Twenty-seven RCTs (n= 7,890) were included in this review: 16 placebo-controlled trials (n=3,743) and 11 active comparison trials (n=4,147).
Placebo-controlled trials. The 16 trials in this section of the review looked at: sulphonylureas (five trials); α-glucosidase inhibitors (five trials); thiazolidinediones (three trials); glinides (two trials); and GLP-1 agonists (one trial). After adjusting for baseline HbA1c and when compared with placebo, sulphonylureas achieved a significantly greater reduction in HbA1c levels than did thiazolidinediones. None of the other intra-drug comparisons were statistically significant.
When only patients failing metformin monotherapy were included, there was no significant difference in the performance of sulphonylureas as against α-glucosidase inhibitors; no other statistical comparisons were possible.
Active comparison trials. The 11 trials in this section compared: thiazolidinediones and sulphonylureas (four trials); insulin and sulphonylureas (two trials); efficacy of agents within the same drug classes (two trials); premixed insulin with basal insulin analogues (two trials); and pioglitazone and a dual PPARα/γ activators (only one trial; not included in the analysis). Direct comparisons reported that sulphonylureas were significantly superior to thiazolidinediones in reducing HbA1c levels, 0.17 (95% CI: 0.16, 0.18, p<0.05). No significant differences were reported between insulin and sulphonylureas.