Regional Disparity in Uninsurance Rate
Regional disparity in the uninsurance rate impact of COVID-19: a spatial machine learning approach
Objective: This study investigates the regional disparities in the impact of the COVID-19 pandemic on the uninsured rate across U.S. counties. The research aims to identify the influence of Medicaid expansion and other socioeconomic factors on health insurance coverage. The key focus is understanding how these variables contributed to changes in uninsured rates during the pandemic and exploring spatial variations in these effects.
Method and Data: The study employs a spatial machine learning approach using data from 3,108 U.S. counties for 2019, 2020, and 2021. The analysis integrates several methodologies, including Getis-Ord Gi* Hotspot analysis, spatial weights matrices, and geographically weighted regression (GWR). The data is sourced from the County Health Rankings dataset, and key variables include Medicaid expansion status, unemployment rates, and demographic factors such as race and education level. Spatial random forests are used to predict the importance of variables, while GWR is utilized to estimate the spatial distribution of uninsured rates.
Results: The results reveal significant spatial variations in the uninsured rates. The Getis-Ord Gi* Hotspot analysis identifies clusters of counties where Medicaid expansion either significantly decreased or, in some cases, increased uninsured rates. The spatial machine learning model highlights Medicaid expansion, Hispanic population percentage, and education level as critical factors influencing the uninsured rate. The geographically weighted regression demonstrates that these variables exhibit spatial heterogeneity, with their effects varying significantly across regions. These findings suggest that Medicaid expansion’s impact is not uniform, being strongly influenced by local socioeconomic and demographic conditions.