Does EuroSCORE II perform better than its original versions? A multicentre validation study.

Area of Science:

• Cardiovascular Surgery
• Medical Statistics
• Risk Prediction Models

Background:

• The European system for cardiac operation risk evaluation (EuroSCORE) is a standard tool for predicting in-hospital mortality after cardiac surgery.
• EuroSCORE II is a recently developed update to improve upon previous versions.
• Validation and comparison of EuroSCORE II against original versions are crucial for clinical application.

Purpose of the Study:

• To validate the performance of the EuroSCORE II risk prediction model.
• To compare the predictive accuracy of EuroSCORE II with the original additive and logistic EuroSCORE models.
• To assess the impact of removing statistically non-significant factors on EuroSCORE II's performance.

Main Methods:

• Analysis of data from 12,325 consecutive patients undergoing major cardiac surgery over a 6-year period.
• Assessment of discriminatory power using the c-index and comparison of scores via Delong, bootstrap, and Venkatraman methods.
• Evaluation of calibration using calibration curves and associated statistics.

Main Results:

• The overall in-hospital mortality rate was 2.2%.
• All evaluated algorithms demonstrated high and similar discriminatory power (Area Under the Curve ≈ 0.82).
• EuroSCORE II exhibited fair calibration up to 30% predicted mortality, with over-prediction beyond this threshold. Removing non-significant factors did not alter performance.

Conclusions:

• EuroSCORE II is a validated predictor of perioperative mortality in cardiac surgery patients.
• The model shows optimal calibration for predicted mortality up to 30%.
• EuroSCORE II does not significantly outperform older versions in higher risk tertiles and could be simplified by removing non-significant factors without compromising performance.