Spatial Analysis of County Level Influenza Vaccination Coverage Estimates

INTRODUCTION: The Centers for Disease Control and Prevention (CDC) reported 43.3% of adults (\u3e18) received a flu vaccination in the 2016-17 season (FluVaxView, 2018). Despite long-standing recommendations and the enormous burden caused by the virus, vaccination coverage for flu is consistently low across seasons. AIMS: This study examines influenza vaccination coverage among adults at the county level, in the contiguous United States (U.S.), to identify clusters of high and low coverage as well as assess access to care variables that could impact coverage levels. METHODS: County level influenza vaccination coverage estimates from the 2014 Behavioral Risk Factor Surveillance System (BRFSS) were examined to identify statistically significant clusters using global (Moran’s I) and local (Local Indicators of Spatial Association (LISA)) spatial clustering tests in GeoDa 1.10. To examine spatial effects of area provider supply and population demand variables, regression analyses were performed in GeoDaSpace. Independent variables were obtained from the Area Health Resource File (AHRF) and the National Council for Prescription Drug Program (NCPDP). RESULTS: The Moran’s I value comparing coverage estimates to the spatially lagged values was 0.48, indicating the coverage estimates are not spatially random. Five hundred sixty five counties were considered ‘High-High’, four hundred sixty three were ‘Low-Low’, and sixty seven had high coverage but were surrounded by counties with low coverage, designated as ‘High-Low’. ‘Low-High’ counties are defined as counties with low coverage surrounded by counties with high coverage. Forty counties were considered ‘Low-High’ in this analysis. The final regression model used a regime to stratify counties by states that expanded Medicaid under the Affordable Care Act and those that did not. Among counties in states that did not expand Medicaid, the variables per capita pharmacies, per capita clinics, and percent less than sixty five years with no health insurance had no statistically significant associations with the outcome. In counties that were part of the Medicaid expansion, only per capita pharmacies and per capita clinics were not significant. DISCUSSION: Per capita health care providers were statistically significant and positively associated with influenza vaccination coverage in both regime models, but had a larger regression coefficient in the expanded Medicaid regime model