Date of Award


Document Type

Open Access Dissertation


Epidemiology and Biostatistics


The Norman J. Arnold School of Public Health

First Advisor

Jihong Liu


Childhood obesity is a serious public health challenge. The underlying causes behind the rising levels of childhood obesity might be driven by prenatal and early life factors. Recently, several studies have examined the association between gestational weight gain (GWG) and the risk of obesity in the pediatric population. Findings on the association of both inadequate and excessive GWG with offspring obesity are inconsistent and vary by maternal prepregnancy body mass index (BMI) status. Inconsistent findings also exist for the association of GWG and BMIZ as a continuous outcome. Nevertheless, existing studies mainly focus on the BMIZ upper centiles using logistic regression or the mean using linear regression, which do not capture associations across the entire distribution of BMIZ. Second, child growth in early life as a critical period contributing to lifetime health has been recognized in various populations. It has long been established that the increased growth following intrauterine growth retardation, known as catch-up growth, is associated with increased risks for obesity and insulin resistance. However, catch-up growth only affects a small fraction of all births and it does not accurately summarize the variations in the rates of infant growth in the population. Furthermore, some infants might experience different growth patterns which also predispose them to long-term health risks. In recent years, latent growth modelling approaches have received more attention due to advances in statistical software and analytical packages. This method is particularly useful to identify homogeneous subpopulations with similar growth patterns. Although BMI Z score is optimal for assessing a child’s static weight status in a single occasion, the best scales for measuring weight changes are raw BMI or BMI percentage. To our knowledge, so far few studies have used raw BMI or BMI percentage. Furthermore, infant growth could not happen in isolation and deviant BMI growth patterns during infancy might contribute to future risks of adverse health consequences, known as the hypothesis of the Developmental Originals of Health and Disease.

Therefore, we proposed three main aims in this dissertation and utilized data from a birth cohort of Infant Feeding Practices Survey study (2005-2007) and its Year Six Follow-Up Study to examine those aims. In Aim1, we examined the association between meeting the Institute of Medicine (IOM) GWG guidelines and offspring obesity at age six and the potential moderating role of maternal pre-pregnancy BMI status. We additionally examined association between GWG categories and offspring BMIZ across the deciles of BMIZ at age six and the potential moderating role of maternal pre-pregnancy BMI using quantile regression analysis which provided the estimates of interest beyond the mean. In Aim 2, we identified the underlying infant BMI trajectory using latent class growth analysis and examined its correlates including prenatal factors such as GWG, smoking during pregnancy and early life factors such as breastfeeding practices. Finally, in the third aim, we examined the association between the identified BMI trajectories during the first year of life and the risk of obesity at age six. Results from the first aim suggest that maternal pre-pregnancy BMI played an important moderating role on the association of meeting IOM GWG guidelines and obesity and BMIZ. Excessive GWG had an increased risk of childhood obesity at age six, and this positive association is more pronounced among mothers who have normal weight before pregnancy. Furthermore, we found that heterogeneous associations exist between GWG and BMIZ indicating that covariates might impact the associations differently across the distribution of BMIZ. In the second aim, we identified three BMI trajectories during infancy labelled as “low-stable” (81.6%), “high-stable” (15.6%), and “rising” (2.8%). Our findings suggest that distinct BMI trajectories are evident among children during the first year of life. Infants born to overweight mothers, minority mothers, and those who smoked during pregnancy had high-stable or rising BMI trajectories in early life and those who were breastfed according to guidelines were protected from being in the rising trajectory. Finally, we found that infants in the high-stable trajectory had an increased risk of obesity at age six.

This finding suggests that a child’s BMI trajectory during the first year of life provides additional information regarding his or her risk for obesity at school ages. Obesity prevention program should start as early as infancy and pay special attention to those children with sub-optimal growth trajectories in infancy. The findings from this dissertation suggest that both prenatal factors such as maternal weight before pregnancy and weight gain during pregnancy, and infant growth during their first year of life are critical factors to be considered for future obesity risk. Future studies are needed to warrant our findings and worth of exploring the underlying biological mechanisms.

Available for download on Tuesday, December 12, 2017

Included in

Epidemiology Commons