Abstract WP336: Patient Characteristics and in Hospital Complications Associated With High Cost Ischemic Stroke Admissions

Abstract

Introduction: Stroke poses a major public health burden. We sought to determine the clinical and demographic variables associated with high cost among hospitalized ischemic stroke (IS) patients. Methods: Using our local Get with the Guidelines Stroke database, we identified 1,578 IS patients admitted from 2010- 2013 and linked them to administrative claims data (EPSI, inc). Patients in the highest cost quartile (n=394) were compared to all others (n=1184) using descriptive statistics and multilevel logistic regression models. All financial data are relative costs, reported as medians [IQR] multiplied by a constant. Results: The median relative cost in the top quartile was 4 times higher than that for all other patients. In univariate analyses, the groups differed substantially (Table 1). In multivariable models, high cost patients were more likely to have discharge ICD 9 codes of 433.11 or 434.91 (IS patients with carotid or cerebral artery occlusion), higher serum creatinine, fasting blood glucose and NIHSSS. They were more likely to receive IV or IA reperfusion, remain NPO during their stay or develop hospital acquired pneumonia, and less likely to transition to comfort care. The C statistic for a model with NIH stroke scale only performed well (c= 0.77) even when compared to a model with all variables present on admission (c= 0.83) or the fully adjusted model (c= 0.86). Conclusion: Many patient level demographic and clinical characteristics available on admission predict high cost, even after adjustment for stroke severity. Cost management opportunities may exist for targeted interventions, perhaps through geographic co-location or specialized stroke units.