Combining clinical and radiological features improves prediction of bowel ischemia in patients with CT findings of pneumatosis intestinalis

Read the full article See related articles

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Background: Pneumatosis intestinalis on CT presents a diagnostic dilemma, because it could reflect bowel ischemia or benign finding. Purpose: To determine radiological and clinical features that can predict bowel ischemia in patients with pneumatosis intestinalis on CT. Materials and Methods: Patients with “pneumatosis” in abdominal CT reports performed between 1/1/2002 and 12/31/2018 were retrospectively included. Pneumatosis intestinalis was confirmed by review of images. Radiological features of pneumatosis, laboratory data, clinical signs and symptoms were collected. Pathologic pneumatosis intestinalis (PPI) was defined as presence of ischemic (viable or dead) bowel on surgery or death during admission or within 30 days of discharge due to ischemia. Univariate statistical analysis was used to identify features associated with PPI, followed by multivariate logistic regression models. Results: A total of 313 consecutive patients with pneumatosis intestinalis (162 (52%) men, median age 67 years, IQR 55 – 78 years) were included. Pathologic pneumatosis intestinalis was present in 114/313 (36%) patients. Presence of arterial or venous thrombosis, porto-mesenteric gas, fat stranding, and location in the small bowel were significantly associated with PPI. A combined clinical and radiological model, which included age, WBC, creatinine, abdominal distention, rebound or guarding, shock, presence of porto-mesenteric gas and fat stranding showed an AUC of 0.85 for prediction of PPI, higher than models using clinical (AUC=0.80, p=0.005) or radiological factors (AUC=0.80, p<0.0001) alone. Conclusion: Improved prediction of pathological pneumatosis intestinalis can be achieved by a model incorporating both clinical and radiological features (AUC=0.85)rather than by either clinical (AUC=0.80) or radiological (AUC=0.80) features alone.

Article activity feed