Clinical SYNTAX Score Can Predict Acute Kidney Injury following On-Pump but Not Off-Pump Coronary Artery Bypass Surgery

Abstract

Background: The complexity of coronary artery disease is usually a neglected factor in risk stratification systems. We aimed to analyze the discriminative ability of the clinical SYNTAX score (CSS) for acute kidney injury (AKI) following on- and off-pump coronary artery surgery. Methods: A total of 193 patients were reviewed in this study. Patients were divided into two groups according to the surgical procedure (group I: off-pump coronary artery bypass grafting, n = 89; group II: on-pump coronary artery bypass grafting, n = 104). Preoperative demographic data, the CSS and postoperative renal functions were evaluated. The postoperative AKI classification was made using the RIFLE (Risk, Injury, Failure, Loss of function, and End-stage renal disease) criteria. Results: Postoperative AKI occurred in 14 of 89 patients (15.7%) in group I and in 29 of 104 patients in group II (27.8%; p = 0.046). The CSS did not vary much between the groups (31.52 ± 13.08 vs. 29.89 ± 15.70; p = 0.638). In group I, the CSS was not different between patients with AKI and those without AKI (30.167 ± 3.93 vs. 31.91± 14.75; p = 0.78). In group II, the CSS was 36.85 ± 18.33 in patients with AKI and 28.02 ± 12.32 in those without, and the difference was significant (p = 0.02). The discriminative ability of the CSS for postoperative AKI using the AUC analysis was 0.500 in group I and 0.840 in group II. Conclusion: The CSSmay be a simple and successful means of risk prediction of postoperative AKI in on-pump coronary artery surgery.

Development and Validation of a Personalized Model With Transfer Learning for Acute Kidney Injury Risk Estimation Using Electronic Health Records

“…The Table shows a comparison of the AUROC for the models in 20 well-studied AKI subgroups from the literature (eTable 13 in the Supplement ). 17 , 20 , 22 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 , 66 , 67 The personalized model with transfer learning was superior to each of the current models, significantly outperforming the global model in 16 subgroups, the subgroup model in 11 subgroups, and the subgroup model with transfer learning in 9 subgroups. For example, among patients older than 65 years, AUROC was 0.76 (95% CI, 0.74-0.77) for the personalized model with transfer learning, 0.73 (95% CI, 0.72-0.75; P < .001) for the global model, 0.71 (95% CI, 0.70-0.72; P < .001) for the subgroup model, and 0.73 (95% CI, 0.72-0.75; P < .001) for the subgroup model with transfer learning.…”