Twenty diagnostic cohort studies (11 prospective; seven retrospective; two unclear) were included in the review (957 patients; median sample size 33). Overall study quality was considered to be moderate (the median study quality was 11 points; range 6 to15 points) and publication bias was detected (p=0.05).
Entire vascular tree (18 studies): Results for the detection of occlusion or stenosis greater than 50% using CT angiography were 95% for sensitivity (95% CI: 92, 97) and 96% for specificity (95% CI: 93, 97). Sub-group analysis found that studies performed on a 16-slice or 64-slice multidetector CT scan were significantly more accurate than those on a 2-slice or 4-slice scan (p=0.002). High levels of heterogeneity were found in this analysis.
Per segment for each anatomic region: For aortoiliac arteries (five studies) the summary estimates were 96% for sensitivity (95% CI: 91, 99) and 98% for specificity (95% CI: 95, 99). For femoropopliteal arteries (five studies) the summary estimates were 97% for sensitivity (95% CI: 95, 99) and 94% for specificity (95% CI: 85, 99). For distal runoff in the tibial arteries (six studies) the summary estimates showed 95% sensitivity (95% CI: 85%, 99%) and 91% for specificity (95% CI: 79%, 97%). One study provided results for the femoral artery up to and including the tibial artery with 98% sensitivity and 96% specificity. High levels of heterogeneity were reported throughout the analyses.
CT angiography correctly diagnosed occlusion in 94% of segments, detected more than 50% stenosis in 87% of segments, and designated a segment without a significant stenosis in 96% of segments. Underestimation of occlusion was found in 6% of segments (5% were detected as more than 50% stenosis). Understaging occurred in 9% and overstaging in 4% of segments, although the text (cited here) and abstract percentages differed in this respect.