| Objective:To investigate the risk factors for delayed abdominal bleeding after pancreaticoduodenectomy(PD)and to establish a risk prediction model,which can help identify individuals at high risk for delayed abdominal bleeding after PD at an early stage,and to target individualized interventions to reduce the incidence of bleeding and effectively improve patient prognosis.Methods:The clinical data of 216 patients with PD admitted to the First Affiliated Hospital of Xinjiang Medical University from 2019-01 to2022-06 were retrospectively collected.Patients were divided into bleeding and non-bleeding groups according to the presence or absence of delayed abdominal bleeding,and the patients’preoperative data,intraoperative data,and postoperative data were analyzed univariately and multifactorially using SPSS software to derive independent risk factors for postoperative delayed abdominal bleeding in PD,and risk prediction models were established and evaluated using each independent risk factor,and finally Nomogram column line plots were drawn using R software and Bootstrap method was used for internal testing.The effectiveness of the line graph model was evaluated by the area under the curve(AUC)and the receiver operating characteristic(ROC)curve was also plotted.Results:A total of 216 patients were included in the study,26 in the bleeding group and 190 in the non-bleeding group.Univariate and multifactorial analyses showed that history of alcohol consumption,preoperative elevated total bilirubin,early postoperative bleeding,postoperative pancreatic fistula,postoperative abdominal infection,and elevated leukocyte count and C-reactive protein in the 5th/6th postoperative d were independent risk factors for late postoperative bleeding in PD.The above independent risk factors were used as predictors to establish a risk prediction model,and the Hosmer-Lemeshow goodness-of-fit test was used to evaluate the prediction model,and the test results showed that x~2=2.553,P=0.959>0.05.The ROC curve of the prediction model was further plotted,and the AUC was calculated as 0.888,P<0.001.R software was used to plot the Nomogram column line plot and performed internal validation,the mean absolute error was 2.1%,and the calibration curve and the ideal curve basically overlapped.Conclusions:History of alcohol consumption,elevated preoperative total bilirubin(>119umol/L),early postoperative bleeding,postoperative pancreatic fistula,postoperative abdominal infection,elevated leukocyte count(>12.12×109)at the 5th/6th d postoperatively,and elevated C-reactive protein(>90mg/L)at the 5th/6th d postoperatively were independent risk factors for delayed abdominal bleeding after PD.The risk prediction model of delayed abdominal bleeding based on the results of multifactorial logistic regression analysis can provide some reference value in clinical practice. |
