Statistical Analysis
The database was established by EpiDate3.1 software, the data were input twice in parallel. The final analysis database is formed after logical error checking and sorting of the input data and analysis and processing of outliers. Continuous variables were expressed as means ±SDs; if the data conformed to a normal distribution, the two groups were compared using an independent samples t test, and multiple groups were compared using variance analysis. The least significant difference (LSD) was used for pairwise comparisons among those with intragroup differences. For nonnormally distributed data, Wilcoxon rank-sum tests were used for comparisons between two groups, and Kruskal-Wallis H tests were used for comparison between multiple groups. Categorical variables were described as percentages (rates); comparisons between two groups were performed using chi-square tests, and comparisons between multiple groups were performed using crosstabulation analysis. After we created the propensity score, patients were matched on the propensity score in a 1:2 ratio. We used calipers of width 0.2 of the standard deviation of the logit of the propensity score, as recommended in literature. Balance diagnostics were done by comparing the absolute standardized differences of the propensity scores, with values greater than 0.1 indicative of imbalance. P<0.05 was considered statistically significant. All analyses were performed in SPSS version 23.