Key result
Elevated stress hyperglycemia ratio linked to ~45% higher 30-day mortality in ICU AF patients.
Why the study?
The stress hyperglycemia ratio correlates with poor outcomes in patients with AF, but its prognostic value for mortality across different glycemic phenotypes remained unclear.
Does an elevated stress hyperglycemia ratio predict 30-day all-cause mortality in critically ill patients with atrial fibrillation across different glucose metabolic states?
Cohort (n=4,365)
No
Does an elevated stress hyperglycemia ratio predict 30-day all-cause mortality in critically ill patients with atrial fibrillation across different glucose metabolic states?
Hazard Ratio: 1.45 (95% CI 1.15–1.82)
p-value: p=<0.001
In critically ill patients with atrial fibrillation, an elevated stress hyperglycemia ratio independently predicts 30-day all-cause mortality specifically in those with normal glucose regulation, rather than in those with prediabetes or diabetes.
Background The stress hyperglycemia ratio (SHR) reflects acute glycemic fluctuations and correlates with poor outcomes in patients with atrial fibrillation (AF). However, its prognostic value across different glycemic phenotypes remains unclear. This study aims to explore the heterogeneous predictive effects of SHR on mortality using interpretable machine learning techniques. Methods Patients with AF were screened in the Medical Information Mart for Intensive Care IV (MIMIC-IV) database and stratified by Glycated hemoglobin A1c (HbA1c) levels into three groups: normal glucose regulation (NGR, n = 1388), prediabetes (Pre-DM, n = 1156), and diabetes mellitus (DM, n = 1821). The study further categorized the SHR into tertiles, with 30-day all-cause mortality serving as the primary endpoint. Various statistical analyses, including Kaplan-Meier, landmark, restricted cubic spline, and Cox regression analyses, were performed. Additionally, sixteen machine learning algorithms, accompanied by SHapley Additive exPlanations (SHAP) interpretation, were utilized for the evaluation of risk factors. Results A total of 4,365 patients were included in the study, with a mean age of 73.1 ± 11.2 years, of which 62.1% were male. Overall, 501 patients (11.48%) died within 30 days of Intensive Care Unit (ICU) admission. Elevated SHR was found to predict mortality across all three subgroups in both unadjusted and partially adjusted models. After full multivariable adjustment, high SHR remained an independent risk factor solely in the NGR patients (Hazard Ratio [HR] = 1.88, 95% Confidence Interval [CI] 1.15–3.07, P = 0.011), with no significant association observed in the Pre-DM (HR = 1.27, 95% CI 0.78–2.07, P = 0.338) and DM subgroups (HR = 1.24, 95% CI 0.84–1.82, P = 0.275). Nonlinear associations between SHR and mortality were identified across all phenotypes, and machine learning analysis demonstrated good discrimination (Area Under the Curve [AUC] > 0.85), with SHAP analysis confirming that SHR carried greater prognostic weight in NGR individuals. Conclusions The SHR serves as a critical prognostic tool for AF mortality, offering enhanced predictive accuracy compared to individual metrics, particularly in patients with NGR. This integrated approach could inform personalized glycemic management strategies, potentially improving clinical outcomes.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2026) conducted a cohort in Atrial fibrillation in critically ill patients (n=4,365). Highest tertile of stress hyperglycemia ratio (SHR T3) vs. Lowest tertile of stress hyperglycemia ratio (SHR T1) was evaluated on 30-day all-cause mortality (HR 1.45, 95% CI 1.15-1.82, p=<0.001). Elevated stress hyperglycemia ratio independently predicted 30-day mortality in ICU atrial fibrillation patients overall (HR 1.45, P<0.001), driven by normoglycemic patients (HR 1.88, P=0.011).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: