Abstract
PURPOSE: Delirium is an acute fluctuating impairment of attention and awareness, common in acute ischemic stroke (AIS). This study aimed to evaluate the prognostic significance of delirium for neurological function at 3 months post-stroke, and develop a predictive model integrating delirium and biomarkers to enhance prognostic accuracy. METHODS: We conducted a prospective cohort study of patients admitted to the stroke unit (n=722). All patients were screened for daily delirium during clinical care. Plasma biomarkers were measured within 24 hours after admission. The main outcomes were evaluated with the 3-months modified Rankin Scale (mRS). RESULTS: Delirium developed in 10.2% of patients during the acute phase of stroke. Patients with post-stroke delirium (PSD) was significantly older (median age 74 vs 68 years, P<0.001), more likely to have pre-stroke cognitive impairment (14.9% vs 4.8%, P=0.001), a higher prevalence of cardiovascular history (35.1% vs 16.2%, P<0.001). PSD was also associated with higher scores of NIHSS (14.3 vs 9.1, P<0.001) and greater scores of mRS (3.0 vs 1.5, P<0.001) at admission. PSD patients showed worse outcomes, with elevated NIHSS and mRS scores at discharge and 3-month follow-up, as well as higher mortality rates (5.4% vs 1.4%, P=0.025). Biomarker analysis revealed increased plasma levels of inflammatory (white blood cells, neutrophils, C-reactive protein) and coagulation biomarkers (fibrinogen, D-dimer) in PSD patients, particularly those with poorer outcomes (P<0.01). Our model, which incorporated delirium and biomarkers of inflammation and coagulation dysfunction, demonstrated strong predictive accuracy for adverse outcomes at 3 months with an AUC of 0.779 (95% CI=0.736-0.822), with clinical utility confirmed by decision curve analysis. CONCLUSION: PSD is a strong independent predictor of poor 3-month outcomes in AIS, including higher mortality and disability. Our findings highlight the critical role of inflammation and coagulation dysfunction in the pathogenesis of PSD. Furthermore, we present the clinical utility of a predictive model integrating delirium and relevant biomarkers to assess the risk of adverse outcomes at 3 months, suggesting potential targets for intervention.