We present the outputs of all models presented in this analysis. We ran logistic mixed models for the base, base plus, and multivariate models. We used a matern kernel based on the patients home address in order to account for spatial autocorrelation in our analysis. We present coefficient effect size (Coefficient), standard error (SE), lower confidence interval (CI_low), upper confidence interval (CI_high), p-value (p), and model component type (Component) for all models. The coefficient effect size have been exponentiated, therefore represents a multiplicative risk ratio. In this situation a value greater than 1 represents increased risk and a value less than 1 represents decreased risk. For example, for a quantitative trait, a coefficient estimate of 1.2 represents 20% increase risk per unit increase of the quantitative trait. All coefficient estimates are reported on the original scale of the variable. All census tract variables are percentages, therefore the effect sizes represent increase or decrease of risk per 1% increase of the census tract variable.
unemployed_pct - percentage of individuals in census tract that are unemployed
pop_dens - population density of census tract (number of people per square km.)
Using C24010 - Sex by Occupation for the Civilian Employed Population 16 Years and Over
Individuals that received viral polymerase chain reaction (PCR) testing for SARS-CoV-2 in the MGB system between February 1, 2020 and July 25, 2020. All MGB employees and individuals under the age of 18 are excluded. Outcome is infection with SARS-CoV-2.
Individuals that tested positive for SARS-CoV-2. All MGB employees and individuals under the age of 18 are excluded. Outcome is inpatient hospitalization related to SARS-CoV-2 among those testing positive.
Individuals that were hospitilized due to SARS-CoV-2. All MGB employees and individuals under the age of 18 are excluded. Outcome is death related to SARS-CoV-2 among those hospitalized.