Development and Validation of a Two-step Model to Predict Outcomes After Endovascular Treatment for Patients With Acute Ischemic Stroke

Background: Physicians and patients are eager to know likely functional outcomes at different stages of treatment after acute ischemic stroke (AIS). The aim of this study was to develop and validate a 2-step model to assess prognosis at different time points (pre- and posttreatment) in patients with AIS having endovascular thrombectomy (EVT). Methods: The prediction model was developed using a prospective nationwide Chinese registry (ANGEL-ACT). A total of 1676 patients with AIS who underwent EVT were enrolled into the study and randomly divided into development (n=1351, 80%) and validation (n=325, 20%) cohorts. Multivariate logistic regression, least absolute shrinkage and selection operator regression, and the random forest recursive feature elimination algorithm were used to select predictors of 90-day functional independence. We constructed the model via discrimination, calibration, decision curve analysis, and feature importance. Results: The incidence of 90-day functional independence was 46.3% and 40.6% in the development and validation cohorts, respectively. The area under the curve (AUC) for model 1 which included 5 pretreatment predictors (age, admission National Institutes for Health Stroke Scale score, admission glucose level, admission systolic blood pressure, and Alberta Stroke Program Early Computed Tomography score) was 0.699 (95% confidence interval [CI], 0.668-0.730) in the development cohort and 0.658 (95% CI, 0.592-0.723) in the validation cohort. Two treatment-related predictors (time from stroke onset to puncture and successful reperfusion) were added to model 2 which had an AUC of 0.719 (95% CI, 0.688-0.749) and 0.650 (95% CI, 0.585-0.716) in the development cohort and validation cohorts, respectively. Conclusions: The 2-step prediction model could be useful for predicting the functional independence in patients with AIS 90-days after EVT.