Volume 54 | Number 4 | August 2019

Abstract List

Gary Pickens Ph.D., Zeynal Karaca Ph.D., Teresa B. Gibson Ph.D., Eli Cutler Ph.D., Michael Dworsky Ph.D., Brian Moore Ph.D., Herbert S. Wong Ph.D.


To estimate the effects of the health insurance exchange and Medicaid coverage expansions on hospital inpatient and emergency department () utilization rates, cost, and patient illness severity, and also to test the association between changes in outcomes and the size of the uninsured population eligible for coverage in states.

Data Sources

Healthcare Cost and Utilization Project State Inpatient and Emergency Department Databases, 2011‐2015, Nielsen Demographic Data, and the American Community Survey.

Study Design

Retrospective study using fixed‐effects regression to estimate the effects in expansion and nonexpansion states by age/sex demographic groups.


In Medicaid expansion states, rates of uninsured inpatient discharges and visits fell sharply in many demographic groups. For example, uninsured inpatient discharge rates across groups, except young females, decreased by ≥39 percent per capita on average in expansion states. In nonexpansion states, uninsured utilization rates remained unchanged or increased slightly (0‐9.2 percent). Changes in all‐payer and private insurance rates were more muted. Changes in inpatient costs per discharge were negative, and all‐payer inpatient costs per discharge declined <6 percent in most age/sex groups. The size of the uninsured population eligible for coverage was strongly associated with changes in outcomes. For example, among males aged 35‐54 years in expansion states, there was a 0.793 percent decrease in the uninsured discharge rate per unit increase in the coverage expansion ratio (the ratio of the size of the population eligible for coverage to the size of the previously covered population within an age/sex/payer/geographic group).


Significant shifts in cost per discharge and patient severity were consistent with selective take‐up of insurance. The “treatment intensity” of expansions may be useful for anticipating future effects.