Quantitative synthesis
The data of included literature was undertaken using Review Manager V.5.4. Random effects models were used as a measure of the outcomes with high heterogeneity, otherwise, use the fixed effects models. We calculated the MD with a 95% CI and pooled the means and SD from continuous results, OR with a 95% CI from dichotomous data. I2 value and Q test was used to assessed the heterogeneity between all statistical tests, which was showed with the forest plots. Heterogeneity was regarded as to be low if I2<=30%, high if I2>30%14. Galbraith’s diagram performed the heterogeneity created by the circle beyond the lower and upper lines. Subgroup sensitivity analysis was executed by excluding one study for the purpose that evaluated the effect on the result. We applied STATA V.12.1 to subsequent operations. The publication bias was assessed with Egger’s test and funnel plot. All the reported p values were two-side and it was considered statistically significant merely it less than 0.05.