| Background and objective: Endoscopic retrograde cholangiopancreatography(ERCP)has become a common minimally invasive technique in clinical diagnosis and treatment of cholangiopancreatic diseases through continuous improvement since it was launched in 1968.However,with the promotion of ERCP,related postoperative complications have been gradually found,in which post-ERCP pancreatitis(PEP)is the most common complication after ERCP,which has a significant impact on the rehabilitation of patients.In order to reduce the incidence of PEP,domestic and foreign experts continue to explore its pathogenesis and influencing factors.Current studies have shown that risk factors including sphincter of Oddi sphincter dysfunction(SOD),difficult intubation,guide wire into the pancreatic duct and other risk factors are closely related to the occurrence of PEP,which has potential predictive value.At the same time,it was also found that prophylactic pancreatic duct stent implantation and rectal use of non-steroidal antiinflammatory drugs could reduce the incidence of PEP.Therefore,it is of great significance to detect early high-risk patients with PEP and take preventive treatment.However,there is no unified standard to classify the risk grade of patients.The purpose of this study is to explore the risk factors related to PEP,and on this basis,to build a risk prediction model of the probability of postoperative pancreatitis,in order to guide clinicians to take timely preventive measures to alleviate the pain of patients.Methods: From September 2019 to December 2020,all patients who received ERCP in our hospital were selected to establish a model group.From September 2021 to December 2021,the patients with ERCP were collected in our hospital,and the verification group was established.Collect general data,clinical data and operative details,and observe whether patients develop pancreatitis after operation.The cases were screened according to the inclusion criteria and exclusion criteria,and all the included cases in the model group were randomly divided into training group and test group according to the proportion of2:1.In the training group,the included factors were analyzed by single factor logistic regression analysis,and the risk factors were extracted.Then the independent risk factors and their risk ratio(odds ratio,OR)and 95% confidence interval(CI)of PEP were obtained by multi-factor logistic regression analysis,and the risk prediction Nomogram was established according to the regression coefficient of logistic regression analysis.The effectiveness of the model was verified internally and externally in the training group,test group and verification group,and the model was evaluated from three aspects: discrimination,calibration and clinical benefit by using Receiver operating characteristic curve(ROC),calibration curve and Decision Curve Analysis(DCA).Results: A total of 558 cases were included in the model group and 139 cases were included in the verification group.In the model group,there were 35 cases in PEP group and 523 cases in non-PEP group.The incidence of PEP was 6.3%.The average age of PEP group was 62.94 ±14.35 years,Body Mass Index(BMI)was 27.82±2.85,and that of non-PEP group was 65.08 ±13.64 years old,BMI was 24.04±3.65.There were significant differences between the two groups in terms of BMI(P < 0.001),previous history of pancreatitis(P =0.015),bile duct stricture(P =0.004),double guide wire assisted intubation(P=0.002),preincision of papillary sphincter(P = 0.007),balloon dilatation of bile duct sphincter(P = 0.002),guide wire inserted into pancreatic duct(P < 0.001)and development of pancreatic duct(P = 0.003).The model group was randomly divided into two groups according to the proportion of 2:1.There were 372 cases in the training group and 186 cases in the test group.The comparison of baseline data between groups had no statistical significance(P > 0.05),and the balance between groups was good.Univariate logistic regression analysis showed that there were seven risk factors: BMI(P < 0.001),bile duct stricture(P =0.008),difficult intubation(P =0.039),double guide wire assisted intubation(P = 0.001),bile duct sphincter balloon dilatation(P = 0.042),guide wire inserted into pancreatic duct(P =0.004)and development of pancreatic duct(P < 0.001).Multivariate regression analysis showed that five independent risk factors associated with PEP were BMI(OR,1.266;95%CI,1.128-1.421;P < 0.001),bile duct stricture(OR,3.088;95%CI,1.071-8.906,P=0.037),balloon dilatation of bile duct sphincter(OR,3.110;95%CI,1.049-9.224;P=0.041),and guide wire inserted into pancreatic duct(OR,3.290;95%CI,1.063-10.179;P=0.039),development of pancreatic duct(OR,5.724;95%CI,1.794-18.264;P=0.003).According to the regression coefficient of each risk factor,the prediction model was established and displayed with line nomogram.In the training group,the area under the ROC curve(AUC)was 0.848,the sensitivity was 0.809,the specificity was 0.829,and the accuracy was 0.827.In the test group,the area under the ROC curve was0.881,the sensitivity was 0.720,the specificity was 0.928,and the accuracy was 0.734.In the verification group,AUC,sensitivity,specificity and accuracy were 0.933,1.000,0.860,0.871,respectively.It shows that the model has a good discrimination to the occurrence of PEP.The calibration curve range of the corrected model fluctuates near the ideal curve,indicating that the model has a good accuracy in the actual prediction,but the accuracy in the external verification is not as good as the internal verification.The Decision Curve Analysis(DCA)curves of the three groups are all above the reference line,and the net return is positive,indicating that the model has potential clinical value.Conclusion: The risk prediction model established in this study can accurately predict the occurrence of PEP in patients at the early stage.BMI,bile duct stricture,biliary sphincter balloon dilatation,guide wire inserted into pancreatic duct and development of pancreatic duct were predictive factors for the occurrence of PEP.The net benefit can be obtained by taking preventive measures to patients according to the risk prediction model,and the model has potential clinical practical value. |
PDF Full Text Request