A Retrospective Multicenter Study Examining Clinical and Imaging Predictors of Emergent Surgery in Small Bowel...

Operative management for small bowel obstruction (SBO) is indicated for patients with concerns about bowel ischemia or failure of conservative therapy; however, most patients respond to nonoperative management.

Our primary objective was to identify clinical and imaging predictors of operative management in SBO, stratified by the timing of intervention, with an emphasis on emergent operative management within 24 h of presentation. Secondarily, we sought to explore which patient populations may be candidates for future prospective evaluation of computed tomography (CT)-avoidance pathways, generating hypotheses regarding patients who may safely undergo an initial trial of nonoperative management without emergent CT imaging.

This was a multicenter retrospective review of patients with SBO. We performed univariate and multivariate explanatory modeling of clinical and CT imaging variables to calculate odds ratios for operative management. We generated receiver operating characteristic curves to compare the performance of clinical and imaging variables in predicting emergent surgical intervention.

A total of 4478 patients with SBO were included, with 10.3% (n = 463) undergoing emergent surgical intervention within 24 h of presentation. The CT variables (odds ratio; 95% confidence interval) most predictive of emergent operative management were closed-loop obstruction (9.64; 6.13-15.2), internal hernia (7.72; 5.77-10.3), and mesenteric swirl (6.91; 5.12-9.29). Among non-CT clinical variables, rebound tenderness (6.13; 4.04-9.21), tap/shake tenderness (5.44; 3.89-7.54), and abdominal guarding (4.28; 3.29-5.54) increased the likelihood of emergent surgery. In contrast, a history of prior SBO (0.29; 0.23-0.36) and prior abdominal surgery (0.42; 0.33-0.53) decreased the likelihood of emergent surgery. Receiver operating characteristic curves demonstrated that a model using clinical features alone was outperformed by a model including clinical and imaging variables in predicting emergent operative management, although with similar performance when optimizing for higher sensitivity.

Clinical features alone may be predictive in identifying a subset of patients with SBO who require emergent operative management. Future research is needed to refine these predictive models and develop scoring systems that incorporate other diagnostic modalities, such as ultrasound, to both diagnose SBO and identify low-risk patient populations. Patients identified as low risk based on clinical variables alone may then be able to avoid initial CT imaging and proceed directly to a trial of nonoperative management. © 2025 Elsevier Inc. All rights are reserved, including those for text and data mining, AI training, and similar technologies.