Stress hyperglycemia ratio improves 28-day mortality prediction beyond GRACE score in critically ill patients with acute coronary syndrome: a retrospective cohort study from MIMIC-IV

BACKGROUND: The stress hyperglycemia ratio (SHR) is increasingly recognized as a reliable biomarker for adverse outcomes. However, its prognostic value specifically for the standardized endpoint of 28-day mortality in critically ill patients with ACS, and whether it provides incremental predictive power over the Global Registry of Acute Coronary Events (GRACE) score, remains unclear. METHODS: This retrospective cohort study analyzed 2,763 critically ill ACS patients from the MIMIC-IV database. Multivariable Cox regression evaluated the association between SHR and 28-day mortality. RCSs were used to assess the dose-response relationship. The incremental value of adding SHR to the GRACE score was quantified by the AUC, NRI, and IDI. Machine learning algorithms with SHAP interpretation were employed for complementary validation. RESULTS: Elevated SHR was independently associated with increased 28-day mortality (adjusted HR 1.60, 95% CI: 1.30-1.96, P < 0.001) after adjustment for confounders. Spline analysis revealed a continuous linear relationship. Integrating SHR yielded an improvement in the predictive discrimination of the GRACE score (AUC increased from 0.796 to 0.816, P < 0.001). More importantly, it optimized risk stratification, as evidenced by significant improvements in reclassification metrics (NRI 0.037, IDI 0.023; both P < 0.001). Machine learning analysis ranked SHR among the top 10 predictors. CONCLUSIONS: SHR serves as an independent predictor of 28-day mortality in critically ill ACS patients. By providing incremental prognostic value to the GRACE score, SHR facilitates refined risk stratification, offering a practical and accessible tool for the early identification of high-risk individuals in the ICU.