Developand Validate A Clinical Model For Predicting The Risk Of Post-ERCP Pancreatitis

Background: The occurrence of post-ERCP pancreatitis(PEP)has been plagued by endoscopy doctors everywhere since ERCP was reported.There has been no universally accepted standard for judging what kind of patient needs preventive measures or early treatment.Our goal is developing and verify a PEP risk prediction model and validate its applicability in clinical decision-making.Provide help for clinical workMethods: We retrospectively included patients who underwent ERCP surgery at the hepatobiliary surgery department of the Chongqing Medical University Second Affiliated Hospital from August 2018 to August 2020.The reasons for these patients included biliary stones,biliary and pancreatic tumors,and pancreatic duct stones.and other diseases.According to each patient’s previous medical history,basic information and intraoperative operation.We established a clinical model for predicting the incidence of PEP and presented in the form of a nomogram.At the same time,the data of patients in the same hospital’s gastroenterology department was used as a validation sample to externally validate the model.Calculate the degree of discrimination and calibration to evaluate its accuracy and versatility.Finally,use the decision analysis curve to evaluate the net benefit of using this model,and intuitively show its clinical utility.Results: After performing univariate analysis,we found that gender,age,previous pancreatitis,previous cholecystectomy,operation time and multiple pancreatic duct intubation related to the incidence of PEP(p<0.05).And there is no obvious collinearity among the predictors,so these above risk factors are used as predictors to establish a model.In the model we made,the AUC was 0.869(95% confidence interval,0.79-0.94).The Hosmer-Lemeshow test was used to test the goodness of fit in the model we made and the p value was 0.117.The area under the ROC curve calculated by external validation is 0.869(95% confidence interval,0.79-0.94).And then,draw the calibration curve.The calibration intercept is 1.181 and the slope is 1.383.Finally,drawing the DCA curve which found that this model has clinical application value in predicting the incidence of PEP in all patients whose risk threshold is less than 0.78 after ERCP surgery.Conclusion: We have developed a reasonable model to predict the risk of PEP in patients undergoing ERCP surgery.The accuracy of the prediction is acceptable and the clinical benefit is good.However,because the data we used for development and validation are all from the same medical center which lack the consideration of universal versatility in the world,and it still needs multiple verification by subsequent researchers.