Conventional visual assessment of coronary arteriogram is fraught with large interobserver variance and disagreement with pathologic findings. Thus quantitative coronary angiography had been recently developed to meet the requirement of more reproducible measurement of severity of coronary artery disease. This study was designed to evaluate the interobserver variability of quantitative coronary angiography and its usefulness in clinical application.

Three independent observers analysed coronary angiogram of 31 consecutive patients by visual assessment and quantitative measurement using computer based algorithm.

1) There was considerable disagreement between 3 observers in the identification of significant coronary stenosis. Complete agreement was achieved in only 29/61 (47%) coronary lesion and agreement of more than 2 observers in 41/61(67%) lesion. 2) In visual assessment, the largest interobserver variance was found in acute marginal artery and distal circumflex artery, whereas the least variance was observed in proximal right coronary artery. 3) The average interobserver variance was 13.1% in visual assessment, 14.9% in geometric analysis, and 10.5% in video densitometric analysis. There was no significant difference between these values. 4) Visual assessment was noted to overestimate the severity of stenotic lesion, by 5.6% compared to geometric analysis and by 11.8% compared to video-densitometric analysis.

These findings suggest that identifying significant lesion is major source of interobserve variability in both visual and quantitative analysis of coronary arteriogram. In addition, suboptimal image quality was responsible for the inability of quantitative analysis to reduce the variance. These factors seem to be major limitation of quantitative coronary arterirogam.