ERCP-related perforations: a population-based study of incidence, mortality, and risk factors

BACKGROUND: Perforations related to endoscopic retrograde cholangiopancreatography (ERCP) are rare but feared adverse events with highly reported morbidity and mortality rates. The aim was to evaluate the incidence and outcome of ERCP-related perforations and to identify risk factors for death due to perforations in a population-based study. METHODS: Between May 2005 and December 2013, a total of 52,140 ERCPs were registered in GallRiks, a Swedish nationwide, population-based registry. A total of 376 (0.72%) were registered as perforations or extravasation of contrast during ERCP or as perforation in the 30-day follow-up. The patients with perforation were divided into fatal and non-fatal groups and analyzed for mortality risk factors. The case volume of centers and endoscopists were divided into the upper quartile (Q4) and the lower three quartile (Q1–3) groups. Furthermore, fatal group patients’ records were reviewed. RESULTS: Death within 90 days after ERCP-related perforations or at the index hospitalization occurred in 20% (75 out of 376) for all perforations and 0.1% (75 out of 52,140) for all ERCPs. The independent risk factors for death after perforation were malignancy (OR 11.2, 95% CI 5.8–21.6), age over 80 years (OR 3.8, 95% CI 2.0–7.4), and sphincterotomy in the pancreatic duct (OR 2.8, 95% CI 1.1–7.5). In Q4 centers, the mortality was similar with or without pancreatic duct sphincterotomy (14% vs. 13%, p = 1.0), but in Q1–3 centers mortality was higher (45% vs. 21%, p = 0.024). CONCLUSIONS: ERCP-related perforations are severe adverse events with low incidence (0.7%) and high mortality rate up to 20%. Malignancy, age over 80 years, and sphincterotomy in the pancreatic duct increase the risk to die after a perforation. The risk of a fatal outcome in perforations after pancreatic duct sphincterotomy was reduced when occurred at a Q4-center. In the case of a complicated perforation a transfer to a Q4-center may be considered.