Development and External Validation of a Predictive Tool for Early Neurological Deterioration in Acute Minor Vertebrobasilar Occlusive Stroke Managed Medically.

Abstract

We sought to construct a scoring model to predict early neurological worsening (END) among minor vertebrobasilar occlusive strokes (National Institutes of Health Stroke Scale score <6) following best medical management. Between January 2019 and June 2024, 137 patients were recruited from a single center. END was characterized by a ≥ 4-point rise in the National Institutes of Health Stroke Scale within 72 hours, without intracranial hemorrhage. Logistic regression analysis identified predictors for END, which were used to construct a scale. External validation was performed using a separate cohort from another hospital during the same study period (n = 81). Model performance was assessed using receiver operating characteristic curve analysis, area under the curve (AUC), calibration plots, and Brier score. END occurred in 41 patients (29.9%). Independent predictors for END were fluctuating symptoms (adjusted odds ratio [aOR] = 1.64, assigned 1 point), proximal vertebrobasilar occlusion (aOR = 2.59, assigned 2 points), and Tmax >6-second volumes (aOR = 1.87, assigned 1 point) (all P < 0.001). The predictive scale showed excellent discrimination in the derivation cohort (AUC = 0.921, 95% confidence interval: 0.882-0.965) and good calibration (Brier score = 0.103). The derivation and validation cohorts were similar in most baseline characteristics. Significant differences were observed in symptom evolution, intravenous thrombolysis usage, and occlusion sites. External validation in an independent cohort yielded an AUC of 0.857 (0.825-0.934) and a Brier score of 0.130. This externally validated scale provided a practical tool to identify high-risk END patients for targeted interventions.