Associate Professor University of Arkansas for Medical Sciences Little Rock, Arkansas, United States
Background: From the 2003–04 through 2018–19 school years, 105,577 Arkansas children completed all body mass index (BMI) measurements at Kindergarten and Grades 2, 4, 6, and 8. Arkansas children have among the highest rates of obesity in the U.S., and during the 2018–19 school year (pre-COVID-19 pandemic), 15.8% of children in Kindergarten were experiencing obesity, with 80% expected to maintain this status through Grade 8. Objective: The objective is to identify trajectories of BMI growth from Kindergarten through Grade 8 to understand demographic and socioeconomic characteristics associated with trajectory assignment. Design/Methods: The standardized comparative measure was the BMI percentile (pscore) obtained from the CDC age by gender referent population curves. A series of latent class growth models (LCGM) were fitted with increasing numbers of non-linear growth trajectories. A final model was chosen when all trajectories had at least 5,000 children, had distinct separation, and intuitive latent trajectory labels could be assigned. A series of logistic regression models compared children assigned to pairs of trajectories with demographic (e.g., gender, race/ethnicity) and socioeconomic (e.g., free/reduced lunch status, census block group median income) characteristics included as covariates. Results: In total, a final LCGM contained eight distinct latent growth trajectories that exhibited good class separation (entropy=0.88) and better fit than a seven-trajectory model (Vuong-Lo-Mendell-Rubin likelihood ratio test: p< 0.001). The highest populated trajectory contains 38,883 children and is characterized by steady high BMI pscore measurements from Kindergarten through Grade 8. Comparatively, 7,678 children were assigned to a growth trajectory that has consistently low BMI pscores over the same five measurement periods. Other trajectories ranged from 6,554 and 17,033 in child assignment. Trajectory population differences were observed when comparing pairwise trajectories. For example, compared to White children, Black and Hispanic children are 2.9 (confidence interval [CI]=1.9-4.3) and 2.1 (CI=1.4-3.2) times as likely, respectively, to be assigned to the less favorable High BMI pscore trajectory than the Low BMI trajectory. Compared to children who do not receive a free or reduced price lunch, those who do are 1.2 (CI=1.1-1.3) times as likely to be assigned to the High over Low trajectory.
Conclusion(s): Distinct patterns of BMI growth are able to be determined in a population of children, and underlying demographic and socioeconomic reasons are associated with assignment to distinct growth trajectories.
