Variability and Reproducibility of the SYNTAX Score for Triple-Vessel Disease

Abstract

BackgroundThe SYNTAX score (SX) is an angiographic grading system to determine the burden and complexity of coronary artery disease (CAD) and to guide operators as to the appropriateness of percutaneous coronary intervention (PCI) vs coronary artery bypass grafting (CABG). However, variability of the SX may exist since the assessment relies on individual clinicians to visually interpret lesion severity and characteristics. We therefore aimed to assess SX variability and reproducibility among interpreting physicians. MethodsFifty patient angiograms were randomly selected from a registry of patients with multi-vessel CAD (treated with PCI or CABG) completed at our institution during the years 2011–2018. Each angiogram was evaluated by 6 clinicians on 2 separate occasions (minimum 8 weeks between occasions) for a total of 600 SX. Our goal was to evaluate both inter- and intra- observer reliability of SX scores. Variation in both raw score as well as risk classification (low, intermediate or high SX) was observed. Inter- and intra-observer reliability were assessed using the intra-class correlation coefficient (ICC), Cohen's weighted Kappa, and Fleiss' Kappa. ResultsSYNTAX scores on both assessments and across all 6 cardiologists had a mean score of 25.3. On the first assessment, the ICC for the inter-observer reliability of SX scores was 0.61 (95% CI: 0.50, 0.73). Across the 6 observers, only 16% of angiograms were classified in the same risk classification by all observers. 34% of angiograms had less than a majority agreement (3 or less observers) on risk classification. The weighted Kappa for intra-observer reliability of risk classification scores ranged from 0.30 to 0.81. Across the 6 observers, the proportion of angiograms classified as the same risk classification between each observer's 1st and 2nd assessment ranged from 46% to 84%. ConclusionThis study shows a wide inter- and intra- user variability in calculating SX. Our data indicates a significant limitation in using the SX to guide revascularization strategies. Further studies are needed to determine more reliable ways to quantitate burden of CAD.