Thirty nine RCTs (17,860 patients) were included in the review. With the exception of one trial, all other trials scored from 3 to 5 on the Jadad scale, which suggested moderate to high quality.
Change in glycated haemoglobin: Glucagon-like peptide-1 analogues showed similar changes in glycated haemoglobin compared with basal insulin and biphasic insulin. However, glucagon-like peptide-1 analogues showed a statistically significant greater reduction in glycated haemoglobin compared with all other agents: sulfonylureas (OR -0.20%, 95% CI -0.34 to -0.04), glinides (OR -0.31%, 95% CI -0.61 to -0.02), thiazolidinediones (OR -0.20%, 95% CI -0.38 to -0.00), alpha-glucosidase inhibitors (OR -0.36%, 95% CI -0.64 to -0.07), and dipeptidyl peptidase-4 inhibitors (OR -0.32%, 95% CI -0.47 to -0.17).
Hypoglycaemia (38 RCTs): Comparisons between active agents showed that there was a statistically significant reduced risk of hypoglycaemia with: thiazolidinediones compared with sulfonylureas (OR 0.05%, 95% CI 0.02 to 0.13) and glinides (OR 0.04%, 95% CI 0.01 to 0.17); dipeptidyl peptidase-4 inhibitors compared with sulfonylureas (OR 0.13%, 95% CI 0.08 to 0.21) and glinides (OR 0.11%, 95% CI 0.03 to 0.32); and glucagon-like peptide-1 analogues compared with sulfonylureas (OR 0.10%, 95% CI 0.05 to 0.21) and glinides (OR 0.09%, 95% CI 0.02 to 0.29). There was a statistically significant increased risk of hypoglycaemia for basal insulin compared with thiazolidinediones (OR 10.57%, 95% CI 2.41 to 51.26), dipeptidyl peptidase-4 inhibitors (OR 4.22%, 95% CI 1.24 to 14.87) and glucagon-like peptide-1 analogues (OR 5.20%, 95% CI 1.77 to 16.45).
Change in body weight (23 RCTs): There was a statistically significant reduction in body weight with: alpha-glucosidase inhibitors compared with sulfonylureas (OR -3.18, 95% CI -4.17 to -2.17) and glinides (OR -2.41, 95% CI -3.63 to -1.22); dipeptidyl peptidase-4 inhibitors compared with sulfonylureas (OR -1.93%, 95% CI -2.35 to -1.53) and glinides (OR -1.16, 95% CI -2.07 to -0.30); and glucagon-like peptide-1 analogues compared with sulfonylureas (OR -3.81, 95% CI -4.44 to -3.24) and glinides (OR -3.06, 95% CI -4.08 to -2.11). There was a statistically significant increase in body weight for thiazolidinediones compared with glinides (OR 1.07, 95% CI 0.05 to 2.03), alpha-glucosidase inhibitors (OR -3.47, 95% CI -4.52 to -2.41), dipeptidyl peptidase-4 inhibitors (OR -2.23, 95% CI -2.81 to -1.66), and glucagon-like peptide-1 analogues (OR -4.12, 95% CI -4.84 to -3.44). There was also a statistically significant increase in body weight for biphasic insulin compared with glinides (OR 2.01, 95% CI 0.41 to 3.56), alpha-glucosidase inhibitors (OR 4.41, 95% CI 2.79 to 6.05), dipeptidyl peptidase-4 inhibitors (OR 3.17, 95% CI 1.82 to 4.52), glucagon-like peptide-1 analogues (OR 5.06, 95% CI 3.74 to 6.41), and basal insulin (OR 2.02, 95% CI 0.69 to 3.35).
Results for comparisons with placebo, and non-significant results for comparisons between different drugs were reported in the review.
Results from meta-regression, sensitivity analyses and direct meta-analyses were generally consistent with network results, although there were some differences between network and direct meta-analyses for change in haemoglobin levels.