Direct versus indirect signs of traumatic aortic injury revealed by helical CT: Performance characteristics and interobserver agreement

OBJECTIVE, The purpose of this study was to evaluate the relative value of and interobserver agreement on direct versus indirect (hematoma) signs of t raumatic aortic injury using helical CT. MATERIALS AND METHODS, From April 1994 through January 1997, 40 patients wh o were suspected to have traumatic aortic injury and who underwent contrast -enhanced helical CT had subsequent proof or exclusion of aortic injury. Al l available CT scans of these patients were combined with CT scans of 13 ra ndomly chosen patients that had been initially interpreted as negative, and clinical follow-up showed no evidence of aortic injury. Two emergency radi ologists and a nonemergency radiologist who were unaware of clinical outcom e performed independent review of these cases to evaluate for mediastinal h ematoma, periaortic hematoma, and direct signs of aortic injury. RESULTS. Direct signs of injury were seen on helical CT by both emergency r adiologists in all 17 cases of aortic injury with no false-positive interpr etations. The nonemergency radiologist failed to observe subtle direct sign s in two cases of aortic injury, but patient management would not have been adversely affected. All observers had more false-negative interpretations for both mediastinal hematoma and periaortic hematoma than for direct signs . Interobserver agreement was higher for direct signs (kappa = .93) than fo r either mediastinal hematoma (kappa = .65) or periaortic hematoma (kappa = .71). CONCLUSION. In this study, helical CT revealed direct signs of traumatic ao rtic injury that were more accurate and more often observed than were indir ect signs. Emphasis on direct signs should improve confidence in using heli cal CT to evaluate traumatic aortic injury.