Abstract
Risk stratification remains critical in cardiac surgery, enabling clinicians to predict adverse outcomes and guide perioperative management. The European System for Cardiac Operative Risk Evaluation (EuroSCORE) II, introduced in 2011, incorporates 18 key variables to provide an evidence-based approach to risk assessment. However, evolving surgical techniques, changing patient demographics, and emerging evidence reveal limitations in the model's predictive capabilities. Important factors such as frailty, race, liver dysfunction, left ventricular dimensions, and advanced cardiac function metrics are not incorporated, reducing its accuracy in diverse and high-risk populations. Additionally, the model does not fully account for key conditions, such as infective endocarditis, where high-risk features like embolic events and abscesses significantly impact surgical outcomes. Simplified categorisation of procedures and the binary assessment of coronary artery disease overlook critical complexities, such as lesion severity and procedural variability. Advanced parameters like global longitudinal strain (GLS), SYNTAX, and Model for End-Stage Liver Disease (MELD) scores could enhance the model's granularity and predictive power. Furthermore, integrating machine learning into future iterations of EuroSCORE offers the potential to capture non-linear interactions and continuously refine predictions. These updates could pave the way for a 'EuroSCORE III' better aligned with modern surgical practices, offering improved precision in risk stratification, more personalised clinical decision-making and optimised patient outcomes.