Journal Article

Relative hyperglycemia and new-onset atrial fibrillation after coronary artery bypass grafting: a focus on stress hyperglycemia ratio

Ning, Chengdong, Wei, Junying, Cao, Heshan, Cheng, Mingjin, Gui, Long, Zheng, Min

2025

Overview

Background New-onset atrial fibrillation (NOAF) after coronary artery bypass grafting (CABG) is associated with an increased risk of adverse outcomes. The stress hyperglycemia ratio (SHR), a novel biomarker reflecting relative hyperglycemia, has an undetermined role in NOAF risk assessment. This study aimed to evaluate the association between SHR and NOAF after CABG, and to further develop an interpretable machine learning (ML) model to validate its potential value in clinical risk stratification. Methods CABG patients were retrospectively identified from the Medical Information Mart for Intensive Care (MIMIC) database. Baseline characteristics from the first postoperative ICU day were extracted. The association between SHR and NOAF was assessed using logistic regression and restricted cubic spline methods. Feature selection was performed via multivariable logistic regression, the Boruta algorithm, and LASSO regression, with the intersection of selected variables used to construct 15 ML prediction models. Finally, the SHapley Additive exPlanation (SHAP) method was applied for interpretability analysis of the best-performing model, which was subsequently deployed as an online tool. Results A total of 3,498 eligible CABG patients were enrolled, with an NOAF incidence of approximately 18.6% (651 patients). Multivariate logistic regression analysis revealed that, after adjusting for confounding factors, SHR was independently associated with the risk of NOAF after CABG (OR = 1.39, 95% CI: 1.04–1.85, P  = 0.027). Restricted cubic spline analysis demonstrated a positive correlation between SHR and NOAF risk, with no significant non-linear relationship observed ( P for non-linearity > 0.05). The intersection of the three feature selection approaches identified eight key features, with SHR consistently emerging as an important predictor of NOAF. Among the 15 ML models developed, the XGBoost model exhibited the best performance with an area under the curve (AUC) of 0.813, while removing SHR reduced AUC to 0.779. SHAP analysis further confirmed the positive contribution of elevated SHR levels to NOAF risk. Conclusion This study demonstrates that SHR is significantly associated with NOAF after CABG, with a higher SHR serving as an independent risk factor. As a novel serological biomarker, SHR shows substantial clinical potential for early risk stratification and management optimization.

• Publisher: BioMed Central,BioMed Central Ltd,BMC
• Subject: Aged / Angiology / Atrial fibrillation / Atrial Fibrillation - diagnosis / Atrial Fibrillation - epidemiology / Biomarkers - blood / Blood Glucose - metabolism / Blood Transfusion Medicine / Cardiac Surgery / Cardiology / Complications / Complications and side effects / Coronary artery bypass / Coronary Artery Bypass - adverse effects / Coronary artery bypass grafting / Coronary Artery Disease - diagnosis / Coronary Artery Disease - epidemiology / Coronary Artery Disease - surgery / Databases, Factual / Diagnosis / Female / Humans / Hyperglycemia / Hyperglycemia - blood / Hyperglycemia - diagnosis / Hyperglycemia - epidemiology / Incidence / Internal Medicine / Machine Learning / Male / Medicine / Medicine & Public Health / Middle Aged / New-onset atrial fibrillation / Predictive Value of Tests / Reproducibility of Results / Retrospective Studies / Risk Assessment / Risk Factors / SHAP / Stress hyperglycemia ratio / Stress, Physiological / Surgery / Time Factors / Treatment Outcome