To determine the risk of complications requiring treatment following male circumcision by health-care professionals and to explore the impact of participant characteristics, type of circumcision and study design.

We identified studies through systematic searches in online databases (MEDLINE, EMBASE and CENTRAL) and hand searches. We performed random-effects meta-analysis to determine risk of circumcision complications and mixed-effects metaregression analyses to explore the impact of participant characteristics, type of circumcision and study design. Methods were prespecified in a registered protocol (Prospero CRD42020116770) and according to PRISMA guidelines.

We included 351 studies with 4.042.988 participants. Overall complication risk was 3.84% (95% confidence interval 3.35-4.37). Our meta-analysis revealed that therapeutic circumcisions were associated with a 2-fold increase in complications as compared to nontherapeutic (7.47% and 3.34%, respectively). Adhesions, meatal stenosis and infections were the most frequent complication subgroups to therapeutic circumcisions. Bleeding, device removals and infections occurred more frequently in nontherapeutic circumcisions. Additionally, adjusted metaregression analyses revealed that children above 2 years, South American continent, older publication year and smaller study populations increased complication risk. Type of circumcision method, provider and setting were not associated with complication risk. Sensitivity analyses including only better-quality studies reporting indication, age at circumcision, treatment for complications, full-text articles, and adequate follow-up clinically for a minimum of one month or through databases confirmed our main findings while accounting better for heterogeneity.

Circumcision complications occur in about 4 per hundred circumcisions. Higher risks of complications were determined by therapeutic circumcisions and by childhood age when compared to infant. Future studies should assess therapeutic and childhood circumcisions separately.