Interpreting the evidence: how accurate is coronary computed tomography angiography?

Abstract

Coronary CT angiography (CTA) has evolved rapidly into a powerful diagnostic tool. More than 30 accuracy studies have reported accuracy results in >2000 patients. A meta-analysis of 29 studies found per-patient accuracy of 96% sensitivity, 74% specificity, 83% positive predictive value, and 94% negative predictive value. Several clinical studies support the safety and accuracy of coronary CTA for acute chest pain, after inconclusive stress testing, and in preoperative evaluation of patients before cardiac valve surgery. Accuracy studies suffer from selection bias because of the inclusion only of patients previously selected to undergo invasive angiography. This increases the incidence of true disease, raising apparent sensitivity and lowering negative predictive value, although the latter remains high at 94%. CTA has relatively low accuracy for the quantitative assessment of stenosis severity. CTA accuracy studies show high figures for sensitivity and negative predictive value in detection of coronary lesions. CTA less accurately shows lesion severity, and intermediate-grade lesions require physiologic evaluation. Clinical studies support the effectiveness of CTA for exclusion of significant coronary disease.