Statistical analysis
A comprehensive meta-analysis was performed to calculate Relative risk (RR) or mean difference and their 95% confidence interval (CI) by Revman 5.3. Statistical heterogeneity was assessed by Cochran’s Q statistic and the I2statistic. The fixed-effects model was used to pool studies when statistical heterogeneity was absent and when substantial heterogeneity was defined as I2 >50% or chi-squared test P <0.1; otherwise, the random-effects model was employed. For individual trials with no events in one or both groups, a continuity correction of 0.5 was added to each cell17,18.