Assessing Spatial & Racial Disparities in COVID-19 Mortality

How do social determinants of health help explain inequities in COVID-19 mortality?

US Covid Atlas researchers have been using exploratory spatial data analytic (ESDA) techniques to better understand the disproportionate impacts of the pandemic on communities of color nationwide. We’re pleased to share our newest research article, “Assessment of Structural Barriers and Racial Group Disparities of COVID-19 Mortality with Spatial Analysis,” published today in JAMA Network Open.

In a previous post, we talked about existing evidence on racial and spatial disparities, challenges in quantifying structural racism in the COVID-19 context, and our responses to these challenges. In this post, we summarize the main findings and important take-aways from the full study.

Co-location Analysis and Concentrated Longitudinal Impact Counties

We started with co-location analysis to identify counties with consistently high COVID-19 death rates that had a high proportion of residents of a particular racial or ethnic group. If a county a) had mortality rates in the top 20% for at least 100 days in the first year of the pandemic, and b) were in the top 20% of adult populations of a particular ethnic or racial group, then this county was identified as a concentrated longitudinal impact (CLI) county. We identified 347 CLI counties for Black or African American population, 198 CLI counties for Hispanic or Latinx population, and 33 CLI counties for non-Hispanic White. Comparing the social determinants of health measures of these counties with other non CLI counties, we found:

• Black or African CLI counties spanned rural, suburban, and urban areas, primarily in the South and Southeast. They experienced numerous disadvantages, including higher income inequality and more preventable hospital stays.
• Hispanic or Latinx CLI counties were located mainly in urban areas and had high percentages of uninsured residents. These counties were clustered in the Southwest, and southern Florida.
• Non-Hispanic White CLI counties were primarily located in rural areas, with high percentages of elderly populations, having limited access to quality health care. These counties were found across the Midwest and rural Appalachian regions.

Spatial Regime Regression and SDOH Associated with COVID-19 Mortality in Rural, Suburban, and Urban Areas

Recognizing that distinct patterns of CLI counties would emerge across rural, suburban, and urban counties, we used spatial regime regression to test the associations between SDOH measures and COVID-19 mortality rates in different communities. The results confirmed the spatial heterogeneity in the sense that most predictors have significantly different coefficients across rural, suburban, and urban areas.

• In urban areas, counties with high mortality rates are those having more immigrant populations with traditional family structures and multiple accessibility stressors (e.g., crowded housing and lack of health insurance).
• In suburban areas, high mortality rates were associated with high poverty rates along with high percentages of older populations and/or people with a disability.
• In rural areas, high mortality rates were linked to limited access to quality health care.
• Access to the Internet was associated with mortality rate in all communities, highlighting the importance of technological access to tele-health services, ordering groceries online, or applying for public assistance.

To sum, we found associations between different SDOH measures and COVID-19 mortality varied across different racial and ethnic groups and community types. This points to the importance of place-based intervention in tackling health disparities. We also urge the disaggregated COVID-19 data at more granular spatial levels for future research.

Read the Article

You can read the full open-access (no paywall) article in JAMA Network Open here.

For a top-line summary of our key findings and research methods, check out our research brief.

If you have any questions about the data or methods used in this research or how to understand the changing COVID landscape, contact me at qinyunlin@uchicago.edu or reach out to US Covid Atlas on Twitter.