Figure 4.

Meta-analysis for fasting glucose and insulin. (a) Meta-analysis using random effect method, outcome fasting insulin. Meta-analysis was performed using post mean value (or median value) and calculating mean differences (with their 95% CI). Impact on fasting insulin reduction is showed considering all studies together. I² is high. All estimates are expressed in mIU/L; (b) meta-analysis using random effect method, outcome fasting glucose. Meta-analysis was performed using post mean value (or median value) or mean change within groups and calculating mean differences (with their 95% CI). Impact on fasting glucose reduction is showed by overall estimate with null heterogeneity. All MDs are expressed in mg/dL.