Design, Inclusion Criteria, Recruitment Setting and Procedure
The reporting of the HCP adhered to the Transparent Reporting of Evaluations with Non-Randomised Designs (TREND) statement [21] and was registered on Research Registry (trial number 2852). Following approval from Ulster University’s Research Ethics Committee, two schools from Northern Ireland (NI) were identified for a 2 (groups) × 2 (time-points) wait-list controlled trial. This entailed purposively selecting the intervention and control groups and staggering the delivery of the HCP across two school semesters whilst collecting data at the same time ([22] see Fig. 4). To reduce the potential for contamination, the control school Principal delayed announcement of the HCP until the following school semester, and both schools were unaware of their school’s data being compared during the intervention.
An inclusion criterion was based on the Multiple Deprivation Measure in Northern Ireland [23]. This index has seven domains of socio-economic deprivation including income, services and crime. Having identified schools of low SES on the measure, two schools with likewise demographics (i.e. mixed gender, urban, size) were approached for recruitment. Both school Principals agreed and invited all Primary five pupils to participate. Participant assent and parental consent were gained prior to conducting the research.
A group of trained researchers conducted baseline (week 0) and post-intervention (week 11) measurements (discussed below) with the pupils under quiet classroom conditions. The classroom teacher was present at all times.
Intervention
The HCP was delivered for 2 h and 15 min each week during school curriculum time for a 10-week period (i.e. 22.5 h of instruction in total). The intervention was in addition to general physical education classes and included (i) weekly hour-long practical sessions delivered by a trained sport student volunteer in tandem with, and under the supervision of classroom teacher and (ii) a ‘Daily Mile’ that involved the classroom teacher leading a 15-min walk every school day. SDT [1, 9] informed several aspects of the programme described below.
The weekly sessions consisted of a series of active discussions and physical tasks that contained messages around the health benefits of physical activity. The student volunteers received a teaching resource detailing language and techniques consistent with needs-supportive tenets in SDT [16], e.g. ‘acknowledge the activities were challenging and congratulate the children for trying their best’. Likewise, the classroom teachers also received a teaching resource including the above language and walking activities that would facilitate autonomy-choice for the children. For example, the ‘mirror image’ activity entailed walking partners completing the Daily Mile in tandem with a choice to mirror each other’s movements.
Student volunteers completed a two-day SDT training programme. The training was focused on facilitating the student volunteers’ understanding of a needs-supportive instructional style [16]. Their training included a discussion regarding the students’ experiences of Duda’s [24] empowering vs disempowering climate and a video evaluation of an authoritative-command vs autonomy-supportive teaching style using a rater proforma (see Fig. 1). The students were then presented with vignettes in which children were in need of competence or autonomy-support and were required to produce needs-supportive techniques to enhance engagement. Finally, the students completed a peer-teaching quality assessment of a Healthy Choices Programme session and were assessed in line with an adapted version of Reeve et al.’s [16] teacher observation sheet (see Fig. 2). In the case where improvement was recommended, the student volunteer was asked to reassess their understanding of the aims of the HCP and to engage the vignettes they encountered during training.
The classroom teachers completed a one-day training event in which they were guided on an autonomy-supportive teaching style during the Daily Mile and supervision of the weekly sessions. To link an autonomy-supportive teaching style with relevant teaching practices in the Northern Ireland Key Stage Two Curriculum [25], ‘active learning’ techniques were used. Active learning entails creating a learner-centred environment, in which the children are encouraged to participate in the direction of a lesson through questions, activity choice and feedback [25]. The teachers were asked to develop active learning techniques they could utilise throughout the weekly sessions and Daily Mile (e.g. use of questions, positive feedback, allowing the students to choose content).
In addition, to enhance the children’s relatedness support, parents and/or guardians participated in an insight afternoon. Through consultation, it was decided to update the parents on the HCP through information flyers and videos uploaded to the school’s online ‘parent space’.
Outcomes
Objective MVPA during school days (i.e. Monday to Friday) and school hours (i.e. 9 am–3 pm, see [26] for time category classification) was measured using Actigraph accelerometers (GT3X and GT1M, Pensacola). The accelerometers were fitted onto the children’s waists with an elasticated belt and positioned on the midaxillary line above the right hip. The devices recorded data in 5 s epochs, a valid capturing period for 8–9-year-old children’s movement [27]. Wells et al.’s [28] wear-time criterion was applied, including at least 8 h wear per-day for a minimum of three weekdays. Children meeting the criteria at both time-points were selected as the ‘valid sample’. Time spent in health-enhancing MVPA intensities [1] were calculated using Evenson cut points [29] deemed the most valid and reliable for 8–9-year-old children [27]. Accelerometer counts of < 20 min of consecutive zeroes, or > 15,000 were removed, as they are considered biologically implausible [27]. For analyses, one variable reflecting the children’s average school-day MVPA was created.
