Does food reward predict changes in weight and body composition during multidisciplinary interventions in adolescents with obesity? Bayesian and frequentist meta-analyses.

Background: A predisposition to elevated food reward may hinder weight loss success during multidisciplinary interventions. However, this has not been consistently demonstrated in adults, nor at all in children.

Objective: To test the associations between explicit and implicit food reward and preference at baseline and prospective changes in weight and body composition in adolescents with obesity.

Methods: A meta-analysis of 6 clinical trials in adolescents with obesity was undertaken using frequentist and Bayesian linear mixed models. Participants from each study took part in similar 12-week multidisciplinary interventions. Liking and wanting for foods varying in fat content and sweet taste were assessed at enrolment, and both anthropometrics were tracked from enrolment to post-intervention.

Results: In a grand sample of N = 132 adolescents with obesity, liking and wanting for high-fat foods did not significantly predict changes in weight or fat mass. However, implicit wanting for sweet foods predicted changes in standardised body and lean mass, such that a greater wanting for sweet foods was associated with greater loss of body (p = 0.039, η²p = 0.05) and lean mass (p < 0.001, η²p = 0.13) in both frequentist and Bayesian analyses.

Conclusions: Baseline implicit wanting for sweet (high carbohydrate, low protein), but not energy-dense, foods may be more strongly related to prospective changes in lean mass than fat mass during weight loss in adolescents with obesity. Further research is needed to clarify whether low protein intake can account for this effect.