Association between stress hyperglycemia ratio and post-stroke delirium in critically ill patients with ischemic stroke: an MIMIC-IV database analysis.

BACKGROUND: Post-stroke delirium (PSD) is a common complication following cerebrovascular accidents and correlates with adverse clinical outcomes. The stress hyperglycemic ratio (SHR) is a recognized prognostic marker for cerebrovascular diseases, but its predictive value for PSD remains uncertain. This research aimed to explore the association between SHR and PSD in critically ill patients with ischemic stroke (IS). METHODS: Data were obtained from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database. PSD assessment was performed via the Confusion Assessment Method for the Intensive Care Units (CAM-ICU). Participants diagnosed with IS were categorized into four quartile groups (Q1-Q4) according to SHR levels. We evaluated the relationship between SHR and PSD using logistic regression modeling and restricted cubic spline (RCS) analysis. Subgroup analyses were conducted to validate the consistency. RESULTS: Among 1389 patients (mean age 70.53 ± 14.96 years, 50.04% male), 30.60% developed PSD. The incidence of PSD increased with higher SHR quartiles (Q4: 44.83% vs. Q1: 20.50%, P < 0.001). Multivariate logistic analyses indicated that SHR was independently associated with PSD both as a continuous variable (OR 1.69, 95% CI 1.01-2.86; P = 0.047) and a quartile-based variable (Q4: OR 1.64, 95% CI 1.07-2.54; P = 0.024). The RCS curves showed a linear association between SHR and PSD development (P for non-linearity = 0.286). Subgroup analyses suggested potential interactions between SHR and age, gender, and sepsis status (all P for interaction < 0.05). CONCLUSIONS: Elevated SHR levels are independently associated with an increased risk of PSD in critically ill patients with IS, highlighting its role as a possible prognostic indicator. Early management in high-risk individuals may enhance clinical outcomes. Additional prospective studies are necessary to validate these results.