Metformin and health outcomes: An umbrella review of systematic reviews with meta-analyses

Background The objective was to capture the breadth of outcomes that have been associated with metformin use and to systematically assess the quality, strength and credibility of these associations using the umbrella review methodology. Methods Four major databases were searched until 31 May 2020. Meta-analyses of observational studies and meta-analyses of randomized controlled trials (RCTs) (including active and placebo control arms) were included. Results From 175 eligible publications, we identified 427 different meta-analyses, including 167 meta-analyses of observational studies, 147 meta-analyses of RCTs for metformin vs placebo/no treatment and 113 meta-analyses of RCTs for metformin vs active medications. There was no association classified as convincing or highly suggestive from meta-analyses of observational studies, but some suggestive/weak associations of metformin use with a lower mortality risk of CVD and cancer. In meta-analyses of RCTs, metformin was associated with a lower incidence of diabetes in people with prediabetes or no diabetes at baseline; lower ovarian hyperstimulation syndrome incidence (in women in controlled ovarian stimulation); higher success for clinical pregnancy rate in poly-cystic ovary syndrome (PCOS); and significant reduction in body mass index in people with type 1 diabetes mellitus, in women who have obesity/overweight with PCOS and in obese/overweight women. Of 175 publications, 166 scored as low or critically low quality per AMSTAR 2 criteria. Conclusions Observational evidence on metformin seems largely unreliable. Randomized evidence shows benefits for preventing diabetes and in some gynaecological and obstetrical settings. However, almost all meta-analyses are of low or critically low quality according to AMSTAR 2 criteria.

Automated summary

The study found that observational evidence on metformin is largely unreliable, while randomized evidence shows benefits for preventing diabetes and in some gynaecological and obstetrical settings. However, almost all meta-analyses are of low or critically low quality according to AMSTAR 2 criteria.