Abstract 049: Derivation and Validation of a Lifestyle CVD Risk Score for the Prediction of CVD Among Middle-aged Women and Men
Background: The AHA has endorsed the use of risk prediction models to assess CVD risk in the primary prevention setting. Modifiable lifestyle factors, which play a key role in the primordial prevention of CVD, and may prevent over 80% of CHD and 50% of ischemic strokes, are not components of most existing prediction models. We derived and validated a CVD risk prediction model based on modifiable lifestyle factors among asymptomatic middle-aged women and men.
Methods and Results: We derived and validated a Lifestyle CVD Risk Score among 61,267 women in the Nurses’ Health Study (NHS) (mean age: 52 years) and 35,126 men in the Health Professionals Follow-up Study (HPFS) (mean age: 52 years) who were free of CVD, diabetes and cancer in 1986. Data on lifestyle factors were ascertained through questionnaires. CVD (CHD and ischemic stroke) was confirmed by medical records. Through 2008 (median follow-up: 21.9 years, women; 21.8 years, men), there were 2287 cases of CVD (68% CHD) in women and 2304 cases (82% CHD) in men. In the derivation step, we used the Bayes Information Criterion to create parsimonious 20-year risk prediction models, from Cox proportional hazards models, among a random 2/3 of participants in each cohort separately. First, a composite diet score was created from a linear combination of sugar-sweetened beverages [serv/d], red+processed meat [serv/d], cereal fiber [g/d], nuts [serv/d] and fruits+vegetables [serv/d] derived from multivariable models for incident CVD. The Lifestyle CVD Risk Score included diet score, smoking, BMI, moderate or vigorous exercise, alcohol intake and age. The score was validated in the remaining 1/3 of participants in each cohort, where it demonstrated good fit (p, lack of fit=0.62 women; 0.51 men), discrimination (Harrell’s C-Index: 0.72; 95%CI: 0.70, 0.73[women]; 0.76; 95%CI: 0.74, 0.77[men]) and calibration, based on calibration plots. Among 45-year olds with a healthy lifestyle (non-smokers, BMI=23.0, exercise=3.5 hr/wk; diet score=90th percentile; moderate alcohol intake), the 20-year risk of CVD was 1.01% in women and 1.35% in men. In comparison, 45-yr-old current smokers, with a BMI of 35.0, who reported no exercise, had a diet score in the 10th percentile, and drank no alcohol, had a 20-year risk that was 7-fold higher in women and 6-fold higher in men.
Conclusions: The Lifestyle CVD Risk Score is an innovative tool that estimates an individual’s 20-year risk for ischemic CVD and may be used to assess CVD risk in the primordial prevention setting. In combination with existing prediction models that focus on clinical risk factors, this score, based on modifiable lifestyle factors, may provide the most complete assessment of cardiovascular health, and help attain the AHA Impact Goal of improving cardiovascular health by 20% by 2020 by highlighting lifestyle factors that need improvement.
- © 2013 by American Heart Association, Inc.