Value of early repeated abdominal CT in selective non-operative management for blunt bowel and mesenteric injury

• DOI: 10.1007/s00330-019-06212-w
• Publication Date: 2019-04-25T08:04:47Z
• Authors: F. Lannes; U. Scemama; A. Maignan; L. Boyer; L. Beyer-Berjot; S. V. Berdah; K. Chaumoître; M. Leone; T. Bège
• Keywords: Neuroradiology; Interventional radiology; McNemar's test

To evaluate the performance of an early repeated computed tomography (rCT) in initially non-operated patients with blunt bowel and mesenteric injuries (BBMI).This was a monocentric retrospective observational study from 2009 to 2017 of patients with a BBMI on initial CT (iCT). Patients initially non-operated on were scheduled for a rCT within 48 h. Initial CT and rCT diagnostic performance were compared based on a surgical injury prediction score previously described. For statistical analysis, we used the chi-square analyses for paired data (McNemar test).Eighty-four patients (1.9% of trauma) had suspected BBMI on iCT. Among these patients, 22 (26.2%) were initially operated on, 18 (21.4%) were later operated on, and 44 (52.4%) were not operated on. The therapeutic laparotomy rate was 85%. Thirty-four patients initially non-operated on had a rCT. The absolute value of the CT scan score increased for 15 patients (44.1%). The early rCT diagnostic performance, compared with iCT, showed an increase in sensitivity (from 63.6 to 91.7%), in negative predictive value (from 77.4 to 94.7%), and in AUC (from 0.77 to 0.94).In initially non-operated patients with BBMI lesions, the performance of an early rCT improved the sensitivity of lesion detection requiring surgical repair and the security of patient selection for non-operative treatment.• Selective non-operative treatment for hemodynamically stable patients with blunt bowel and/or mesenteric injuries on CT is developing but remains controversial. • An early repeated CT improved the sensitivity of lesion detection requiring surgical repair and the security of patient selection for conservative treatment.