Abstract 24: Predicting Persistent Upper Extremity Impairment in Ischemic Stroke Patients

Abstract

Background: Motor impairment after ischemic stroke is common and has variable recovery that depends on patient factors and stroke severity. An important challenge in rehabilitation research is enrolling patients who may benefit from interventions to improve recovery because they will not recover with standard care. Hypothesis: We can accurately predict persistent upper extremity motor impairment at 90 days after acute ischemic stroke. Methods: The primary outcome was an NIHSS arm domain score of 2-4 at 90 days in patients with a 24-hour NIHSS arm score of 1-4, which we termed persistent arm impairment. With the NINDS tPA trial dataset we used LASSO regression to select baseline demographics and 24-hour NIHSS score domains for a predictive model. We gave one point each for age ≥60 years and 24-hour NIHSS values of worst arm=4, worst leg>2, facial palsy=3, and total NIHSS≥10. The optimal cutpoint for a positive Persistent UPPer extremity Impairment (PUPPI) Score was ≥3 points. We validated the PUPPI score in the ALIAS Part 2, IMS-III, and DEFUSE 3 trials. Results: We included 431, 383, 331, and 71 patients from the NINDS tPA, ALIAS Part 2, IMS-III, and DEFUSE 3 trials. PUPPI was most common in the NINDS tPA trial (62%) and least common in ALIAS (31.3%). The PUPPI Score accurately predicted PUPPI with an area under the receiver operating curve (AUC) of >0.75 for all trials (Table 1). The positive predictive value was 74.6%, 90.9%, 86.1%, and 74.5% in the NINDS tPA, ALIAS Part 2, IMS-III, and DEFUSE 3 trials (Table 1). Conclusion: The PUPPI score uses readily available information to provide accurate prediction of patients who will have persistent upper extremity motor impairment at 90 days from stroke onset.