The STEP study is a cluster randomized, controlled six-month trial to examine the effects of a SBS-PA program implemented during the designated gross-motor playtime during the preschool day in preschool-aged children. Outcomes of interest include total preschool-day PA, percent of time spent in MVPA, and sedentary time during gross-motor playtime and total preschool-day. Preschool centers were randomly assigned to one of two groups: SBS-PA or traditional unstructured PA (UPA). All children within a given preschool center were allowed to participate in the assigned intervention. This study protocol is in compliance with the Declaration of Helsinki and has been reviewed and approved by a University of Massachusetts, Amherst human research review board.
Participant recruitment and eligibility
A total of 15 preschool centers with a minimum of two full-day classrooms that serve low-SES areas in Springfield, Massachusetts area were approached to participate in STEP. Ten of the 15 preschool centers approached met the screening criteria and agreed to participate in the study. Reasons preschool centers declined participation included unwillingness to be randomized and inability to consistently provide 30 min of morning and afternoon gross-motor playtime per preschool day. Although not a requirement for eligibility, we assessed the existing PA policies for each of the 10 preschool centers enrolled in the study for randomization/stratification purposes. Existing PA policies were assessed using a modified version of the Environment and Policy Assessment and Observation Audit Tool (EPAO). The assessment included written PA policy, teacher training in PA and involvements with children during gross-motor playtime, and indoor and outdoor PA equipment.
Within each preschool center, between 25 – 30 children were individually recruited for the assessment portion of the study. Children were eligible for the assessment portion of the study if they: 1) attended one of the 10 participating preschool centers; 2) were between 2.9 – 5 years of age on the date of baseline assessment; 3) did not have a condition limiting their participation in MVPA (e.g., unable to participate in routine gross-motor playtime at preschool); 4) unable to wear the activity monitor, and 5) attend preschool full-time. Children that were excluded due to eligibility criteria or whose parents did not wish to sign the informed consent were still allowed to participate in the intervention protocol; however no assessment data were collected on them.
The STEP study utilized a cluster-randomized design because within a preschool center, classrooms sometimes share resources during gross motor playtime and thus, contamination between intervention arms would be unavoidable. Preschool centers were randomized into the treatment (SBS-PA) or control (UPA) group, stratified by preschool center size and existing PA policy. Randomization was performed using the Efron procedure .
Due to the PA requirements set forth by the Massachusetts State Department of Early Education & Care, all participating preschool centers in the current study were mandated to provide 60 minutes of gross-motor playtime per day. Typically preschool centers divide this playtime into two 30-min sessions—one in the morning and one in afternoon. For this study, both the treatment and control interventions were implemented during the designated morning and afternoon gross motor playtimes five days per week for six months.
The study intervention was developed based on the social ecological model and the Meta-Volition Model (MVM) . The MVM’s fundamental premise is that organizational practice and policy change is the key lever for population health behavior change. Of particular relevance for this project, the MVM posits that as classroom teachers experience the ease of implementation and benefits of the intervention for their own work (e.g., improved student attentiveness after PA breaks), this will reinforce and cement their adherence. These teachers will also comment favorably to other teachers, spurring engagement of additional classrooms and, ultimately, institutionalization within the center. MVM is built upon Social Cognitive Theory [13, 14], and Diffusion of Innovations . Social cognitive theory asserts that behavior develops, is altered, and maintained through the interplay of personal, behavioral, and environmental factors [16, 17]. The Social Cognitive Theory construct most pertinent to the MVM is observational learning from role models  to increase self-efficacy and behavioral adoption [12, 16, 17, 19]. Classroom teachers are thought to be the second largest category of role models among preschoolers, after family members. The involvement of classroom teachers accomplishes many things: it makes activity participation more appealing and culturally adaptable; it provides highly-valued but inexpensive incentives for participation that may help to reinforce the activities and may help generalize them to other settings; and it represents a more feasible and sustainable practice and policy change in low-resource schools than those requiring expensive equipment or expert personnel.
Additionally, these theoretical models include personal factors: the children’s and teachers’ interest and participation in SBS-PA; the children’s belief that they have the ability to perform behaviors that will secure desired outcomes (self-efficacy); and outcome expectation (the intervention will result in more fun for the students and more manageable students for the teachers). Preschoolers have developed some aspect of self-efficacy because at this age they know that their personal action can produce certain results. Behavioral factors include the knowledge and skills available and needed to perform the intervention; and the degree of competence attained in using these skills. The structured activities utilized in this intervention will be developmentally appropriate for the motor skills of preschool-aged children. Of the three determinants of behavior development, environmental factors (social or physical) will have the greatest relative influence on behavior in this intervention. With respect to this intervention, the social environment includes modeling by peers and adults (teachers). Preschoolers are able to pay attention and therefore peer and adult modeling of the intervention will enhance the children’s self-efficacy for learning and their ability to successfully perform the activities. The physical environment deals with the structurally integrated approach of the intervention (changes in routine, e.g., to plug in a DVD/CD or rearrange furniture). The intervention task has built-in reinforcement because the activities are designed to be fun.
