2.6 Statistical analysis
A meta-analysis of the summary statistics from individual trials was
performed. The statistical analysis was completed by an independent
statistician (C-CF). The differences in dichotomous variables and
outcome endpoints were reported as an odds ratio (OR) or risk ratio (RR)
with 95% confidence intervals (CIs). The continuous variables were
analyzed using weighted mean differences (WMD) or standard mean
differences (SMD). Fixed- and random-effects models used weighting based
on an inverse variance, which was calculated according to DerSimonian
and Laird11. The
between-study heterogeneity was reflected by I 2>50%, with a P <0.05 deemed statistically
significant. In cases of statistical heterogeneity, sensitivity analyses
were performed to assess the contribution of each study to the pooled
estimate by sequentially excluding individual trials and recalculating
the pooled RR estimate for the remaining
studies12. When the
pooled analysis still yielded significant heterogeneity, the
random-effects model was used. The statistical analysis was performed
using the Review Manager (version 5.3; Copenhagen : The Nordic Cochrane
Center, The Cochrane Collaboration, 2014).