Statistical analysis
The statistical analyses were conducted using the Review Manager version 5.4 software (the Cochrane Collaboration 2014, Nordic Cochrane Centre Copenhagen, Denmark; https://community.cochrane.org/). In this study, significant heterogeneity was indicated by a P-value of < 0.10 in the Cochrane Q test or an I² value of > 50%, which led to the use of random-effects models and the exploration of a potential source of heterogeneity. When these tests were negative for heterogeneity, fixed-effects models were chosen to calculate pooled HRs through the inverse-variance method. In the pooled analysis, the adjusted HRs and 95% CIs were converted to the natural logarithms (Ln [HR]) and their corresponding standard errors (Ln [upper CI]-Ln [lower CI])/3.92), which were pooled by a DerSimonian and Laird random-effects model with an inverse variance method.