Statistical Analysis
All statistical analyses were performed using SPSS version 13.0 (SPSS
Inc., Chicago, IL). Continuous variables of baseline patient and
hospital characteristics, treatment variables, and complications were
compared using Student t test and discrete variables by the Fisher Exact
Test. Predictors of developing severe COVID-2019 were determined by
performing univariate logistic regression on each demographic variable
using survey statistics; those with a p value of < 0.01 in
univariate analyses were included in the multiple variable logistic
regression model using forward prediction. Area under the receiver
operating characteristic curves (AUC) assessed the discriminatory
capacity of each logistic regression model. Thereafter, hierarchical
multivariate logistic regression models were constructed with all
defined predictor variables to evaluate outcomes. A P value of
< 0.05 was considered statistically significant.