Thirty-one studies (2,092 patients, 2,368 vessels) were included.
All values for ranges in sensitivity and specificity were read from SROC plots and so are approximate.
For QCA compared with FFR, threshold 0.75 (18 studies), there was substantial heterogeneity. The sensitivity ranged from 30 to 100% and the specificity from 0 to 90%. The pooled sensitivity and specificity were 78% (95% CI: 67, 86) and 51% (95% CI: 40, 61), respectively.
For QCA compared with FFR, threshold 0.80 (16 studies), there was substantial heterogeneity. The sensitivity ranged from 30 to 100% and the specificity from 0 to 88%. The pooled sensitivity and specificity were 76% (95% CI: 64, 85) and 54% (95% CI: 40, 68), respectively.
For FFR (threshold 0.75) compared with noninvasive imaging (17 studies), there was substantial heterogeneity although less than for the comparisons of QCA with FFR. The sensitivity ranged from 40 to 100% and the specificity from 36 to 100%. The pooled sensitivity and specificity were 76% (95% CI: 69, 82) and 76% (95% CI: 71, 81), respectively.
Studies that compared FFR with SPECT (15 studies) showed a pooled sensitivity of 75% (95% CI: 66, 82) and pooled specificity of 77% (95% CI: 70, 83). Concordance was greater for studies that compared FFR with dobutamine echocardiography (6 studies): the pooled sensitivity and specificity were 82% (95% CI: 62, 92) and 74% (95% CI: 66, 81), respectively.
The exclusion of three studies that enrolled patients exclusively or predominantly following myocardial infarction, two studies that included patients with restenosis, or one study that measured hyperemia differently to the other studies, did not alter the results.