The Affordable Care Act Medicaid Coverage Expansion and Changes in Hospital Inpatient Utilization

Tuesday, June 14, 2016
Lobby (Annenberg Center)

Author(s): Eli Cutler; Michael Dworsky; Christine Eibner; Zeynal Karaca; Brian J. Moore; Gary Pickens; Herbert S. Wong

Discussant: Thomas DeLeire

Objective. Estimate effects of the Affordable Care Act Medicaid expansion on hospital inpatient volumes and outcomes, for all-payers, Medicaid and the uninsured.

Data Sources. State Inpatient Databases from the Healthcare Cost and Utilization Project, 2011-2014; State population estimates from the Census Bureau; and unemployment rate estimates from the Bureau of Labor Statistics.

Study Design. This is a retrospective cohort study using a sample of hospitals within states.  A difference-in-differences regression design estimated Medicaid expansion affects using cross-state variation in outcomes for all-payers, Medicaid and the uninsured. Outcomes were discharge volumes, chronic condition discharge percentages, discretionary and preventable hospitalization percentages, length of stay, estimated cost, and a case-mix index.  Expansion effects, estimated using regression models, represent percentage changes in outcomes in expansion states compared to non-expansion states before and after the Affordable Care Act Medicaid expansion.

Principal Findings. Discharge volume expansion effects for all-payers were mainly small and insignificant; those for Medicaid were large, positive and significant; and those for the uninsured were large, negative and significant.  Uninsured expansion effects for chronic condition percentages were negative and significant; those for Medicaid and all-payers were inconsistent or insignificant.  Expansion effects for discretionary discharge percentages among Medicaid patients and the uninsured were positive and significant, while those for all-payers were small and insignificant.  Uninsured expansion effects for preventable discharge percentages were negative and significant; expansion effects for Medicaid and all payers were small and insignificant.  For the uninsured, expansion effects of length of stay, cost, and the case-mix index were significant and negative. Corresponding expansion effects for Medicaid were inconsistent and smaller, and there were no significant all-payer expansion effects for any of these metrics.

Conclusions. The hypothesis that discharge volumes would increase for Medicaid, while uninsured discharges would decline, was corroborated.  However, expansion effects for total discharge volumes were small and not statistically significant, contrary to expectation.  Results for uninsured discharges and other outcomes suggest that people with chronic conditions, those at highest risk for preventable hospitalizations and those consuming the most hospital resources were differentially likely to transfer from uninsurance into Medicaid or private coverage.  One explanation consistent with our findings is that Medicaid has no impact on inpatient utilization relative to uninsurance and compositional changes in uninsured and Medicaid discharges are purely driven by selection.