Validity and variability in visual assessment of stenosis severity in phantom bifurcation lesions: a survey in experts during the fifth meeting of the European Bifurcation Club.

Objectives: To investigate the adequacy of visual estimate regarding the percent diameter stenosis (DS) in bifurcation lesions.
Background: Quantitative coronary angiography (QCA) is more accurate and precise compared to visual estimate in assessing stenosis severity in single-vessel lesions.
Methods: Thirty-six experts in the field of bifurcation PCI visually assessed the DS in cine images of five precision manufactured phantom bifurcation lesions, experts being blinded to the true values. Expert DS estimates were compared with the true values and they were also used to define the Medina class of each individual bifurcation. Results were pooled together both for proximal main vessel (PMV), distal main vessel (DMV) and side-branch (SB) segments and for vessel segments with similar DS values.
Results: Individual performance was highly variable among observers; pooled values and range of accuracy and precision were 2.79% (-6.67% to 17.33%) and 8.69% (4.31-16.25%), respectively. On average, DS was underestimated in the PMV (-1.08%, P = 0.10) and overestimated in the DMV (3.86% P < 0.01) and SB segments (5.58%, P < 0.01). Variability in visual estimates was significantly larger in lesions of medium severity compared to the clearly obstructive ones (P < 0.01); the latter were consistently overestimated. Inter-observer agreement was moderate (κ = 0.55) over the entire number of estimates. However, if the segments with true DS = 0% were excluded, agreement was diminished (κ = 0.27). Inter-observer agreement in Medina class was rather low (κ = 0.21). True bifurcation lesions were misclassified as non-true ones in 14/180 estimates.
Conclusions: Visual assessment by experts is more variable and less precise in the analysis of bifurcation lesions compared to bifurcation QCA software.
(Copyright © 2011 Wiley Periodicals, Inc.)