Figure 3.

Forest plot comparing antidiabetic therapies for the occurrence of major cardiovascular adverse events (MACE) in an additive effects network meta-analysis with a random-effects model. The reference group was considered as placebo. Number of pairwise comparisons was 140 (140 studies with >100 individuals per arm), number of treatments (n = 12), number of designs (n = 25) with 7 active components. Heterogeneity / inconsistency analysis showed tau2 = 0; tau = 0; I² = 0% [0.0; 6.5%]; Q = 111.72 (p = 0.91). Among TZDs (thiazolidinediones) only pioglitazone was included.