“Data is synthesized both quantitatively and qualitatively depending on type of literature, research question, expected heterogeneity and poolability of included studies. To perform the quantitative synthesis, we conduct meta-analysis utilizing both continuous outcome data (mean, standard deviation) and binary outcome data (events, proportions and percentages). The DerSimonian and Laird random or fixed effects models are used to generate the summary measures of effect in the form of mean differences (MD), standardized mean difference (SMD), risk ratio (RR), odds ratio (OR), risk differences (RD), hazard ratio (HR) and ratio of risk ratio (RRIntervention/RRControl).

To evaluate statistical stability, robustness of results and further explore observed statistical heterogeneity in pooled estimates, we conduct various sub-group, sensitivity, and meta-regression analyses based on type of pooling method and various population and study level factors that may influence the pooled effect estimates. We also provide estimates of absolute effects (absolute risk reduction, risk increase and number needed to treat or harm) to have a better understanding of evidence and to inform clinical implications of findings. The Cochrane’s Q statistic is used to detect statistical heterogeneity, where p<0.05 indicates a high level of statistical heterogeneity between studies; and I2 statistic is employed to quantify the statistical heterogeneity between studies, where I2 30-50% represents moderate and I2 50-90% represents substantial heterogeneity.”