Obesity-attributable risk of cardiovascular disease (CVD) in the United States: a Bayesian network analysis.

Background: Obesity and related complications present clinical and financial burdens for patients and health care systems. Quantification of the impact of obesity on cardiovascular disease (CVD) would aid in disease management. This study aimed to apply Bayesian networks to nationally representative data to analyze the relationship between obesity and CVD. Methods: Self-reported medical history, risk factor status, and physiological measurements were longitudinally adapted from the National Health and Nutrition Examination Survey (1999-2018). Respondents were required to have weight history and ≥1 CVD outcome, including myocardial infarction, stroke, heart failure, or overall CVD. The cohort was restricted to respondents ≥35 years to ensure recent weight measurements were sufficiently captured. Population attributable fractions were estimated via the traditional approach (assessing direct effects of exposures on outcomes) and Bayesian networks (assessing indirect effects of interdependent relationships) for each CVD outcome. Results: Obesity was reported in 31 %-32 % of respondents with any CVD outcome (n = 29,388-29,529). CVD outcomes attributed to obesity through the traditional approach were estimated to be <4 %. Using Bayesian networks, estimates of myocardial infarction, stroke, heart failure, and CVD cases attributable to obesity were 21 %, 16 %, 38 %, and 19 %, respectively. For all Bayesian networks, >80 % of the total effect of obesity was attributable to its effect on intermediate medical conditions. Conclusions: These data highlight obesity as a significant CVD risk factor. This work has provided valuable insight into the relationship between obesity and CVD-related outcomes, indicating a need for greater clinical attention to obesity to lessen the burden of downstream complications.