Twelve studies were included in the review (n=1,187): two that compared ultrasonography with CT (n=127); six of CT (n=557); and four of ultrasonography (n=503). Study quality was described as moderate overall and methodological study details were often under reported. Inclusion criteria were defined in 50% of studies. The time interval between execution of reference standard and index tests was unclear in nearly all studies. A vague description of the execution of reference standard was given in slightly more than half of the studies. Additionally, in all studies the index test results were incorporated into the reference standard and only one study reported inconclusive test results.
Mean summary sensitivity and specificity estimates of ultrasonography (six studies) and CT (eight studies) were not significantly different. Mean summary sensitivity estimates for ultrasonography were 92% (95% CI 80% to 97%) compared with 94% (95% CI 87% to 97%) for CT, p=0.65. The mean summary specificity estimates for ultrasonography were 90% (95% CI 82% to 95%) compared with 99% (95% CI 90% to 100%) for CT, p=0.07. There was evidence of statistically significant heterogeneity for both sensitivity and specificity. The ROC plot showed inter-study heterogeneity in diagnostic performance. Removal of outliers for CT sensitivity and specificity eradicated statistical heterogeneity, but had no significant influence on summary estimates. Summary likelihood ratios were not significantly different between ultrasonography and CT (LR+ p=0.07, LR- p=0.53).
Sensitivity for the identification of alternative diseases (eight studies) ranged between 33% and 78% for the ultrasonography studies and between 50% and 100% for the CT studies.
One of the head-to-head studies reported higher sensitivities than the other (100% versus 85% for ultrasonography and 98% versus 91% for CT) and higher specificities than the other (100% versus 84% for ultrasonography and 100% versus 77%), but these differences were not significant. One study reported good kappa agreement between ultrasonography and CT findings. Kappa agreement was good for depicting pericolic fat inflammation, bowel wall thickening and pericolic abscesses.