Karoline Mortensen1, Paula H Song. 1. Department of Political Science, Rice University Houston, Texas 77005-1892, USA. kmort@rice.edu.
Abstract
OBJECTIVE: To examine differences in Emergency Department (ED) utilization between Medicaid enrollees and the uninsured. DATA AND METHODS: We use nationally representative data for adults age 19-64 from the 2004 Medical Expenditure Panel Survey. We calculate descriptive statistics highlighting differences in distributions of measurable characteristics and logit regressions describing differences in the coefficients of the characteristics between the 2 groups. Blinder-Oaxaca nonlinear decomposition methods are applied to quantify the effect that differences in measurable characteristics between the groups have on the differences in ED utilization, as well as examine the magnitude of the effect on utilization of differences in behavior between the groups. RESULTS: Twenty-seven percent of Medicaid enrollees had an ED visit, whereas 10% of those uninsured the entire year had an ED visit. Roughly half of the utilization gap can be explained by differences in measurable characteristics. Although predisposing factors such as demographic and socioeconomic characteristics explain little of the gap, the variables capturing health limitations reflecting need explain a more substantial amount of the gap. Specifically, the higher presence of chronic conditions reported by Medicaid enrollees accounts for 18 percentage points of the gap, and reporting fair or poor health status explains 10 percentage points of the disparity. CONCLUSIONS: Decomposition analysis results show that differences in measurable characteristics between the groups explain half of the disparity in visits. The remaining portion of the gap in ED utilization is driven by differences in unmeasured or unobserved characteristics between the groups such as care-seeking behavioral differences.
OBJECTIVE: To examine differences in Emergency Department (ED) utilization between Medicaid enrollees and the uninsured. DATA AND METHODS: We use nationally representative data for adults age 19-64 from the 2004 Medical Expenditure Panel Survey. We calculate descriptive statistics highlighting differences in distributions of measurable characteristics and logit regressions describing differences in the coefficients of the characteristics between the 2 groups. Blinder-Oaxaca nonlinear decomposition methods are applied to quantify the effect that differences in measurable characteristics between the groups have on the differences in ED utilization, as well as examine the magnitude of the effect on utilization of differences in behavior between the groups. RESULTS: Twenty-seven percent of Medicaid enrollees had an ED visit, whereas 10% of those uninsured the entire year had an ED visit. Roughly half of the utilization gap can be explained by differences in measurable characteristics. Although predisposing factors such as demographic and socioeconomic characteristics explain little of the gap, the variables capturing health limitations reflecting need explain a more substantial amount of the gap. Specifically, the higher presence of chronic conditions reported by Medicaid enrollees accounts for 18 percentage points of the gap, and reporting fair or poor health status explains 10 percentage points of the disparity. CONCLUSIONS: Decomposition analysis results show that differences in measurable characteristics between the groups explain half of the disparity in visits. The remaining portion of the gap in ED utilization is driven by differences in unmeasured or unobserved characteristics between the groups such as care-seeking behavioral differences.