Statistical Analyses

The data included in this study were divided into three groups: The data from hospital A-D for 2015 to 2019 were divided into the training set and internal validation set in a ratio of 7:3; the data from hospital A-D for 2020 and 2021 were used as external validation set1; and the data from hospital E were used as external validation set2.

Statistical analyses were performed using SPSS 22.0 and R (version 4.0.2, Vienna, Austria). Univariate logistic regression analyses and least absolute shrinkage and selection operator (LASSO) regression analyses were performed to identify independent factors. A nomogram was constructed on the basis of these independent factors. The discriminatory values of the models were assessed on the basis of the concordance index (C-index). The area under the curve (AUC) of the receiver operating characteristic (ROC) curve was used to evaluate the accuracy of the nomogram. Calibration curves were established to evaluate the calibration of the nomogram. In addition, decision curve analysis (DCA) and clinical impact curve (CIC) were used to demonstrate the clinical benefit of the nomogram. The multiple imputation method was used to fill in the missing continuous variables. The enumerative data were expressed as rate and percentage, and the chi-square test was used for comparisons between groups. The quantitative variables did not show a normal distribution and were represented by median and interquartile range [M (Q25-Q75)], and comparisons between groups were performed with the Mann-Whitney U test. All statistical analyses were two-sided, and statistical significance was set at P < 0.05.