Estimating Covid-19 Survival Rate and Inferring Case Severity with Respect to Milwaukee County Policy Change Using Logistic Regression

Abstract

Coronavirus disease 2019 (COVID-19) is a global issue, and it is affecting 170 countries in very different ways. In the United States, a lot of efforts have been made nationally and by individual states to curb the spread and severity of COVID-19. Policy changes and recommendations have been met with variable success across the country. There is a wealth of information on where COVID-19 infection and death are prevalent, and there are several articles discussing disparities in those outcomes among different populations. However, those findings are not necessarily tied to a policy change or, in the weeks that follow a change, a description of the corresponding change in COVID-19 prevalence and severity, if any. In this thesis, we will use univariate logistic regression and the cumulative logit model to identify the population in Milwaukee County most at-risk for death from COVID-19 with respect to age, race, and sex, using confirmed COVID-19 case and death data from the Wisconsin Department of Health Services. We will then break the data apart into time intervals of approximately two months to see if these risks were more or less severe as a function of policy changes made regarding social distancing, requiring a mask, and limiting non-essential work interactions.