Admission glucose level and clinical outcomes in the NINDS rt-PA stroke trial

Background: Hyperglycemia during acute ischemic stroke may augment brain injury, predispose to intracerebral hemorrhage (ICH), or both. Method: To analyze the relationship between admission glucose level and clinical outcomes from acute ischemic stroke, the authors performed multivariate regression analysis with the National Institute of Neurological Disorders and Stroke recombinant tissue plasminogen activator (rt-PA) Stroke Trial data. Neurologic improvement was defined as improvement on the NIH Stroke Scale by 4 or more points from baseline to 3 months, or a final score of zero. Favorable outcome was defined as both Glasgow Outcome score of 1 and Barthel Index 95 to 100 at 3 months. Symptomatic ICH was defined as CT-documented hemorrhage temporally related to clinical deterioration within 36 hours of treatment. Potential confounding factors were controlled, including acute treatment (rt-PA or placebo), age, baseline NIH Stroke Scale score, history of diabetes mellitus, stroke subtype, and admission blood pressure. Results: There were 624 patients enrolled within 3 hours after stroke onset. As admission glucose increased, the odds for neurologic improvement decreased (odds ratio [OR] = 0.76 per 100 mg/dL increase in admission glucose, 95% CI 0.61 to 0.95, p = 0.01). The relation between admission glucose and favorable outcome depended on admission mean blood pressure (MBP): as admission MBP increased, the odds for favorable outcome related to increasing admission glucose levels progressively decreased (p = 0.02). As admission glucose increased, the odds for symptomatic ICH also increased (OR = 1.75 per 100 mg/dL increase in admission glucose, 95% CI 1.11 to 2.78, p = 0.02). Admission glucose level was not associated with altered effectiveness of rt-PA. Conclusions: In patients with acute ischemic stroke, higher admission glucose levels are associated with significantly lower odds for desirable clinical outcomes and significantly higher odds for symptomatic ICH, regardless of rt-PA treatment. Whether this represents a cause and effect relationship remains to be determined.