4
Geographic Variation in Uninsurance and Changes in Hospital Inpatient Utilization in the Pre-ACA vs. Post ACA Periods
Data Sources.Eighteen State Inpatient Databases from the Healthcare Cost and Utilization Project, 2011-2014; State population estimates from the Census Bureau; unemployment rate estimates from the Bureau of Labor Statistics; population health insurance estimates from the American Community Survey, 2013-14 for adults 18-64.
Study Design.A retrospective cohort study using a sample of hospitals within states. A hospital fixed effects regression design estimated ACA expansion effects on uninsured discharges using both hospital market variation in uninsurance rates and cross-state differences in ACA coverage expansion participation. States were stratified into those not opting for ACA Medicaid expansion, expansion states with pre-ACA childless adult Medicaid coverage and expansion states without such coverage. Hospital markets were assigned to quartiles based on the uninsurance rate in their primary markets. Outcomes were total, chronic condition, discretionary and preventable discharge volumes for adults age 19-64. Hospital market uninsurance rate quartile and state expansion group effects on uninsured discharge volumes were estimated.
Principal Findings.Between 2013 and 2014, observed uninsured discharges decreased by 6.2% in non-expansion states, 57.2% in expansion states without pre-ACA childless adult coverage and 32.6% in expansion states without such coverage; corresponding changes in estimated population uninsurance rates for adults 18-64 in our study states were -3.4%, -6.1% and -3.3%. For hospitals in the two expansion state groups, uninsured discharge effects decreased across hospital market uninsurance quartiles, as expected. Effects were more pronounced for hospitals in states with pre-ACA childless adult coverage than those without such coverage. Hospital market uninsurance quartile effects were weak and inconsistent for non-expansion states, except for hospitals in the highest uninsurance rate quartile. Comparisons of effects across state groups, within hospital market uninsurance rate quartiles, were generally as expected: non-expansion state hospitals had the smallest uninsured discharge volume effects, while hospitals in expansion states without pre-ACA childless coverage had the largest effects. Effects for chronic condition, discretionary and preventable discharges were similar to those for total discharges.
Conclusions.Hospitals in expansion states with pre-ACA childless adult Medicaid coverage had the strongest relationship between uninsurance rates in their markets and changes in uninsured discharge volumes: hospitals in markets with higher average pre-ACA uninsurance rates experienced larger decreases in post-ACA uninsured discharges. A similar relationship held for expansion states without pre-ACA childless adult coverage, but estimated effects were much smaller in magnitude, to our surprise. Hospitals in non-expansion states exhibited, at best, a weak relationship between market uninsurance rates and uninsured discharges. Since non-expansion states and those with pre-ACA childless adult coverage experienced population uninsurance rate decreases of a similar amount, the question remains of why uninsured discharge decreases in the two groups are so different.