Effects of a Multi-Pronged Intervention on Children’s Activity Levels at Recess: The Aventuras para Niños Study^,

Abstract

Latino children spend more time in sedentary activities than other American children, and only ~1 in 5 Latino children in public elementary and middle schools meet all 6 fitness standards in statewide fitness testing. Schools that facilitate physical activity (PA) by maintaining playgrounds and providing physical education classes have children who are more active and less overweight. The aims of the present study were to examine the extent to which several social and physical environmental changes in school settings resulted in observed changes in area characteristics and children’s activity levels during recess. Thirteen elementary schools serving predominately Mexican American children were randomized into control or activity and nutrition environmental intervention conditions. Playgrounds and activities were restructured in 6 intervention schools to promote more PA. After 1 y, there were no overall statistical differences between treatment groups in PA or sedentary behavior in these settings and results did not differ by gender. Changing the social and physical environments to promote children’s moderate-to-vigorous PA is important to the design of active and healthy recess environments. The present results are not conclusive as to the link between these interventions and actual behavior, but show sufficient promise for further population and setting specific research.

Introduction

Overweight and obesity rates continue to rise among U.S. children and adolescents (1, 2). Obesity is now the most prevalent chronic disorder among youth and is especially pronounced among children of lower socioeconomic status and Latino and African American parents (3, 4). Obesity compounds threats to the health of U.S. Latino youth and adults, who are at higher risk for chronic health conditions, including cardiovascular disease (5) and type 2 diabetes (6–8). Whereas in the past, obese children were targeted for intensive clinical interventions or special recreational programs, the epidemic of obesity onset has necessitated public health planners to take a population wide approach based on the concept of population attributable risk (9). The assumption in the population attributable risk approach is that environmental and other changes that can affect entire populations may reduce overall morbidity and mortality.

Reducing sedentary activities and promoting moderate-to-vigorous physical activity (MVPA)⁴ constitute important targets for preventing obesity, and public schools are a critical venue for promoting MVPA and reducing sedentary behavior (10). Latino children spend more time in sedentary activities than do other American children and are less physically fit, with only ∼1 in 5 Latino children and youths in public elementary and middle schools meeting all 6 fitness standards in statewide fitness testing (11). Schools that facilitate physical activity (PA) by maintaining their playgrounds and providing physical education classes have children who are more active and less overweight (12). Assessing changes in potentially modifiable factors and their relationships to children’s PA is important to the dissemination of recess interventions aimed at increasing children’s PA.

Our purpose in this study was to use direct observation in 13 predominantly Mexican-American elementary schools in low-income neighborhoods to examine the extent to which a social and physical environmental intervention delivered by promotoras (Community Health Advisors) resulted in observed changes in area characteristics and children’s activity levels during recess.

Methods

Study design

This study involved the analysis of pre- and postintervention observational data collected as part of the “Aventuras para Niños” study (APN; Adventures for Children). The APN study was a community intervention trial designed to prevent and control childhood obesity in children aged 5–7 y at enrollment. Schools from lower income neighborhoods (median household income < U.S. $30,000/y) were invited to participate in the program if they met the following eligibility criteria: 1) a Latino enrollment of at least 70%; 2) had not participated in other obesity related programs or special physical education training within the past 4 y; and 3) had defined enrollment boundaries (as opposed to being a charter or magnet school drawing students from a broad region). Thirteen public elementary schools met the inclusion criteria and were randomly assigned to 1 of 4 conditions: micro-environment only (home), macro-environment only (school and community), both types of intervention (“micro + macro” [home, school, and community]), or a no-treatment control. This paper focuses specifically on intervention components conducted in the schools to promote active play during recess (micro + macro and macro only). The schools had an average enrollment of 667 children (range = 438–1136) and encompassed 3 school districts in south San Diego County, California. The study was approved by the Institutional Review Board at San Diego State University.

