Forty-one RCTs (n=3,074) were included in the review: 33 parallel trials and 8 crossover trials.
The authors reported that there was no evidence of publication bias, although this analysis was not shown. They also reported that there was no statistically significant heterogeneity between the studies; this was also not shown.
Systolic BP (21 trials, n=1,667): the pooled WMD showed no significant difference between the groups (WMD -1.09 mmHg, 95% confidence interval, CI: -3.01, 0.82). This was also the case in the trials using the highest doses of metformin (at least 2,550 mg/day): WMD 0.54 mmHg (95% CI: -1.94, 3.02).
Diastolic BP (19 trials, n=1,609): the pooled WMD showed no significant difference between the groups (WMD -0.97 mmHg, 95% CI: -2.15, 0.21). This was also the case in the trials using the highest doses of metformin (at least 2,550 mg/day): WMD -0.17 mmHg (95% CI: -1.62, 1.27).
HbA1c: there was a small but statistically significant increase in glycaemic control with metformin compared with the comparator (-0.74, 95% CI: -0.84, -0.65).
Plasma triglycerides (37 trials, n=2,891): the pooled WMD showed a significantly greater reduction in concentration of plasma triglycerides in the metformin groups than in the control groups (WMD -0.13 mmol/L, 95% CI: -0.21, -0.04). Sensitivity analyses showed that metformin significantly reduced triglycerides compared with control treatment only in the following: studies with the greatest difference in glycaemic control in favour of metformin; studies using the two highest range of doses of metformin; studies with intermediate length of follow-up; studies that compared metformin with insulin therapy; and studies in which metformin was administered as monotherapy.
Plasma total cholesterol (38 trials, n=2,973): the pooled WMD showed a significantly greater reduction in concentration of plasma total cholesterol in the metformin groups than in the control groups (WMD -0.26 mmol/L, 95% CI: -0.34, -0.18). Sensitivity analyses found no relationship between this effect and any of the variables investigated.
Plasma LDL cholesterol (24 trials, n=1,867): the pooled WMD showed a significantly greater reduction in concentration of plasma LDL cholesterol in the metformin groups than in the control groups (WMD -0.22 mmol/L, 95% CI: -0.31, -0.13). Sensitivity analyses generally found no relationship between this effect and any of the variables investigated, with the exception of a stronger effect of metformin in those studies using a higher dose.
Plasma HDL cholesterol (29 trials, n=2,037): the pooled WMD showed no significant difference between the groups (WMD 0.01 mmol/L, 95% CI: -0.02, 0.03).