Forty five studies were included in the review, of which 41 were diagnostic accuracy studies that enrolled at least 2,491 participants and analysed at least 2,334. Five were prognostic studies that enrolled at least 1,785 participants (1,778 analysed). The quality of the 21 full text studies was reasonably good with an appropriate reference standard and avoidance of verification bias.
The pooled sensitivity of 64-slice CT was 99 per cent (95% CrI: 97%, 99%, 18 studies). Specificity was 89 per cent (95% CrI: 83%, 94%) for patient-based detection of significant CAD (at least 50 per cent stenosis). The median negative predictive value (NPV) was 100 per cent (range 86 per cent to 100 per cent) and median PPV 93 per cent (95% CIr: 64%, 100%). Eleven of 718 patients (2 per cent) could not be assessed as scans were unreadable. There was no evidence of statistically significant heterogeneity.
For segment-level analysis pooled sensitivity was 90 per cent (95% CrI: 85%, 94%, 17 studies). Specificity was higher (97%, 95% CrI: 95%, 98%). The median NPV was 99 per cent (range 95 per cent to 100 per cent). The median PPV was 76 per cent (95% crI: 44%, 93%). Pooled estimates for bypass graft analysis were 99 per cent sensitivity (95% CrI 95%, 100%) and 96 per cent specificity (95% CrI: 86%, 99%). Stent analysis pooled sensitivity was 89 per cent (95% CrI: 68%, 97%) and specificity 94 per cent (95% CrI: 83%, 98%). Statistically significant heterogeneity was detected for both sensitivity (I2 = 80.1%) and specificity (I2 = 95.1%).
Results were also reported at the level of individual arteries.
Prognostic studies reported good NPV in 64-slice CT in low-risk participants with suspected acute coronary syndrome at short-term follow-up (three studies). Another study reported no significant increase in numbers of diagnostic interventions in the year following introduction of 64-slice CT. A final study reported that invasive CA was avoided in 398 of 486 patients (82 per cent) over six months following introduction of 64-slice CT.