2.3.3. Statistical analysis
All calculations were performed by Review Manager soft-ware (version 5.3, Nordic Cochrane Centre) and guided by Implement Meta-Analysis with Non-Comparative Binary Data in RevMan Software(9).I2 statistic was used to assess heterogeneity.I2 of less than 25% was low heterogeneity, of 25%~50% was medium heterogeneity, and more than 50% was high heterogeneity. Because of the high heterogeneity of this study, we used random effects model to pool the study-specific frequencies and 95% confidence interval (95% CI) of clinical features or outcomes. Cochrane Q testP < 0.05 was considered statistically significant. Funnel plots were used to assess publication bias. Subgroup analysis was used to assess sensitivity.