Reliability of new scores in predicting perioperative mortality after mitral valve surgery

Objective: The study was designed to validate euroSCORE II and ACEF (age, creatinine, and ejection fraction) scores in patients undergoing isolated or associated mitral valve surgery and compare them with logistic euroSCORE and Society of Thoracic Surgeons scores. Methods: Data on 3441 consecutive patients undergoing isolated or associated mitral valve surgery in a 6-year period were retrieved from 3 prospective institutional databases. Discriminatory power was assessed with the C index. Calibration was evaluated with calibration curves and associated statistics. Results: In-hospital mortality was 3.4%. Discriminatory power was uniformly good (for euroSCORE II: area under curve, 0.79; 95% confidence interval, 0.74-0.84; for logistic euroSCORE: area under the curve, 0.78; 95% confidence interval, 0.74-0.83; for ACEF: area under the curve, 0.73; 95% confidence interval, 0.69-0.79) but significantly higher in euroSCORE models (P < .05 for Delong, bootstrap, Venkatraman methods). Calibration pattern was slightly better for the ACEF score, although related summary statistics (unreliability, Hosmer-Lemeshow test, Spiegelhalter z-test for calibration accuracy) were not significant even for euroSCORE II. The euroSCORE II demonstrated a performance similar to Society of Thoracic Surgeons score. Logistic euroSCORE confirmed the progressive trend toward overprediction previously demonstrated in the general cardiac surgical population (summary statistics P < .05). Analysis of score performances in the surgical group studied showed results comparable to the global population. Conclusions: The euroSCORE II and ACEF scores are good predictors of perioperative mortality in patients undergoing isolated or associated mitral valve surgery, with better discrimination for the first and better calibration for the second. No algorithm seems suitable for risk estimation in mid and high-risk patients.