Development of short bouts structured PA intervention: (tutti fruitti instant recess)
Tutti Fruitti Instant Recess (TFIR) was based on the Instant Recess® (IR) program, which has been utilized in different age populations [12, 20–23]. IR is 10-min physical activity routines, set to music with simple movements that engage major muscle groups. IR is available in audio (CD) or audiovisual (DVD) format, and is utilized to incorporate active breaks into routine organizational settings (e.g., the workplace, church activities, classrooms) for adults and youth. The SBS-PA intervention (TFIR) consisted of 10-min routines recorded onto DVDs that featured music and movements appropriate for preschool-aged children. A total of 16 audiovisual DVDs were created—ten existing IR routines that were originally geared toward adults (http://www.toniyancey.com) were modified for the current study, and six new TFIR routines were created by the research staff. The TFIR program was created in collaboration with a physical education specialist. All routines were designed to be simple, easy to learn and included low- to moderate-impact aerobic movements. Within each routine, movements (choreographed to music set at 100–120 beats/min) started with lower body actions before the addition of upper body movements. Each 10-min TFIR routine consisted of a warm-up (1 min), moderate- intensity activity (8 min) and a cool-down (1 min). Warm-up and cool-down movements were set to the same musical tracks, while the 8-min moderate-intensity movements were set to two 4-min tracks. Each TFIR DVD was accompanied with a movement-by-movement verbal and photo guide to assist classroom teachers in implementing the intervention.
Implementation of SBS-PA intervention
The TFIR DVDs were implemented during the first 10 min of the designated 30-min time for gross motor play (in the classroom setting). To implement the TFIR protocol, classroom teachers and students followed along with the TFIR DVD on a portable DVD player (which was provided to each classroom by the study staff for the duration of the study). TFIR DVDs were intended to be viewed by the classroom teacher while the students watched and followed the teacher’s lead. After the completion of the 10-min TFIR DVD, the students were allowed to engage in their usual gross motor playtime activities (unstructured play) for the remaining 20 min. The simplicity of the TFIR protocol enhanced teachers’ ability to implement the intervention. The TFIR protocol was implemented during the morning and afternoon designated gross motor playtime. Two TFIR DVDs (one for morning and one for afternoon playtime) were distributed to classroom teachers each week to ensure that preschool centers were implementing the same TFIR DVD during the intervention times. Each week, teachers were instructed to implement the assigned morning and afternoon DVD at the respective gross motor playtime. Each set of TFIR DVDs with its accompanying picture and verbal movement guide were dropped off the Friday preceding their implementation to give classroom teachers time to become familiar with the routine for following week. The previous week’s TFIR DVDs were picked up on Monday. Throughout the 6-month study, each morning and afternoon set of TFIR DVDs were repeated every eight weeks for a total viewing of three times.
The control intervention (UPA) consisted of traditional long bouts (30 min) of gross motor playtime and included supervised free time (unstructured) for the preschoolers to play on their own or with other children. During the gross motor playtime (either indoors or outdoors), children were provided with their usual play equipment, which was the current standard in the participating preschools. Children were allowed to choose from a variety of activities like running and jumping, bouncing, catching, and playing on structures or obstacle courses.
Teacher training and process evaluation
To enhance the fidelity to the intervention, within each preschool center all classroom teachers participated in a two hour in-service training session. The training sessions were held separately for staff at each preschool center. Training for teachers assigned to the SBS-PA intervention, included information on the importance of PA and how to implement the TFIR protocol. During the training session, teachers were taught how to lead preschoolers in TFIR activities and were guided through three different TFIR routines. Training of the teachers from the control centers focused on the importance of allowing their students to play freely during the allocated gross motor playtime. All participating teachers were expected to follow their assigned protocols during designated gross motor playtime within each normal preschool day throughout the study. Research staff members were available throughout the 6-month protocol to assist classroom teachers in implementing their assigned intervention. At least once each week, research staff observed each participating classroom to determine if both the SBS-PA and UPA interventions were being implemented as designed.
