Abstract

Introduction: Various stroke scales exist to predict outcomes after acute ischemic stroke (AIS). We propose a simplified novel scale based on individual variables previously used in other scales to predict outcome after AIS. Methods: 166 consecutive patients presented with AIS between 2015 and 2018 who received mechanical thrombectomy. We collected the following variables: age, sex, NIHSS, HTN, diabetes, atrial fibrillation, stroke subtype, CHF, cancer, renal dialysis, preadmission disability, systolic blood pressure, 3-item stroke severity scale, last known well (LKW) to tPA administration, LKW to ED arrival, use of thrombolytic therapy, presence of visual field defect, level of consciousness, and visible hypodensity on CT (none, <1/3 MCA territory, or ≥1/3 MCA territory). All other missing data was addressed using multiple imputation. Logistic regression analysis was performed to find their association with poor prognosis (mRS 2 or greater). Because we desired a relatively simple model for clinical use, we fit models using the least absolute shrinkage and selection operator (LASSO) to both select important predictors and prevent overfitting. Results: The LASSO selected sex, age, 3-item stroke severity, and visible hypodensity on a CT to be in the predictive model. Females, older individuals, more severe strokes, and patients with a visible hypodensity covering ≥1/3 of the MCA territory were found to be associated with increased risk of a 90 day MRS of 2 or greater. This model had reasonable discriminative ability in line with many of the other stroke scores studied (internally validated AUC = 0.69, 95% CI [0.55, 0.72]). A risk threshold of 0.78 maximized the sum of specificity (0.90) and sensitivity (0.52). This model was further discretized into an easy to use risk score (Table 1). Conclusion: We propose a simplified novel stroke prognostication scale with high specificity in predicting poor prognosis in stroke patients receiving mechanical thrombectomy.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call