Objectives: To assess intraobserver and interobserver variation in computed tomography coronary angiography (CTA) in 3 patient cohorts at very low, low-to-intermediate, and intermediate-to-high likelihood of coronary artery disease (CAD).
Methods: One hundred thirty-three patients underwent 64-channel CTA. Coronary arteries were analyzed by 2 experienced blinded observers and reported as having 0%, 1% to 29%, 30% to 49%, 50% to 69%, 70% to 99%, or 100% stenosis. Intraobserver and interobserver agreement was calculated at cohort level and combined.
Results: Overall intraobserver and interobserver agreement was good (kappa = 0.74 and kappa = 0.78, respectively). Segmental agreement for stenoses 50% or greater and 70% or greater was greater than 96%. Disagreements were more likely in the presence of noneccentric calcification for both intraobserver (odds ratio = 0.45 and 0.22) and interobserver (odds ratio = 0.40 and 0.10) measurements.
Conclusions: Interobserver and intraobserver variability for the detection of coronary stenoses on CTA is good and justifies routine clinical use. The presence of noneccentric calcium and mixed plaque morphology are important causes of disagreement.