Data Analysis
Statistical analyses were conducted using STATA MP-Parallel Edition
Statistical Software, Version 18, College Station, TX: StataCorp LP. Ap -value of 0·05 was considered statistically significant.
Descriptive statistics included mean and standard deviation for
continuous-level data, median and interquartile range for ordinal data,
and frequency and proportion for nominal data. Comparative analyses of
demographic and clinical characteristics and study outcomes according to
ABO blood type included Chi-Square Test of Homogeneity or Fisher’s Exact
Test, Kruskall-Wallis H Test, and one-way Analysis of Variance (ANOVA).
Relative risk of COVID-19 infection, disease severity, and disease
outcomes (ICU admission, intubation, and mortality) according to ABO
blood type was measured using risk ratio (RR), which was estimated using
log-binomial regression.9 Crude risk ratios (cRR) were
initially estimated. Afterwards, significant confounders were screened,
analyzed, and controlled using a 10% change-in-estimate
criterion10 to estimate the adjusted risk ratio (aRR).