Construction of a predictive model for postoperative biliary tract infection in patients with obstructive jaundice based on clinical characteristics and biliary drainage methods

OBJECTIVE To investigate the risk factors for biliary tract infection (BTI) after biliary drainage in patients with obstructive jaundice (OJ) based on clinical characteristics and biliary drainage methods, and to construct and validate a nomogram model for risk prediction.

METHODS A total of 168 patients with OJ treated in the Hepatobiliary Surgery Department of Ningbo Municipal Hospital of TCM from Jan. 2020 to Jan. 2025 were included as study subjects. They were divided into a non-infected group (n=97) and an infected group (n=71) based on the occurrence of BTI after biliary drainage. Lasso regression and logistic regression analysis were used to explore the risk factors for BTI after surgery in patients with OJ. The "rms" package in R programming language version 4.2.6 was used to construct the nomogram model, and decision curves, calibration curves and receiver operating characteristic (ROC) curves were plotted to evaluate the model's effectiveness, calibration and discrimination.

RESULTS Diabetes history (OR=3.314), drainage method (percutaneous transhepatic cholangial drainage, OR=2.666), preoperative jaundice duration (OR=2.920), alkaline phosphatase (OR=1.064), total bilirubin (OR=1.028), NLR (OR=1.110) and high biliary obstruction (OR=4.770) were identified as risk factors for BTI after surgery in patients with OJ (P < 0.05). The importance of these risk factors, ranked by the feature_importances_ function of the logistic regression model, was as follows: high biliary obstruction, drainage method, preoperative jaundice duration, diabetes history, NLR, total bilirubin and alkaline phosphatase. Based on these seven risk factors, a nomogram model for predicting BTI risk after surgery in patients with OJ was constructed. The area under the ROC curve (AUC) of the model was 0.851, with a concordance index of 0.849. The decision curve suggested that the model had good predictive performance within a probability threshold range of 0.050 to 0.920. The calibration curve of the model closely aligned with the ideal curve.

CONCLUSIONS The risk factors for BTI after surgery in patients with OJ include diabetes history, drainage method, preoperative jaundice duration, alkaline phosphatase, total bilirubin, NLR and high biliary obstruction. The constructed nomogram model for predicting BTI after surgery in patients with OJ demonstrates good predictive performance.