Visual versus quantitative assessment of the severity of coronary artery stenoses: Can the angiographer's eye be reeducated?

The percent reduction of intraluminal diameter of 102 coronary stenoses before (n = 26), immediately after (n = 24), and 6 months after (n = 52) coronary balloon angioplasty as assessed by two observers experienced in quantitative coronary angiography and by automated measurement by means of the Cardiovascular Angiographic Analysis System (CAAS) was compared. Two hundred twenty-seven still frames were selected for analysis by CAAS and displayed for assessment by each of the observers, who were unaware of CAAS results. Comparisons with CAAS measurements were made for all still frames ("per view" analyses) and for each coronary stenosis after averaging the values obtained for a given coronary segment ("per stenosis" analyses). Intraobserver variability was tested on 21 still frames from 10 stenoses, assessed 6 weeks apart by one observer. Intraobserver variability was 5% (r = 0.98); interobserver variability was 6% (r = 0.94). Per view analyses showed that observer measurements correlated well with CAAS measurements (r = 0.89 and r = 0.90); the accuracy was -0.9% and -0.7% for observers 1 and 2, and the precision was 9% and 8%, respectively; for per stenosis analyses, the accuracy was -0.5% and -0.8% (r = 0.92 and 0.94) and the precision 7% and 6%, respectively. There was no difference in the accuracy and precision of visual assessment for lesions with 〈50%- or 〉50%-diameter stenoses. The occurrence of restenosis was likewise detected visually with a high sensitivity and specificity compared with CAAS. Visual assessment of coronary obstructions by observers trained in quantitative coronary angiography techniques can yield measurement accuracy and precision of sufficient standard for clinical practice and may be applicable in some clinical angiographic studies.