Abstract
Cerebral hemorrhage is a major cause of mortality and disability. This study investigates the association between glycemic variability (GV) and short- and long-term clinical outcomes such as poor outcomes at discharge, mortality at 90 days and 1 year and intensive care unit (ICU) /hospital length of stay (LOS) in ICU patients with critically ill cerebral hemorrhage. This retrospective analysis examined 732 ICU patients with non-traumatic cerebral hemorrhage from the Medical Information Mart for Intensive Care (MIMIC)-IV database. GV was quantified as the ratio of standard deviation to mean glucose during ICU stay. To assess associations between GV and clinical outcomes (poor outcomes at discharge, 90-day and 1-year mortality, ICU/hospital LOS), the study employed logistic regression, Cox proportional hazards models, and linear regression. Additionally, non-linear relationships were explored through restricted cubic spline analysis. The investigation further incorporated subgroup and sensitivity analyses to ensure robustness of findings. To evaluate the incremental predictive value of GV, the study utilized receiver operating characteristic (ROC) curve analysis, net reclassification improvement, and integrated discrimination improvement, thereby providing a comprehensive assessment of GV's clinical utility. Higher GV was significantly associated with increased risk of poor outcomes at discharge and 90-day and 1-year mortality in both patient groups. GV showed a linear association with poor outcomes at discharge but a non-linear association with 90-day and 1-year mortality. GV thresholds of ≥ 0.11 for non-traumatic cerebral hemorrhage increased mortality risks. Cohort showed non-linear relationships between GV and ICU/hospital LOS. GV's impact was stronger in non-hypertensive and male patients. Adding GV to existing severity scores improved their predictive ability for adverse outcomes. In patients with non-traumatic cerebral hemorrhage admitted to the ICU, GV demonstrates an independent association with poor outcomes over both short-term and long-term time horizons. Furthermore, GV is associated with extended durations of both ICU and overall hospital stays in this patient population. These findings underscore the importance of glycemic control in this patient population and suggest that GV could be a valuable prognostic indicator and potential therapeutic target.