Much of the motivation for expanding health insurance coverage in the US stems from the disadvantage experienced by the millions of individuals who are uninsured.  Recently, however, some research suggests that a large uninsured population in a community may affect access and quality of medical care for all residents, even those with insurance coverage.  The costs associated with widespread uninsurance thus “spillsover” to the insured population.  We extend this line of research by investigating whether the number of uninsured individuals in a community is associated with the cost of emergency room care and, if so, who bears the cost.  Emergency rooms are a particularly important context in which to examine the possibility of a spillover effect because the poor and uninsured frequently use them as a costly substitute for primary care.  Further, emergency departments must serve everyone regardless of ability to pay and, therefore, may have particularly high rates of uncompensated care.  The associated costs may be passed on to insured patients and their insurance providers.  The ACA was designed both to insure those without coverage and to “bend the cost curve” of medical services.  Determining the extent to which community uninsurance rates affect the health care system at large is important to achieving both these goals. 

We use data from the Medical Expenditure Panel Survey (MEPS), a nationally representative sample of the US non-institutionalized population.  Our unit of analysis is the emergency room visit.  Visits are linked to person-level characteristics from the MEPS and county-level characteristics from the Small Area Health Insurance Estimates (SAHIE).  We employ a generalized linear modeling (GLM) approach predicting total expenditures on each visit, as well as expenditures by source of payment (private insurance, public insurance, and out-of-pocket).  The key independent variables for assessing spillover are the number of uninsured individuals per emergency room in a county, and its interactions with indicators for individual-level insurance status (covered with private insurance, public insurance, or uninsured).  We control for a variety characteristics at the event-level (type of visit, diagnostic codes), person-level (e.g. income, education, race, age, gender, chronic conditions and subjective health) and county-level (poverty rate, unemployment rate, vacancy rate, and the supply of medical services). 

Preliminary results suggest that privately insured individuals who live in areas with a large uninsured population pay more, on average, than their counterparts in counties where being uninsured is less common.  An increase of 1000 uninsured individuals per emergency room from the median is associated with a $10 increase in the amount paid at a typical emergency room visit, net of all the variables in the model.  A standard deviation increase in the number of uninsured people per emergency room (about 10,000) from the mean is associated with a $140 increase in the amount paid at a typical emergency room visit.  These findings suggest that the ACA’s expansion in insurance coverage may benefit not only those who are currently uninsured, but those with private insurance too.