Meta-analysis
Heterogeneity was assessed using the I2 index, with
I2 > 50% indicating significant
heterogeneity21. Continuous variables of interest were
described with means, standard deviations, and sample size to estimate
confidence intervals (CIs). Dichotomous variables were described using
numerators and denominators to estimate proportions and CIs. The random
effects model was used to estimate summary statistics for variables of
interest, where individual studies were treated as a random variable. To
estimate the difference in variables of interest over time, only
publications that contained both baseline and follow-up means and
standard deviations were used. We calculated risk ratios using the
Mantel-Haenszel method and the mean difference with inverse variance.
All statistical analyses were performed using Review Manager 5.3
software (Cochrane).