The City of Moreland, a local government municipality (population of 135,205 in 2006) is located 8.5 km north west of the central business district of Melbourne, in South Eastern Australia. Of the 31 Melbourne municipalities, this area ranked seventh in social disadvantage at the time of the study.
Census data from 2011 indicated that the majority of the residents in Moreland (58%) spoke English (only) at home compared with 71% of the population average across the Melbourne Statistical District. This municipality also has one of the highest levels of residents who belong to the Catholic and Islamic faiths (36% and 10%, respectively, compared to 30% and 4% across Melbourne). However, there is marked variation in demographic and economic background across the municipality, and it has shifted over time towards a higher socio-economic profile as housing demand and inner-urban location has resulted in families with higher median incomes moving into the area [17,18,19].
fun ‘n healthy in Moreland! was funded by three departments of the state government (Sport and Recreation, Health, and Education). Intervention staff (Community Development Workers) were employed by Merri Community Health ServicesFootnote 1 and research and evaluation staff were employed by Deakin University (2004–7) – relocating to University of Melbourne (2007–9). This model ensured that Merri Community Health Services had influence and leadership on the design and implementation of the intervention, in partnership with the schools, maximising the chance of high impact and sustainability. The implementation of the trial and evaluation study was governed by a project team comprising both researchers and Merri Community Health staff who consulted regularly with school staff, families and community leaders. Key research decisions were referred to the full team of 11 investigators, including representation from each discipline and content area relevant to the conduct of the study. Additional advice was provided by an internal Merri Community Health Services Staff Advisory Committee during the development of the study, and an external committee of government stakeholders which met annually.
The design and implementation of the intervention was underpinned by the WHO Health Promoting Schools Framework, an evidence informed decision making process, and the International Obesity Task Force ‘10 guiding principles for obesity prevention’, which state that health promotion initiatives be empowering, participatory, holistic, inter-sectoral, equitable, sustainable and multi-strategy. The Health Promoting Schools Framework (HPSF) is based on health promotion theory and is consistent with a socio-environmental theoretical framework . HPSF has been widely used and developed to assist schools to address health issues over the past decade [6, 21]. The advantage of the HPSF is that it is designed to guide multilevel interventions to account for environmental, sociocultural and individual influences on health behaviours. It allows for a community participatory approach  which was extended in this study to include models of cultural competence to guide the engagement of the culturally diverse community in the Moreland area [23,24,25]. “Cultural and linguistic competence is a set of congruent behaviours, attitudes, and policies that come together in a system, agency, or among professionals that enables effective work in cross-cultural situations” .
Schools were supported to develop fun ‘n healthy programs according to the fixed requirement of a whole school combined focus on increasing fruit, vegetable and water consumption, increasing physical activity and encouraging positive self-esteem in children. Within the intervention schools, the school community determined the exact content of the program strategies, based on interventions that had demonstrated evidence of implementation or success in previous studies, or innovative programs which had a strong likelihood of success. The fun ‘n healthy in Moreland! study offered schools the support of Community Development Workers (CDWs) for the 3.5 year intervention period from Jan 2006 – June 2009 who acted as knowledge brokers, providing information and guiding schools' customised development of intervention program strategies and their efforts to resource them. Three full time CDWs provided support to 4 schools each in the first 2 years. This then reduced to 2 full time CDWs providing targeted support to schools based on need. This support ensured that the strategies followed health promotion principles in creating a supportive and sustainable environment, customised for the school community to achieve changes in relation to the school system, policy, curriculum, environment, and child behavior and health outcomes. The CDWs were in turn supported by the Research Program Manager (LGi) to enable shared problem solving and links with evidence-informed approaches.
The aims of the intervention were to:
Reduce overweight and obesity and improve child health and wellbeing
Improve child and family dietary intake, increase child and family physical activity and reduce child sedentary behaviours
Improve knowledge and skills of school staff, family and children regarding sustainable strategies for healthy eating, physical activity and environmental changes
Develop sustainable positive changes in school, home and community environments (system integration, policies, physical, social, and community connections)
Examine contextual and programmatic features of the intervention that impact on results.
