Abstract
BACKGROUND: Glycemic variability (GV) is a key determinant of outcomes in critically ill patients, yet its impact on cerebral infarction patients in intensive care units (ICUs) remains underexplored. This study evaluates the association between GV and clinical outcomes, including discharge outcomes, 90-day and 1-year mortality, and ICU/hospital length of stay (LOS). METHODS: This retrospective study of 778 cerebral infarction patients from the MIMIC-IV database assessed GV, calculated as the glucose standard deviation-to-mean ratio during ICU stays. Regression models evaluated GV's impact on discharge outcomes, mortality, and ICU/hospital LOS, with adjustments for confounders. Restricted cubic spline analyses identified risk thresholds, while sensitivity and subgroup analyses validated findings. Predictive performance was assessed using AUC, NRI, and IDI, and multiple imputation methods addressed missing data. RESULTS: Higher GV was significantly linked to adverse outcomes. Patients in the highest GV quartile had increased risks of poor discharge outcomes (adjusted OR: 1.83; 95% CI: 1.03-3.32; P = 0.042), 90-day mortality (adjusted HR: 1.51; 95% CI: 1.03-2.22; P = 0.036), and 1-year mortality (adjusted HR: 1.53; 95% CI: 1.07-2.18; P = 0.018). RCS analysis identified critical GV thresholds (≥ 11% for 90-day and ≥ 10% for 1-year mortality). Subgroup analysis revealed stronger associations between GV and poor outcomes in non-diabetic patients (adjusted OR: 1.89; 95% CI: 1.24-2.88; P = 0.003) compared to diabetic patients (adjusted OR: 0.81; 95% CI: 0.53-1.25; P = 0.337). Sensitivity analyses confirmed the robustness of findings across imputation methods. CONCLUSIONS: GV independently predicts poor outcomes in ICU cerebral infarction patients. Integrating GV metrics into clinical workflows may improve risk stratification and guide interventions. Future research should validate these findings and explore strategies to reduce GV.