A Meta-Analytic Review of the Youth Fit For Life Intervention for Effects on Body Mass Index in 5- to 12-year-old Children

Abstract

A meta-analytic review of 16 studies of the Youth Fit For Life obesity prevention intervention found an overall significant body mass index (BMI) reduction effect (r=.07) in the intended 5- to 12-year-old age range. Possible moderation of effects on BMI by participants' age, ethnicity, gender, intervention administration format, and publication status was tested. Only ethnicity was found to be a significant moderator, with minority participants having significantly larger effects than White participants. The overall effect size was larger than overall effects on BMI in a recent comprehensive meta-analysis of 64 child and adolescent obesity prevention intervention studies (r=.04) (Stice, Shaw, & Marti, 2006), and overall effects in a subset of 45 of those studies with ages consistent with the Youth Fit For Life studies. In that case, the effect size (r=.01) was not significantly different from zero. When moderators were contrasted between meta-analyses, Youth Fit For Life studies had larger effects for participants who were ages 8 through 12 years, and when physical activity participation was mandated. Implications for child obesity prevention research and interventions were discussed.

Keywords:

• Obesity
• meta-analysis
• intervention studies
• Body Mass Index
• children

Notes

1. The following summary statistics were available for each study: (a) a mean change between Week 1 and Week 12, (b) the normative mean change, (c) the standard deviation of change in the sample, and (d) the average correlation between BMI at Week 1 and Week 12. Gain score SDs were the SD of the gain scores divided by the square root of the product of 2 multiplied by 1 minus the Week 1 to Week 12 correlation (Lipsey & Wilson, 2001).

2. The mean effect size of the 15 studies with available correlations using the Youth Fit For Life intervention was significantly larger than zero, r=.06 (95%CI=.02, .10), d=.15, p<.001, and had a range of r=.02 to r=.17. Heterogeneity analysis indicated significant variability among the studies, Q=160.71, p<.001. No moderators were significant predictors of effect size; the ethnicity effect that was significant in the main analysis was marginally significant (z=1.84, p=.06).

3. The weight computed for the inverse variance weighting macros that estimate mean effect sizes and moderator regression models was derived by dividing 1 by the squared SE (Lipsey & Wilson, 2001).

4. We also examined the regressions as fixed effect models. Ethnicity remained significant (z=1.98, p=.04); in addition, Youth Fit For Life administered as PE (z=7.03, p<.001) and average participant age (z=2.24, p=.03) were significant, with non-PE administration and older participants demonstrating larger effects. The distinction between fixed and random effect models is somewhat arbitrary in this situation as we had all studies to which we were generalizing findings, however, theoretically, we are generalizing to a larger set of future studies. In actuality, the report of our results as random effect models for comparisons with Stice et al. (2006) was a primary goal of the study.