Study Design
The detailed protocol for this pilot trial has been reported elsewhere [16]. As the theory of Ecological Dynamics was introduced post hoc for designing this study, it was not introduced in the protocol [16]. The efficacy of the ecological blended “Quantity + Quality” intervention was evaluated using a small-scale randomized controlled trial (RCT), with a three-arm design which focused on approximately 9-year-old (4th grade of primary school) students from one Hong Kong primary school, as previous research has reported a significant increase in sedentary behaviour in children aged 11 years and older [17]. This public school was located in the New Territories, Hong Kong, and according to the school report [18], all the 4th grade participants’ families had a middle socioeconomic level. The study received ethics approval from Survey and Behavioural Research Ethics of the Chinese University of Hong Kong (Reference No. SBRE 18-108) and all the participants provided written parental consent forms prior to participation.
Randomization was conducted using Google random number generator to randomly assign the students to the intervention classes or control condition using a 1:1:1 ratio. Outcome data were collected at baseline, post-intervention and 3-month follow-up for all the variables. Baseline data collection was conducted before the intervention began in January 2019, the post-intervention measure for all the students was performed successively 13-week after the completion of the intervention during the end of semester in July 2019, and the follow-up measurement in all participants was performed 3 months after at the start of the next semester (October 2019). Reporting of the trial follows the CONSORT statement (Additional file 1).
Participants and Intervention
A sample size of 20 in each group (recruiting 24 with an assumed 20% attrition) would have at least power of 80%, an alpha of 0.05 and a moderate effect size (f = 0.25) [19] calculated by G-power software. All the grade four students (n = 133) were invited and distributed parent consent forms, and 81 students (response rate = 61%) agreed to participate and returned the forms. The children were excluded if they had a disability that prevented periods of standing or had an injury or illness that limited performing normal daily tasks. All the participants were randomly assigned to either of the three conditions (Fig. 1), which included a blended PL group (combining sit–stand desks and play-based recess; SSPlay), a single play-based group (Play), and a control group (CG).
Thirty height-adjustable sit–stand desks (Askisi 720, SMART Inc., USA) were placed in the SSPlay students’ classroom of the school for two academic semesters. Similar to Hinckson, Salmon [20]’s descriptions, the equipment could be moved up and down manually with the use of a lever and allowed the children to work in a seated or standing position. Prior to the intervention phase, a 3-h briefing session regarding the instructions on how to administer the sit–stand desks and PA-based recess was held for all the teachers and parents in order to support them in the development of classroom environment within schools. The research plan of breaking up prolonged sitting every 15 min during two regular classes (each class before the recess) per day could ensure all children in SSPlay group use the sit–stand desks for at least 1 h per day on average across the week [21]. Stools or chairs were retained in the classroom for them to feel free to choose whether they sat or stood when using the sit–stand desks. Teachers encouraged and led each child to stand when 15-min prolonged sitting was reached in the classroom (“Quantity”, e.g. both sit-to-stand transitions and stand-to-sit transitions). Besides, SSPlay and Play children also participated in a play activity during recess time, in which the unstructured outdoor interactive games were introduced to children and led by PE interns during recess (“Quality”, e.g. extra play-based opportunities). The extra play-based activities were up to 15 min in duration and twice a day across the week, which includes games such as skipping rope, shuttlecock kicking, hide-and-seek in the specific area, and supplemented with several minutes of cooling down. The students who were assigned to the control condition adhered to their regular class schedules and lesson delivery. They used their standard classroom desks in the classroom, with no experimental changes made to their classrooms.
Measures
Children’s height and weight were measured by trained appraisers using standardized procedures, with children in light clothing and shoes removed, using TANITA measuring boards (Tanita RD-545-sv, Tokyo, Japan) and Seca (model 770 scales, Hamburg,
Germany). Body mass index (BMI, kg/m2) was then calculated from the measured weight (nearest 0.1 kg) and height (nearest 0.1 cm), with the standard equation (body weight [kg]/height [m]2).
