Prognostic Modeling for an Efficacy and a Safety of Thrombolysis in Acute Ischemic Stroke

ABSTRACT

BACKGROUND: The aims of this study were to develop and internally and externally validate a prognostic model that can predict the benefit and harm of thrombolysis in patients with acute ischemic stroke and that may be used promptly in an emergency setting. METHODS: The data of a consecutive series of patients who were hospitalized to Seoul National University Bundang Hospital within 12 hours of stroke onset between January 2004 and March 2008 and with relevant ischemic lesions on diffusion-weighted MRI were used to develop and internally validate the prognostic model. The external validation was performed using the data of patients from five participating centers of the Clinical Research Center for Stroke that had been collected between April 2008 and September 2009. The score on the modified Rankin Disability Scale at 3 months was selected to determine the efficacy outcome, and the occurrence of symptomatic hemorrhagic transformation was used to evaluate the safety outcome. Prognostic models were constructed with logistic regression, and both internal and external validations were performed. RESULTS: The discriminative abilities of the efficacy model (C statistic=0.880) and the safety prognostic model (C statistic=0.864) were confirmed. External validation of both models revealed remarkably little degradation in the discrimination power (C statistic=0.835 and 0.822 for the efficacy and safety models, respectively). CONCLUSIONS: This study shows that the efficacy and safety prognostic models developed with basic clinical variables were reliably validated with independent data. Both models may be helpful to clinicians in the emergency setting to identify patients who would benefit from thrombolysis.