School-based intervention

The intervention, designed to last ~5 semesters, was based on the Social Ecological Theory (13), which holds that the physical and social environments in which people live are important factors to target for individual behavior change. The influence of environmental conditions has repeatedly been shown to be a major force behind the rapidly rising rate of overweight and obesity (14–16). Cohen et al. (17) identified 4 types of structural factors beyond individual control that influence behavior: 1) availability of protective or harmful programs or products; 2) physical structures; 3) social structures and policies; and 4) media and cultural messages. The APN school intervention components were designed to affect the 4 categories of influences as they relate to the promotion of active play during physical education and at recess: play facilities (the school grounds themselves), equipment, PA opportunities at school, and school policies. In some cases, school changes were implemented by APN staff whereas others were implemented by promotoras (i.e. community advocates).

The promotora model has been identified as one of the most culturally appropriate and well received intervention methods for Latinos (18–20). APN staff recruited and trained the school-based promotoras and then worked with them to implement the school intervention components. The number of school-based promotoras varied from 6 for the first wave of schools to 9 with wave 2 enrollment and finally a core group of 4. Promotoras received an 8-session training at the outset, which included basic information on obesity, nutrition, and PA, as well as details of the project goals in schools. Throughout the intervention period, promotoras attended weekly meetings with a project coordinator for feedback, troubleshooting, and continued training. They worked 15–20 h/wk and were paid hourly. Importantly, during implementation of the intervention activities outlined below, the promotoras in this condition (macro-only and micro + macro) and project staff did not have contact with participating children or their families or even knew who they were. Information and programs were provided to the entire school, or, at minimum, to targeted grade levels. The intervention components designed to increase active play at recess included new and previously developed programs: Peaceful Playgrounds painted game designs (21), recess walking clubs, and Super Aventuras, which are described below.

Peaceful playgrounds.

Prior to intervention activities, the intervention schools had some type of game markings outlined in white paint on the playgrounds, such as basketball and 4-square courts. The markings were sparse at some schools and rarely were colorful. APN purchased stencils, blueprints, and school licenses for the Peaceful Playgrounds program, which was developed by a San Diego County educator as a method to keep more children, safe, occupied, and out of trouble during recess (22). The APN staff identified active games that used play area designs, tailored the designs with input from teachers, and then worked with principals and district staff to have initial area outlines painted with striping machines. APN staff and promotoras, with the help of a few parents, used rollers and brushes to paint 6-color shapes, numbers, and letters to fill out the designs. All teachers were given a set of simple rules for active games that could be played on the newly painted designs.

Recess walking clubs and #x201CSuper Aventuras.#x201D

Initially, the school-based promotoras organized “walking clubs" for children during the recess time following lunch. The children, sometimes accompanied by a promotora, walked laps around a designated area of the school grounds. At first, the promotoras tracked how many miles each child accumulated, but that was cumbersome with such large numbers. During the second year, that system was abandoned and promotoras walked along with groups of children and simply tracked who had participated. Participating children sometimes received stickers and other incentives such as water bottles, jump ropes, and small balls.

The walking clubs were replaced by a new program called “Super Aventuras,” which was designed by staff to give children more interesting options for recess activities. This program allowed students to rotate between activity stations staffed by the promotoras. Up to 3 stations operated at a time, including parachute games, aerobic dance, rope jumping, and a locomotor movement course named after animals (e.g. bunny hop, cheetah chase, frog jump). Signs were made to illustrate the movements, and cones and other colorful markers were used to mark the circular course. Charts, with the names of students by class, were mounted temporarily on a fence, and students participating were given a small sticker to place by their name. At the end of the year, students were given certificates and prizes for participation, but there were no other incentive prizes. Super Aventuras was held at each of the 6 schools twice per month. At 1 school, volunteer staff from a nearby sports equipment store came regularly to help. At other schools, the promotoras recruited several parents to help supervise the activities.

Additional promotoras' role in the schools.