Well-being was measured using the 7-day recall Kidscreen-27 questionnaire [30]. Kidscreen-27 has demonstrated excellent psychometric properties with children aged 8–18 [30] and was recently validated with Irish children of low SES [31]. Kidscreen-27 assesses seven physical, social and psychological well-being dimensions [31], and for analyses, a single variable reflecting the total of the 27-items was created.
To assess the degree to which the children felt their teachers supported their need for autonomy, a modified version of Standage, Duda and Ntoumanis’s [32] Physical Education (PE)-adapted Learning Climate Questionnaire was employed. As the HCP involved physical activity outside of PE (i.e. through the Daily Mile and weekly sessions), the items were modified to reflect autonomy-support during physical activity classes. The scale included six items and responses preceded with the stem: ‘In physical activity classes my teacher…’, and were scored using a 7-point Likert scale ranging from ‘strongly disagree’ to ‘strongly agree’. A confirmatory factor analysis (CFA) revealed support for a single latent factor (χ2 = 13.961 (9) p = .124; CFI = .947; TLI = .912; RMSEA = .063). A scale total was created for analyses.
Children’s perceptions of psychological needs satisfaction (i.e. autonomy, competence and social relatedness) in the context of physical activity were assessed using an age-appropriate questionnaire [33]. The questionnaire included 18 items scored a 5-point Likert scale ranging from ‘not like me at all’ to ‘really like me’ and encompassed three 6-item subscales for autonomy, competence and relatedness. After the omission of the two negatively worded items (item 4 autonomy, and item 12 competence), a CFA within the sample revealed a good-fitting three-factor model with covariance paths between the latent variables (χ2 = 152.789 (99) p = .000; CFI = .920; TLI = .903; RMSEA = .065). A total needs satisfaction variable was created for analyses.
Four dimensions of SDT’s motivation continuum were measured using an age-appropriate questionnaire [33]. The questionnaire included 12 items encompassing four 3-item motivation subscales (i.e. intrinsic motivation, identified regulation, introjected regulation and external regulation) answered on a 5-point likert scale ranging from ‘not like me at all’ to ‘really like me’. A four-factor model consisting of two latent co-varying factors (i.e. identified with intrinsic motivation and introjected regulation with external regulation) yielded an unacceptable fit. However, correlating three items (i.e. item in 1 intrinsic motivation with item 2, and 10 in identified regulation; and item 11 in introjected regulation with item 12 external regulation) theoretically aligned with Ryan and Deci’s [9] conception of autonomous and controlled motivation in SDT, subsequently yielded an acceptable fit (χ2 = 81.982 (45), p = .001; CFI = .937; TLI = .907; RMSEA = .077). Scale totals for each dimension were created.
Data Management
Raw data from each individual questionnaire was manually inputted into SPSS (Version 22; IBM Corp., NY). Ten percent of questionnaires were checked as a quality assurance procedure. The expectation maximisation algorithm was conducted on each independent scale to estimate missing data after Little’s Missing Completely at Random test confirmed that the data was missing at random on both time-points (p > .05).
Statistical Analyses
Two models subscribing to Fortier, Duda, Guerin and Teixeira’s [11] SDT model for health interventions were specified. The aim of testing the models was to determine if changes in the children’s perceptions of autonomy-support (from teachers) would indirectly affect changes on the primary outcomes of MVPA (model 1) and well-being (model 2) through needs satisfaction and intrinsic motivation (see Fig. 3).
The independent variable (X) was coded as a dichotomous variable (control = 0 and intervention = 1). Difference scores were created by subtracting post-intervention scores from baseline. MVPA and well-being were coded as dependent variables (Y). Mediator 1 (M1) refers to autonomy-support, mediator 2 (M2) as needs satisfaction, and mediator 3 (M3) as intrinsic motivation. Intrinsic motivation was selected as M3 because it is assumed and has been empirically found to yield the most adaptive outcomes in terms of increasing MVPA in children [3] and well-being [13] (see Additional file 1 wherein identified regulation, introjected regulation and external regulation were selected as M3).
The procedures described by Hayes [34] were used, testing one direct effect between X on Y (\( \overset{\acute{\mkern6mu}}{c} \)) and seven singular or serial indirect effects between X on Y through M1, M2 and M3. Hayes’ model also examines three direct and three indirect effects between X on the three mediators. The results can confirm if the effect of X (intervention) on Y (outcomes) is either (i) not significant, (ii) fully explained by the mediators (i.e. full mediation), (iii) partially explained through the mediators (i.e. partial mediation) or (iv) indirectly explained through the mediators (i.e. indirect effects) [35].
Two figures were produced specifying beta (β) coefficient values for each direct path and r2 values related to the proportion of total variance predicted in model 1 and model 2. A table was created to detail the completely standardised effect sizes and confidence intervals for each of the seven indirect effects of the intervention on the dependent variables. If confidence intervals did not cross zero, the indirect relationship was interpreted as statistically significant [36]. For improved accuracy, the models were tested with 5000 bootstrap samples [35]. Analyses were conducted using Hayes’ [37] PROCESS macro for SPSS (Version 22; IBM Corp, NY).