Reproducibility of a consensus panel in the interpretation of coronary angiograms

This report describes the consistency of coronary angiogram evaluation by a four man panel of experts rendering "consensus opinion." The panel evaluated films from 38 patients at two grading sessions separated by an interval of seven months. Fourteen patients' films were selected at random for duplicate evaluation. These contained 186 lesion sites. The panel was 95% consistent in designating significant stenosis (greater than or equal to 70%). "Consensus" panel reading appears more consistent than "group opinion" panel reading. The over-all standard deviation of the difference in panel reading was 14%. The panel was most consistent in evaluating the right coronary artery, proximal LAD, and proximal circumflex. In the left main segment two of fourteen duplicate evaluations showed major discrepancy.