Background and study aims: Determinants of complications after endoscopic retrograde cholangiopancreatography (ERCP) have not yet been completely characterized.
Patients and methods: Data were collected from an endoscopic database. Univariate analysis and multivariate logistic regression analysis were used to generate the best model of independent predictors of post-ERCP pancreatitis.
Results: The database included 1239 ERCP examinations carried out to investigate suspected choledocholithiasis over a five-year period. From these, 45 patients who developed post-ERCP complications were compared to a random sample of 486 patients who had undergone an uncomplicated ERCP for suspected choledocholithiasis. Univariate analysis demonstrated significant differences between the two patient groups for the following factors: age, using a cut-off point of 59 years (27% vs. 51%, P = 0.002), pancreatic channel opacification (73% vs. 58%, P = 0.05), and absence of common bile duct stones (41% vs. 24%, P = 0.03). Using multivariate logistic regression, the best model for predicting post-ERCP pancreatitis in patients undergoing sphincterotomy included age under 59 years (P = 0.04), and absence of a common bile duct stone (P = 0.004). The model yielded probabilities of developing post-sphincterotomy pancreatitis that ranged from 2.8% if no predictor was present, to 27% when both predictors were present. Among patients in whom a sphincterotomy was not performed, the only significant independent predictor found was pancreatic channel opacification (P = 0.05).
Conclusion: Age under 59 years, pancreatic channel opacification, and an absence of common bile duct stones at ERCP are all independent predictors of post-ERCP pancreatitis.