PA was measured by ActiGraph GT3X + (Actigraph LLC, Pensacola, FL, USA), which were worn on the children’s waists for seven consecutive days. Monitors were initialized prior to data collection and were set to begin collecting data at the start of the school day on the Monday of every week. Data were collected in 10 s epochs to account for children’s natural activity levels, which generally occurred in short bouts [22] as it was shown to present the most acceptable classification accuracy for accelerometer use among children. Evenson cut-points (MVPA ≥ 2296 counts min − 1) were applied to intensity levels. As suggested by Aadland, Andersen [23], the non-wear period refers to a 45 or 60-min consecutive zero count-criterion in paediatric studies. The monitors had to be worn for at least 400 min/day for a minimum of 4 days, with at least one valid weekend included [24]. The accelerometers could be removed only during water activities, such as showering or swimming, and the participants had to fill in the information provided in the log sheets.
Physical literacy was measured using the Chinese version of the Canadian Assessment of Physical Literacy, second edition (CAPL-2, Chinese) [25], which was the first to comprehensively assess children’s PL in the Chinese context, comprising of four domains: Daily Behaviour (30 points); Physical Competence (30 points); Knowledge and Understanding (10 points); and Motivation and Confidence (30 points). The total achievable score for this assessment was 100 points. The whole CAPL-2 (Chinese) model was reported a good fit with construct validity: Chi-square (χ2 = 70.16, df = 43, p < 0.05), root mean square error of approximation (RMSEA) = 0.04, 90% CI (0.024–0.062), comparative fit index (CFI) = 0.94, Tucker–Lewis index (TLI) = 0.90, to be adopted to evaluate children’s PL [25].
Children’s aspects of cognitive function were assessed by two computer-based tasks, all of which were performed using the Inquisit 5 platform. Participants were required to perform the tasks in a quiet room under the supervision of an instructor who was trained prior to the testing. The classical version of the Wisconsin Card Sorting Test (WCST) with the standard number of 128 cards [26] was adopted to measure cognitive flexibility and working memory. This task consisted of 4 key cards and 128 response cards. Participants were instructed to sort the response cards, shown at the bottom of the screen, according to the characteristics of the key cards presented on the screen’s upper side, comprising the following categories: colours (red, green, yellow and blue); forms (triangle, star, cross and circle); and numbers [1,2,3,4]. The instructor was permitted to provide instructions relating to the categories either prior to or during the task, while feedback on “correct” or “incorrect” was presented after each selection. It took each participant approximately 20 min to complete the task. Both total and perseverative errors were recorded as target variables, since an increase in any of these variables indicated cognitive flexibility impairment [27]. The Tower of London Task, a widely administered neuropsychological assessment, was used for measuring the planning and problem-solving [28]. The task consisting of a practice trial and 12 test trials required the participants to move beans to solve problems. When presented with a graph on the screen showing three vertical pegs with graded heights and each holding beans (either 3, 2 or 1), the participants had to move the beans so as to be identical to the goal graph, without violating the rules [29]. It took each participant approximately 20 min to complete the task. Both the total correct and total move scores were derived for analysis, given that these variables were found to be influenced by sport-related exercises [29].
Statistical Analysis
According to the CAPL-2 (Chinese) [25], missing raw scores in Physical Competence domain could be replaced using the recommended algorithm stated in the manual; therefore, the techniques were adopted to replace missing values for PL in the Physical Competence domain. For the measure of PA, individual information-centred approach was adopted for substituting missing data points [30]. This method was demonstrated as an effective method and superior to the group information-centred methods for handling accelerometer missing data when 7 days of data were collected. For each domain of PL, the missing values belonging to the Physical Competence domain and the Daily Behaviour domain could be calculated according to the fraction that the CAPL-2 (Chinese) manual provided. A maximum of one protocol could be completely missed and still have a calculated score [25].
Descriptive statistics were expressed as means and standard deviations for continuous variables, and gender was shown with percentage in male. All data were imported into SPSS version 23 for analysis. An α level of 0.05 was used for all statistical tests. Shapiro–Wilk and Levene tests were used to check the normality and homogeneity of data for both univariate and multivariate normality. Multivariate analysis of variance (MANOVA) test was used to assess between group comparisons at baseline, followed by the post hoc pairwise comparison with the Bonferroni adjustments. A two-factor mixed-design ANCOVA was conducted to assess the change in dependent variables over the 3 time points between groups, separately. Adjustments were made for sex, age and BMI category. Effect sizes (ESs) using partial eta squared were calculated and reported, large effect with η2 ≥ 0.14, medium effect with 0.14 > η2 ≥ 0.06, and η2 < 0.06 indicating small effects [31].