The accuracy of computed tomography in the diagnosis of blunt small-bowel perforation

Computed tomography (CT) is commonly used to evaluate patients with possible blunt intra-abdominal injury. One of its reported weaknesses is failure to demonstrate intestinal trauma. However, CT accuracy in identifying blunt small-bowel perforation has not been adequately assessed.

We tracked 883 consecutive stable trauma victims who had abdominal CT because of equivocal physical findings. Initial "wet reading" results were compared with laparotomy findings and patient outcome.

Small-bowel perforation occurred in 26 patients (3%). Twenty-four had CT abnormalities suggesting the injury. Twelve had CT findings considered diagnostic: contrast extravasation (n = 5) and/or extraluminal air (n = 11). One additional patient was thought to have free air on CT, but had no intestinal injury at laparotomy. Another 12 patients had CT scans that were non-diagnostic but suggestive: free fluid without solid organ injury (n = 10), or small-bowel thickening (n = 4) or dilatation (n = 3). Two patients with small-bowel injuries had normal CT scans. Of 857 patients without small-bowel disruption, 802 had normal abdominal CT scans, and 55 had 67 CT findings suggesting intestinal injury. Thus, CT diagnosed small-bowel perforation with a sensitivity of 92%, a specificity of 94%, and negative and positive predictive accuracies of 100% and 30%, respectively. The test had an overall accuracy (validity) of 94%.

Blunt small-bowel injury is uncommon. When it is present, abdominal CT is usually abnormal. CT findings in intestinal perforation can be subtle and nonspecific. Any unexplained abnormality on CT after blunt abdominal trauma may signal the presence of intestinal perforation and warrants close clinical observation or further diagnostic tests.