Setting, Design, & Procedures
In the fall of 2005, all schools (n = 8) participating in an after-school program alliance of the Lawrence Public School District, Douglas County Cooperative Extension Service, Lawrence Boys and Girls Club, and community partners were considered for inclusion in the study. Of the eight sites, seven were invited and adopted the program. One site was not invited to participate because the after-school program was not on the elementary school grounds.
A three-year group-randomized controlled trial was conducted with random assignment at the school level after a baseline year of assessment (Figure 1). The study used a nested cross-sectional design with a baseline year (2005-2006), and two subsequent intervention years (2006-2007, 2007-2008). For each year of the study, new children in fourth grade and in after-school programs participated in the study. By using a "repeated cross-section" methodology the outcomes were tracked for the same places rather than for the same individuals [12]. If this study used a longitudinal design and attempted to follow students over three years, it is likely that participant dropout would have exceeded 30%. Although drop out would likely occur due to various reasons unrelated to the intervention, such as the family moving to another school district, change within an entire population can be best studied in a repeated cross section methodology when losses of participants due to movement or dropout are a concern [13].
After the baseline year, sites were stratified into two groups (High Socioeconomic Status (SES)/Low Diversity; Low SES/High Diversity) based on the percentage of students who qualified for free and reduced lunch, and the percentage of students who were nonHispanic white or of diverse race/ethnicity. Following stratification, within each matched group, the principle investigator used a random number generator to blindly randomize sites to the two-year HOP'N after-school intervention (n = 4) or to the control condition (n = 3). After baseline and the randomization procedure, the research team was not blind to the randomization.
In each study year, after-school participants in the fourth grade group underwent two data collection protocols. The first protocol was implemented in both the fall (Pretest) and the spring (Posttest) during after-school time. At least two research assistants traveled to the site and measured children on height and weight in a private setting. Height and weight were assessed while wearing light clothing and no shoes. Research assistants also administered a survey to children and sent a survey home to parents.
For the second protocol, after-school programs were observed by at least two research assistants on six days. Both control and intervention after-school sites were observed on a Monday, Tuesday, and Thursday (in random order) for no more than one day each month in the fall (three days) and in the spring (three days) semesters of each year. The second site visit protocol began with a research assistant attaching an ActiGraph accelerometer to each student's right hip via an adjustable elastic belt and recording the time of attachment and the identification number of the accelerometer and student. Then throughout the after-school day, one research assistant recorded the type of session offered to the children. The session activity was coded as academic, enrichment, recreational (active, nonactive), or snack. For sessions that offered PA, the other research assistant used the System for Observing Fitness Instruction Time (SOFIT) and categorized the active recreation sessions as either organized or free play. For sessions that offered a snack, all foods and beverages were recorded. At the end of the after-school program, or when a student left the early, the research assistant removed each accelerometer and noted time of removal.
Independent of these two data collection protocols, a third data collection effort involved all fourth graders (with informed parental consent) at each school regardless of their participation in the after-school program. Research assistants measured children's height and weight late in the spring semester of each year of the study during the school day. This fourth grade data set allowed us to examine the impact of HOP'N on the school as a whole. This school-wide sample also enabled us to determine whether the after-school program participants were representative of the overall school population [11]. The Institutional Review Board at Kansas State University approved all procedures in 2005.
Participants
After-school participants
Children were included in the after-school sample if they were enrolled in the after-school fourth grade group, provided informed parental consent, and agreed to participate in the fall height and weight assessment. In the U.S., fourth graders are approximately nine years old. Because each after-school program, in addition to fourth graders, may have included third graders and some fifth and six graders to increase enrollment in the fourth grade group, we excluded children from the study if they were not in third or fourth grade and if they participated in the study in a previous year.
In-school participants
Similar to the after-school sample, each year a new group of children entered fourth grade and were eligible for participation. Children were included in the school sample (n = 716) if they were enrolled in a fourth grade school classroom in the fall of that year, had informed parental consent, and assented to participate in the height and weight assessment.
