ABSTRACT
Obesity is a pervasive public health problem with origins in childhood, and despite advances in treatment for
obesity, primary prevention is essential to prevent morbidity and early death. There is an urgent, unmet need to
predict which infants and young children are at highest risk of obesity. The first years of life provide a promising
sensitive period to address excess, rapid infant weight gain and subsequent obesity and cardiometabolic risk.
Previous work demonstrates associations between risk factors and rapid infant weight gain, but novel growth
characteristics, such as magnitude and timing of infancy BMI peak, may provide better insight into obesity risk
when combined with known risk factors. Primary care visits in the first years of life, with brief and frequent
contact with parents eager for guidance, provide an ideal setting for both risk prediction and interventions. The
overarching hypothesis is that childhood obesity prevention strategies in the clinical settings can be improved
through predictive models and clinical decision support designed to incorporate modifiable risk factors and
infancy growth patterns. The overall objectives of this proposal are to improve clinically-relevant prediction of
childhood obesity and design a clinical decision support tool that incorporates real-time risk prediction. Aim 1
identifies associations between known risk factors during infancy and novel infancy growth patterns. Aim 2
develops, compares, and validates predictive models using novel infant growth characteristics, including
individual and group-based patterns. Aim 3 tests implementation of a risk prediction and clinical decision
support tool within the electronic health record. The outlined research aims and career development plan
provides Charles Wood, MD, MPH the skills to achieve his overall career goal of becoming an independent
investigator focused on obesity prevention in the primary care setting focused on the first years of life. Dr.
Wood’s training plan includes experiential learning and didactic coursework to achieve the following short-term
training goals: 1) master approaches to repeated measures analysis and predictive modelling construction and
validation; 2) execute data linkage and harmonization using electronic health record (EHR) sources; 3) learn
and practice optimal use of EHR data for research; 4) incorporate state-of-the-art approaches to designing
clinical decision support tools; and 5) continue to develop career and professional skills. Dr. Wood will receive
focused mentorship and consultation from a team of experts of pediatric outcomes research (Dr. Smith),
primary care obesity interventions (Dr. Perrin), obesity and population health (Dr. Skinner), epidemiology of
childhood growth (Dr. Woo), repeated measures analysis (Dr. Kuchibhatla), and EHR use for research (Dr.
Goldstein). The rich research environment at Duke University will allow Dr. Wood to fulfill his research and
career development plans and begin to address his long-term goals of comparing risk prediction strategies for
childhood obesity and conducting prospective observational and interventional trials in the primary care setting.