Prediction of post-ERCP related pancreatitis

INTRODUCTION: The etiology of post ERCP pancreatitis is unknown, although numerous associations have been postulated. Identification of specific risk factors may lead to insights in etiology and prevention of this common complications. METHODS: Data on 6,071 consecutive ERCPs were entered into the Duke GI-Trac database. Complications were reported as part of a QA exercise. Candidate predictor variables evaluated were: age, gender, hepatobiliary/ pancreatic (HPB) neoplasm, pancreatogram obtained pancreas divisum, presence of pseudocyst, bile duct stones, sphincterotomy, prior post-ERCP pancreatitis, manometry, bile duct size. All pts with active pancreatitis were excluded. A predictive model was generated utilizing the stepwise logistic regression technique. RESULTS: 201 pts (3%) developed pancreatitis. Univariate analysis showed that cancer, bile duct size, prior ERCP pancreatitis and pseudocyst were all p>.05 and thus not entered into a predictive model. The remaining 7 variables were entered into a stepwise logistic regression model. All independent variables were dichotomous except age which was continuous. All 7 variables remained in the final model. Interactions were >.20. Odds ratios were: manometry 3.48 pancreas divisum 2.89 pancreatogram 2.85 gender (female) 1.79 stones sphincterotomy 2.07 (any vs. none) 0.41 age 0.98 The overall χ² for the model = 38.6 (p<.0002) CONCLUSIONS: 1) We identified 7 risk factors for post ERCP related pancreatitis. 2) The risk decreases with age (ie, a 30 yr old has 2.7 times the risk of an 80 yr old). 3) The presence of bile duct stones appears protective. 4) Identification of preprocedure risk factors may lead to more accurate informed consent. 5) These data can be used to develop algorithms of HBP disorder risk/benefit.