Multiple logistic regression model to predict bile leak associated with subtotal cholecystectomy

Area of Science:

• Gastroenterology
• Surgical Oncology
• Hepatobiliary Surgery

Background:

• Bile leakage is a known complication following subtotal cholecystectomy (STC).
• Currently, no validated prediction models exist to identify patients at high risk for post-STC bile leakage.
• This study addresses the need for a predictive tool to manage STC outcomes.

Purpose of the Study:

• To develop and evaluate a multivariable prediction model for bile leakage after subtotal cholecystectomy (STC).
• To identify independent risk factors associated with post-STC bile leakage.

Main Methods:

• Analysis of prospectively collected data from 81 patients who underwent STC by a single surgeon.
• Classification of STC into open-tract and closed-tract variants.
• Application of multivariable logistic regression to identify predictors and generate a prediction model, with performance assessed by discrimination and calibration statistics.

Main Results:

• Twenty-eight patients (35%) experienced bile leakage post-STC, with 18 (64%) requiring surgical intervention.
• Independent predictors for bile leak were open-tract STC (OR, 7.07) and acute cholecystitis (OR, 5.449).
• The model demonstrated good predictive accuracy with an AUC of 82.11%.

Conclusions:

• Open-tract STC and acute cholecystitis are significant predictors of bile leakage following STC.
• The developed model shows promise in predicting post-STC bile leaks.
• Further multicenter studies are recommended for external validation and refinement of prediction models.