Using spatial Bayesian models to estimate associations between structural racial discrimination and disparities in severe maternal morbidity

Introduction: The United States maternal health crisis is especially profound in Georgia, disproportionately affecting Black birthing people. In Georgia, 35% of all births are to Black-identifying people, with rates of severe maternal morbidity (SMM) significantly exceeding national averages. Methods: The sample population comprised Georgia linked live birth/fetal death certificate and hospital discharge data for deliveries from 2013 to 2021. Structural racial discrimination (SRD) was defined at the county level using four domains. We estimated county-specific rates of SMM using Bayesian conditional autoregressive Poisson models and compared the rate difference in SMM for Black versus white birthing people across domain-specific strata of SRD. Results: The sample included 709,335 deliveries to Black and white birthing people. The prevalence of SMM was higher among births to Black individuals compared to white counterparts (3.2% vs. 1.7%), with a mean risk difference of 13.1 per 1000 deliveries between race groups. Results of the SRD-SMM regression demonstrated larger Black-white racial disparities in counties with the highest concentration of resource deprivation compared to the highest concentration of affluence (B: 4.5, 95% CI: 1.1, 8.0). Similarly, in counties with a greater polarization of Black and low-income residents, the disparity was larger compared to counties with greater racial and income homogeneity (B: 3.84, 95% CI: 0.22, 7.44). Conclusions: Our results highlight the nuanced relationship between structural racism and health outcomes in Georgia. SMM rates were higher among Black birthing people compared to their white counterparts.