Dietary information improves cardiovascular disease risk prediction models

Area of Science:

• Epidemiology
• Cardiovascular Disease Research
• Nutritional Science

Background:

• Limited data exist on cardiovascular disease (CVD) risk prediction models incorporating dietary predictors.
• Existing models often overlook the impact of diet on CVD risk assessment.

Purpose of the Study:

• To construct and evaluate CVD risk prediction models that integrate dietary information alongside traditional risk factors.
• To determine the added value of dietary predictors in improving the accuracy and reclassification of CVD risk.

Main Methods:

• Utilized data from prospective cohort studies involving 9026 adults (40-69 years) free of CVD and cancer at baseline.
• Employed Cox proportional hazard regression to build traditional, office-based, and diet-containing risk models.
• Evaluated models using Akaike information criterion (AIC), C-statistics, integrated discrimination improvement (IDI), and net reclassification improvement (NRI).

Main Results:

• Diet-containing models incorporating predictors like poultry, legumes, soft drinks, and green tea showed significant improvements.
• Adding dietary predictors reduced AIC, increased relative IDI by 38-53%, and improved NRI compared to traditional and office-based models.
• Calibration plots indicated better agreement for high-risk individuals when dietary predictors were included.

Conclusions:

• Dietary information is a valuable component for enhancing the accuracy of CVD risk prediction models.
• The integration of dietary data offers potential for more precise CVD risk assessment and targeted interventions.