Twenty-two RCTs were included in the review (n=9,325 participants; range 64 to 1159). Five thiazolidinedione-based regimen and two exenatide-based regimen RCTs were open label studies. Allocation concealment and randomisation methods were adequately described in three double blind RCTs and three open label trials. Inclusion/exclusion criteria were adequately described in all trials. Two trials did not report intention-to-treat analysis. Withdrawals were high for each drug (3 to 38% for thiazolidinediones and 15 to 31% for exenatide).
Glycated haemoglobin (HbA1c) was significantly reduced with thiazolidinedione-based regimens (WMD -0.80%, 95% CI -1.10 to -0.50; 20 study arms; n=6,587 participants) and exenatide-based regimens (WMD -0.60%, 95% CI -1.04 to -0.16; five study arms; n=2,002 participants) compared with control treatments. These were associated with significant heterogeneity (I2>95%). Funnel plots suggested the presence of publication bias.
Thiazolidinedione-based regimens (OR 2.27, 95% CI 1.22 to 4.24; nine study arms; 3816 participants) and exenatide-based regimens (OR 2.90, 95% CI 1.28 to 6.55; five study arms; n=1,965 participants) were both significantly better at achieving HbA1c targets of less than 7% compared with control treatments. These were associated with significant heterogeneity (I2>91%).
Thiazolidinedione-based regimens were associated with significantly lower fasting plasma glucose concentrations than control treatments (WMD -29.58mg/dL, 95% CI -39.27 to -19.89), but fasting plasma glucose concentrations were not significantly different between exenatide-based regimens and control treatments.
Body weight was significantly reduced with exenatide-based regimens (WMD -2.74kg, 95% CI -4.85 to -0.64), but was not significantly different between thiazolidinedione-based regimens and control treatments.
There was no significant relationship between thiazolidinedione-based regimens or exenatide-based regimens and control treatments for the risk of non-severe hypoglycaemia, The most commonly reported adverse effects in the exenatide studies were gastrointestinal; there was a significant increase in nausea (OR 9.02, 95% CI 3.66 to 22.23), vomiting (OR 4.56, 95% CI 3.13 to 6.65) and diarrhoea (OR 2.96, 95% CI 2.05 to 4.26) compared with control treatments.
The results of the sensitivity analyses were similar.
The results of subgroup analyses were also reported.