Accuracy and interobserver variability of coronary cineangiography: a comparison with postmortem evaluation

Abstract

The accuracy of interpretation of coronary cineangiography by two independent observers was tested against postmortem findings in 27 patients who died within 6 months of cardiac catheterization. Variations in cineangiographic interpretations between the angiographers were also evaluated. Two patients had normal coronary arteries, while the remaining 25 patients had significant coronary artery disease. Significant stenosis was defined as 75% or greater reduction in luminal diameter. Of 326 coronary segments that could be evaluated postmortem, 15% could not be evaluated cineangiographically. The respective overall accuracy of the two observers was 89 and 88% with an accuracy of 96 and 100% for the left main coronary artery, 91 and 93% for the left anterior descending artery, 84 and 86% for the right coronary artery and 89 and 79% for the left circumflex coronary artery. Cineangiographic assessment of luminal status distal to a significant proximal lesion was possible in more than 70% of major vessels with accuracy levels of 86% for both observers. Of 96 distal vessels inadequately opacified cineangiographically, 49 (52%) were found to be free of significant lesions. Both angiographers agreed in their assessment of 86% of the 340 coronary segments. Interobserver agreement was significantly better for the left main, right and left anterior descending coronary arteries than for the left circumflex coronary artery (p less than 0.05). Accuracy was 93% for 244 segments that were adequately opacified and assessed the same by both angiographers. Cineangiography can thus be used to evaluate coronary anatomy with a high degree of accuracy and minimal interobserver variability.