Initial non-surgical versus upfront surgical management of ERCP-related perforations by Stapfer type: a systematic...

ERCP-related perforation is uncommon but potentially fatal, and optimal initial management remains uncertain across Stapfer subtypes. We compared initial non-surgical management with upfront surgery overall and by Stapfer type, retaining patients requiring rescue surgery within their original strategy.

Databases were searched to 30 June 2026. Treatment success was standardised as survival to discharge without operative rescue after initial non-surgical management and survival to discharge after upfront surgery. Random-effects risk ratios (RRs) were pooled overall and by Stapfer type; certainty was assessed using GRADE.

Twenty-three retrospective studies included 681 perforations. In 15 reconstructable cohorts, initial-strategy success was 227/273 (83.2%) after initial non-surgical management and 75/89 (84.3%) after upfront surgery (RR 0.97, 95% CI 0.81-1.17; I² = 25%). Mortality was 16/273 (5.9%) vs. 14/89 (15.7%) (RR 0.53, 95% CI 0.25-1.15). Length of stay was heterogeneous and clinically inconclusive (MD 1.03 days, 95% CI - 3.35 to 5.40; I² = 83%; prediction interval - 11.24 to 13.30). Type I success (RR 1.04, 95% CI 0.77-1.41) and mortality (RR 1.47, 95% CI 0.56-3.86) were non-significant. Contemporary endoscopic closure succeeded in 10/10 highly selected Type I cases (100%; exact 95% CI 69.2-100.0%), but was too small to establish superiority. Type II success (RR 0.69, 95% CI 0.41-1.16) and mortality (RR 1.44, 95% CI 0.23-8.90) were also non-significant. Type III and IV non-surgical success was 46/49 (93.9%) and 34/34 (100%). Rescue surgery occurred in 14.3, 12.9, 5.4, and 0% of Types I-IV. Certainty was very low.

Very-low-certainty evidence suggests management should be tailored to Stapfer subtype, physiological stability and source control, with prompt operative rescue when non-surgical management fails.