WEIGHT GAIN PREVENTION IN YOUNG ADULTS: PREDICTORS OF WEIGHT CHANGE AND BEHAVIORAL TYPOLOGIES

This dissertation examined the role of four modifiable health behaviors (tobacco and alcohol use, diet, and physical activity) as predictors of weight change over a two-year period and identified distinct behavioral typologies among a population of young adults. This study used baseline and 24-month data from The Choosing Healthy Options in College Environments and Settings study, a randomized controlled trial designed to prevent unhealthy weight gain in young adults (aged 18 – 35) attending 2-year community colleges in the Twin Cities, Minnesota. Aim 1. A multivariate linear regression model was fit to understand the association between 24-month BMI and baseline health behaviors, controlling for covariates (n=365). To aid interpretation of the results, post-hoc analyses applied logistic modeling by dichotomizing the outcome variable to those who maintained/lost less than 3% weight and those who gained 3% weight between baseline and 24-months. Results suggested that binge drinking was significantly predictive of lower 24-month BMI in both models. Linear results suggested those who reported higher amounts of leisure physical activity and current smokers experienced an increased 24-month BMI. The latter associations did not hold in the logistic model. Individual behaviors assessed were generally non-significant, providing some evidence that behaviors may work synergistically, rather than independently, to promote weight change. Aim 2. This aim identified three typologies using Latent Class Analysis (n=441): Class 1: “active, binge-drinkers with healthy dietary intake,” Class 2: “non-active, moderate-smokers, and non-drinkers with poor dietary intake,” and Class 3: “moderately active, non-smokers, non-drinkers with a moderately healthy dietary intake.” Using adjusted multinomial logistic regression models to understand predictors associated with class membership, evidence suggested there were meaningful differences for BMI and age between latent classes. For each additional unit increase in BMI, individuals were more likely to be in Class 2 versus Class 3. For each additional year in age, individuals were more likely to be in Class 1 versus Class 2. Understanding these typologies may help public health professionals create more effective interventions in community colleges by targeting certain behaviors to help reduce the onset and improve management of existing chronic diseases.Doctor of Philosoph