Observer variabilities for the diagnosis of coronary artery disease using anatomical and functional testing: the impact of certification

To compare the reproducibility in reporting of coronary computed tomography angiography (CCTA) or cardiovascular magnetic resonance imaging (CMR) by certified readers for CCTA and CMR by the German Society of Cardiology (DGK) versus that by non-certified readers.

The study included 40 randomly selected CCTA and vasodilator stress CMR patient datasets. For CCTA, the degree of lumen narrowing, plaque composition, and high-risk plaque features were assessed. For CMR, wall motion and perfusion abnormalities and late gadolinium enhancement (LGE) were rated. All measures were conducted by segments and for individual patients. Intraclass correlation coefficients (ICC) were calculated to assess agreement between non-certified (n = 4) vs. DGK-certified readers (n = 4).

ICC for assessment of luminal narrowing, plaque composition, and high-risk features were, respectively, 0.65 (95% confidence intervals [CI] 0.59–0.69), 0.64 (95%CI 0.45–0.80), and 0.45 (95%CI 0.22–0.66) for non-certified versus 0.78 (95%CI 0.74–0.81), 0.88 (95%CI 0.79–0.93), and 0.89 (95%CI 0.81–0.95) for DGK-certified readers (p < 0.001 for all). ICC for the assessment of wall motion, perfusion, and LGE were, respectively, 0.41 (95%CI 0.35–0.48), 0.27 (95%CI 0.18–0.38), and 0.48 (95%CI 0.41–0.54) for non-certified versus 0.71 (95%CI 0.67–0.75), 0.71 (95%CI 0.67–0.75) and 0.67 (95%CI 0.62–0.71) for DGK-certified readers (p < 0.001 for all). The agreement was excellent among DGK-certified readers for obstructive CAD (≥ 70% lumen narrowing) assessed by CCTA and high for abnormal perfusion and for LGE by CMR in a per-patient analysis (0.88; 95%CI 0.79–0.94 and 0.84; 95%CI 0.71–0.92), respectively.

Substantially better CCTA and CMR reporting was observed for DGK-certified cardiologists, who achieved high agreement for diagnosing the presence or absence of obstructive CAD by CCTA and abnormal perfusion by CMR. Since important clinical decisions may be based on these readings, our data support quality-controlled education programs for advanced cardiac imaging.