Accuracy of computed tomography (CT) scan in the detection of penetrating diaphragm injury.

BACKGROUND: The use of computed tomography (CT) to identify injury after penetrating torso trauma has become routine in the hemodynamically stable patient. The diaphragm has been a historically difficult structure to evaluate, however, and missed injuries to the diaphragm may result in significant morbidity. With the increasing use of multidetector row CT (MDCT), we hypothesized that CT would be an accurate detection modality to identify patients with diaphragm injuries. METHODS: We retrospectively reviewed the admission CT of consecutive patients admitted for penetrating injury to the torso during a 4-year period. The CT scans were reviewed and classified into three categories: positive (P), negative (N), or equivocal (Eq). Data from the medical records of these patients were abstracted to identify demographics, injury-specific data, length of stay, length of follow-up (LOFU), and operative findings. RESULTS: There were 803 patients who met inclusion criteria. Mechanism of injury was gunshot wound in 36% and stab wound in 64%. Mean length of stay was 4 days (+/-6.6) and mean length of follow-up was 43 days (+/-184). CT was read as P in 57, N in 710, and Eq in 36 patients. Diaphragm injury was detected in 67 patients overall and was excluded in 736. For the entire study population, sensitivity and specificity were calculated as 94.0% (95% CI = 88.4-99.7) and 95.9% (94.5-97.4) with an overall accuracy of 95.8% (94.4-97.2) if the CT scan was used to exclude diaphragm injury ([P and Eq] vs. N). Sensitivity and specificity were 82.1% (72.9-91.3) and 99.7% (99.4-100) if CT was used to detect diaphragm injury (P vs. [N and Eq]). One hundred and forty-eight patients underwent operative procedures in which the diaphragm was evaluated. Diaphragm injury was identified in 50 (38 P, 4 N, 8 Eq) and was surgically excluded in 104 patients (2 P, 93 N, 9 Eq). Three hundred and eighty-four patients were lost to follow-up; including 348 who had negative finding on CT. There were no known missed diaphragm injuries during the study period or in follow-up. CONCLUSIONS: Injuries to the diaphragm occur commonly after penetrating torso trauma. MDCT scan is an accurate test to detect diaphragm injury. When MDCT is equivocal, further investigation is required to evaluate the diaphragm.