Citation
Salvia, Margaret G. (2024). Eating Behaviors in Adolescence and Young Adulthood and Adult Cardiometabolic Disease Risk.
Abstract
Cardiometabolic diseases, including type 2 diabetes (T2D), hypertension, and high cholesterol, are significant public health concerns. Understanding risk factors and mechanisms of disease progression can shape prevention and intervention efforts to improve health outcomes and reduce the burden of these diseases. Cardiometabolic diseases develop over the course of several years, and exposures and health behaviors earlier in life can impact risk in adulthood. Adolescence and young adulthood are potentially important developmental windows when dietary patterns and eating behaviors can impact long-term health trajectories. It is also a vulnerable time for the onset of disordered eating behaviors, which emerging research suggests may be associated with cardiometabolic disease risk. In the first project (Chapter 2), we evaluated prospective associations between binge eating patterns and type 2 diabetes, hypertension, and high cholesterol in the Growing Up Today Study cohort. In the second project (Chapter 3), we explored plasma metabolic profiles related to experiencing frequent binge eating behaviors in adolescence and constructed a metabolites-based score representing downstream phenotypes. We then examined associations between these profile scores with T2D risk. In the final project (Chapter 4), we evaluated a range of eating and health behaviors in machine learning models (Support Vector Machines, Random Forest, and XG Boost) that explicitly incorporated social determinants of health in the Add Health cohort. We operationalized three feature sets and assessed model performance in the classification of prevalent T2D, undiagnosed T2D, and the prediction of 10-year incident T2D. In the Growing Up Today Study cohort, 14.2% of participants reported binge eating with loss of control monthly or more often in adolescence and young adulthood. Ever binge eating was associated with an increased risk of impaired glucose tolerance and T2D (RR=1.93, 95% CI: 1.37, 2.73, p.001), and a dose-response relationship was observed, such that increased binge frequency was associated with higher risk. In analyses stratified by sex at birth, males were found to have a higher risk than females for hypertension and high cholesterol. Finally, sustained binge eating duration over several years was associated with higher risk of cardiometabolic disease. In the metabolomics analysis, a subset of 20 metabolites were identified, and several metabolites were involved in lipid metabolism, had known associations with T2D, or were markers of oxidative stress. The metabolites-based score constructed from this subset was positively associated with T2D risk; compared to individuals in the lowest quartile, those with the highest metabolites-based scores had a two-times increased risk of T2D (OR=2.03, 95% CI: 1.48, 2.81). In the Add Health study cohort, 407 (10.4%) of participants had T2D as of the fifth survey wave, though 35.1% of these individuals did not report receiving a diabetes diagnosis from a healthcare provider and were considered cases of undiagnosed T2D. We found XG Boost models tended to have the strongest performance in classifying T2D. In addition to health behaviors and eating behaviors, mental health and social determinants of health were found to be important factors influencing disease risk. This dissertation shows binge eating behaviors are associated with increased risk for cardiometabolic diseases. Oxidative stress and altered lipid metabolism may play a role. Health behaviors, mental health, and social determinants of health are important components of disease risk.
URL
https://dash.lib.harvard.edu/handle/1/37377933Reference Type
Thesis/Dissertation
Book Title
Population Health Sciences
Author(s)
Salvia, Margaret G.
Editor(s)
Hu, Frank
Series Author(s)
Mattei, Josiemer
Year Published
2024
Volume Number
Ph.D.
Publisher
Harvard University
Reference ID
10352
