Aims: The association between metformin use and neurodegenerative disease (ND) onset remains controversial. In this systematic review and meta-analysis, we aimed to determine the relationship between metformin use and ND risk based on data from population-based cohort studies.
Methods: Articles were systematically searched in PubMed, EMBASE and Cochrane Library databases. Pooled relative risks (RRs) with 95% CIs were obtained using a random-effects model. Subgroup analyses, sensitivity analyses and meta-regression were performed to identify the sources of heterogeneity and strengthen the results.
Results: Twelve population-based cohort studies involving 194,792 participants (94,462 metformin users and 100,330 metformin non-users) were eligible for inclusion in this meta-analysis. The pooled RR of NDs reached 0.77 (95% CI 0.67-0.88) when comparing metformin users with non-users. The effects were more prominent in long-term metformin users (≥4 years) (RR 0.29, 95% CI 0.13-0.44) and studies from Asian countries (RR 0.69, 95% CI 0.64-0.74). The effect estimates were stable when stratified by subtypes of NDs, study designs, and control definitions (p for interaction >0.05). Meta-regression did not identify the coefficients as the sources of heterogeneity (all p > 0.05).
Conclusions: This systematic review and meta-analysis found that metformin use, especially long-term use, was associated with lower ND risk. However, because there was substantial heterogeneity among studies, high-quality randomized controlled trials are still needed to confirm this finding.
Keywords: diabetes mellitus; meta-analysis; metformin; neurodegenerative diseases; population-based cohort studies.
© 2022 Diabetes UK.