Refitting EuroSCORE II for 120-day mortality after coronary artery bypass grafting using nationwide registry data

Abstract

Abstract Objectives The EuroSCORE II risk prediction model is the current gold standard in Europe to predict in-hospital mortality after cardiac surgery and is widely used in clinical decision-making and outcome reporting. A risk prediction model specifically for patients undergoing elective or urgent isolated coronary artery bypass grafting (CABG), a commonly performed cardiac procedure, may lead to a better performance for individual risk prediction. The aim of this study is to validate and refit the EuroSCORE II for 120-day mortality, using international multicentre data from 2 national quality registries. Methods All elective or urgent isolated CABG procedures performed between 2013 and 2019 in adult patients were included using nationwide data from the Netherlands Heart Registration and SWEDEHEART. The performance of the EuroSCORE II was analysed in the Dutch cohort and refitted by estimating a new intercept and coefficient for the linear predictor of the original EuroSCORE II. This refitted model was validated in the Swedish cohort and tested against the original EuroSCORE II model. Results In total 45 016 procedures from the Netherlands were included for validation and refitting of the model. The refitted model showed similar discrimination to the EuroSCORE II and could be marginally optimized by adding the interaction between age and sex, resulting in a C-statistic of 0.80. This refitted model was validated using 16 374 procedures from the Swedish cohort. The refitted model showed improved calibration and discrimination compared to the original EuroSCOREII. Conclusions We present a refitted risk prediction model for 120-day mortality after elective or urgent isolated CABG using high-quality nationwide data.