Statistical Analysis
Statistical analysis was performed by STATA software version 14
(StataCorp, Texan, USA). All included patients stratified as they
received atorvastatin or not during hospitalization. The parametric or
non-parametric distribution of the quantitative data was analyzed using
the Kolmogorov-Smirnov test. Data description of parametric quantitative
variables reported as means ± standard deviation and non-normally
distributed data reported as median (interquartile range=IQR).
Qualitative variables reported as frequency (percentage). Continues
variables compared using independent t-test or Mann-Whitney U test for
normally and non-normally distributed data in bivariate analysis,
respectively. Categorical variables analyzed using Chi-square or
Fisher’s exact test (in the situation in which more than 25% of the
categories had frequencies below five). A P-value of less than 0.05 is
considered significant.
The crude association between atorvastatin administration and occurrence
of outcomes including the need for invasive mechanical ventilation and
in-hospital mortality was performed using the univariate COX
proportional hazards regression model. For the selection of the best
predictors, variables with a P-value of less than 0.2are considered to
be analyzed in multivariable COX regression analysis using a stepwise
selection approach. Confounders were selected based on the
recommendation of previously published epidemiological studies that
reported the probable prognostic value of underlying conditions and also
medications that were being used 16. This included
age, gender, body mass index (BMI) in demographics, hypertension,
diabetes mellitus, coronary artery disease, chronic respiratory
conditions, malignancies, immunocompromised, chronic kidney disease in
comorbidities. Also, we adjusted the model for using beta-blockers and
ACEIs or ARBs 17,18. the model was adjusted for the
medications which were used to treat COVID-19. Patients with negative
time to event removed from the analysis. A 95% confident interval of
hazard ratio was reported. The proportional hazard assumption for COX
analysis was tested using scaled Schoenfeld residuals and the P-value of
0.05 or more considered as no serious violations of the proportional
hazards assumption.