Predictors of In-Hospital Mortality After Percutaneous Cholecystostomy in Acute Cholecystitis

Background Acute cholecystitis is a frequent cause of emergency surgical admission. Percutaneous cholecystostomy (PC) is commonly used as a minimally invasive treatment option in patients considered unsuitable for early cholecystectomy. However, predictors of mortality and the role of radiologic disease severity in determining outcomes remain incompletely understood. Methods We conducted a retrospective cohort study including consecutive adult patients who underwent image-guided PC for acute cholecystitis at a tertiary referral hospital between January 2022 and December 2025. Clinical, laboratory, radiologic, and procedural variables were analyzed. Multivariable logistic regression was used to identify independent predictors of in-hospital mortality. A radiologic severity score (RSS) was constructed using predefined imaging findings including pericholecystic fluid, gallbladder perforation, emphysematous cholecystitis, and gallbladder wall thickness ≥ 7 mm. Associations between RSS and inflammatory response or clinical outcomes were evaluated. Results A total of 266 patients were included (mean age 64.9 ± 17.1 years; 56.4% male). The overall in-hospital mortality rate was 7.5%. Independent predictors of mortality included age (OR 1.07 per year), malignancy (OR 9.19), LDH (OR 1.006), and post-procedural day-3 CRP (OR 1.016). The predictive model demonstrated excellent discrimination with an AUC of 0.914. Internal validation using cross-validation yielded an AUC of 0.895. Higher RSS values were significantly associated with reduced CRP decline following PC (p = 0.041), suggesting slower inflammatory resolution. However, RSS was not significantly associated with mortality or hospital length of stay. Conclusion In patients undergoing PC for acute cholecystitis, mortality is largely determined by baseline patient vulnerability and early inflammatory dynamics. Radiologic severity reflects local inflammatory burden and may complement laboratory markers in predicting inflammatory recovery. Trial Registration Not applicable