Stress Hyperglycemia Ratio Outperforms Glycemic Variability in Predicting Mortality Among Acute Myocardial Infarction Patients With Reduced Ejection Fraction: A Retrospective Cohort Study.

AIMS

Stress hyperglycemia ratio (SHR) and glycemic variability (GV) are established markers of glucose metabolism dysregulation. This study compared their predictive values for mortality and arrhythmic events in patients with reduced ejection fraction following acute myocardial infarction (AMI).

MATERIALS AND METHODS

We analyzed 1933 AMI patients with reduced ejection fraction from the MIMIC-IV database (v3.1, 2008-2022). The primary endpoint was in-hospital mortality, with secondary endpoints including 1-year all-cause mortality, ventricular tachycardia/ventricular fibrillation (VT/VF), and cardiac arrest. Multivariate logistic regression models evaluated associations with in-hospital outcomes, while Cox proportional hazards models assessed 1-year mortality. Restricted cubic spline models examined non-linear relationships between SHR and outcomes, with discriminative ability compared using area under the receiver operating characteristic curve (AUC) analysis.

RESULTS

Among patients (mean age 67.3 years), 401 (20.7%) died during hospitalization and 662 (34.2%) within one year. After adjustment, SHR showed the strongest association with in-hospital mortality (odds ratio [OR]: 1.51, 95% confidence interval [CI]: 1.24-1.82) compared to GV (OR: 1.00, 95% CI: 0.99-1.01). For 1-year mortality, SHR maintained superior performance (hazard ratio: 1.40, 95% CI: 1.19-1.65), with the highest tertile significantly associated with increased risk. ROC analysis confirmed SHR's superior predictive capacity for both mortality endpoints and VT/VF, while none of the indices significantly predicted cardiac arrest. SHR's predictive value was more pronounced in non-diabetic patients.

CONCLUSIONS

In post-AMI patients with reduced ejection fraction, SHR demonstrated superior predictive value for mortality compared to GV, supporting its incorporation into risk stratification models for individualized glucose management.