Two randomised controlled trials (both completed, sample sizes of 263 and 84) were included in the review and meta-analyses. The quality of the two trials was poor and they were at high risk of bias; a breakdown of the results was provided. One control group received placebo and the other received glimepiride.
The trial with the glimepiride control group reported significant reductions in mean change in HbA1c from baseline for both the metformin (-0.71%, p=0.0002) and glimepiride arms (-0.54%, p=0.001).
The placebo-controlled trial found a statistically significant greater reduction of HbA1c from baseline with metformin compared with control (MD -1.10, 95% CI -1.19 to -1.01). This trial also demonstrated a significant reduction in fasting plasma glucose with metformin (-42.9mg/dL) over the placebo control (+21.4mg/dL) with a mean difference of -64.80 in favour of metformin. A significant decrease in low-density lipoprotein levels was reported for the metformin group (data not shown).
Meta-analyses of the two trials showed that there were no statistically significant differences between metformin and control groups for mean changes in HbA1c from baseline (MD -0.63%, 95% CI -1.56 to 0.30; Ι²=99%) and mean changes in body mass index from baseline (MD -0.46, 95% CI -2.27 to 1.34; Ι²=0%).
Further results for each trial were reported separately in the review; this included adverse event results from three additional retrospective cohort studies. The authors did not consider any of these data amenable to meta-analysis.