The promotoras were seen as the “eyes and ears" of the project. They were expected to implement or observe the implementation of program activities and to provide program progress feedback to teachers at least monthly. In addition to delivering and collecting materials, they provided prompts to the teachers and feedback to project staff. However, it soon became clear that because of the differences in background and training that the promotora model was not appropriate for working with teachers. Language and cultural barriers also impeded the development of a sense of being peers, a relationship at the core of the promotora model, and monthly face-to-face contact was too time consuming for teachers’ schedules. The school coordinator eventually took over responsibility of communicating with teachers. To collect materials from the teachers, colorful plastic crates were placed in the teachers’ lounges with a hanging folder for each teacher.

Evaluation

One purpose of the APN school-based intervention was to increase children’s PA. PA was measured by direct observation, which allowed for the recording of activity levels simultaneously with current contextual influences, including physical and social environmental conditions. Specifically, our research question asked “Are children more active in an activity-promoting environment?” and ”Do area characteristics change with a social and environmental change intervention?”

Observers used System for Observing Play and Leisure Activity in Youth (SOPLAY), a method designed to obtain data on the number of youths and their PA levels during leisure opportunities (10). SOPLAY is based on momentary time sampling and has been used to assess children’s leisure time behavior and playground characteristics in a study of 24 middle-schools (10) and in community parks (23). All potential areas for leisure time PA in each school were identified and measured prior to data collection at baseline and at the 1-y follow-up. Agreement among assessors was established on the location, size, and boundaries of each target area, and maps detailing target areas and where observers should stand when observations were made. A total of 137 outdoor areas were targeted for observation, averaging 10.5/school (SD = 2.8). Indoor spaces such as gymnasiums and multipurpose rooms were not available in the schools.

Training.

Three paid, bilingual, female research assistants (RA) collected the data at each time point. The RA memorized operational definitions of the behavior dimensions and their subcategories first and then learned the general procedures for recording data. Videotaped examples and role-playing were used to demonstrate each category during training, which was followed by practice observations on school playgrounds. RA training included how to reduce reactivity from children and adults. The average training program took ∼16 h, with training for an RA continuing until she exceeded an inter-observer agreement score of 80% on video segments on a training DVD. Additional review and training sessions ~1 h in length were conducted each school semester.

Systematic SOPLAY scans of target areas were made during 3 measurement periods during recess (over a 3-mo period) at baseline and at the 1-y follow-up. All activity areas were observed in a specific order during each observation period. During each visit to a target area, observers scanned the space visually from left to right and entered a code representing each student’s activity level into a hand-held mechanical recording device at an approximate rate of 1 child/s. During a scan, the activity of each student in a target area was coded as sedentary (i.e. lying down, sitting, or standing), walking, or vigorous. These activity codes have been validated by both heart rate monitoring (24) and accelerometry (25). Separate scans were made for girls and boys. Simultaneous entries were also made for the contextual characteristics of each area, including its accessibility, usability, and whether supervision, organized activities, and equipment were provided. The average length of time for each recess period was: 15.8 min (range 0–30 min) for before school recess, 18.5 min (range 15–30 min) for morning recess, and 47.3 min (range 45–60 min) for lunchtime.

Data analyses

The unit of analysis was an individual visual observational scan of a predetermined target area. The sample size was 1386 scans at baseline and 1206 scans at the 1-y follow-up. Counts were tallied for children engaged in sedentary, walking, and vigorous behavior in each area to obtain a summary score (for boys and girls separately). An additional summary score, MVPA, was created by summing the walking and vigorous frequencies. Because of the high variability in the total number of children in the observed areas (baseline range = 1–97; 1-y follow-up range = 1–104), count data from observations were expressed as the percentage of children in each activity level during a scan. This was done by dividing the number of children observed as either sedentary, walking, or vigorously active by the total number of children in the area. These proportions were calculated separately for boys and girls. Data were excluded for scans that did not contain any children (516 excluded) and scans that contained ≤4 children (459 excluded), resulting in a final sample size of 853 scans and 764 scans at baseline and follow-up, respectively.

