Predicting outcome of IV thrombolysis-treated ischemic stroke patients The DRAGON score

Objective: To develop a functional outcome prediction score, based on immediate pretreatment parameters, in ischemic stroke patients receiving IV alteplase. Methods: The derivation cohort consists of 1,319 ischemic stroke patients treated with IV alteplase at the Helsinki University Central Hospital, Helsinki, Finland. We evaluated the predictive value of parameters associated with the 3-month outcome and developed the score according to the magnitude of logistic regression coefficients. We assessed accuracy of the model with boot-strapping. External validation was performed in a cohort of 330 patients treated at the University Hospital Basel, Basel, Switzerland. We assessed the score performance with area under the receiver operating characteristic curve (AUC-ROC). Results: The DRAGON score (0-10 points) consists of (hyper) Dense cerebral artery sign/early infarct signs on admission CT scan (both = 2, either = 1, none = 0), prestroke modified Rankin Scale (mRS) score >1 (yes = 1), Age (>= 80 years = 2, 65-79 years = 1, <65 years = 0), Glucose level at baseline (>8 mmol/L [>144 mg/dL] = 1), Onset-to-treatment time (>90 minutes = 1), and baseline National Institutes of Health Stroke Scale score (>15 = 3, 10-15 = 2, 5-9 = 1, 0-4 = 0). AUC-ROC was 0.84 (0.80-0.87) in the derivation cohort and 0.80 (0.74-0.86) in the validation cohort. Proportions of patients with good outcome (mRS score 0-2) were 96%, 88%, 74%, and 0% for 0-1, 2, 3, and 8-10 points, respectively. Proportions of patients with miserable outcome (mRS score 5-6) were 0%, 2%, 5%, 70%, and 100% for 0-1, 2, 3, 8, and 9-10 points, respectively. External validation showed similar results. Conclusions: The DRAGON score is valid at our site and was reliable externally. It can support clinical decision-making, especially when invasive add-on strategies are considered. The score was not studied in patients with basilar artery occlusion. Further external validation is warranted. Neurology (R) 2012; 78: 427-432