Abstract

Background: The Predicting Risk of CVD EVENTs (PREVENT) equations are a new set of sex-specific and race-free equations developed by the American Heart Association to calculate risk of cardiovascular disease, including stroke as an independent outcome. Objective: To describe the ten-year risk of stroke (ischemic and hemorrhagic) using the PREVENT equation and assess its validity in predicting stroke among racially and ethnically diverse adults. Methods: The Northern Manhattan Study is a community-based observational cohort of male and female participants who are of non-Hispanic White (NHW), non-Hispanic Black (NHB), and Hispanic background. Participants were aged 40 years or older and without a history of stroke at the baseline visit between 1993 and 2001. Information was collected on blood lipids, systolic blood pressure, body mass index, estimated glomerular filtration rate, smoking status, diabetes status, and anti-hypertensive or lipid lowering medication use. Using the base model PREVENT equation for stroke, we calculated the ten-year risk of stroke and categorized participants based on levels of risk. Participants were followed and stroke events occurring within ten years were adjudicated by an independent panel of clinicians. We calculated the stroke discordance rate by comparing the predicted to observed number of strokes overall, by sex, and by race/ethnicity. Results: Among 2,229 participants, mean age was 65 years, with 61% female, 17% NHW, 20% NHB, 60% Hispanic, and 2% Other. Overall, 25% had a <2.5% risk of stroke; 34% had a 2.5 to <5.0% risk of stroke, 25% had a 5.0 to <7.5% risk of stroke, 11% had a 7.5 to <10.0%, and 5% had a ≥10% risk of stroke. Men compared with women and NHB compared with NHW adults had a higher risk of stroke, p<0.05. Over ten years, 106 strokes were predicted, and 136 strokes were observed (discordance rate: 22%). Stroke was more likely to be underpredicted among females (discordance rate: 24%) than males (discordance rate: 20%), p<0.01 and among NHB (discordance rate: 42%) or Hispanic adults (discordance rate: 18%) compared with NHW adults (discordance rate: 0%), p <0.01. Conclusion: In this sample of urban dwelling adults, the base model PREVENT equation for stroke was most accurate in predicting stroke among NHW adults but performed with less accuracy among NHB and Hispanic adults. Additional factors, such as the social deprivation index, may be important for estimating stroke risk in NHB and Hispanic populations.