Background: How precise and reliable is ultrasonography as a primary tool for injury assessment in blunt abdominal trauma?
Methods: A systematic review and meta-analysis were conducted of prospective clinical trials of ultrasonography for blunt abdominal trauma. Publications were retrieved by structured searching among databases, review articles and major text books. Authors and experts in the field were contacted for original and unpublished data. For statistical analysis, summary receiver operating characteristic curves (SROCs) were computed using weighted and robust regression models, with Q* denoting the shoulder of the curve. Post-test probabilities were calculated as a function of pooled likelihood ratios (LRs).
Results: Thirty of 123 trials enrolling 9047 patients were eligible for final analysis. With respect to targeting organ lesions, ultrasonography showed a summary Q* value of 0.91 (inverse variance weights, 95 per cent confidence interval (c.i.) 0.76-1.07); negative predictive values ranged from 0.72 to 0.99. A similar SROC slope was calculated for screening for free fluid (Q* = 0.89 (95 per cent c.i. 0.73-1.05)). Ultrasonography detects the presence of organ lesions, but fails to exclude abdominal injuries (random effects negative LR 0.23 (95 per cent c.i. 0.18-0.28)). Given a pretest probability of 50 per cent for blunt abdominal injury, a post-test probability of nearly 25 per cent remains in the case of a negative sonogram.
Conclusion: Despite its high specificity, ultrasonography has an unexpectedly low sensitivity for the detection of both free fluid and organ lesions. In clinically suspected abdominal trauma, another assessment (e.g. helical computed tomography) must be performed regardless of the initial ultrasonographic findings.