The Latent Class Structure of Exercise and Eating Behavior

Abstract

The overall goal of this study is to inform community-based health promotion initiatives related to obesogenic behavior. Three aspects of behavior change (increasing fruit and vegetable intake, reducing fat intake, and increasing exercise) are considered simultaneously in this study. The relationship between chronic disease status and obesogenic behavior was also evaluated. Chronic diseases addressed in this study are hypertension, diabetes, and high cholesterol as these conditions require ongoing management of dietary and exercise behavior to prevent relapse and worsening. Using data from a county health department's adaptation of the Behavioral Risk Factor Surveillance System (n=7014), Latent Class Analysis (LCA) was employed to identify subgroups of individuals based on their stages of change for multiple eating and exercise behaviors. Latent class analysis (LCA) is a statistical technique used to identify subgroups of individuals where the grouping variable is not directly observed. The latent class structure of obesogenic behaviors was explored separately for each gender-race group, as different latent classes were expected to emerge in these subgroups. Covariates were then incorporated into the latent class model as grouping variables in order to assess the relationship between indicators of chronic illness and latent class membership. Eleven unique latent classes of exercise and eating behavior emerged from the race-gender latent class models. Latent class membership probabilities varied substantially by race and gender. Effects of chronic disease status, and overall health status also varied across race-gender groups. The LCA approach allows one to look at multiple obesogenic behaviors simultaneously. A latent class approach provides a vehicle by which public health practitioners can design and market community level interventions based on the behavioral characteristics of subgroups of the target population.