Initial Cardiovascular Disease Risk Assessment in Nonclinical Settings: Framingham-Simplified Model Predictive...

The Framingham risk model is recommended for estimating 10-year cardiovascular disease (CVD) risk in Canada. Its simplified version (Framingham-SP) replaces lipids with body mass index, eliminating the need for laboratory testing. The predictive performance of the Framingham-SP in Canada is unknown. This external validation study assessed the discrimination and calibration of the Framingham-SP model in a prospective cohort from Quebec, Canada.

The study sample included 3692 women and 3671 men free of CVD at baseline (1999-2001; follow-up until 2011). Incident CVD events were identified using medico-administrative databases. Multiple imputation was used to handle missing data. Discrimination was assessed using C-statistics. Calibration was evaluated using calibration-in-the-large, absolute difference, and discordance.

Over the 10-year follow-up, 211 CVD events occurred among women and 453 occurred among men. For women, the predicted 10-year CVD risk was 4.07% (95% confidence interval [CI]: 3.94-4.20), and the observed risk was 5.72% (95% CI: 5.01-6.51). The C-statistic was 0.72 (95% CI: 0.68-0.76). For men, the predicted risk was 11.73% (95% CI: 11.44-12.02), and the observed risk was 12.34% (95% CI: 11.32-13.44). The C-statistic was 0.72 (95% CI: 0.69-0.74). Overall, calibration was lower among women, particularly those in intermediate and high predicted risk categories.

The Framingham-SP model demonstrated acceptable predictive performance in this Canadian cohort. Its simplified format may help support preliminary CVD risk assessment and lifestyle counselling, including in nonclinical settings, while remaining complementary to guideline-based clinical evaluation. Recalibration may improve performance among high-risk women.