Radiology: Volume 249: Number 2—November 2008
Mostafa Atri, MD, FRCPC John M. Hanson, MBBCh Lenny Grinblat, MD Nicole Brofman, MD Talat Chughtai, MD George Tomlinson, PhD
Purpose:To determine the accuracy of multidetector computed tomography (CT) in the detection of surgically important blunt bowel and/or mesenteric injury, to identify and describe the most reliable CT features of bowel and/or mesenteric injury, and to evaluate the performance of readers with different levels of expertise.
Methods: Institutional review board approval was obtained for this retrospective case-control study of 96 subjects with lapa-rotomy-confirmed findings: 54 consecutive patients with bowel and/or mesenteric injury (surgically important and unimportant) (32 male patients, 22 female patients; mean age, 40.4 years ± 17.6 [standard deviation]; range, 16-86 years) and 42 matched patients without bowel and/or mesenteric injury (22 male patients, 20 female patients; mean age, 36.8 years ± 20.1; range, 14-84 years) who underwent four-detector CT prior to surgery. A second-year radiology resident, an abdominal imaging fellow, and a staff abdominal radiologist, blinded to patient outcome, independently reviewed CT studies and recorded the probability of bowel and/or mesenteric injury on a five-point scale. Sensitivity and specificity were calculated for each reviewer, and areas under the receiver operating characteristic curve (AUCs) were compared. Thirty-eight (40%) of 96 patients had surgically important bowel and/or mesenteric injury, and 58 (60%) of 96 patients had either no or surgically unimportant bowel and/or mesenteric injury. Sensitivities of the three reviewers in the diagnosis of surgically important bowel and/or mesenteric injury ranged from 87% (33 of 38) to 95% (36 of 38); specificities ranged from 48% (28 of 58) to 84% (49 of 58).
Results: The only significantly better AUC belonged to the staff radiologist for surgically important mesenteric injury (P = .01). Bowel wall defect, extraluminal contrast material, thick large bowel, mesenteric vessel beading, abrupt termination of mesenteric vessels, and mesenteric vessel extravasation showed the best positive likelihood ratios for surgically important bowel and/or mesenteric injury; absence of peritoneal fluid showed the best negative likelihood ratio.
Conclusion: Multidetector CT findings accurately reveal surgically important bowel and/or mesenteric injury and have a high negative predictive value.