Forty-one studies (n=17,810 patients) were included: 32 published and nine unpublished studies. Only three studies provided an appropriate description of randomisation. Seven studies were adequately blinded, 33 studies provided an appropriate description of withdrawals and 37 studies carried out an intention-to-treat analysis.
Results of meta-analyses were reported on summary forest plots; exact numerical results were unclear for most comparisons as was the number of studies that contributed to each meta-analysis. There was substantial heterogeneity between studies for HbA1c (I2=94.6%, p<0.01). When data for all placebo-controlled studies were combined there was a significant improvement in HbA1c. Effects were similar when results were stratified according to dipeptidyl peptidase-4 agent, whether the dipeptidyl peptidase-4 agent was administered as monotherapy or combined therapy, duration of therapy, baseline HbA1 and duration of diabetes. Dipeptidyl peptidase-4 inhibitors showed a similar effect to thiazolidinediones, but metformin and sulphonylureas were significantly more effective than dipeptidyl peptidase-4 inhibitors.
There was no significant difference in incidence of hypoglycaemia or for BMI between dipeptidyl peptidase-4 agents and placebo. dipeptidyl peptidase-4 agents were associated with a significantly lower risk of hypoglycaemia than sulphonylureas and a significantly lower BMI compared to thiazolidinediones.
Risk of adverse events was similar for dipeptidyl peptidase-4 inhibitors and placebo and was significantly lower than sulphonylureas (OR 0.64, 95% CI 0.51 to 0.80; two trials), metformin (OR 0.78, 95% CI 0.61 to 1.00; two trials) and α-glucosidase inhibitors (OR 0.51, 95% CI 0.39 to 0.67; two trials). There was no significant difference in risk of death or cardiovascular events compared to control.
There was no evidence of publication bias (p=0.13).