Interobserver variability of the SYNTAX score by interventional cardiologists

Abstract

Purpose: In spite of being a complex calculation, reports of interobserver variability between interventional cardiologists (IC) of the SYNTAX score (SS) are limited. The objectives of this study are: a) report the interobserver variability of the SS between ICs, b) detect the main source of discrepancy. Methods: 98 consecutive angiograms with left main and/or three vessel disease were selected. Two ICs calculated the total SS. Results were analyzed categorically with linear weighted Kappa (k), quantitatively with concordance correlation coefficient (CCC) and visually through Bland Altman plot. The number of lesions observed by patient, presence of left main disease, number of bifurcations and total occlusions were analyzed by k. After establishing significant discrepancy of SS as more than the 50th percentile of the total error committed, logistic regression was performed in order to detect which characteristics were associated with high discrepancy. Results: The median SS was 28.5 (IQ: 22-36) and 29 (IQ: 22-41) for both ICs. The k was 0.70 (95% CI 0.59 to 0.81) and 0.64 (95% CI 0.54 to 0.74) for tertiles and deciles respectively. The CCC was 0,82 (CI 95% 0,75 - 0,87). Visual analysis is compatible with good agreement between observers (Figure). The k of lesions was 0,76, left main disease 0,83, bifurcacions 0,55 and occlussions 0,95. The 50th percentile of discrepancy was 4 points. Only the disagreement on the number of lesions per patient was associated with a discrepancy of more than 4 points (OR 3.72, 95% CI 1.54 to 8.99). ![Figure][1] Bland Altman Conclusions: We detected an acceptable level of agreement of the SS between ICs. Although variability of lesions per patient was acceptable, their disagreement was the only factor associated with a significant error in the assessment of the SS. [1]: pending:yes