Abstract WP144: Predictors of In-hospital Speech Therapy Referral in Aphasic Acute Ischemic Stroke Patients Within a Population

Abstract

Background: Aphasia is a major cause of disability following ischemic stroke (IS). We sought to determine clinical characteristics of patients with aphasia that are associated with speech therapy evaluation (STE) in the acute setting within a large, biracial population. Methods: Adult IS patients in 2010 in the Greater Cincinnati/Northern KY region (population 1.3 million) were ascertained from all local hospitals via ICD-9 codes 430-436, using retrospective chart review. Aphasia was defined by symptoms or examination documented in the medical record. Demented patients were excluded. Logistic regression was used to determine associations between STE and clinical characteristics. We ran our model twice, including and excluding patients who died in-hospital or shifted to comfort/hospice care. Age and NIHSS were evaluated for potential nonlinear effects using spline functions. Results: In 2010, there were 1997 non-demented IS subjects, with 21% black, 54% female, and mean age of 68 (SD 15) years. Of these, 728 (36%) presented with aphasia, of whom 514 (71%) were assessed by ST. Age, race, gender, and academic hospital were not significantly associated with STE. In-hospital death and hospice referral were negative predictors (OR 0.12 and OR 0.19; both p<0.01); excluding these 202, 463 (79%) of the remainder were assessed. Factors predictive of STE in both models were stroke/neurology consult (OR 2.56 (95% CI 1.55, 4.21), p<0.01), dysarthria (OR 2.34 (1.46, 3.75), p<0.01), and NIHSS. NIHSS had a non-linear relationship, with moderate strokes most likely to be evaluated and mild and severe strokes less likely (Figure). Discussion: Of non-moribund IS patients presenting with aphasia, 1 in 5 did not receive STE during acute hospitalization. While moderate stroke patients had a high rate of STE, mild and severe strokes showed a bias towards less consultation. Our results suggest that limited ST resources are not always directed at those with potentially the greatest need.