External Validation of the ASTRAL and DRAGON Scores for Prediction of Functional Outcome in Stroke

• DOI: 10.1161/strokeaha.116.012802
• Publication Date: 2016-05-13T03:20:00Z
• Authors: Charith Cooray; Michael Mazya; Matteo Bottai; Laura Dorado; Ondrej Skoda; Danilo Toni; Gary A. Ford; Nils Wahlgren; Niaz Ahmed
• Keywords: Male; calibration; prognosis; registries; ROC curve; stroke; Cardiology and Cardiovascular Medicine; Neurology (clinical); Advanced and Specialized Nursing; Brain Ischemia; 03 medical and health sciences; 0302 clinical medicine; Predictive Value of Tests; Reference Values; Humans; Thrombolytic Therapy; Prospective Studies; Registries; Aged; registries; Reproducibility of Results; Recovery of Function; Middle Aged; calibration; stroke; ROC curve; 3. Good health; Stroke; Treatment Outcome; ROC Curve; Area Under Curve; Calibration; Female; prognosis; Algorithms

Background and Purpose— ASTRAL (Acute Stroke Registry and Analysis of Lausanne) and DRAGON (includes dense middle cerebral artery sign, prestroke modified Rankin Scale score, age, glucose, onset to treatment, National Institutes of Health Stroke Scale score) are 2 recently developed scores for predicting functional outcome after acute stroke in unselected acute ischemic stroke patients and in patients treated with intravenous thrombolysis, respectively. We aimed to perform external validation of these scores to assess their predictive performance in the large multicentre Safe Implementation of Thrombolysis in Stroke-International Stroke Thrombolysis Register.

Methods— We calculated the ASTRAL and DRAGON scores in 36 131 and 33 716 patients, respectively, registered in Safe Implementation of Thrombolysis in Stroke-International Stroke Thrombolysis Register between 2003 and 2013. The proportion of patients with 3-month modified Rankin Scale scores of 3 to 6 was observed for each score point and compared with the predicted proportion according to the risk scores. Calibration was assessed using calibration plots, and predictive performance was assessed using area under the curve of the receiver operating characteristic. Multivariate logistic regression coefficients for the variables in the 2 scores were compared with the original derivation cohorts.

Results— The ASTRAL showed an area under the curve of 0.790 (95% confidence interval, 0.786–0.795) and the DRAGON an area under the curve of 0.774 (95% confidence interval, 0.769–0.779). All ASTRAL parameters except range of visual fields and all DRAGON parameters were significantly associated with functional outcome in multivariate analysis.

Conclusions— The ASTRAL and DRAGON scores show an acceptable predictive performance. ASTRAL does not require imaging-data and therefore may have an advantage for the use in prehospital patient assessment. Prospective studies of both scores evaluating the impact of their use on patient outcomes after intravenous thrombolysis and endovascular therapy are needed.