2.5. Statistical analysis

Data are presented as count and relative frequency or median and interquartile range. The distribution of variables was graphically assessed using histograms. Binary variables were compared using chi-square analysis or Fisher's exact test, as appropriate. Metric variables were compared using the Mann–Whitney U test. Binary logistic regression analysis and Cox regression analysis were performed, and factors that were considered clinically relevant and did not fulfil the criteria for multicollinearity were included in the model. We used a hierarchical backward stepwise approach. The initial model was gradually reduced. Variables that caused a change in parameter estimates by >10% or were statistically significant at a 0.05 level remained in the model. Missing values were addressed by using the multiple imputation method. Statistical analysis was conducted using IBM SPSS Statistics Version 24.0 (IBM Corp., Armonk, NY). Generally, a p-value <0.05 was considered statistically significant.

The study was prepared in accordance with the Strengthening the Reporting of Observational studies in Epidemiology recommendations (see supplementary material).²⁴