2.5. Data synthesis

Revman 5.3 software provided by the Cochrane Collaboration was used for data analyses. The model used to pool the data depends on the existence and extent of heterogeneity. If the I² statistics were less than 50%, the heterogeneity could be accepted, and the fixed-effect model was chosen. If the I² statistics exceeded 50%, the random-effects model was used. When heterogeneity among studies was obvious (I² > 50%), the sources of heterogeneity would be investigated. For binary outcomes, the pooled relative risk (RR) with 95% confidence interval (CI) was used as the effect measure. For the continuous outcome, weighted mean differences (WMD) or standardized mean differences (SMD) was used as the effect measure, both with 95% confidence intervals (CI). The approach to incorporating cross over trials is to take all measurements from the 2 intervention periods and analyse these as if the trial were a parallel-group trial.^[27] Publication bias would be assessed by funnel plot and the Eggers test if the group included more than 10 studies. The Eggers test was performed by software Stata 14.0 (Stata Corp, College Station, Tex).