SESSION TITLE: Cardiac and Thoracic Surgery Posters
SESSION TYPE: Original Investigation Poster
PRESENTED ON: Wednesday, October 30, 2013 at 01:30 PM - 02:30 PM
PURPOSE: EuroSCORE II has been recently developed, but its performance of early mortality after cardiac surgery has been poorly examined. The aim of this study was to evaluate EuroSCORE II and to compare its performance with original logistic EuroSCORE (LES) and additive EuroSCORE (AES) in predicting early mortality after aortic valve replacement.
METHODS: We analyzed the data from 340 consecutive patients who underwent AVR, with or without CABG, for aortic stenosis at our institution between 2002 and 2012. Calibration and discrimination ability of these risk algorithms were assessed.
RESULTS: Observed early mortality was 3.24% (n=11). In overall patients, mean predicted mortality for EuroSCORE II, LES, and AES were 3.13% (O/E ratio=1.04, p=0.62), 7.09% (O/E ratio=0.46, p<0.001), and 6.28% (O/E ratio=0.51, p<0.001), respectively. Therefore, EuroSCORE II was well calibrated, but LES and AES significantly overestimated the risk. Patients were then stratified into approximately equal-size tertiles according to risk score. EuroSCORE II significantly underestimated mortality for the low-risk (O/E ratio=1.71) and moderate-risk (O/E ratio=1.28) tertiles, whereas it slightly (not significantly) overestimated mortality for the high-risk tertile (O/E ratio=0.84). Both LES and AES significantly overestimated mortality in all tertiles. Hosmer-Lemeshow test did not show bad calibration for all algorithms. C-index as discrimination power for EuroSCORE II was slightly lower (C-index 0.66, 95%CI 0.46 to 0.85) than that for LES (C-index 0.72, 95%CI 0.55 to 0.88) and AES (C-index 0.71, 95%CI 0.54 to 0.87).
CONCLUSIONS: EuroSCORE II is better calibrated than LES and AES, but does not improve the discrimination power when compared with LES and AES. Larger validation studies are necessary to confirm these findings.
CLINICAL IMPLICATIONS: Calibration of the new EuroSCORE II was good, but its discrimination power was suboptimal. EuroSCORE II partially improved the risk prediction performance in AVR when compared with original version.
DISCLOSURE: The following authors have nothing to disclose: Kenji Kuwaki, Atsushi Amano, Hirotaka Inaba, Taira Yamamoto, Terumasa Morita, Shizuyuki Dohi, Takeshi Matsumura, Akie Shimada, Atsumi Oishi, Kishio Kuroda
No Product/Research Disclosure Information