Date of Award
Doctor of Philosophy in Psychology
James O. Prochaska
Health risk behaviors (HRB) have been shown to impact non-communicable chronic disease (NCI) over the lifespan, which are among the leading causes of mortality within the United States. Treatment of NCI includes both pharmacological and behavior change intervention. Though behavior change recommendations are common in the treatment of NCI, our understanding of the long-term trajectory of HRB and how they cluster are not yet fully understood. The current study seeks to elucidate the clustering of specific health behavior changes over the lifespan, bolstering the potential for taking preventative measures as well as fostering positive health behavior change more broadly in a limited treatment setting. Data from the Framingham Heart Study were used to examine 4 health risk behaviors: tobacco use, alcohol misuse, fruit and vegetable intake, and physical activity. Methods: Data were analyzed using latent transition analysis to explore patterns of health behavior clustering as well as transition patterns of participants between clusters over time. Grouping variables of gender, age, and waist to height ratio were also included to explore potential moderating effects. Results: Three health behavior clusters were identified: unhealthy, healthy, and energy balance deficient. The unhealthy class had the highest likelihood of not meeting any of the guidelines for health behavior recommendations. Conversely, the healthy class had a high likelihood of meeting recommendations for all four HRB. The energy balance deficient class showed a high likelihood of meeting recommendations for alcohol and not smoking but were less likely than the healthy group to meet physical activity or fruit and vegetable recommendations. These clusters were consistent across timepoints while participant membership in clusters changed over time. Overall, participants were most likely to move into the healthy class from Time 1 to Time 2 and then most likely to move to the energy balance deficient class from Time 1 to Time 2 and then most likely to move to the energy balance deficient class from Time 2 to Time 3. Gender and age were predictive of cluster membership at Time 1 and waist to height ratio was not a good predictor of cluster membership at Time 1. Conclusions: Health behaviors tend to cluster and these clusters tend to appear consistently over time. People tend to change their health behaviors and often change more than one. Behaviors most likely to cluster are addictive behaviors such as drinking alcohol or smoking tobacco, while physical activity and dietary intake may also cluster together. This information provides insight into the potential for improving healthy lifestyle behaviors when multiple health risk behaviors are present. Personal information, such as demographic characteristics, can have important implications for health behavior, but waist to height ratio may not be a reliable indicator for health behavior assumptions or recommendations.
Baumann, Nathan L., "LONGITUDINAL CLUSTERING OF HEALTH BEHAVIORS WITHIN THE FRAMINGHAM HEART STUDY" (2022). Open Access Dissertations. Paper 1400.