Stroke Risk Prediction Models Do Not Perform as Well in Black Individuals as in White Individuals

Previously published models, this new pooled analysis, and machine-learning methods all demonstrated racial disparities in model performance.

Having unbiased and accurate methods to predict stroke risk is essential to optimize resource allocation and screening activities and to reduce the potential for under- and overtreatment. Previous models have relied on traditional regression techniques rather than newer machine-learning algorithms and have been tuned primarily to predict a composite cardiovascular outcome that includes stroke, rather than being stroke-specific or necessarily focused on subgroups by age, sex, and race. Therefore, researchers harmonized and pooled data from 4 large NIH-sponsored cohort studies spanning more than 3 decades: Framingham Offspring, Atherosclerosis Risk in Communities (ARIC), Multi-Ethnic Study for Atherosclerosis (MESA), and Reasons for Geographi…

Reviewing Author

Anthony S. Kim, MD

Disclosures

Grant/Research SupportNIH/National Institute of Neurological Disorders and Stroke; NIH/National Center for Advancing Translational Sciences

Anthony S. Kim, MD

Disclosures

Grant/Research SupportNIH/National Institute of Neurological Disorders and Stroke; NIH/National Center for Advancing Translational Sciences

Stroke Risk Prediction Models Do Not Perform as Well in Black Individuals as in White Individuals

Topics

Reviewing Author