Childhood obesity remains one of the most serious threats to the public’s health, with 1 in 3 children and adolescents overweight or obese (body mass index (BMI) ≥ 85th percentile) [
In parallel, comparative effectiveness research is being discussed within the national health reform debate as a mechanism for improving healthcare quality and decreasing healthcare spending [
It is estimated that 8.4 million children attend after-school programs (ASP), and an additional 18.5 million would do so if a program was available [
There is limited published research on ASPs designed to increase physical activity. Systematic reviews suggest that it is possible to improve activity levels, physical fitness, body composition, and blood lipids in the after-school setting [
To assess the comparative effectiveness of this community-driven ASP as a pediatric obesity prevention intervention, we compared it to the routine aftercare available to working parents in the community and asked two research questions: (1) Are children in the alternative ASP more physically active than children in the standard ASP? (2) Do the operating costs associated with these programs differ?
This study was guided by principles of community-based participatory research (CBPR). CBPR is an important research approach that equitably involves community members who are affected by the issue being studied in all phases of the research process [
The study design was an observational prospective cohort study and a natural experiment in Nashville, TN, USA. The “naturally occurring” event, the parks department’s new ASP, was the intervention, and children attending this community ASP formed the intervention group (
The two ASPs followed similar formats, and operated from 3–6 PM every day public schools were open. Both ASPs included time for snack, homework, and play and did not focus on a single activity (e.g., tutoring, chess, and team sport). The community ASP was set in a community recreation center and involved staff-led games. The school-based ASP was set in a school cafeteria and involved opportunities for arts and crafts and playing on the playground. The main differences between the two ASPs were (1) format of active play time (adult-led versus unstructured) and (2) location (community recreation center versus public school). Refer to Table
Comparison of after-school programs.
Community (intervention) | School-based (comparison) | |
---|---|---|
Location | (i) Community recreation center | (i) Public school cafeteria |
| ||
Who | (i) Ages 5–14 yrs | (i) Only open to students at that elementary school (5–10 yrs) |
| ||
Program format | (i) 3–6 PM | (i) 3–6 PM |
(ii) Transportation from neighborhood public schools to the community center |
(ii) Transportation not necessary | |
(iv) Snacks provided | (iv) Snacks provided | |
(v) Homework help provided | (v) Homework help provided | |
(vi) Staff-led activities (children select activity) | (vi) Unstructured play time (children select activity) | |
| ||
Stated physical activity goal | (i) 60 minutes of activity/day | (i) 45 min of moderate activity 3/week |
| ||
Physical activities |
(i) Staff leads students through activities: |
(i) Staff supervises for safety: |
| ||
Nonphysical activities |
(i) Arts and crafts | (i) Arts and crafts |
| ||
Physical activity resources |
(i) Playground |
(i) Playground |
| ||
Cost | (i) Free of cost to families |
(i) $46.50/week paid by family |
All measures were collected at the ASPs at three time points over approximately 12 weeks (February–May 2010), with six weeks separating each wave of measurement. The measurement period was selected based on the Cochrane Review that states that obesity prevention interventions should last at least 12 weeks for behavior change to be observed [
Physical activity was assessed using ActiGraph GT1M accelerometers (ActiGraph, Pensacola, FL, USA) only during ASP programming time. Accelerometry is considered an objective measure of physical activity [
Freedson’s age-dependent cut points were used to determine time spent in sedentary, light, moderate, and vigorous activity [
Daily percentage of time spent in each level of physical activity (i.e., sedentary, light, moderate, and vigorous) was determined by dividing the minutes spent in each activity level by the sum of minutes the ActiGraph was worn in a day (i.e., time in attendance at the ASP). Children spent varying amounts of time in ASPs depending on their family needs. Thus, the continuous outcome measures were the proportion of time spent in LMVPA (light-moderate-vigorous physical activity) or MVPA (moderate-vigorous physical activity) out of total time in attendance, rather than the number of minutes the program was open, to allow for a meaningful comparison within individuals and across groups. Daily percentages were averaged across days to create individual participants’ physical activity (PA) scores at each measurement period.
