Interobserver concordance in aortic regurgitation grading: a real-world comparison between experts and trainees using quantitative echocardiographic parameters

Abstract Background Accurate grading of aortic regurgitation (AR) severity via transthoracic echocardiography (TTE) is essential for guiding timely intervention and preventing irreversible left ventricular (LV) dysfunction. Current guidelines support the use of quantitative parameters due to their stronger correlation with long-term prognosis and surgical outcomes. However, their applicability in clinical practice may be limited by technical and interpretative variability. Purpose To evaluate the concordance of AR severity grading in a real-world setting, comparing assessments performed by a team of experts in advanced cardiac imaging techniques (EXACT-team) and trainees with intermediate echocardiographic skills (TRAIN-team). Methods We analyzed 99 consecutive patients referred to our tertiary echocardiography outpatient clinic for AR evaluation. Patients were excluded if they had incomplete TTE studies, congenital heart disease, acute AR, or AR secondary to aortic dissection or endocarditis. The EXACT-team included two cardiologists with over 5 years of TTE experience in a certified echocardiography lab, while the TRAIN-team comprised two cardiology residents with at least 1 year of TTE experience. All observers independently reviewed the TTEs, blinded to clinical information, and graded AR severity according to current European guidelines. Results The mean patient age was 70.9 ± 14.4 years. Quantitative assessment was feasible in 77 cases using pressure half-time (PHT), 68 with vena contracta (VC), and 30 with effective regurgitant orifice area (EROA). Overall interobserver agreement for AR grading between EXACT and TRAIN teams was moderate (κ = 0.51) and fair (κ = 0.39), respectively, with Fleiss’ Kappa across all raters indicating moderate agreement (κ = 0.47). PHT-based grading yielded fair agreement for both teams (EXACT: κ = 0.30; TRAIN: κ = 0.39) and moderate overall agreement (κ = 0.36), with 3 major (grade 1 vs 3) and 5 moderate-to-severe (grade 2 vs 3) discrepancies. VC-based grading demonstrated substantial agreement for the EXACT team (κ = 0.70) and almost perfect agreement for the TRAIN team (κ = 0.86), with a substantial overall Fleiss’ Kappa (κ = 0.77). EROA-based grading showed substantial agreement within the EXACT (κ = 0.64) and TRAIN teams (κ = 0.74), with moderate overall concordance (κ = 0.55). Conclusions AR grading is complex and a higher level of experience is needed for a more comprehensive evaluation. VC demonstrated the highest interobserver concordance, highlighting its reliability for AR grading across varying level of experience in routine echocardiographic practice.