This review concluded that multi-detector computed tomography performed well for detecting and excluding significant coronary artery stenosis and performed better in proximal than distal segments. These conclusions should be interpreted cautiously due to limitations in search strategy and the number and size of studies. Weaknesses in the analytical methods made the conclusion about differential performance unlikely to be reliable.

To assess the diagnostic performance of 40-slice and 60-slice multi-detector computed tomography angiography compared with invasive coronary angiography for the detection of significant (50 per cent or greater) coronary artery stenosis.

PubMed and EMBASE were searched from January 2005 to December 2007. Search terms were reported. The search was limited to peer-reviewed articles published in English. The references of retrieved articles were screened manually for additional studies.

Prospective or retrospective studies assessing the diagnostic performance of computed tomography angiography (performed with a 40- or 64-MDCT scanner with a detector element width of 0.625 mm or less) for the diagnosis of significant coronary artery stenosis (defined as 50 per cent or more) in patients with suspected coronary artery disease were eligible for inclusion. Eligible studies needed to use invasive coronary angiography as the reference standard to establish diagnosis. Included studies were required to report absolute numbers of true-positive, false-negative, false-positive and true-negative findings (or data allowing their calculation). Analyses needed to be performed on a per-patient or per-segment basis (significant per-patient coronary artery disease was defined as significant stenosis in at least one segment).

Studies were excluded if they had 50 per cent or more participants with known coronary artery disease, balloon angioplasty, stents or coronary artery bypass graft. Also excluded were studies that did not identify non-evaluable scans and segments, and studies with frequent extrasystole heartbeats.

The mean age of participants in included studies was 60 years (range 54 to 69 years). The mean heart rate was 64 beats per minute (range was 58 to 73 beats per minute). Prevalence of coronary artery disease ranged from 18 per cent to 85 per cent.

The authors did not report how many reviewers performed the inclusion screening.

The authors did not state that they assessed study validity.

Absolute numbers of true and false negatives, true and false positives and numbers of uncertain or non-evaluable results were extracted per-patient and per-segment (proximal and distal). The main outcomes used in the review were sensitivity and specificity.

The Bypass Angioplasty Revascularization Investigation (BARI) model was used to convert the different segment classification systems used in studies to a uniform model; categorisation into proximal and distal segments followed the American College of Cardiology/American Heart Association (ACC/AHA) model.

Data were extracted independently by two reviewers; disagreements were resolved by consensus.

Pooled estimates of sensitivity and specificity and their 95% confidence intervals (CIs) were calculated per patient and per segment using a random-effects (DerSimonian-Laird) model weighted by sample size. Analyses included a correction for non-evaluable segments. Heterogeneity was assessed using the I² test. Sensitivity and specificity were also estimated separately for proximal and distal segments.

Summary receiver operating characteristic (SROC) curves were also constructed. Regression analyses were used to assess the dependence of sensitivity upon clinical variables such as disease prevalence and heart rate.

Twenty two studies were included in the meta-analysis. A total of 1,331 patients and 10,561 segments were analysed. The percentage of patients excluded from studies after CT angiography ranged from 0 to 13.5 per cent. The percentage of non-evaluable proximal segments ranged from 0 to 15.4 per cent and for distal segments from 0 to 19.3 per cent.

Patient level data: The pooled estimate of sensitivity was 97.7 per cent (95% CI: 96.2%, 98.7%), I² = 0.0% and specificity was 91.0% (95%CI: 88.5, 93.1%), I² = 23.4%. If non-evaluable CT angiography investigations were included, specificity was reduced significantly if non-evaluable participants were classified as having significant coronary artery stenosis; sensitivity was reduced significantly if they were classified as not having significant stenosis. Regression analyses indicated that the prevalence of coronary artery stenosis did not significantly affect sensitivity, but that there was a significant negative correlation between heart rate and sensitivity.

Segment level data: The overall pooled estimate of sensitivity was 90.8 per cent (95% CI: 89.0, 92.4%), I² = 86.2% and specificity was 95.7 per cent (95% CI: 95.2, 96.1%), I² = 93.0%. For proximal segments, sensitivity was 94.2 per cent (95% CI: 92.3, 95.7%), I² = 56.0%, and specificity 94.1% (95% CI: 93.4, 94.8%), I² = 89.7%. For distal segments, sensitivity was 84.8% (95% CI: 81.1, 88.0%), I² = 85.0 and specificity 96.9% (95% CI: 96.4, 97.4%), I² = 86.0%. Sensitivity was significantly higher and specificity significantly lower in proximal compared with distal segments (p < 0.05).

Further data were reported for specific arterial segments.

The performance of 40-slice and 64-slice multi-detector computed tomography was good to excellent for detecting or ruling out significant (50 per cent or more) coronary artery stenosis. Results were better for proximal than for distal coronary artery segments.

The review addressed a clearly stated research question defined by appropriate inclusion criteria. The search strategy was limited to two bibliographic databases and included methodological terms for diagnostic accuracy studies, which have been shown to reduce the sensitivity of searches, so it was possible that relevant data were missed. Searches were restricted to published English-language studies, leaving the review open to both publication and language bias. During data extraction, measures were taken to reduce the potential for error and/or bias, but it was unclear whether similar measures were applied to study selection. No assessment of the methodological quality of included studies was reported.

The meta-analytic methods applied to the patient-level analysis were broadly appropriate. Adequate details of the individual studies were reported. But, the segmental analysis was problematic on a number of levels. Data for all segments had I² values that indicated high levels of heterogeneity. This heterogeneity persisted when data were stratified by proximal and distal segments, making the estimation of pooled sensitivities and specificities of doubtful value. The analysis, as presented, allowed the inclusion of multiple data points (segments) per patient and appeared to assume independence. More sophisticated methods of analysis that take account of the inclusion of multiple data points per patient may have been more appropriate in this scenario.

The authors' conclusion that multi-detector computed tomography generally performed well for the detection or exclusion of significant coronary artery stenosis reflected the data presented, but should be viewed with some caution given that it is based on 20 small data sets and that relevant studies may have been omitted. The conclusion that multi-detector computed tomography performed better in proximal than in distal segments was much weaker, given the limitations of the analysis and that the data presented indicated only that sensitivity may be higher for proximal segments.

The authors did not make recommendations for research or practice.

The study was supported by the EC-EURATOM 6 Framework Programme (2002-2006) and was part of the CT Safety and Efficacy project, FP/002388.

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.