disparities in health care delivery have been well described, but little is known about such disparities for children who seek emergency care. The objective of this study was to test the hypothesis that severity-adjusted emergency department pediatric admission rates are associated with race/ethnicity.
Methods: Secondary analyses were conducted of an established database of 16 emergency departments that participated in a national study to validate the Pediatric Risk of Admission II score, which is used to measure severity of illness. Patients were randomly selected by the coordinating center from daily emergency department visit logs. Crude and severity-adjusted admission rates were compared among the 3 most common races/ethnicities: white, black, and Hispanic. Adjusted admission rates were calculated by using the standardized admission ratio, which was calculated by dividing the observed admissions by the predicted admissions, when predicted was calculated from the Pediatric Risk of Admission II score.
Results: After exclusion of 3 sites that recorded race/ethnicity in <10% of patients, there were 13 sites with 8952 patients in the 3 major race/ethnicity groups. Black and Hispanic patients were similar to each other and different from white patients; therefore, these 2 groups were combined for analyses. Both crude (8.2% vs 5.3%) and severity-adjusted (standardized admission ratio: 1.71 vs 1.1) admission rates were higher in white than in nonwhite patients. Standardized admission ratios were close to 1.0 in both race/ethnicity groups in the higher quintiles of illness severity. In contrast, white patients were admitted at 1.5 to 2 times the expected rate in the lowest 2 quintiles of severity.
Conclusions: There are differences in both crude and adjusted admission rates between white and black/Hispanic patients. The results are more consistent with high rates of discretionary admissions for white patients with low illness severity than with underadmitting severely ill black or Hispanic patients.