School-Based Physical Activity Interventions: A Meta Analysis
Is Version Of
School-based interventions have been proposed as a key strategy in reducing childhood obesity and type 2 diabetes, addressing cardiovascular disease and increasing physical activity. Numerous studies have examined the impact of interventions focusing on physical activity and nutrition education. Twenty published school-based interventions, which included a control group were identified.
These studies were reported in 110 individual publications. Meta analysis wasused to examine the outcomes of twelve interventions; eight interventions were excluded because the data was not available (e.g., confidence intervals or means and standard deviations) to estimate effect sizes. The included studies reported
data from 12,930 children and were reported in fifteen of the 110 publications.
Two methods were used to calculate effect sizes for nineteen dependent variables. One method was the pre to post where the pre test mean was subtracted from the post test mean for both the intervention and control groups and divided by the
control group pre standard deviation. This method produced one effect size for each level of the intervention and control for each dependent variable, yielding 168 effect sizes. Three variables were declared statistically significant; those were moderate-to-vigorous physical activity, mile walk/run, and knowledge. However the effect size for the mile run was small. Intervention and control
was the independent variable in two t-tests for the remaining effect sizes of MVPA and knowledge. Knowledge was significantly better for intervention than control groups, MVPA was not. A second method of calculating the effect size was to compare the intervention to the control at post-test using the control post-test standard deviation. This produced 96 effect sizes. Four were statistically different from zero; mile walk/run, pull-ups, knowledge and total skinfolds. Of those four, only knowledge had a moderate effect size.
While each of these effect sizes represented multiple effect sizes, some were from a single study. Therefore, effect sizes were combined to categories of cardiovascular outcomes (e.g., cholesterol, blood pressure), physical activity (e.g., fitness, mile run) and knowledge. For all studies, knowledge was greater in intervention participants (ES=0.90) as was physical activity (ES=0.76). The composite physical activity effect size (fitness and MVPA) was used as the dependent variable in a regression with total intervention time and Coordinated School Health Program (CSHP).
components as predictors. The regression and both predictors were significant. Further analyses determined that grade (age) and gender were not significant categorical variables influencing the outcomes of interventions.
Considering both methods of calculating effect sizes knowledge and physical activity are efficacious dependent variables because these are sensitive to change. Clearly, schools are well suited to influence both when given the resources to do so. Further, multifaceted approaches to increasing physical activity, such as the CSHP, produce larger effect sizes than single approaches. Future studies of school-based physical activity interventions should consider reporting data so that effect sizes can be calculated, focus on long term outcomes, and explore a variety of components of the Coordinated School Health Program.