Bell 2014

Methods	Study design: quasi‐experimental trial Intervention duration: average of 22 months between initiation of intervention and collection of follow‐up data Length of follow‐up from baseline: average 22 months (between initiation of intervention and collection of follow‐up data) Differences in baseline characteristics: reported Unit of allocation: childcare service Unit of analysis: childcare service
Participants	Service type: preschools and long daycare services Region: Intervention: Hunter New England region, New South Wales, Australia; Control: New South Wales, Australia Demographic/socioeconomic characteristics: Intervention: the Hunter New England region ‐ a geographically large area (130,000 km2) with a demographically diverse population including metropolitan urban and suburban areas, regional services, and rural and isolated remote communities. The region included pockets of wealth and poverty, and an overall socioeconomic status lower than the New South Wales state average. Control: not reported Inclusion/exclusion criteria: all services located within the intervention region were invited to participate. Services were excluded that catered for children with special needs such as intellectual or physical disabilities. Number of services randomised: 583 (287 intervention, 296 control) Numbers by trial group: n (control baseline) = 251 n (control follow‐up) = 191 n (intervention baseline) = 261 n (intervention follow‐up) = 240 Recruitment: Intervention: all services (n = 287) located within the intervention region were invited to participate. Control: a simple random sample of eligible centre‐based childcare services in all other regions of the state of New South Wales were invited to participate in the study as the comparison group (n = 296) Recruitment rate: Intervention: 91%; Control: 85%
Interventions	Number of experimental conditions: 2 (intervention, control) Policies, practices or programmes targeted by the intervention: Healthy eating policies and practices of childcare services including: ‐ Staff training in nutrition ‐ Policy guiding the content of food and drinks provided to children by the service ‐ Policy guiding the content of food and drinks packed for children by parents ‐ Provision of non‐sweetened drinks (milk and water) only to children during care ‐ Parent participation in nutrition policy or programmes ‐ Provision of foods to children consistent with dietary guidelines (for services that provide meals to children) and accreditation requirements Implementation strategies: ‐ Identifying leaders and obtaining their support and endorsement of the programme and targeted policy and practices ‐ Provision of professional development for staff (2 x 6‐hour workshops (1 for staff and service managers, 1 for cooks and service managers) ‐ Small incentives ‐ Resource provision ‐ Performance monitoring and feedback ‐ Follow‐up support (20‐minute phone call once, 5 newsletters) Who delivered the intervention: the intervention was delivered by health service staff who worked with regional representatives of the Department of Community Services and childcare service staff to implement the intervention strategies Theoretical underpinning: the intervention was based on practice change and capacity‐building theoretical frameworks Description of control: from July 2008 onwards, preschool services (not including long daycare services) in New South Wales were able to access implementation support via a government‐supported programme that aimed to promote physical activity and healthy eating for children
Outcomes	Outcome relating to the implementation of childcare service policies, practices or programmes: Service healthy eating policies and practices: ‐ Staff with nutrition training ‐ Services with a policy guiding the content of food and drinks provided to children by the service ‐ Services with a policy guiding the content of food and drinks packed for children by parents ‐ Services providing only water or plain milk to children ‐ Parent participation in nutrition policy or programmes Data collection method: computer‐assisted telephone interview with service managers Validity of measures used: not reported Nutritional quality of lunch menus: ‐ Number of times processed foods high in fat, salt and/or sugar were listed on the menu each day ‐ Number of times sweetened drinks were listed on the menu each day ‐ Number of times water was listed on the menu each day ‐ Number of ‘child size’ servings of fruit listed on the menu each day ‐ Number of ‘child size’ servings of vegetables listed on the menu each day Classification into the following categories: ‐ No high‐fat, ‐salt and/or ‐sugar processed food menu items ‐ No sweetened drink menu items ‐ Water with every eating occasion ‐ 1 child‐size serving of fruit listed on the menu each day ‐ The number of child‐size servings of vegetables listed on the menu each day Data collection method: all services were invited to submit a copy of their current 2‐week menu Validity of measures used: not reported Outcome relating to staff knowledge, skills or attitudes: not applicable Outcome relating to cost: not applicable Outcome relating to adverse consequences: not applicable Outcome relating to child diet, physical activity or weight status: not applicable
Notes	—
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Quasi‐experimental design. High risk of selection bias as intervention services were recruited from a selected area. Control services were randomly selected from a comparison region. There were no details provided regarding the sequence generation procedure used to randomise control services for selection.
Allocation concealment (selection bias)	High risk	Quasi‐experimental design. Intervention services were recruited from a selected area, therefore high risk of bias as no concealment of allocation.
Blinding of participants and personnel (performance bias) All outcomes	High risk	We assumed that due to the nature of the intervention childcare service staff and study personnel delivering the intervention were not blind to the study allocation and therefore there is a potential high risk of performance bias.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Self‐reported policies and practices. There was no blinding of research personnel or participants (service managers) and due to the self‐report of this outcome, risk of bias is considered to be high.
Incomplete outcome data (attrition bias) All outcomes	High risk	There was a large difference in the proportion of services followed up amongst intervention and control groups and the proportion that provided a menu for assessment: Intervention group: 91% of services surveyed at baseline were followed up and 61% provided a menu Control group: 76% of services from the control area (NSW) were followed up and 49% provided a menu Due to the magnitude of difference in the proportions of participants followed up between groups, the risk of bias is assessed as high
Selective reporting (reporting bias)	Unclear risk	No prospective trial protocol or trial registration so it is unclear whether there was selective outcome reporting.
Potential confounding	Unclear risk	Authors state that "Characteristics of services were not adjusted for in the logistic regression model as we were looking at change within services and the baseline score of the services effectively controlled for potential differences in baseline characteristics between the two regions." It is unknown whether this was adequate to reduce bias due to known confounders.
Other bias	Unclear risk	—