One of the most disturbing trends of the COVID-19 pandemic in ethnically diverse countries such as the United Kingdom and the United States (US) is the disproportionate burden of COVID-19-related morbidity and mortality borne by non- white, racial and ethnic minorities. This led to a nation-wide call for collection of COVID data by race/ethnicity. While summary data are available now in many states, individual-level data to use in modeling remain elusive. The higher burden of chronic diseases such as hypertension and diabetes among ethnic and racial minorities in the US is believed to explain some of the disparities in COVID-19 mortality, while socioeconomic characteristics including occupational exposure and housing conditions are thought to partially explain higher rates of infection. However, none of these factors fully account for the ethnic and racial disparities that have been observed, indicating the contribution of structural racism to the inequities revealed by COVID-19. Our novel disparities modeling approach has the ability to bring together disparate data sources, to determine the patterns of mixtures of exposures, and can then link these patterns to outcomes. Data on socioeconomic status (SES) and demographics come from the American Community Survey, comorbidities from the CDC 500 Cities data, and COVID-19 data from the GitHub repository of NY Health Department. We utilize latent class mixture models, based on area level SES and demographic data, and determine how the patterns of each contribute to health outcomes. In a hierarchical model framework, we then use spatio-temporal modeling of the clusters of symptoms and SES exposures to model test-positivity in NYC, using publicly available data. These analyses are carried out in MPlus and via the INLA package in R. Latent class SES models have 4 classes, Low and High Advantage and Disadvantage (LAdv, LDis, HAdv, HDis); comorbidities yields 3 classes, Low, Moderate, and High (L, M, H). Lowest SES is associated with non-Hispanic Black (NHB), Hispanic (Hisp), while areas with highest comorbidities associated with higher NHB and age over 65 population densities. A Bayesian random effects Poisson model demonstrates that test positivity is elevated in areas with higher NHB and Hisp population densities, in areas with M to H comorbidities and in areas with lowest SES. These models demonstrate a significant impact of race/ethnicity beyond the SES and comorbidity differences exhibited. This could be due to lack of individual-level information, or due to occupational differences and ability to socially distance not yet fully captured by our modeling. Our hope is that these data can aid in identifying areas of acute need when determining testing and tracing priorities as well as areas to prioritize as a vaccine or other preventative strategies becomes available.

Citation Format: Kara McCormack, Alexandra Larsen, Nicholas Bachelder, Victoria Seewaldt, Terry Hyslop. Disparities of COVID outbreak in NYC: A spatio-temporal analysis of SES, comorbidities and race [abstract]. In: Proceedings of the AACR Virtual Conference: Thirteenth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2020 Oct 2-4. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(12 Suppl):Abstract nr PO-006.