Analytic approaches to establish the diagnostic accuracy of coronary computed tomography angiography as a tool for clinical decision making

Although 64-slice multidetector coronary computed tomography angiography (CTA) has been reported to have excellent test characteristics for the detection of significant coronary artery disease, current analytic approaches may, not appropriately reflect the process of clinical decision making. Thirty-seven patients (29 men; mean age 63 +/- 11 years) who underwent coronary CTA for clinical indications followed by invasive coronary angiography within 4 weeks were studied. Computed tomography angiograms were analyzed independently for the presence of significant coronary artery stenosis ( >= 50% luminal narrowing) by 2 observers blinded to invasive coronary angiographic results. The diagnostic test performance of coronary CTA was determined with and without inclusion of unassessable segments. Because stenosis could not be excluded in unassessable segments, these segments were counted as positive for stenosis. Sensitivity, specificity, and positive (PPV) and negative predictive values of CTA for detecting significant stenoses on assessable segments were 85% (51 of 60, 95% confidence interval [CI] 76% to 94%), 99% (414 of 416, 95% CI 99 to 100), 96% (95% CI 51 of 53), and 98% (95% CI 414 of 423), respectively. Overall, 13% of coronary segments (70 of 546) were not assessable using CTA (heavy calcium in 48 segments). By including these segments, PPV decreased from 96% to 60% (74 of 123), whereas sensitivity improved from 85% to 89% (95% CI 74 of 83). In conclusion, the clinical utility of coronary CTA may be limited by a low PPV in patients with a high prevalence of coronary artery disease. (c) 2007 Elsevier Inc. All rights reserved.