Intervention
Using Social Cognitive Theory [14] and an ecological developmental systems approach [7], the HOP'N intervention was designed to target the development of the skills and efficacy of adult leaders and children to build healthy after-school environments. The HOP'N intervention model included three levels: a community/government/human service agency (County Cooperative Extension), after-school staff training, and after-school program quality elements. The quality elements included an organized daily PA session for at least 30 minutes, a daily healthful snack that included a FV, and a weekly nutrition and PA education experience (HOP'N Club).
The first level of intervention targeted the development of the community/government/human service agency (County Cooperative Extension office) to coordinate improving after-school programs. The research team provided technical assistance to this agency, 5% salary for the Family and Consumer Science County Agent (who prior to the study coordinated the county nutrition education programs), and salary for a half-time Cooperative Extension assistant. The County Agent hired and supervised the Extension Assistant, conducted local community development work, sat on the school district's Wellness Council, and worked with school food service to improve the quality of the snack. The Extension Assistant coordinated the after-school staff training and delivered the HOP'N Club intervention at each site.
The Cooperative Extension office delivered the second level of intervention (after-school staff training) with the assistance of content expertise from the research team. This level of intervention included three staff training sessions per year (six sessions total), staff monthly meetings with the Extension Assistant, and continuous web support http://www.hopn.org. The training was modeled after the Healthy Places "performance community hub" where participants were encouraged to share and problem solve their implementation challenges [15, 16]. Content for the first intervention year began with basic training on how to implement the HOP'N quality elements. Session content then progressed to goal setting, feedback, and problem solving strategies to achieve change in the after-school environment. Because there was high turnover in after-school staff, the sessions in the second intervention year paralleled year 1 in content.
For the third level of intervention, the after-school staff and the Extension Assistant implemented the HOP'N after school quality elements at each intervention site. The after-school program at each site was approximately 2.5 hours per day. Every day, staff had the goal to implement 30 minutes of organized PA following the CATCH Kids Club PA principles [2]. The project provided the CATCH Kids Club curriculum box [10] and PA equipment. Also, after-school program staff was directed to work with their school's food service to provide FV with every snack. In addition to this "bottom-up approach", the County Extension Office worked with the school district food service to achieve the same FV goal. To assist the program staff, the research team provided a list of healthy snack ideas and content expertise. Snacks were not purchased for the program.
Finally, also part of the third level of intervention, the HOP'N Club was a weekly social-cognitive-theory based curriculum delivered by the Cooperative Extension Assistant to each after-school intervention site for 60 minutes once a week. The curriculum was organized in a notebook form with weekly modules that included learning objectives, behavior change strategy goals, and implementation procedures and scripts. The HOP'N Club child behavioral goals were: Be physically active every day (30 minutes after-school, 60 minutes daily); eat FV at every meal or snack; drink less soda and juice drinks (drink water, no more than 1 can of soda or small cup daily); and cut back on TV and video games (no more than 2 hours a day; remove TV from bedroom). The behavioral goals were reviewed weekly and appeared on club cards, t-shirts, and other materials provided by the project. The first 15 sessions were delivered during the 2006 and 2007 fall semesters and targeted building children's competency to adopt and self-regulate behavior to meet the child behavioral goals. The next 14 sessions were delivered in the spring semesters of 2007 and 2008 and were designed to build children's skills and efficacy to influence their home and community environments.
An example of these skill-building activities was a "house hunt". Children took pictures of their home environment on a scavenger hunt, where they searched for physical environmental opportunities for healthful and unhealthful eating ("Go Foods" or "Whoa Foods") or opportunities for PA or sedentary behavior. Then, in "Cool Contracts" children selected a home environment change goal, role played how to ask their parents to participate in signing a contract to change the home environment, and completed a home environment change contract with their parents. Finally, after implementing environmental changes, the children again took pictures of their home and made a "HOP'N-at-Home" poster, which illustrated their home environmental changes. If parents did not want pictures taken in their home, the children completed the same poster experience by drawing pictures or cutting pictures out of magazines.
