Abstract
Background: EuroSCORE was widely used for predicting outcomes after cardiac surgery, but has become poorly calibrated for contemporary cohorts. EuroSCORE II and AusSCORE, based on an Australasian population, were recently developed. We compared the performance of these three scores for coronary artery bypass grafting (CABG). Methods: All isolated CABG patients at Auckland City Hospital during July 2010–June 2012 were included. EuroSCORE I, EuroSCORE II and AusSCORE retrospectively calculated and compared for discrimination and calibration of outcomes. Results: There were 818 patients followed for 1.4± 0.6 years. Mean EuroSCORE I, EuroSCORE II and AusSCOREwere4.5± 5.0%, 2.6± 3.1%and0.9± 1.3%.Mortality at 30 days, follow-up and composite surgical morbidity were 1.6% (13), 2.9% (24) and 17.8% (146). C-statistic of these scores for 30-day mortality were 0.675 (95% confidence interval 0.531–0.819), 0.642 (0.503–0.780) and 0.661 (0.516–0.807), while the Hosmer–Lemshow test for calibration were p= 0.061, 0.150 and 0.420 respectively. C-statistics for mortality at follow-up were 0.639 (0.525–0.752), 0.604 (0.483–0.752) and 0.593 (0.480–0.705); and for morbidity 0.678 (0.631–0.726), 0.634 (0.582–0.686) and 0.645 (0.593–0.698). EuroSCORE I was the best model at detecting stroke (c= 0.736) and ventilation> 24h (c= 0.712), and EuroSCORE II for deep sternal wound infection (c= 0.720) and return to theatre (c= 0.626). Conclusion: EuroSCORE II and AusSCORE estimates were closer to the observed 30-day mortality than EuroSCORE I, but their discriminative ability for outcomes were not superior. Revision of risk models to fit contemporary surgical outcomes is important for calibration, but discrimination is more difficult to improve upon.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call