Association for Public Policy Analysis & Management

Panel: Novel Data Sources & Methods for Examining State-Level Disparities in Health Insurance, Cost, and Access to Care

(Health Policy)

Extensive changes in federal healthcare policy over the past decade, including the Affordable Care Act of 2010 and the Children’s Health Insurance Program Reauthorization Act of 2009, have led to major declines in uninsurance and increased access to care. At the same time, significant policy decisions are being made at the state level resulting in large variations in access between states. For instance, the ACA Medicaid expansion has only been implemented in 31 states and DC. Analyses at the national level, while important, may mask disparities at the state-level. Therefore, state-level data and estimates are needed in order to understand whether potential gains achieved through federal policy are also observed across the states. Furthermore, it is unclear whether population-level gains have translated into reductions of between- and within-state disparities by, for example, income, race/ethnicity, immigrant status, and geography. The papers in this panel use novel data sources and methods to examine state-level disparities in health insurance coverage, cost, and access to care. The first two papers use large national datasets to produce state-level estimates of disparities in uninsurance, affordability, utilization, and access to care. The latter two analyses use state-specific health insurance survey and/or administrate data in two states that fully implemented the Affordable Care Act – Oregon and Minnesota. The first paper uses restricted data from the National Health Interview Survey (NHIS), which unlike public-use NHIS data includes state identifiers. Whereas previous work in the NHIS had been limited to national estimates, access to these restricted identifiers allows for identification of state-level disparities. The second paper estimates state-level uninsurance for citizen children in undocumented immigrant families using a multiple imputation method to predict documentation status into the American Community Survey. Model predictions originate from the Survey of Income & Program Participation (SIPP), the only nationally representative survey that includes measures of documentation status but it is not state-representative. The American Community Survey, on the other hand, is a large survey that is designed to be representative at the state level. The third paper uses a novel method, the OHSU Health Insurance Coverage Model, to combine administrative data from Medicare, Medicaid, and state insurance division enrollment data with population estimates and survey estimates of the uninsured to arrive at more timely estimates of coverage. This approach also allows for the estimation of county-level estimates, another important sub-group analysis that draws attention to disparities that may be masked at a higher level of data. The final paper uses the Minnesota Health Access Survey – a statewide survey conducted since the 1990s with the goal of tracking health insurance and access for the entire MN population as well as across key policy-relevant sub-groups. The availability of representative samples of these subgroups allows for examination of within-state changes in disparities in coverage and affordability indicators by age, race and ethnicity, and income.