CMStatistics 2020: Start Registration
View Submission - CMStatistics
Title: Measuring community vulnerability in spatiotemporal disease mapping of COVID-19 Authors:  Rachel Carroll - University of North Carolina Wilmington (United States) [presenting]
Abstract: Community vulnerability is an important measure to consider in modeling the spread and impact of COVID-19. This spatially dependent factor can be described by several variables or reduced to a composite measure. Several well established composite measures of community vulnerability exist, including the social vulnerability index and the area deprivation index. A new, additional measure of vulnerability specifically related to COVID-19 was developed in recent months - the COVID-19 community vulnerability index. We compare these four methods of accounting for community vulnerability in the modeling of COVID-19. The statistical model used was a spatiotemporal Bayesian Poisson regression model, and models were compared to fit via the deviance information criterion and the Watanabe-Akaike information criterion. Results suggest a better model fit when including any of the vulnerability measures, and all indices generally suggest a higher risk of COVID-19 in areas with more vulnerability. The COVID-19 community vulnerability index has some benefit in terms of model fit compared to the others considered, particularly when examining COVID-19 deaths. Still, insights on specific variable relationships with COVID-19 are not accessible when using composite measures. In conclusion, it is important to adjust for community vulnerability in modeling of COVID-19, but the best measure depends on the goal of the assessment.