Association of stress hyperglycemia ratio with 1-year all-cause mortality in patients with acute myocardial infarction and chronic kidney disease: a retrospective two-cohort study

Patients with acute myocardial infarction (AMI) complicated by chronic kidney disease (CKD) represent a high-risk population in whom early risk stratification remains challenging. The stress hyperglycemia ratio (SHR) reflects acute glycemic stress independent of chronic glycemic status; however, its independent prognostic value in this population has not been systematically evaluated in externally validated cohorts. This retrospective two-cohort study included patients with AMI and CKD from the MIMIC-IV database and an independent external coronary care unit cohort. The primary endpoint was 1-year all-cause mortality. Multivariable Cox proportional hazards models assessed the SHR-mortality association. Incremental predictive value was evaluated using discrimination, reclassification, and decision curve analyses. Machine learning models were additionally developed and externally validated for exploratory risk prediction. We included 606 patients from MIMIC-IV and 263 from the external cohort. In fully adjusted Cox models, higher SHR was independently associated with 1-year mortality (HR 1.14, 95% CI 1.03–1.26), with a graded association across quartiles (highest vs. lowest: HR 1.64, 95% CI 1.17–2.32). The prognostic effect was most pronounced during the early post-admission phase. Adding SHR to the baseline model yielded a non-significant improvement in overall discrimination (C-index 0.699 to 0.706; ΔAUC 0.008, P = 0.058), but improved correct reclassification of non-events. The XGBoost model achieved strong internal discrimination (AUC 0.872) but demonstrated attenuated external calibration (AUC 0.718; Brier score 0.253; calibration slope 0.298), underscoring the need for cohort-specific recalibration. Elevated SHR is independently associated with 1-year all-cause mortality in patients with AMI and CKD, particularly during the early post-admission phase. As a readily calculable index of acute metabolic stress, SHR may serve as a clinically accessible adjunctive biomarker to complement early risk stratification in this high-risk population, pending prospective validation and local model recalibration prior to broader clinical implementation.