A socio-ecological approach promoting physical activity and limiting sedentary behavior in adolescence showed weight benefits maintained 2.5 years after intervention cessation

Abstract

Background: Obesity in youth remains a major public health issue. Yet no effective long-term preventive strategy exists. We previously showed that a school-based socio-ecological approach targeting behavior and social/environmental influences on physical activity (PA) prevented 4-year excessive weight gain in 12-year olds. In this study, we investigated if this efficacy persists 30 months after intervention cessation.

Methods and findings: The program targeted students, family, school and the living environment to promote/support PA and prevent sedentary behavior (SB). A total of 732 students from eight randomized middle schools completed the 4-year trial. At the 30-month post-trial follow-up, body mass index (BMI), fat mass index (FMI), leisure PA (LPA), home/school/workplace active commuting, TV/video time (TVT), and attitudes toward PA were measured in 531 adolescents. The beneficial effects of the intervention on the excess BMI increase (+0.01 vs +0.34 kg m(-2) in the intervention and control groups, respectively) and on the overweight incidence in initially non-overweight students (4.3% vs 8.6%; odds ratio=0.48 (95% confidence interval: 0.23-1.01)) were maintained at the post-trial follow-up. LPA was not maintained at the level achieved during the trial. However, we still observed a prevention of the age-related decrease of the adolescents' percentage reporting regular LPA (-14.4% vs -26.5%) and a higher intention to exercise in the intervention group. The intervention promoted lower TVT (-14.0 vs +13.6 min per day) and higher active commuting changes (+11.7% vs -4.8%). Trends in higher BMI reduction in students with high initial TVT and in the least wealthy group were noted. TVT changes throughout the follow-up predicted excess BMI and FMI changes.

Conclusions: Long-term multilevel approach targeting PA and SB prevents excessive weight gain up to 30 months after intervention cessation. The efficacy may be higher in the most sedentary and least wealthy adolescents. Healthy PA-related behavior inducing long-lasting weight effects can be promoted in youth providing that an ecological approach is introduced in the prevention strategy.

Figure 1

Trial chart.

Figure 2

Evolution of excess BMI during the trial period and at the post-trial follow-up. Values are presented as adjusted means±s.e.m. Black and white squares represent the control and intervention groups, respectively. Excess BMI was calculated by subtracting the gender- and age-specific median BMI of the French reference curves from BMI. Analyses were done using mixed linear models taking into account the cluster randomization and the repeated individual data over time, with different baseline covariables as fixed effects.

Figure 3

Differences across groups in excess BMI change at the post-trial follow-up according to different baseline characteristics. The black boxes represent the effect sizes (adjusted mean differences in excess BMI changes) and the horizontal dotted lines represent the 95% confidence intervals, separately for each of the categories of the baseline characteristic of interest. The P-values represent the intervention effect and the heterogeneity across subgroups of baseline characteristics. Excess BMI was calculated by subtracting the gender- and age-specific median BMI of the French reference curves from BMI. Analyses were done using mixed linear models taking into account the cluster randomization and the repeated individual data over time, with different baseline covariables as fixed effects.

Figure 4

Excess BMI and FMI at the post-trial follow-up according to tertiles of overall changes in television/video time during the total follow-up period. Values are presented as adjusted means±s.e.m. for each tertile of overall TV/video time changes. Medians of TV/video changes are, respectively, -56 min per day; 5 min per day and +52 min for each tertile. Excess BMI was calculated by subtracting the gender- and age-specific median BMI of the French reference curves from BMI. Analyses were done using mixed linear models taking into account the cluster randomization, with different baseline covariables as fixed effects.