Agent-Based Simulation of Social Determinants of Health for Equitable COVID-19 Intervention

In the COVID-19 pandemic, one-size-flts-all interventions have been implemented based on COVID-19 disease models which simulate disease spread on a more generalized scale, lacking specificity for communities in different settings. This approach, not considering the important local health indicators Social Determinants of Health (SDOH), renders inequities and disparity in intervention effectiveness at the local level. This research answers the following questions: how specific SDOH risk profiles impact COVID-19 outbreak severity and how should interventions be implemented to achieve net positive health impact? A novel agent-based disease model was developed using NetLogo to simulate COVID-19 transmission and intervention using relevant SDOH in specific localities. The model is fitted with COVID-19 variant-specific constants such as susceptibility, mortality rate, recovery time, incubation period, mask efficacy, vaccine efficacy, and reinfection rate. Those constants are further calibrated with SDOH such as healthcare access (vaccination and booster rates) and social context (population size, population density, racial profile, and age demographics). Model inputs also include intervention used (mask mandate, testing and isolation, lockdown) and compliance rate to such interventions. The model was validated in Westchester County, NY for two different time periods with Alpha and Omicron yielding 84.2% and 68.5% accuracy respectively. Sensitivity analysis demonstrated: 1) a higher elderly population, lower young population, lower vaccination rate, and higher Hispanic and Black population were all factors that increased outbreak severity; 2) all variants had similar death rate after reaching ~25% of population vaccinated; and 3) boosters affected Omicron more than other variants, especially in reducing breakthroughs. Scenario analyses were conducted for four U.S. counties: Hunterdon, NJ; Levy, FL; Monterey, CA; and Coles, IL. These analyses showed that 1) informed interventions based on localities' SDOH would dramatically reduce inequity, 2) interventions have higher impact in localities with higher risk SDOH, and 3) weighing other health and social economic consequences against predicted COVID-19 mortality can achieve holistic equity. The model enables local officials to assess the type, intensity, and timing of interventions to achieve maximum health outcomes. They can weigh the benefits of interventions against the socioeconomic or other risks of inequity to local populations. This research empowers local officials in diverse settings with an accessible modeling tool to remain nimble, stay conscious of health disparities, and better focus limited resources in health related decisions for their communities.