Latent Class Analysis of Health Risk Behaviors Among Adolescents

Thursday, April 25, 2013
Exhibit Hall Poster Area 2 (Convention Center)
Youngdeok Kim, Saori Ishikawa and Minsoo Kang, Middle Tennessee State University, Murfreesboro, TN

Background/Purpose: Physical activity and sedentary behaviors are associated with higher risk of obesity among adolescents. Little is known for the joint associations of physical activity and sedentary behaviors with obesity. The purpose of this study was to examine the relationships of the obesity with the latent sub-populations of adolescents based on their health risk behaviors.

Method: Data from a total of 13,827 adolescents who participated in Youth Risk Behaviors Survey (YRBS) during 2010-2011 were analyzed. Six health risk behaviors (i.e., regular participations of physical activity, strength exercise, physical education classes, sports team activity, and TV watching and computer usage) were used to predict the latent sub-populations. Latent class analysis (LCA) with the covariates (i.e., gender and grade) and a distal outcome of obesity (i.e., BMI ≥ 95th percentile by age and gender) was performed using Mplus v6.2. Complex sampling scheme of YRBS was accounted for all analyses.

Analysis/Results: LCA with three latent sub-populations showed relatively good data-model fits. The first latent sub-population (35.7%) consisted of adolescents who were highly physically active and less sedentary. The second sub-population (18.2%) were highly physically active and sedentary, and the third sub-population (46.1%) were modestly physically active and sedentary. The probability of being obese were 6.5%, 19.9%, and 14.8% for each sub-population.

Conclusions: The results clearly implied that reducing sedentary behaviors and increasing physical activities may reduce the risk of obesity among adolescents. Furthermore, sedentary behaviors were shown to be more crucial health risk behaviors related to obesity than physical activities.