Twenty four studies were included in the review. Four studies directly compared CT and MRI. Ten studies that used MRI (n=628 participants) and 18 studies that used CT (n=1,024 participants) were included in the meta-analyses. Most of the included studies (16 out of 24) were prospective, but only eight of these recruited consecutive series of patients. Only four studies reported blinded interpretation of test results. All except two studies provided a clear description of the test.
For CT, pooled sensitivity was 0.42 (95% CI 0.26 to 0.56) and pooled specificity was 0.82 (95% CI 0.80 to 0.83). For MRI, pooled sensitivity was 0.39 (95% CI 0.22 to 0.56) and the pooled specificity was 0.82 (95% CI 0.79 to 0.83).
For CT, estimated Q* value of was 0.77 (95% CI 0.69 to 0.83) and for MRI estimated Q* value was 0.77 (95% CI 0.73 to 0.80); there was no significant difference in performance between CT and MRI.
Average prevalence of lymph node metastases in studies included in this analysis was 0.17 for CT and 0.3 for MRI. When these numbers were used as pre-test probabilities and combined with pooled estimates of positive and negative likelihood ratios, post-test probabilities of a positive test were 0.31 (95% CI 0.23 to 0.40) for CT and 0.47 (95% CI 0.30 to 0.58) for MRI. Post-test probabilities for a negative test were 0.12 (95% CI 0.10 to 0.16) for CT and 0.23 (95% CI 0.18 to 0.29) for MRI.
Diagnostic accuracy of CT (as indicated by DOR) was higher in the five studies with greater than 50 participants. No other subgroups showed significant differences.