Specifically, the logic of the approach was underpinned by a hypothesis that changes in the school environment in terms of policies, programs, curriculum, physical environment and parent engagement would result in changed parent and child knowledge and behaviours, and with sufficient time lead to improvements in health and wellbeing and weight status of children.
School selection and recruitment
Schools were eligible to participate in the study if they were located in the Moreland municipality and exclusively covered the primary (elementary) school-aged group, aged 4–13 years (n = 36 schools). All school principals of primary schools in the Moreland municipality were contacted by phone by the Research Project Manager (LGi) and invited to participate in the study.
A Plain Language Statement detailing the study and research process, and a school principal consent form, were provided to all schools who expressed interest in being involved. Schools which returned the consent form were included in the study, resulting in a sample of 24 schools (65%) (Fig. 1). All children attending the consenting schools and their parent/guardian were invited to participate.
Following recruitment and baseline data collection, schools were randomised using computer-generated random numbers to either actively engage with the fun ‘n healthy in Moreland! program (intervention arm) or continue with normal school activities and programs for healthy eating and physical activity (comparison arm). Intervention schools were provided with a memorandum of understanding which clearly articulated the parameters of the intervention and the respective rights and responsibilities of each participating organisation (school, community health service and university).
Evaluation measures and processes
A mixed method evaluation was conducted using a repeated cross-sectional design for the collection and analysis of quantitative data. Eligible participants were in the study schools at the time of each measurement occasion at three time points: baseline (2004–5), midway (2007) and completion (2009). This contrasts with a cohort design that uses the same participants at all measurement occasions. Longitudinal analysis was only feasible on a nested sample of 350 students because of the turn-over in the school population in the 5 years of the study.
A pilot study of the data collection was initially carried out in an inner-urban, a suburban, and a rural primary school in Victoria, Australia in 2003 to test the feasibility and acceptability of the processes and measurement tools. The tools were subsequently refined for the main study collection of data at both school and individual child/parent levels Child questionnaires were completed by children in grades three to six (approximate age range 8 to 12 years). Two versions of the questionnaires were produced with one tailored to grade three and four children, the other for grade five and six children. A body image sensitivity protocol was also developed to minimize any potential harm in relation to body dissatisfaction .
The individual measures were as follows:
The pre-specified primary outcome, BMI was measured by:
BMI z-score calculated using direct measure of child height and weight to generate BMI, and then z-scores against the WHO reference curves . Project staff were trained in standardised child height and weight measurement and a process developed that was sensitive, confidential and avoided value judgements . Weight in light clothing without shoes was recorded to the nearest 0.1 kg using digital scales and height to the nearest 0.5 cm using rigid stadiometers. All measures were taken twice and the mean value used. Where two readings differed by more than 0.4 kg or 4 cm, a third reading was taken and the two closest values used to calculate the mean.
Fruit and vegetable intake and sweet drink consumption were measured by:
Parental report through parent questionnaires covering issues such as family food habits , and usual intake of fruit, vegetable, dairy and drink consumption 
Child report through child questionnaire assessing food behaviours 
Direct Assessment of school foods: Lunch box survey whereby fruit, vegetables and drinks in children’s lunchboxes were recorded
24-h food record  which was distributed to parents on a weekday for description of food and drink consumed by child for the next 24 h period.
Participation in sedentary activity, physical activity and activity intensity was measured by:
Parental report in parent questionnaires covering issues such as family physical activities and child sedentary and physical activities and level of active transport (Physical activity questions changed from baseline to follow up to reduce burden and increase comprehensibility)
Child report through child questionnaire covering issues such as family physical activities and child sedentary and physical activities and level of active transport (Physical activity questions changed from baseline to follow up to reduce burden and increase comprehensibility)
Child experience was measured by:
Child-report through child questionnaire of quality of life using the 10-item version of the international self-reported measure of quality of life, KidScreen 
Child focus groups to explore children’s concepts of health and strategies to promote health in the home and school environments
Impacts on the school, home and community environments were measured by:
School reported audit of the school food and physical activity environment, including physical activity facilities, canteen and fundraising policies and practices 
Principal exit interviews to identify barriers and enablers to the school experience and likelihood of sustainability
Teacher-reported school- and class-based nutrition and physical activity initiatives and level of support
: SOPLAY (System for Observing Play and Leisure Activity in Youth)  based on momentary time sampling techniques using systematic and periodic scans of individuals and contextual factors within pre-determined target areas. The instrument permits comparison of physical activity levels in different play environments .