Separate analyses were conducted for each dependent variable (i.e. sedentary, walking, vigorous activity, and MVPA) and separately for boys and girls. Independent samples t tests were used to examine whether the percentage of boys and girls engaged in sedentary activity, walking, vigorous activity, or MVPA changed from baseline to the 1-y follow-up by treatment condition. These analyses were repeated and stratified by area characteristics. Linear regression analyses were used to test for treatment effects over time on each of the 4 activity variables by testing for a treatment-by-time interaction. In addition, analyses were conducted to test for differential effects by gender and area characteristics (i.e. treatment-by-time-by-gender interaction and treatment-by-time-by-area characteristic interaction). Chi-square was used to test for treatment group differences in area characteristic proportions. Logistic regression was used to test for treatment effects in the area characteristics from baseline to follow-up (area characteristic was binary dependent variable and independent variables included a treatment-by-time interaction). Lastly, linear regression was used to test for treatment-by-time interactions for each activity variable within each area characteristic, stratified by gender. All analyses were conducted using PASW version 17 software.

Results

Schools readily accepted the structural and environmental changes that were designed to promote more PA, including the playgrounds painted game designs, recess walking clubs, SPARK-based active physical education, and “Super Aventuras.” Thus, the fidelity to the implementation of the interventions was fairly high.

Nevertheless, these structural and environmental changes did not necessarily translate into observed behavior change. At baseline, assessors observed and coded the activity levels of a total of 12,639 children (5955 girls and 6684 boys) during recess. At the 1-y follow-up, they observed and coded the activity levels of a total of 10,807 children (4943 girls and 5864 boys) during recess. The number of children per scan did not differ between intervention and control conditions at baseline and follow-up (range 10–12 per scan; P > 0.05). Table 1 shows that (among girls) there was a significant reduction in the percentage of children observed walking from baseline to the 1-y follow-up in both intervention and control conditions. Among boys, there was a significant increase in the percent of children observed sedentary and a significant decrease in the percent of children observed in MVPA in the intervention group only. Regression analyses showed no significant treatment-by-time interactions for percent sedentary (β = 0.04; P = 0.42), percent walking (β = −0.05; P = 0.27), percent vigorous (β = 0.004; P = 0.92), and percent MVPA (β = −0.04; P = 0.42). There were also no significant treatment-by-time-by-gender interactions for percent sedentary (β = 0.006; P = 0.87), percent walking (β = 0.04; P = 0.19), percent vigorous (β = −0.04; P = 0.22), and percent MVPA (β = −0.006; P = 0.87).

Table 1.

Baseline and 1-y follow-up changes in the mean proportion of girls and boys in defined activity levels during recess, stratified by condition¹

	Intervention		Control
	Baseline	1-y	Baseline	1-y
Girls, %	n = 148	n = 206	n = 255	n = 160
Sedentary	36.7	40.9	34.9	37.1
Walking	34.8	28.2**	34.5	29.0**
Vigorous	28.6	31.0	30.6	34.0
MVPA (walking + vigorous)	63.3	59.2	65.1	62.9
Boys, %	n = 154	n = 223	n = 296	n = 175
Sedentary	28.1	33.9*	29.3	33.9
Walking	38.6	35.4	34.0	34.4
Vigorous	33.4	30.7	36.8	33.0
MVPA (walking + vigorous)	71.9	66.1*	70.8	67.4

¹Different between baseline and 1-y follow-up at < 0.05* and P < 0.01**.

