Active City of Liverpool Active Schools and SportsLinx Project: a Clustered Randomised Controlled Trial (A-CLASS)

Study design

Treatment arms in this four-arm parallel-group cluster randomised controlled trial (RCT) included a high intensity physical activity (HIPA) intervention group, a fundamental movement skill (FMS) intervention group, a physical activity signposting (PASS) intervention group and a control group (usual practice).

Recruitment

- Organisation level

Sixteen primary schools were targeted by the research team in May-July 2006. The schools were targeted based on their size (student enrolment >400 primary school; >250 junior school), current afterschool club provision (limited), school sport facilities available for use (bi-weekly) and socioeconomic status of the school postcode classified by postcode in the Index of Multiple Deprivation as deprived (IMD > 40). Signed consent was sought and obtained from the head teachers of the sixteen schools for pupil recruitment, study contact, laboratory visits during school time and access to school facilities. Eight from the sixteen schools were randomly selected to take part using random number allocation.

- Individual level

All children in year 5 in consenting schools received a verbal and written overview of the study through a researcher led study information session held at the respective schools. Children were given 2 weeks to express interest via the return of a written assent form (children) and parental/guardian consent form to their school teacher. Medical questionnaires were distributed to all children who agreed to take part in the study, and they were assessed for stature and body mass in order to calculate body mass index (BMI; kg/m2). BMI was used as guide to target the children who were overweight or may be nearing overweight according to BMI cut-off thresholds stipulated by Chinn and Rona (2004). Generally, 20-25 children with the highest BMI within each school who were free from the presence of chronic disease, metabolic disorders, motor or co-ordination difficulties and prescribed medications including steroids inhaled by asthma sufferers were then enrolled to participate in the project. There was no racial or gender bias in the selection of participants.

Group assignment and Intervention

Participating schools were randomly allocated by a computer generated procedure to one of four treatments to reduce risk of contamination effects across the trial.

Data collection

Measurements took place at three time points: at baseline (month 0; September to mid-October 2006); end-intervention (9 months; June to mid-July 2007); and follow-up (3 years; October 2009). At each time point participants attended University laboratories for individual assessments. At each time point participants' habitual physical activity was objectively assessed using accelerometers, and participants' fundamental movement skills were assessed at their respective school. Prior to laboratory visits, participants were instructed to fast for a minimum of 8 hours and avoid strenuous exercise for 24 hours.

Sample size

It was feasible to recruit eight schools and randomly assign two schools to each of the four arms. With an estimated number of consenting Year 5 pupils of 20 per school, the planned sample size was around 40 participants per arm. Allowing for 10% attrition at 9 months and a design effect of 1.2 to account for school-level clustering (ICC of 0.01), our effective sample size was 30 participants per arm. This sample size provides approximately 80% power at 2P=0.05 to detect a targeted difference between intervention and control (3 planned comparisons) at the 9-month timepoint of 3 units for the total FMS skills score (based on an between-subjects standard deviations (SD) of 7 units and a reliability of r=0.8 over the timeframe of the intervention; ANCOVA model adjusting for baseline FMS score). With an anticipated attrition of up to 50% for the 2-year follow-up timepoint, our effective sample size was only 16 participants per arm. Group comparisons at this timepoint are therefore defined as exploratory, as we have relatively low power to detect the same effect size (around 50%).

Statistical analyses

There are too few clusters per arm to account robustly for the hierarchical data structure using linear mixed (multilevel) modelling, generalised estimating equations, or clustered robust standard errors. Therefore, data will be analysed at the individual level, with standard errors inflated by the square root of the design effect. The change in FMS score from baseline to post-intervention (9-months) will be compared between arms using a regression model (ANCOVA), adjusting for baseline FMS score and sex. There are 3 planned comparisons comprising each of the 3 interventions vs. usual practice. Point estimates will be derived together with uncertainty expressed as 90% confidence intervals. For the primary outcome, a purely exploratory sub-group analysis will be conducted using a sex-by-intervention group interaction term to examine the potential for differential intervention effects in boys vs. girls. Secondary outcomes will be analysed using the same general modelling approach, but with no inferential emphasis placed on the results. Where appropriate, a principled method will be applied to address missing data (e.g., multiple imputation or full information maximum likelihood).