Abstract 3790: The TIA Admission Score - A Tool to Determine Observation Versus Inpatient Admission Status for TIA
BACKGROUND: Centers for Medicare and Medicaid Services has criteria for admission to observation vs. inpatient status, one of which is expected discharge within 24 hours. Validated scores predicting the likelihood of stroke after TIA are useful in supporting admission, but do not assist with assigning admission status. We developed a scoring system that predicts which patients presenting with TIA would have a high probability of LOS >23 hours.
METHODS: Patients presenting to our stroke center from 7/08-12/10 with a diagnosis of TIA were identified from a prospective stroke registry containing demographic and clinical characteristics. A scoring system was created to estimate the probability of a LOS >23hrs based on patient demographics, past medical history, and current medication. Potential predictors of LOS >23hrs with sensitivity ≥65% or (area under the curve) AUC ≥0.6 were evaluated to determine if they demonstrated significant, independent associations with LOS >23hrs using logistic regression, adjusting for weekend admission. Sensitivity and specificity were used to evaluate the score’s ability to predict observation status (LOS ≤23hrs, score ≤2), as well as inpatient status (LOS >23hrs, score ≥4).
RESULTS: Out of 223 patients with TIA (52% F, 58% AA, median age 62) 110 (49%) had a hospital stay of >23 hours. Patients with LOS >23hrs were older (p=.036), more frequently had a history of coronary artery disease (CAD) (p<.001) and congestive heart failure (p=.059), and more frequently reported taking blood pressure (BP) medications (p=.008). No significant differences were observed in the two groups with regards to the remaining demographic and clinical variables. The TIA admission score ranges from 0-6, consisting of age >70yrs (2 points), presence of CAD (2 points), current use of BP medications (1 point), and being African American (1 point). As shown in Figure 1, 85% of patients with a score of 6 had a LOS>23hrs, as compared to 50% in patients with a score of 3, and 29% of patients with a score of 1. A score of ≤2 showed a sensitivity of 68% (specificity 79%) for identifying observation status, whereas a score of ≥4 showed a sensitivity of 79% (specificity 58%) for identifying inpatient status.
CONCLUSIONS: In our population of TIA patients, clinical characteristics available on presentation identified admission to observation and inpatient status with acceptable sensitivities and specificities. Validation is required in other populations. If validated, this score could assist providers in determining whether TIA patients should be admitted under observation or inpatient status.
- © 2012 by American Heart Association, Inc.