2.3. Statistical Analyses

We entered the information into an electronic spreadsheet and double-checked it during data transfer. We analyzed the data using the STATA, version 14.0 (College Station, TX, USA, 2015). We conducted the mortality analysis in two stages, descriptive and analytical. First, we performed the Kolmogorov–Smirnov test in the descriptive step with the continuous variables. We assessed the means and standard deviations for normal distributions and the median, interquartile range (IQR), or minimum and maximum for non-normal distributions. The categorical variables were described in absolute and relative frequencies. We compared the underlying and immediate causes of death among the age groups by using the Pearson chi-square test.

We conducted bivariate and multivariate analyses to examine the association of the factors related to overall survival. In the bivariate analysis, each independent variable was linked to overall survival and we obtained the unadjusted hazard ratio (HR) and the respective 95% confidence interval (CI). Subsequently, we included variables with p values of <0.20 in a Cox proportional hazards model to control confounding variables [4]. The results of the multiple regression were described as Adjusted HR (HRadj) and 95% CI. In addition, we provide a graph of the Kaplan–Meier curves of the statistically significant variables in the final model. We considered variables with p values of <0.05 in the multiple regression analysis to be statistically significant.