Body weight was measured after voiding while children wore light clothing without shoes. Calibrated digital scales (Detecto, Webb City, MO, USA, Model#758C) were accurate to the nearest 0.1 kg. Body height without shoes was measured to the nearest 0.1 cm with the scale’s stadiometer. BMI percentile, adjusted for age and gender, was calculated using these measurements [
Body composition was measured by the RJL Systems BIA Quantum II (RJL Systems, Clinton, MI, USA) after voiding. Standard procedures for whole body bioelectrical impedance measurement were used [
Children were asked to complete a 1/2 mile run as fast as possible on a running track [
Parents completed a survey asking about child’s date of birth (used to calculate age), gender, race/ethnicity, and name of school.
Because children were not randomly assigned, preexisting differences between groups were potential confounders. Therefore, we compared children enrolled in the ASPs to test for differences on basic demographic and process variables, using bootstrap
To assess change in PA over time, a conditional linear latent growth model was used with random intercepts and slopes that were free to covary and time varying error variances. The model was estimated using Mplus version 6.11 [
We used the cost analysis guidelines for research evaluation proposed by Levin and McEwan [
Of the 91 children who attended the ASPs, baseline demographics were obtained from parents of 83 children. The analytic sample included the 82 children with PA data from at least one time point; one child in the school-based ASP did not provide at least 3 days of PA data in any measurement period and was not included in the analyses. Of the 82 participants, 62 had data for all three time points, 16 had data for two time points, and 4 had data for only one time point.
The baseline sample was 65% female and 7.9 years of age (
Between-group comparison of baseline and process measures.
Community ASP ( |
School-based ASP ( |
| ||||||||
---|---|---|---|---|---|---|---|---|---|---|
Mean/% | SD | Min | Max | Mean/% | SD | Min | Max |
|
| |
Child characteristics | ||||||||||
Male | 43% | 25% | 0.10 | 0.44 | ||||||
Hispanic ethnicity | 26% | 11% | 0.10 | 0.45 | ||||||
Black | 47% | 31% | 0.14 | 0.57 | ||||||
White | 26% | 56% |
|
|
||||||
Age at baseline (yrs) | 8.79 | 1.67 | 5.57 | 12.08 | 7.96 | 1.55 | 5.45 | 10.34 | 0.023 | 0.12 |
BMI percentile** | 74.74 | 23.60 | 8.40 | 99.60 | 73.87 | 21.03 | 11.60 | 99.40 | 0.86 | 1.00 |
Body fat percentage | 29.26 | 11.27 | 5.80 | 54.30 | 29.92 | 8.08 | 15.90 | 48.60 | 0.77 | 1.00 |
Fitness ( |
6.29 | 1.09 | 4.23 | 9.41 | 6.08 | 1.13 | 4.23 | 8.51 | 0.40 | 0.99 |
| ||||||||||
Process measures | ||||||||||
Waves of data collection per child | 2.83 | 0.52 | 1.00 | 3.00 | 2.92 | 0.37 | 1.00 | 3.00 | 0.40 | 0.94 |
Minutes activity monitor worn per measurement period | 108.74 | 20.37 | 57.60 | 143.00 | 105.83 | 27.02 | 33.20 | 149.25 | 0.59 | 1.00 |
**Underweight:
Between-group comparison of time spent in physical activity (model-implied estimates).
Community ASP | School-based ASP | Group difference | ||||
---|---|---|---|---|---|---|
% Time |
|
% Time |
|
% Time |
|
|
Baseline | ||||||
LMVPA | 78.4 |
|
75.8 |
|
2.6 | 0.30 |
MVPA | 30.1 |
|
24.2 |
|
5.9 | 0.06 |
| ||||||
Change per measurement period (6 weeks)* | ||||||
LMVPA | 3.0 | 0.006 | −3.4 | 0.002 | 6.4 |
|
MVPA | 2.8 | 0.006 | −1.6 | 0.12 | 4.4 | 0.002 |
The linear latent growth model implied that, on average, children in the community ASP became more active over time (average change between Baseline-Week 6 and Week 6-Week 12), compared to the children in the school-based ASP (Table
Percent of time spent in physical activity (LMVPA) after-school. Notes: lines show mixed model outcome slopes; points show observed means ± standard error.
