Association of life-course body mass index with cardiovascular disease risk: differential contributions of cardiometabolic and behavioral pathways in the CARDIA study

Background Body mass index (BMI) is associated with cardiovascular disease (CVD) risk and has been robustly linked to multiple established behavioral and biological cardiometabolic risk factors. The importance of a life-course perspective on BMI and related cardiometabolic factors in population health research is increasingly recognized. Yet, estimates from well-defined population-based longitudinal cohorts with several decades of follow-up during adulthood remain limited. Moreover, the pathways linking excess body weight to CVD risk are multifactorial and may differ across the BMI spectrum. Purpose To evaluate the continuous association between BMI and CVD risk and examine how accounting for individual cardiometabolic risk factor trajectories alters its observed shape. Methods We analyzed participants from the Coronary Artery Risk Development in Young Adults (CARDIA) study, a prospective U.S. population-based cohort initiated in 1985–1986 with up to 10 repeated assessments of BMI, moderate-to-vigorous-intensity physical activity (MVPA), and cardiometabolic risk factors through 2020–2022. The analytic sample included 4,942 participants aged 18–30 years at entry with ≥3 study visits (55% women; 51% Black participants). Fatal and nonfatal CVD events (coronary heart disease, heart failure, or stroke) were ascertained through August 31, 2022. A Bayesian multivariate joint modeling framework was applied, combining linear mixed-effects models for longitudinal trajectories of BMI, MVPA, LDL cholesterol, fasting glucose, and systolic blood pressure with Cox proportional hazards models for time to incident CVD. Flexible nonlinear associations across the BMI distribution were accommodated using spline terms. Survival models were identical except for the inclusion of individual cardiometabolic factor trajectories in addition to BMI. Results During a median (IQR) follow-up of 36.8 (35.7–37.5) years, 427 incident CVD events occurred. Higher BMI was associated with higher CVD risk in a monotonic but nonlinear manner, with a steeper gradient at lower BMI levels (Figure). Continuous BMI-specific attenuation curves demonstrated that attenuation of the BMI–CVD association after adjustment for various cardiometabolic factors varied across the BMI range. At lower BMI values, MVPA adjustment resulted in near-complete attenuation, whereas at higher BMI values, adjustment for systolic blood pressure and fasting glucose accounted for most of the attenuation. Simultaneous adjustment for all measured factors nullified the association across the BMI spectrum (Figure and Table). Conclusion Across the transition from young adulthood to late midlife, the association between BMI and CVD risk operates mainly through its interrelationships with cardiometabolic risk factors, with physical activity predominating at lower BMI values and blood pressure and glycemia playing a central role at higher BMI values.