Process evaluation using monitoring maps, photos, and audits to track and record changes in school plans, policies and environment, stability of changes, costs of changes, and level of independence from the research team
Parental report through parent questionnaire of parent and spouse/partner demographics and funds expended on nutrition and on physical activities.
An outline of these measures and data collection time points is presented in Table 1. This paper will present results from the anthropometric measures, school questionnaire, principal interviews, parent questionnaire, child questionnaire and lunchbox survey collected at baseline and completion.
The randomisation allocator was blind to school status. However it was not possible for schools and participants to be blind to allocation because of the nature of the intervention. Field staff collecting data were blind to the intervention status of each school. Data collection, however, occurred on school premises and for some schools their intervention status was obvious. Schools were de-identified at data entry prior to data being sent to the analysis team.
We aimed to recruit and randomise 9 schools to each trial arm (18 schools altogether) and sample 127 children from each school at each wave. Using bmi-z score as the outcome, this is large enough to detect a difference of 0.2 with 80% power at the (2-sided) 5% level of significance, assuming a standard deviation of 0.96 and an intra-cluster (intra-school) correlation coefficient (ICC) of 0.017.
Pupil level outcomes
Characteristics are summarised using means and standard deviations for continuous variables and proportions for binary variables. Intervention effects are estimated based on the intention-to-treat principle with participants and schools analysed according the trial arm they were randomised to. Descriptive adiposity scores were generated using WHO cut points, with bmi z-scores as the adiposity outcome to model the intervention effect. For continuous outcomes (e.g. bmi z-score), the intervention effect was estimated using random effects linear regression models fitted by maximum likelihood estimation to allow for clustering. For dichotomous outcomes such as prevalence of overweight/obesity, marginal logistic regression models were fitted using generalized estimating equations with information sandwich (“robust”) estimates of standard error, specifying an exchangeable correlation structure. Both crude analyses and analyses adjusted for prognostic factors were run. As the study used a repeated cross-sectional design, analyses of continuous outcomes were adjusted for baseline level of the corresponding outcome by using the mean score for the school at baseline as a predictor variable in the models. Binary outcomes were adjusted for the proportion with the characteristic of interest in the study cluster at baseline. As the baseline physical activity variables differed to those used at follow up, the baseline school means for the original measures of physical and sedentary activity  were used as surrogate school means in the analyses of follow-up data. Models were also adjusted for child age and sex, socio-economic position ((SEP) measured by maternal education, residential SEIFA (Australian Bureau of Statistics Socio-economic Index for Areas (SEIFA) index of relative socioeconomic disadvantage), and ethnicity (only English spoken at home). Statistical analyses were conducted with STATA 10.1 (Stata Corp LP, College Station, Tex).
Community Development Workers kept records of intervention strategies implemented in schools. Descriptive statistical analyses of school policies, environments and practices were undertaken using school questionnaire data. Interviews were conducted with school principals following the collection of follow up data in order to understand the impact of the intervention on schools, principals’ views on what factors mitigated or enhanced the implementation of the intervention, recommendations in relation to future strategies, and whether the model had been acceptable to the school context and systems. An inductive, thematic analysis was conducted on the interview transcripts to generate insights into principals’ perspectives on the school experience of the study, and development and response to school policies and practices.
A costing of the resources invested in the intervention, including the CDW salaries, school resources and parent expenses was also undertaken. Costs incurred across all schools were split equally between the intervention schools. School-level costs were split equally across the student population. Costs were discounted at 5% and presented in 2009 Australian dollars.
Data sharing is not applicable to this article as this was not part of the original consent arrangements.