Abstract 5: Do Stroke Centers Reduce Mortality for Patients with Acute Ischemic Stroke? An Instrumental Variable Analysis
Background: Although the establishment of stroke centers based on the Brain Attack Coalition recommendations has great potential to improve quality of stroke care, little is known about whether stroke centers improve health outcomes such as mortality. Methods: Using 2005-2006 New York State Statewide Planning and Research Cooperative System data, we identified 32,783 hospitalized patients age 18+ with a principal diagnosis of acute ischemic stroke (ICD-9 433.x1, 434.x1 and 436). We compared in-hospital mortality and up to one year all-cause mortality between New York State Designated Stroke Centers and non-stroke center hospitals. Because patients were not randomly assigned to hospitals, stroke centers might treat different types of patients than other hospitals (a selection effect). We used a “natural randomization” approach, instrumental variable analysis (differential distance was the instrument), to control for this selection effect. To determine whether the mortality difference was specific to stroke care, we repeated the analysis using a different group of patients with gastrointestinal (GI) hemorrhage (N=53,077). Results: Of the 32,783 stroke patients, nearly 50% (16,258) were admitted to stroke centers. Stroke centers had lower unadjusted in-hospital mortality and 30-, 90-, 180-, and 365-day all-cause mortality than non-stroke centers (7.0% vs. 7.8%, 10.0% vs. 12.6%, 14.6% vs. 17.5%, 18.0% vs. 21.0%, 22.4% vs. 26.2%, respectively). After adjusting for patient and hospital characteristics, comorbidities, and the patient selection effect, stroke centers were associated with significantly lower all-cause mortality. The adjusted differences were -2.6%, -2.7%, -1.8%, and -2.3% for 30-, 90-, 180- and 365-day mortality (all p<0.05). The adjusted difference in in-hospital mortality was -0.8% but was not statistically significant. In a specificity analysis of patients with GI hemorrhage, stroke centers had slightly higher mortality. Conclusions: Hospitals that are Designated Stroke Centers had lower mortality for acute ischemic stroke than non-stroke center hospitals. The mortality benefit was specific to stroke and was not observed for GI hemorrhage. Providing stroke centers nationwide has the potential to reduce mortality.