Computed tomographic scanning without oral contrast solution for blunt bowel and mesenteric injuries in abdominal trauma

Association between compliance with methodological standards of diagnostic research and reported test accuracy: meta-analysis of focused assessment of US for trauma

PURPOSE

To study whether compliance with methodological standards affected the reported accuracy of screening ultrasonography (US) for trauma.

MATERIALS AND METHODS

Meta-analysis was conducted of prospective investigations in which US was compared with any diagnostic reference test in patients with suspected abdominal injury. Reports were retrieved from electronic databases without language restrictions; added information was gained with manual search. Two reviewers independently assessed methodological rigor by using 27 items contained in the Standards for Reporting of Diagnostic Accuracy (STARD) checklist and the Quality Assessment of Studies of Diagnostic Accuracy included in Systematic Reviews (QUADAS) instrument. Inconsistencies were resolved by means of consensus. Summary receiver operating characteristics and random-effects meta-regression were used to model the effect of methodological standards and other study features on US accuracy.

RESULTS

A total of 62 trials, which included a total of 18,167 participants, were eligible for meta-analysis. The average proportion of men or boys was 71.7%, the mean age was 30.6 years +/- 10.8 (standard deviation), and the mean injury severity score was 16.7 +/- 8.3. The prevalence of abdominal trauma was 25.1% (95% confidence interval [CI]: 21.1%, 29.1%). Pooled overall sensitivity and specificity of US were 78.9% (95% CI: 74.9%, 82.9%) and 99.2% (95% CI: 99.0%, 99.4%), respectively. Varying end points (hemoperitoneum or organ damage) did not change these results. US accuracy was much lower in children (sensitivity, 57.9%; specificity, 94.3%). Strong heterogeneity was observed in sensitivity, whereas specificity remained constant across trials. There was evidence of publication bias. Initial interobserver agreement with methodological standards ranged from poor (kappa = 0.03, independent verification of US findings) to perfect (kappa = 1.00, sufficiently short interval between US and reference test). By consensus, studies fulfilled a median of 13 methodological criteria (range, five to 20 criteria). In investigations that lacked individual methodological standards, researchers overestimated pooled sensitivity, with predicted differences of 9%-18%. The use of a single reference test, specification of the number of excluded patients, and calculation of CIs independently contributed to predicted sensitivity in a multivariate model. In 16 investigations (1309 subjects), a single reference test was used, which provided a combined sensitivity of 66.0% (95% CI: 56.2%, 75.8%).

CONCLUSION

Bias-adjusted sensitivity of screening US for trauma is low. Adherence to methodological standards included in appraisal instruments like STARD and QUADAS is crucial to obtain valid estimates of test accuracy.

Computed tomographic scanning and selective laparoscopy in the diagnosis of blunt bowel injury: a prospective study

BACKGROUND

We prospectively evaluated whether computed tomographic (CT) scanning and selective laparoscopy (LP) for the diagnosis of blunt bowel injury (BBI) could prevent nontherapeutic laparotomy and delayed diagnosis.

METHODS

Between April 1994 and May 2002, hemodynamically stable patients suspected of having BBI were enrolled in this study. Patients with hemodynamic instability or solid organ injuries with hemoperitoneum were excluded. All patients underwent a physical examination and contrast CT scanning at admission and once again approximately 12 hours (range, 6-24 hours) after admission. LP was performed under general anesthesia in patients who had local peritoneal signs and indirect CT signs (bowel thickening or isolated intraperitoneal fluid) or in whom abdominal pain or tenderness increased or intraperitoneal fluid increased on the repeat CT scan. The indications for a celiotomy were diffuse peritonitis, pneumoperitoneum on the abdominal CT scan, or bowel perforation visible on LP.

RESULTS

During the study period, 399 of 1,074 patients admitted for blunt torso injuries were enrolled in this study. Eleven patients underwent emergency celiotomy and 11 underwent LP immediately after admission to the emergency department. One nontherapeutic laparotomy was performed among the patients who underwent celiotomy. The LPs revealed seven bowel perforations and one mesenteric laceration. After a repeat CT scan, three and seven of the patients underwent laparotomy and LP, respectively. Four bowel perforations were found by LP. The remaining 198 patients were treated conservatively, and no complications related to a delayed BBI diagnosis occurred.

CONCLUSION

CT scanning and selective LP can prevent nontherapeutic laparotomy and delayed diagnosis in patients with suspected BBI.

Computed tomography without oral contrast solution for blunt diaphragmatic injuries in abdominal trauma

OBJECTIVE

The aim of this study was to estimate the sensitivity, specificity, and positive predictive value (PPV) of computed tomography (CT) without oral contrast for diaphragm injuries (DIs) in blunt abdominal trauma.

METHODS

We prospectively enrolled 500 consecutive "trauma-one" patients who received CT imaging and interpretation (CT-Read1) of the abdomen within 45 minutes of their arrival from July 2000 to December 2001. All patients were imaged without oral contrast but with intravenous contrast. Computed tomographic images were reviewed within 24 hours of admission by research radiologists (CT-Read2) blinded to CT-Read1. True DIs were determined hierarchically by either laparotomy or autopsy.

RESULTS

There were 9 patients with laparotomy or autopsy-proven blunt DIs; 8 of these injuries involved the left hemidiaphragm. The CT-Read1 correctly detected only 6 of 9 blunt DIs, thus missing 3 DIs. One of these involved the right hemidiaphragm, whereas the other 2 were left sided. There were no false-positive findings with CT-Read1 for blunt DI. The sensitivity and specificity of CT imaging with respect to DI were 66.7% (95% CI, 29.9%-92.5%) and 100% (95% CI, 99.2%-100%), respectively. The PPV for the test was 1.00 (95% CI, 0.65-1.00).

CONCLUSION

Although the low number of blunt DIs in this study limits its general applicability, CT imaging of the diaphragm without oral contrast appears to perform within the range of reported imaging techniques using oral contrast. Still, CT scanning appears to have an unsatisfactorily low sensitivity to be reliably used in eliminating the diagnosis of blunt DI.