Abstract
Background: The 2018 Tokyo Guidelines (TG18) are used to classify the severity of acute cholecystitis (AC) but insufficient to predict the length of hospital stay (LOS). Methods: For patients with AC, clinical factors and computed tomography features, including our proposed grading system of pericholecystic fat stranding were used for predicting an LOS of ≥7 days in the logistic regression models. Results: Our multivariable model showed age ≥ 65 years (OR: 2.56, p < 0.001), C-reactive protein (CRP) ≥ 2 mg/dL (OR: 1.97, p = 0.013), gamma-glutamyltransferase levels (OR: 2.460, p = 0.001), TG18 grade (OR: 2.89 per grade, p < 0.001), and moderate to severe pericholecystic fat stranding (OR: 2.14, p = 0.012) exhibited prolonged LOS ≥ 7 days. Conclusions: We developed a scoring model, including TG18 grades (score of 1-3 per grade), our grading system of fat stranding (score of 1), CRP (score of 1), and gamma-glutamyltransferase (score of 1), and a cutoff of >3 had highest diagnostic performance.