Prospective, observational validation of a multivariate small-bowel obstruction model to predict the need for operative intervention

Abstract

Background: We published previously a model predictive of the need for exploration in small-bowel obstruction. We aimed to validate and refine the model, hypothesizing that the model would be predictive, would prevent delayed management of strangulation, and would be successfully improved.

Study design: Data from 100 consecutive patients with small-bowel obstruction and concurrent CT were collected prospectively. New features evaluated included obstipation and the absence of colonic gas on CT.

Results: Overall mortality was 8%. Twenty-nine patients had all 4 clinical features, 22 of whom required operative exploration (concordance index = 0.75), confirming the validity of the old model. Intraperitoneal free fluid (odds ratio [OR]: 2.6, 95% CI: 1.0 to 6.9) and vomiting (OR: 1.5, 95% CI: 0.5 to 4.5) were not predictive of operative exploration; however, mesenteric edema (OR: 4.2, 95% CI: 1.1 to 15.8) and lack of the small-bowel feces sign were (OR: 3.5, 95% CI: 1.4 to 8.8). Obstipation was associated with the need for exploration (OR: 2.8, 95% CI: 1.2 to 6.6), but absence of colonic gas was not. A new model was equally predictive of the need for exploration: mesenteric edema (OR: 5.6, 95% CI: 1.5 to 20.7), lack of the small-bowel feces sign (OR: 5.1, 95% CI: 1.9 to 13.6), and obstipation (OR: 3.2, 95% CI: 1.2 to 8.3). The concordance index for this new model was 0.77.

Conclusions: Our current prospective study validated our original model and was successfully improved. Our new model demonstrated equivalent predictive ability and was simpler to use. When all 3 features of the new model are present, strong consideration for early operative exploration should be entertained and may decrease the rate of missed strangulation obstructions.