Derivation and validation of a predictive model for subtotal cholecystectomy

Introduction

Rates of subtotal cholecystectomy (STC) are increasing in response to challenging cases of laparoscopic cholecystectomy (LC) to avoid bile duct injury, yet are associated with significant morbidity. The present study identifies risk factors for STC and both derives and validates a risk model for STC.

Methods

LC performed for all biliary pathology across three general surgical units were included (2015–2020). Clinicopathological, intraoperative and post-operative details were reported. Backward stepwise multivariable regression was performed to derive the most parsimonious predictive model for STC. Bootstrapping was performed for internal validation and patients were categorised into risk groups.

Results

Overall, 2768 patients underwent LC (median age, 53 years; median ASA, 2; median BMI, 29.7 kg/m²), including 99 cases (3.6%) of STC. Post-operatively following STC, there were bile leaks in 29.3%, collections in 19.2% and retained stones in 10.1% of patients. Post-operative intervention was performed in 29.3%, including ERCP (22.2%), laparoscopy (5.0%) and laparotomy (3.0%). The following variables were positive predictors of STC and were included in the final model: age > 60 years, male sex, diabetes mellitus, acute cholecystitis (AC), increased severity of AC (CRP > 90 mg/L), ≥ 3 biliary admissions, pre-operative ERCP with/without stent, pre-operative cholecystostomy and emergency LC (AUC = 0.84). Low, medium and high-risk groups had a STC rate of 0.8%, 3.9% and 24.5%, respectively.

Discussion

The present study determines the morbidity of STC and identifies high-risk features associated with STC. A risk model for STC is derived and internally validated to help surgeons identify high-risk patients and both improve pre-operative decision-making and patient counselling.