Abstract
Background/Objectives: Small bowel obstruction (SBO) is a common surgical emergency associated with significant morbidity and mortality. This retrospective analysis aimed to identify key predictors for the need for surgery in SBO management and to develop a simple clinical risk score to support decision-making. Methods: This retrospective study included 285 patients treated for SBO at the University Hospital Erlangen from 2018 to 2022. Pretherapeutic clinical, laboratory, and imaging data, as well as treatment details and outcome parameters were assessed and analyzed using univariate and multivariate logistic regression to identify significant predictors for the need of surgery. A weighted point-based risk score was then derived from the final model, and its discriminative performance was evaluated using receiver operating characteristic (ROC) analysis. Results: Of the 285 patients, 234 (82.1%) underwent surgery and 51 (17.9%) were successfully managed conservatively. Multivariate analysis identified the following independent predictors for surgery: 0-1 previous abdominal operation (OR 4.7, p = 0.009), serum albumin ≤ 34 g/L (OR 4.5, p = 0.011), free intraperitoneal fluid on imaging (OR 3.6, p = 0.015), air-fluid levels on plain abdominal X-ray (OR 3.5, p = 0.024) and a transition point on CT (OR 11.4, p = 0.002). A weighted score (range 0-6 points) was constructed, assigning 1 point to each of the first four predictors and 2 points to the transition point. The score showed good discrimination for predicting the need for surgery (AUC 0.874). Using a cut-off of ≥3 points, sensitivity was 96.2% and specificity 64.7%. The observed proportion of patients requiring surgery increased from 21.4% in the low-risk group (0-2 points) to 88.6% in the intermediate-risk group (3-4 points) and 97.3% in the high-risk group (5-6 points). Conclusions: The proposed predictors and the weighted risk score may support bedside decision-making in SBO by distinguishing patients who require surgery from those eligible for conservative management, but they require prospective multicenter validation before routine clinical implementation.