Areas characteristics

At baseline, there were no significant differences between intervention and control schools in proportion of areas supervised (79.3 vs. 80.4%; P = 0.70). However, at follow-up, intervention schools had a lower proportion of areas supervised compared with control schools (53.9 vs. 61.2%; P = 0.04). There were no differences in the proportion of areas with equipment between intervention and control schools at baseline (48.3 vs. 54.3%; P = 0.09). However, at follow-up, intervention schools had a lower proportion of equipped areas compared with control schools (59.5 vs. 73.0%; P < 0.01). At both baseline and follow-up, control schools had a higher proportion of areas with organized activities (35.9 vs. 10.7%; P < 0.01) and (57.3 vs. 41%; P < 0.01). Results of logistic regression analyses showed no significant treatment-by-time interaction for supervised areas (OR = 0.79; P = 0.32) or equipment (OR = 0.69; P = 0.09). However, a significant treatment-by-time interaction was observed for activity areas organized (OR = 2.43; P < 0.01), whereby intervention schools had a significantly greater increase in the proportion of areas providing organized activities (30.3% increase) compared to control schools (21.4% increase) from baseline to follow-up.

Table 2 shows the change in proportion of children engaging in each of the activity intensities for intervention and control schools, stratified by gender and area characteristics. In supervised areas, the percentage of girls engaged in walking decreased in both the control group and intervention group (P < 0.01). In supervised areas, the percentage of boys engaged in sedentary behavior increased in the intervention group (P < 0.05), whereas the percentage engaged in vigorous activity decreased in the control group (P < 0.05) and MVPA decreased in the intervention group (P < 0.05). In areas with organized activities, the percentage of girls engaged in walking decreased in the control group (P < 0.01) and the proportion engaged in vigorous activity decreased in the intervention group (P < 0.05). In areas with organized activities, the proportion of boys engaged in sedentary behavior increased in the control group (P < 0.05), the proportion engaged in walking decreased in the control group (P < 0.05), the proportion engaged in vigorous decreased in both the intervention and control groups (P < 0.01 and P < 0.05, respectively), and the proportion engaged in MVPA decreased in the control group (P < 0.05). In areas with equipment, the proportion of girls engaged in walking decreased in the control group (P < 0.05) and the proportion of boys engaged in sedentary behavior increased (P < 0.01) and the proportion engaged in MVPA decreased in the intervention group (P < 0.01). Results of linear regression showed no significant treatment-by-time-by-supervised interactions or treatment-by-time-by-equipment interactions for any of the activity variables. However, a significant treatment-by-time-by-organized interaction was shown for percent sedentary (β = 0.17; P < 0.01), percent vigorous (β = −0.22; P < 0.01), and percent MVPA (β = −0.17; P < 0.01). These analyses were also stratified by gender. There were no significant treatment-by-time-by-supervised interactions for boys or girls in any of the activity variables. However, there was a significant treatment-by-time-by-organized activity interaction for both boys and girls in percent sedentary (β = 0.18 P < 0.01 and β = 0.17, P < 0.01), percent vigorous (β = −0.21, P < 0.01 and β = −0.21, P < 0.01), and percent MVPA (β = −0.18, P < 0.01 and β = −0.17, P < 0.01). Lastly, there was a significant treatment-by-time-by-equipment interaction for girls (β = −0.10, P = 0.04), but not for boys (β = −0.05, P = 0.30) for percent vigorous.

Table 2.

Baseline and 1-y follow-up changes in the mean proportion of children in different activity levels at recess by area characteristics, stratified by gender¹

