This review assessed the accuracy of contrast enhanced coronary computed tomography for the diagnosis of coronary artery stenosis and concluded that accuracy was high. There were a number of concerns regarding the conduct of the review, the accuracy of it's reporting and the reliability of pooled results combining clinically heterogeneous studies, so the conclusions should be treated with some caution.

To assess the accuracy of contrast enhanced coronary computed tomography (CE-CCT) for the diagnosis of coronary artery stenosis.

MEDLINE and EMBASE were searched from 1990 to March 2006; search terms were reported. Reference lists of retrieved articles were scanned and experts in the field contacted to identify published and unpublished studies.

Studies comparing CE-CCT to catheter-based coronary angiography independently of computed tomography (CT) findings to evaluate native coronary arteries were eligible for inclusion if they reported sufficient data to allow diagnostic outcomes to be extracted or calculated. The CT scanners used varied across studies (electron beam, and 4-, 8-, 12-, 16-, 40- and 64-slice). Where reported, the mean age of participants ranged from 53 years to 65 years, the proportion of men from 58 per cent to 96 per cent and the number of coronary segments ranged from 50 to 1,384, with the proportion of segments considered assessable ranging from 57 per cent to 100 per cent. The prevalence of coronary artery disease ranged from 20 per cent to 100 per cent. A threshold of 50 per cent stenosis was used as the criterion for positivity in most studies; 70 per cent and 75 per cent were used in some. The authors stated neither how studies were selected for the review nor how many reviewers performed the study selection.

The authors did not state that they assessed study quality, but blinding of assessors and the consecutive recruitment of patients were noted and discussed.

Sensitivity and specificity for assessable segments, all segments and per patient were extracted or calculated for each study. Additional data were obtained from trialists where required. Data were extracted by three reviewers; whether this was conducted independently, and how disagreements were resolved, was not reported.

It was stated that pooled summary estimates of sensitivity and specificity were calculated using a random-effects model, however, results presented were from a fixed-effect model. Results were presented for assessable segments, all segments and per patient. Pooled diagnostic odds ratios (DORs) and 95% confidence intervals (CI) were also calculated. Heterogeneity was assessed using the I² test. Analyses were conducted to assess the impact of diagnostic threshold. Summary receiver operating characteristic (sROC) curves were produced. Subgroup analyses were conducted to investigate sample size, patient age, proportion of men, prevalence of CAD, blinding and publication date.

Forty-one studies were included in the review (n=2,515; range 20 to 133). Of these, 27 reported consecutive enrollment and 34 blinding to angiography findings when interpreting CE-CCT.

For all CT
The pooled sensitivity was 95 per cent, specificity was 85 per cent and DOR 102 (95% CI: 63, 141) for all segments. The pooled sensitivity was 96 per cent, specificity 95 per cent and DOR 429 (95% CI: 240, 617) for assessable segments. The pooled sensitivity was 99 per cent, specificity 76 per cent and DOR 306 (95% CI: 183, 428) per patient.

For electron beam CT
The pooled sensitivity was 80 per cent, specificity 77 per cent and DOR 13 (95% CI: -52, 77) for all segments. The pooled sensitivity was 86 per cent, specificity 90 per cent and DOR 60 (95% CI: -282, 401) for assessable segments. The pooled sensitivity was 90 per cent, specificity 59 per cent and DOR 12 (95% CI: -228, 252) per patient.

For multidetector CT
The pooled sensitivity was 95 per cent, specificity 90 per cent and DOR 152 (95% CI: 103, 200) for all segments. The pooled sensitivity was 96 per cent, specificity 96 per cent and DOR 591 (95% CI: 365, 817) for assessable segments. The pooled sensitivity was 99 per cent, specificity 85 per cent and DOR 408 (95% CI: 266, 550) per patient.

Results for different types of multidetector CT and sensitivity analyses were presented.

Diagnostic accuracy of CE-CCT is high following advances in technology that enabled increases in accuracy and the proportion of assessable segments. Per patient accuracy may be lower and CT may have more limited clinical utility for those with high risk of CAD.

The authors addressed a clear research question in terms of participants, intervention, reference standard, outcomes and study design. Two relevant databases were searched. Although the authors stated that experts were contacted to identify unpublished studies, the inclusion criteria stated that studies had to be published in peer-reviewed journals and so publication bias could not be ruled out. It was unclear whether language restrictions were applied, so language bias may have been present. Despite the number of included studies, these may not have represented the entire evidence base.

Data extraction seemed to be conducted in duplicate, but it was unclear whether similar methods to reduce error and bias were used during study selection. The authors did not systematically assess study quality, but some aspects of study quality were reported. Most studies had small sample sizes and several reported no consecutive recruitment of patients or blinding of assessors, indicating the poor quality of a proportion of the studies.

The authors stated that a random-effects model was used due to significant statistical heterogeneity to give greater precision than a fixed-effect model (a random-effects model provides a more conservative estimate with wider confidence intervals) yet the forest plots presented were labelled as being derived from fixed-effect meta-analyses. There was a large degree of clinical heterogeneity across studies, and the reliability of the pooled estimates is questionable. Sources of heterogeneity were investigated. The actual figures for the confidence intervals were not reported for the pooled estimates of sensitivity and specificity, despite being presented on forest plots. The populations in studies were restricted and the results may not be generalisable to clinical practice. Given the concerns stated, the reliability of the conclusions should be treated with some caution.

Practice: The authors did not state implications for practice.

Research: The authors stated that large prospective trials were necessary to establish clinical applications of cardiac multidetector CT across different populations.

Not stated.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.