Reliability of assessing change with sequential coronary angiography

Data are presented from a study evaluating the reliability of sequential angiography in estimating changes in coronary lesions. Three panels of three expert angiographers each read on two separate, independent occasions 18 sets of paired angiograms taken 24 months apart. All readers were blinded to the temporal sequence of the films, clinical data, and ventriculography information. The need for simultaneous viewing and reading of the two films, for training sessions, and for allowance to be made for apparent differences arising from boundary definitions prior to final analysis was demonstrated. Under stringent conditions (determination of change based on agreement of at least two out of three panels of physicians) it was possible to ascertain change in coronary atherosclerosis from sequential sets of coronary angiograms with good reliability. However, single-panel readings yielded an unacceptable overestimate of the number of patients with lesion changes.