The Cholesterol Lowering Atherosclerosis Study, a randomized, angiographic clinical trial, has demonstrated the beneficial effect of niacin/colestipol therapy on coronary and femoral atherosclerosis. The primary outcome was a panel-determined consensus score evaluating global coronary changes determined angiographically at 2 years. This article presents an evaluation of interreader agreement in independently assessing the status of native coronary arteries and overall coronary condition. Parameters include 1) identification of the presence of lesions and lesion changes; 2) estimation of lesion severity (percent stenosis) and amount of change in lesion severity, and 3) global assessment of change in coronary status. Readers independently agreed on 1) presence of lesions (82%) and change in lesions (51%); 2) percent stenosis±10% (76%) and change in stenosis±10% (81%); and 3) global assessment of change in coronary status within one step (96%). Results of these analyses may be useful in effectively designing angiographic trials that use a panel of human evaluators as well as computerized methods for angiographic interpretation. {Arteriosclerosis and Thrombosis 1991;ll:385-394) V alid and reliable information about the status of coronary lesions is needed for evaluation of interventional therapy in controlled angiographic trials. When patients are randomized to test or control, the required information should lead to analysis on a per-patient basis to determine trial outcome. For patients with multiple lesions, results for each lesion must be integrated into a single score for that patient.Coronary lesions have been evaluated for clinical research in three general ways: 1) human readers viewing angiographic films alone 1 " 3 or as panel members 4 -6 ; 2) human readers measuring films with calipers 7 or tracing vessel edges by hand for later computer analysis 8 ; and 3) computerized edge-finding after human designation of the search areas.8 " 10 The reliability of film evaluation by humans for changes in coronary status is influenced by 1) technical features, such as film quality, the conditions for film reading, the number of films a reader has recently evaluated, and order of film presentation and 2) coronary pathological and anatomic characteristics, such as lesion size, degree and direction of change in lesions, degree and direction of change in per-patient disease status, degree to which branches overlap, and frequency of lesion formation at branch junctions.
" 13The Cholesterol Lowering Atherosclerosis Study (CLAS) was a randomized, placebo-controlled clinical trial testing the effects of colestipol plus niacin therapy. 4 Coronary, carotid, and femoral angiograms, selected to sample treatment effects as widely as possible, were obtained at baseline and at 2 years. CLAS was monitored by an external advisory committee, which mandated an analysis of coronary vessels at 2 years as a requirement for a proposed trial extension (CLAS-II), in which each patient would remain on his assigned treatment for 4 years and the...