Objectives: We evaluated the predictive performance of the stroke risk functions from the Framingham and the Cardiovascular Health Study in a population-based study in The Netherlands.
Methods: We assessed how well the functions could estimate the number of strokes in the Rotterdam Study, and how well they could discriminate between high- and low-risk persons.
Results: Both functions predicted the number of stroke cases reasonably well, except in persons with a predicted 5-year risk higher than 5%, where they overestimated the risk. The area under the receiver operator characteristic curve was 0.75 (95% confidence interval, 0.71-0.78) for both functions.
Conclusions: Stroke risk functions developed in US populations predict the number of strokes reasonably well in elderly Western European persons at low or modest risk for stroke. The functions can be a useful tool to discriminate between persons at high or low risk for stroke.