Percent of time spent in moderate/vigorous physical activity (MVPA) after-school. Notes: lines show mixed model outcome slopes; points show observed means ± standard error.
The community ASP served 54 children; the school-based ASP served 37 children. Total implementation costs (valued in 2010 dollars) for the 12-week study period were $1,184 per child ($19.25 daily per child) for the community ASP, compared to $1,087 per child ($17.67 daily per child) for the school-based ASP (9% difference; Table
Total implementation cost per participant and program (2010 dollars).
Community ASP |
School-based ASP | |
---|---|---|
|
$781 | $706 |
| ||
|
$380 |
$314 |
| ||
Snacks | $17 | $62 |
| ||
|
$6 | $4 |
| ||
Total direct cost per participant for 12 weeks | $1184 | $1087 |
Daily direct cost per participant | $19.25 | $17.67 |
***The school-based ASP reported a 100% depreciation rate within one year. The community center program also reported 100% depreciation rate for light recreational equipment within one year and a 5-year life span on electronics and large equipment.
With more than 23 million parents of school-aged children employed full-time [
Assuming the improvement in activity was solely due to the intervention; these findings suggest that children attending traditional school-based ASPs, already costing an average of $17.67 per day, would need an additional daily investment of $1.59 per child over 12 weeks to increase their LMVPA by 15.4 percentage points or their MVPA by a model-implied 14.7 percentage points. Cost-effectiveness analyses are often lacking for community-based prevention efforts. The annual cost of childhood obesity-related health expenses in the US is $14.1 billion for outpatient care and $237.6 million for inpatient care which translates to about $5 in healthcare expenses per day per child, without including other relevant long-term costs related to school performance, labor market involvement, quality of life, welfare needs, and so forth [
ASPs have long played a critical role in supporting academic achievement, safety, discipline, and avoidance of risky behaviors [
First, accelerometers do not adequately measure body movements of upper and lower extremities, but they are considered the gold standard for measuring PA under free-living conditions. This should not have biased our results since the limitation of accelerometry was the same across groups. Second, our sample was small but having three waves of data increased statistical power and was sufficient for detecting a significant increase in PA under free-living conditions. Third, despite efforts to select a comparable comparison group and measure potential confounders, we cannot rule out all systematic differences between the two groups. We did rule out the most important possible confounds in the literature: body composition, fitness, age, and gender. It is possible that the difference in racial composition of the groups could explain baseline variance [
Fourth, for the community ASP, there were significant differences between the observed and model-implied averages at Weeks 6 and 12. These discrepancies highlight the fact that the final specified model did not perfectly recreate the observed data. This could have been partially due to missing data for this group at either time point. The discrepancies could also have arisen because the community ASP's growth rate was not linear; yet, a model with only three time points does not have the degrees of freedom to investigate more sophisticated growth parameter specifications (e.g., quadratic). Nonetheless, applying latent growth models has provided further insight into how ASPs might impact children's PA over time (e.g., what effect does ASP type have on PA change over time? What is the typical growth rate of PA for children over time? Do some programs increase the growth rate of certain types of PA (e.g., light, moderate, or vigorous) more than others? What is the functional form of PA change over time?)
An ASP set in a community recreation center and led by recreation staff incorporating structured physical activity opportunities was associated with significant increases to physical activity during ASP time in a multiethnic sample of public school children in 12 weeks, compared to a standard school-based ASP. Utilizing community recreation centers’ built environment and staff could be a promising low-cost proposition to improve health trajectories among school-aged children.
The authors declare that no competing financial interests exist.
The Vanderbilt Energy Balance Laboratory provided technical assistance with accelerometry for the study. The content is solely the responsibility of the authors and does not necessarily represent the official views of the funding agencies. The authors thank Jonathan Dodson for program implementation; Eileen Ruchman for project management; Principal Roxie Ross of Metro Nashville Public Schools; and Paul Widman, Bill Troup, and Tommy Lynch of Metro Nashville Parks and Recreation for their collaboration and support through the Nashville Collaborative. Research reported in this paper was supported by pilot funds awarded to S.B. Gesell from the Vanderbilt Institute for Obesity and Metabolism and the National Center for Advancing Translational Sciences of the National Institute of Health under Award no. UL1 TR000445. S.B. Gesell was supported by the