	Supervised		Organized		Equipped
	Baseline	1 y	Baseline	1 y	Baseline	1 y
Girls, %
Sedentary	I = 38.9	I = 44.4	I = 36.0	I = 48.4	I = 37.3	I = 43.9
	C = 35.1	C = 38.7	C = 29.8	C = 36.1	C = 34.6	C = 36.9
Walking	I = 35.3	I = 27.2**	I = 31.0	I = 30.9	I = 35.5	I = 30.3
	C = 33.9	C = 27.0**	C = 36.6	C = 28.1**	C = 36.2	C = 28.5**
Vigorous	I = 25.8	I = 28.4	I = 33.0	I = 20.6*	I = 27.2	I = 25.8
	C = 31.0	C = 34.3	C = 33.6	C = 35.8	C = 29.3	C = 34.5
MVPA (walking + vigorous)	I = 61.1	I = 55.6	I = 64.0	I = 51.6	I = 62.7	I = 56.1
	C = 64.9	C = 61.3	C = 70.2	C = 63.9	C = 65.4	C = 63.1
Boys, %
Sedentary	I = 31.0	I = 37.9*	I = 35.0	I = 43.3	I = 27.7	I = 36.5**
	C = 30.3	C = 33.4	C = 25.1	C = 33.0*	C = 27.5	C = 32.2
Walking	I = 39.3	I = 37.1	I = 35.6	I = 39.1	I = 40.4	I = 37.4
	C = 32.2	C = 34.9	C = 34.1	C = 33.1*	C = 35.3	C = 34.0
Vigorous	I = 29.7	I = 25.1	I = 29.4	I = 17.5**	I = 32.0	I = 26.1
	C = 37.5	C = 31.6*	C = 40.8	C = 33.9*	C = 37.2	C = 33.7
MVPA (walking + vigorous)	I = 69.0	I = 62.2*	I = 65.0	I = 56.7	I = 72.3	I = 63.5**
	C = 69.7	C = 66.6	C = 74.9	C = 67.0*	C = 72.5	C = 67.8

¹Different between baseline and 1-y follow-up at < 0.05* and P < 0.01**

Discussion

This study demonstrated the feasibility of making structural and other environmental changes to public school environments to potentially increase the levels of children’s PA and ultimately reduce their risk for obesity. In all, these changes did not result in an overall increase in MVPA or a decrease in sedentary behavior in either girls or boys. A potential explanation for not observing a change in the settings where structural changes were made was that the control schools took it upon themselves to purchase substantially more equipment during the course of the study, even though they received no encouragement to do so. Qualitative data indicated that 4 of the 6 control schools were led by principals with a dedication to health and PA, 1 of whom was forthright that he was going to purchase more equipment regardless of receiving resources from the project; another expressed disappointment that his school was not randomized into the intervention. This level of enthusiasm was not matched by intervention schools. Three of these schools presented considerable staffing and attitudinal barriers to project implementation. In any case, school principals uniformly were concerned about intrusions into the tightly scheduled instructional day. That concern was amplified by almost all teachers at the intervention schools and is a guiding principle for selecting school intervention strategies with minimal demands on teacher time or class time. Both boys and girls regardless of treatment condition showed a consistent increase in sedentary behavior and a consistent decrease in walking and MVPA. These results are in accord with an expected age-related decline in children’s PA for both boys and girls (26). Apparently, the intervention was not sufficiently powerful to affect these age-related changes.

A number of research trials have targeted PA and/or dietary change to prevent childhood obesity. Although some have achieved behavior change in specific areas, they have not resulted in weight reductions (27, 28). A review of 22 studies in Asia, Europe, and the Americas concluded that no single program showed evidence of effectiveness in preventing childhood obesity and that future research should use comprehensive strategies incorporating both behavior change and the creation of supportive environments (29). Although the current intervention was designed specifically for a Mexican-American community, the process described could be modeled by other interventionists to develop materials and programs for other ethnic populations or for the Latin American counterparts to the schools and participants in the study.

Limitations.

Despite some promising results, several limitations to the present study and intervention should be discussed. First, schools did not have physical education specialists, who ordinarily would serve in the role of PA champion. Instead, the project had to rely on teachers or parents who would volunteer to help implement the changes without (additional) compensation. Second, resource limitations led to our having only 5 d of observation at baseline and follow-up; thus, we may have missed time-specific programs that were being implemented and may have under-sampled observational data. Lastly, the gender stratified and area characteristic stratified analyses showed no consistent intervention effects. This may be due to reduced statistical power resulting from smaller sample sizes within the stratified comparisons. Nevertheless, future studies seeking to implement population wide interventions and with adequate resources to do so should continue to use this objective environmentally oriented methodology to evaluate program implementation and outcomes.