Risk Factors Analysis And Prediction Model Establishment Of Endoscopic Retrograde Cholangiopancreatography-related Immediate And Delayed Hemorrhage

Objective Endoscopic retrograde cholangiopancreatography(ERCP)has become one of the most commonly used methods for the diagnosis and treatment of pancreaticobiliary diseases and hemorrhage is one of its common complications.ERCP-related hemorrhage can be classified as immediate and delayed hemorrhage depending on the time of occurrence.The aim of this research was to explore the risk factors of ERCP related immediate and delayed hemorrhage respectively,propose corresponding preventive measures,and establish models for predicting hemorrhage.Methods Clinical data of patients who underwent ERCP in Renmin Hospital of Wuhan University from January 2017 to December 2020 were retrospectively collected,including patient related factors,such as gender,age,secondary nipple,diverticulum,indications(choledocholithiasis,cholangitis,bile duct stricture,jaundice and pancreatic diseases),background diseases(biliary and pancreatic tumors,gallstones or cholecystitis,hypertension,diabetes mellitus,coronary heart disease,renal failure and cirrhosis);operation(endoscopic sphincterotomy,endoscopic papillary balloon dilatation,pre-cut,pancreatic duct intubation,placement of the pancreatic duct stent,placement of the biliary stent and nasobiliary drainage);past history(cholecystectomy,subtotal gastrectomy and history of antiplatelet or anticoagulant medication);laboratory test data and other relevant data.IBM SPSS Statistics 25.0 was used for statistical analysis of the data,qualitative data were analyzed by chi-square test or Fisher exact probability method,and quantitative data were analyzed by Mann-Whitney U test,the factors with statistical significance(P<0.05)in univariate analysis were included in the binary logistic regression analysis to obtain independent risk factors for ERCP related immediate and delayed hemorrhage.And independent risk factors were used to establish preliminary models to predict immediate and delayed hemorrhage,in addition,the predictive ability of each model was tested.Results A total of 1020 ERCP cases were included in the research,including 85 cases in the immediate hemorrhage group,30 cases in the delayed hemorrhage group,and 905 cases in the non-hemorrhage group.The mean age of the patients was 61.33(20-100)years old,and there were 563 males(55.2%)and 457 females(44.8%).Most of the patients in the immediate hemorrhage group were mild bleeding,while in the delayed hemorrhage group,there were 7 cases(23.3%)of mild bleeding,7 cases(23.3%)of moderate bleeding,and 16 cases(53.3%)of severe bleeding.The most common symptom of delayed hemorrhage was melena,which occurred in 20 patients(66.7%),in addition,5 patients(16.7%)had postoperative nasobiliary drainage of hemorrhagic fluid,and 5 patients(16.7%)had postoperative hematemesis.Univariate analysis showed that choledocholithiasis(P=0.028),pre-cut(P=0.002),endoscopic sphincterotomy(P=0.033),endoscopic papillary balloon dilatation(P=0.023),endoscopic sphincterotomy with endoscopic papillary balloon dilatation(P < 0.001),and history of antiplatelet or anticoagulant medication(P=0.006)were risk factors for ERCP-related immediate hemorrhage,while cholangitis(P=0.001),jaundice(P=0.002),coronary heart disease(P=0.003),pre-cut(P=0.004),history of antiplatelet or anticoagulant medication(P=0.006),and some laboratory tests,such as high postoperative 24 h amylase(P=0.008),high postoperative 4 h lipase(P=0.011),high postoperative 24 h lipase(P=0.035),low preoperative platelet counts(P=0.044),low postoperative platelet counts(P=0.004),low postoperative hemoglobin(P=0.032),high aspartate transaminase(P=0.035),high total bilirubin(P=0.014),high direct bilirubin(P=0.016),high postoperative urea(P=0.044),low fibrinogen(P=0.007),high prothrombin time(P=0.034)and thrombin time(P=0.003)were risk factors for ERCP-related delayed hemorrhage.Binary logistic regression analysis showed that pre-cut(P=0.001,OR 2.658,95% CI 1.458-4.849),history of antiplatelet or anticoagulant medication(P=0.004,OR 3.757,95% CI 1.517-9.305)and choledocholithiasis(P=0.032,OR 1.782,95% CI 1.052-3.019)were independent risk factors for ERCP-related immediate hemorrhage,while coronary heart disease(P <0.001,OR 10.872,95% CI 3.004-39.351),high postoperative 24 h amylase(P=0.017,OR 4.582,95% CI 1.306-16.007),and low postoperative platelet counts(P=0.007,OR 5.559,95% CI 1.604-19.268)were independent risk factors for ERCP-related delayed hemorrhage.The results showed that the combination of independent risk factors(indicators)with the highest sensitivity,specificity,positive predictive value(PPV)and negative predictive value(NPV)for ERCP-related immediate hemorrhage were pre-cut,choledocholithiasis,history of antiplatelet or anticoagulant medication(88.2%),pre-cut and history of antiplatelet or anticoagulant medication(99.9%),pre-cut and history of antiplatelet or anticoagulant medication(50.0%),pre-cut,history of antiplatelet or anticoagulant medication or common bile duct stones(96.6%).And the factors with highest sensitivity,specificity,PPV and NPV for ERCP-related delayed hemorrhage were coronary heart disease,high amylase 24 h after ERCP or low postoperative platelet counts(88.9%),coronary heart disease,high amylase 24 h after ERCP and low postoperative platelet counts(99.4%),high amylase 24 h after ERCP and low postoperative platelet counts(14.3%),coronary heart disease,high amylase24 h after ERCP or low postoperative platelet counts(99.4%).Conclusions The factors affecting ERCP-related immediate and delayed hemorrhage have a big difference,pre-cut and the history of antiplatelet or anticoagulant medication are common risk factors for ERCP related immediate and delayed hemorrhage,while other risk factors are different,which should be paid attention to separately to reduce the risk of hemorrhage.Prediction models with different combinations of independent risk factors have different advantages in predicting hemorrhage,therefore,it is necessary to comprehensively analyze the corresponding dominant indicators in order to better guide clinical practice.In general,the sensitivity,specificity and NPV of these indicators are relatively high,while the PPV is insufficient.This study provides a basis for the prediction of ERCP related immediate and delayed hemorrhage through risk factors.