Assessments and measures
Participants total preschool day PA was assessed at baseline, and at 3-months and 6-months post-initiation of the intervention with the Actigraph accelerometer (GT1M, Actigraph, LLC, Pensacola, FL), which has been previously validated and used in preschool-age children [4, 24]. The accelerometer was attached to an adjustable elastic belt and worn around participants’ waists at the center of their lower back to be unobtrusive . We asked that children wear the accelerometers during the preschool day from Monday thru Friday during each assessment time point. Accelerometers were placed on children upon their arrival at their preschool center and removed before departing for home. Classroom teachers were instructed on monitor placement and asked to ensure accurate repositioning of the monitor whenever removed. The monitor was programmed to store data at 15-s epochs daily. To reduce the baseline PA accelerometer data, a custom software program was used to process all data using the Sirard et al. 15-s epoch count cut off . The age-specific, 15-s counts cut-offs for 3, 4, and 5 year olds for the different activity intensities were sedentary ≤301, ≤363, ≤398; light 302–614, 364–811, 399–890; moderate 615–1230, 812–1234, 891–1254; and vigorous ≥1231, ≥1235, ≥1255, respectively . For the study population, the preschool-day was defined as 7:00 am – 4:30 pm. Data from the 15-s counts cut-off were converted to average counts/min and the percent of time spent in the various activity intensity thresholds and used for analysis.
The Observational System for Recording Physical Activity in Children-Preschool Version (OSRAC-P) [28, 29] was used to collect information about participant PA during gross motor playtime and the contextual circumstances (location of PA, group composition, and activity initiator) of their PA within their preschool environment. Direct observation of gross motor playtime PA was conducted on one day (in the morning and afternoon) during each assessment time point. Within each preschool center, two randomly selected classrooms were observed. Prior to the beginning of the data collection session, researchers randomly selected eight children (six to be observed, two alternates) to observe over the course of the gross motor playtime session. Children were observed for 15-s intervals and then, during the next 15 s, recorded one code for each of the four variables of interest (PA level, location of PA, initiator of PA, and group composition). Each child was observed for one, five-minute session, with the observer rotating which child was assess every five minutes. An alternate child was used only in the event that the child who was being observed left the observation area. Observational data were collected on HP IPAQ handheld computers and later downloaded into a database and analyzed. All observers underwent training and were required to demonstrate high inter-observer reliability prior to assessing PA using OSRAC-P. Five to 10% of all sessions were analyzed to assess inter-observer reliability.
A self-report questionnaire, completed by parents/guardians, was used to obtain each child’s demographic and SES data. Standing height to the nearest millimeter (direct reading stadiometer) and body weight to the nearest 0.1 kg (digital scale) were assessed with participants wearing light clothing, with shoes removed. Body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in meters. Parents/guardians reported the amount of time their child engaged in sedentary activity (e.g., watching television, playing video games, art work or crafts) using a self-reported survey [4, 30, 31]. Anthropometric measures and sedentary behavior (via questionnaire) were assessed at baseline and 6-month after the initiation of the intervention.
The primary objective of the STEP study is to examine the effects of short bouts of structured PA to increase preschool-age children’s total school-day PA, as well as percent time spent in MVPA during the school day. The preschool center is the unit of randomization and intervention; the individual child is the unit of measurement. The analyses will use the child-level data, accounting for correlation of child responses within class and of classrooms within a center. For all analyses of primary outcomes and mediating variables we will use generalized linear modeling statistical methods. The methods allow assessment of a simple treatment comparison as well as multivariable modeling. Mediator and modifier significance (set a priori at a significance level of 0.25 or below) will be assessed univariately to determine which variables will be eligible for inclusion in the multivariable model. Intervention effects (change in PA) will be assessed with repeated measures mixed model analysis of variance or covariance, with childcare center as a random effect given that we do not expect that the within-group time trends will be homogeneous. Using a mixed effects model allows us to incorporate random effects for the center and class within center (clusters) and thereby model the correct covariance structure of the data and produce more accurate estimates of the effects under study and their standard errors. The distributions of individual variables will be evaluated and the data transformed as necessary to meet the assumptions of normality and homoskedasticity. For categorically scaled measures, we will use the generalized estimating equation methods for logistic regression analysis. Demographic variables will be added to multivariate models along with potential interactions with the intervention. Significant (at the p = 0.05 level) variables and interactions will be retained in the final model. For the baseline characteristics statistical comparison between groups was made with Wilcox Rank Sum tests for scaled variables and Chi-Square tests for categorical variables, with a two-sided alpha = 0.05. All analyses will be computed using SAS (version 9.2).
Sample size and power calculations
Sample size was determined using methods for studies with randomization by cluster . With 25 children per preschool center and five preschool centers per condition, we have 80% power to detect a significant difference in the mean change in number of minutes of PA (2-sided) between the two conditions of between 5.5 and 10.7 min of total school day PA at the 5% significance level, depending on the correlation ranges above. Thus we have adequate power to detect a difference that would be significant both statistically and clinically. Based on the assumption that repeat measures of PA within clusters will only be moderately related after adjusting for repeat measures at the individual level, this power will hold similarly when analyzing the full (3 observation points) study using a random coefficients model .