Child Outcome Measures
Body Mass Index Z Scores (BMIz)
Height was measured using a portable stadiometer (Seca 214 Hamburg, Germany) and weight was measured using high precision digital scales (Seca Corp, Model 770, Hamburg, Germany) that were calibrated daily. Height and weight were measured twice and if the first two measurements differed by more than 1.0 centimeter or 0.1 kg, respectively, a third measure was taken, and the average of the two closest measures was used in the analyses. Unlike adults, BMI values in children do not provide an indication of overweight and obesity. Thus these values must be related to norm reference standards for growth by age and gender as recommended by the Centers for Disease Control and Prevention (CDC) [17]. This is done by using the CDC growth curves and obtaining a z-score for each BMI value. These z-scores are in turn related to the percentiles used to assess overweight and obesity. Participants were classified as "overweight" and "obese" using the age- and gender-specific 85th and 95th percentile BMI values for age and gender.
PA and Sedentary Behavior
Objectively monitored PA was assessed during after-school programs with ActiGraph GT1 M accelerometers (Shalimar, FL). The ActiGraph was programmed to record data every 30 seconds, and ''activity counts'' were processed using cut points developed by Freedson and colleagues [18] to estimate minutes of sedentary activity (< 50 counts per 30 seconds), light activity (≥ 50 counts to 3.9 METS), moderate activity (≥ 4 to 6.9 METS), and vigorous activity (≥ 7 METS) [19].
After-School Environment and HE and PA Opportunities
Details about the observational system developed for HOP'N are available elsewhere [20]. Briefly, after-school programs were observed to determine if activities offered to children were academic, enrichment, recreational (active and non-active), or snack. A PA opportunity was defined as an active recreation session that involved any type of PA and was subcategorized as either organized or free play. Non-active recreation involved activities that were not designed to build skills and included activities such as board games, reading for fun, or computer use for fun. The active recreation sessions were objectively coded using the activity intensity categories of SOFIT to determine the time spent in MVPA [21]. All observers using SOFIT had agreements ≥ 80% for child PA intensity and ≥ 93% for lesson context.
A healthful snack opportunity was defined as offering a FV snack. The type of snack offered during after-school was observed and recorded for FV, total and fat calories, and carbohydrate grams. Nutritional content of the snack was evaluated using actual snack labels, company website nutrition information, and/or the USDA National Nutrient Database. Samples of snacks without labels were collected and weighed, and information was obtained from the USDA database by weight. Only FV offerings are reported here.
Statistical Analyses
The impact of the intervention on the child outcome measures was analyzed using methods to adjust for the lack of independence of the data [13]. Student data were associated with other student data within school sites (i.e., intraclass correlation). To adjust for the clustered data structure, a mixed model three-level design structure (school, year, child) was used to analyze the after-school participant outcomes (fall to spring academic year change in BMIz, fall to spring academic year change in BMI, and accelerometer measured PA across the year) and fourth grade student outcomes (Spring BMIz, Spring BMI).
Separate analyses of covariance using PROC MIXED (SAS Version 9.1, SAS Institute Inc., Cary, NC) were conducted. Variables in each of these analyses were condition (HOP'N, Control), year (baseline, intervention yr 1, intervention yr 2), weight status (overweight/obese, normal), school stratification for randomization (High SES/Low Diversity, Low SES/High Diversity), grade, gender (boy, girl), race/ethnicity (white, nonwhite), and family SES (eligible for free and reduced lunch, not eligible). For the BMI and BMIz change score analyses, the fall (baseline) assessment was also a covariate. For PA, accelerometer monitoring time was additional covariate.
The study was originally powered to detect a .5 kg/m2 difference in BMI between a sample size of 4 intervention and 4 control schools with a reduction in the detectable difference adjusting for age, ethnicity, and gender using 20 students per group. Assuming that the small after-school dropout from fall to spring was due to random factors, the after-school analyses examined the impact of HOP'N for those who initiated the program regardless of their level of attendance and dose of receiving the intervention. Comparisons of correlated response variables evaluating between condition and between year differences were evaluated at p < 0.05, two-tailed tests. Some alpha level adjustment for multiple tests should be made, but because of the complexity of the model there are not methods available to carry out such adjustments. A Bonferroni adjustment could be used but it would be too severe and would cover up possible meaningful comparisons where as an adjustment that would utilize the model and correlations would not [22].
The impact of the intervention on the after-school site PA session opportunities and FV snack offerings was analyzed using a non-parametric Mann-Whitney U test for each group comparison.
