Background: Many childhood obesity intervention studies report mean outcomes but do not explore the variation in responses and the characteristics of those who respond well.
Objective: To identify child and family characteristics associated with improvement in the primary outcome, %BMIp95 , of the Texas Childhood Obesity Research Demonstration project (TX-CORD).
Methods: The 12-month TX-CORD secondary prevention study randomized 549 children, ages 2 to 12 years, with BMI ≥85th percentile to the intensive intervention vs. the comparison program, with measurements at baseline, 3-, and 12-months. A growth mixture model was used to identify mutually exclusive latent %BMIp95 trajectories. Latent class regression tested associations between baseline characteristics and latent class membership.
Results: A 2-class solution emerged after accounting for the effect of intervention randomization. Latent Class 1 participants (86% of sample) were characterized by mild-to-moderate obesity and demonstrated a significantly greater response to the intensive intervention between 0 and 3 months (slope-on-group = -0.931, p = 0.03). A rebound between 3 and 12 months was not significantly different between arms. Latent Class 2 participants (14%), who had severe obesity, demonstrated no difference in response between intervention groups. Characteristics associated with Class 1 membership included younger age (2-5 years vs. 6-12 years: OR 3.70, p = .035) and lower maternal BMI category (< 35 kg/m2 vs. ≥ 35 kg/m2 : OR 7.14, p < .0001).
Conclusions: The optimal target population for the intensive intervention are children who have milder obesity, are younger, and do not have a mother with severe obesity. Children with severe obesity may require different approaches.
Trial registration: ClinicalTrials.gov NCT02724943.
Keywords: BMI percent of the 95th percentile; childhood obesity; community weight program; growth mixture modeling; maternal obesity; weight management intervention.
© 2020 World Obesity Federation.