Multi-model insights into racial and ethnic heterogeneities in early COVID-19 transmission and severity in the U.S.

The COVID-19 pandemic was marked by unequal disease burden across racial and ethnic groups in the US. Using a multi-model ensemble framework, we retrospectively assessed the impact of racial and ethnic heterogeneities in COVID-19 transmission and disease severity on mortality during the winter 2020–2021 surge in California and North Carolina. Models incorporating race and ethnicity projected mortality more accurately than null models, while the ensemble outperformed all individual models on average by weighted interval scores. Eliminating transmission heterogeneities produced larger mortality reductions than eliminating severity differences alone, particularly for Latino populations in California (55% reduction) and Black populations in North Carolina (28% reduction). Reducing severity heterogeneity also lowered mortality, with the greatest benefit among Black populations in both states (21% and 16% reductions in California and North Carolina, respectively). Eliminating both transmission and severity heterogeneity yielded the largest reductions overall, predicting 49–70% lower mortality across racial and ethnic groups in California and 23–40% in North Carolina. Mitigating transmission inequities also produced indirect benefits for non-targeted populations, highlighting the broad impact of reducing disparities in exposure risk and access to protective resources.