Association of County Race and Ethnicity Characteristics With Number of Insurance Carriers and Insurance Network Breadth

Abstract

While the Affordable Care Act (ACA) set out to eliminate insurer discrimination based on preexisting conditions, the ACA health exchanges allow insurers to select what markets to enter and afford them great freedom on how they design their physician networks. Strategic market participation and physician network design based on population race, ethnicity, and health characteristics may give rise to a present-day form of redlining within health insurance markets-ie, a systematic underprovision of insurance plans and in-network practitioners within areas that are populated with higher proportions of non-Hispanic Black residents. To examine if markets with relatively higher non-Hispanic Black populations have systematically fewer insurers and lower network inclusion of physicians residing within these areas. This cohort study conducted a regression analysis of the US ACA health insurance exchange marketplace across 34 states with federal exchanges and physicians located within the 500 most populous US cities in 2014. County-level data were sourced from individual market and issuer enrollment databases and county health rankings; census tract data came from a national database of physician networks in 2014 marketplace plans, US Census Bureau data, and the Centers for Disease Control and Prevention's PLACES database. Adjustment was made for a rich set of county (or census tract) controls and state fixed effects to capture broad market and/or policy differences across states. Analyses were performed in June 2021. The raw count of insurers within a county and the mean percentage of insurance networks that physicians participate in within each census tract. A total of 2270 counties were examined within our first analyses. In the counties analyzed, a mean (SD) of 23.0% (3.2%) of the population was aged 18 years or younger, and a mean (SD) of 11.0% (15.8%) of the population had non-Hispanic Black race and ethnicity. For the second analysis, 16 006 to 25 096 census tracts were examined (depending on physician specialty). With adjustment for population size, age, and race and ethnicity, a 1-SD increase in the county non-Hispanic Black population was associated with a 14.1% reduction in the number of insurers (mean [SE] marginal effect size, -2.18 [0.13]; P < .001). Accounting for additional county-level risk selection controls and state fixed effects, a 1-SD increase in the non-Hispanic Black population was associated with a 2.3% reduction in available insurers (marginal effect size, -0.36 [0.17]; P = .04). For practitioners network breadth inclusion, a 1-SD increase in the non-Hispanic Black population was associated with a 15.8% (marginal effect size, -0.32 [0.01]; P < .001) to 24.7% (marginal effect size, -0.14 [0.02]; P < .001) reduction in the physicians' network participation depending on their specialty. Adjusting for additional state fixed effects yielded estimates of 6% (marginal effect size, -0.08 [0.01]; P < .001) to 13.5% (marginal effect size, -0.12 [0.02]; P < .001) reductions in practitioner network participation. These findings suggest that strategic decisions by insurers may contribute toward markets with higher racial or ethnic minority populations having systematically fewer participating insurers, as well as a higher prevalence of local physicians not included in coverage networks. These findings call for further examination of potential insurance redlining within the ACA marketplaces.