Predicting early biliary infection after stenting of malignant biliary obstruction: model development and internal validation.

ABSTRACT

PURPOSE: To analyze the risk factors and develop a clinical prediction model for early biliary infection (EBI) after percutaneous transhepatic biliary stenting (PTBS) in patients with malignant biliary obstruction (MBO). METHODS: The clinical data of 236 patients with MBO treated with PTBS from June 2012 to June 2021 were retrospectively analyzed. Independent risk factors were analyzed by univariate and multivariate logistic regression, and a nomogram model was constructed based on the results. Discrimination, calibration, and clinical usefulness of this model were further assessed. RESULTS: The technical success rate of PTBS was 100%, and EBI after PTBS was 20.3%. Multivariate logistic regression analysis showed that hilar MBO (P = 0.020), diabetes (P = 0.001), previous surgical or endoscopic intervention (P = 0.007), procedure time > 60 min (P = 0.007), and intraprocedural biliary hemorrhage (P = 0.003) were independent risk factors for EBI after PTBS. A nomogram model was developed to predict the probability of EBI. ROC curves showed good discrimination of the model (area under curve = 0.831). The calibration plot indicated that the predicted probability of EBI by this model was in good agreement with the actual probability of EBI. The DCA curves showed that the net benefit of nomogram-assisted decisions was higher than or equal to the net benefit of treatment for all or none at a wide threshold probability (0-0.8). CONCLUSION: The nomogram model based on the above independent risk factors can predict the probability of EBI and model-assisted treatment decisions contribute to improved clinical outcome. Therefore, MBO patients with probability of EBI > 0.20 based on the model should be recommended for perioperative broad-spectrum antibiotics and close monitoring.