Poor diet is a modifiable risk factor for the development of noncommunicable diseases including stroke, diabetes, and heart disease, accounting for 19% mortality and 10% of morbidity, globally [1]. Population surveys in Australia and internationally indicate that both adults and young children are not consuming the recommended servings of fruit and vegetables and consume more than recommended amounts of discretionary (energy-dense and nutrient-poor) foods [2-5]. As dietary behaviors established in early childhood track into adulthood [6,7], the World Health Organization recommends that population health approaches be undertaken to improve healthy eating behaviors in young children [8,9].

As approximately 662,000 children aged 0 to 5 years attend formal care in Australia [10], childcare services represent an opportune environment in which to intervene to establish healthy eating behaviors. Systematic review evidence, leading health authorities, and governments internationally recommend that childcare services provide foods in line with dietary guidelines [2,8,11-14]. In the state of New South Wales (NSW), Australia, the Caring for Children [15] resource outlines best practice dietary guidelines for the childcare sector. However, research internationally and in Australia suggests that such dietary guidelines are poorly implemented, with childcare services frequently providing foods and drinks inconsistent with guideline recommendations [16-19].

Childcare staff have reported a number of barriers to the implementation of dietary guidelines. Findings from a recent systematic review indicated such barriers to childcare service staff implementation of guidelines related to knowledge, skills, social influences, environmental context, and a lack of resources [20]. These barriers center around the lack of staff training and support to undertake menu planning consistent with guidelines and regulatory standards (eg, child allergies) and challenges associated with self-assessment of a menu to determine the nutritional adequacy [18,21-24] and its compliance with guidelines.

To improve the implementation of dietary guidelines in childcare, strategies that target known barriers to implementation are required. To our knowledge, only 4 controlled trials have been conducted with the aim of improving the provision of foods and beverages to children in childcare in accordance with dietary guidelines [17,19,25,26]. All 4 trials assessed the impact of multistrategy interventions consisting of a combination of educational materials, face-to-face meetings, or audit and feedback; and when compared with control groups, none found significant improvements in the implementation of the targeted dietary guidelines. The implementation support strategies utilized in these previous trials, therefore, appear insufficient to address knowledge and skill barriers to the implementation of dietary guidelines in this setting.

Web-based interventions offer an opportunity to provide implementation support that has the potential to be effective in enhancing childcare service implementation of dietary guidelines. First, childcare services have existing infrastructure (computer and internet access) to support a Web-based intervention [27]; and staff are willing to use such an intervention to support their implementation of healthy eating policies and practices [27]. Second, specific programming within Web-based systems [28] has the potential to integrate active behavior change strategies [29] to target primary barriers to guideline implementation, including resources, audit, and feedback for menus, and automated calculation of menu compliance with guidelines, eliminating the need for manual calculations by service staff. Third, Web-based interventions can be tailored to a particular service’s needs, delivered with high fidelity, at low end-user cost, and are able to address equity issues related to access to dietetic support, particularly for childcare services in rural and remote areas [30,31]. Finally, Web-based systems have the potential to minimize the need for ongoing investment in implementation support (eg, the provision of training and resources) for practice improvements to be sustained.

Despite this potential, the effectiveness of a Web-based intervention to improve childcare service implementation of dietary guidelines has not yet been evaluated [32]. As such, the primary aim of the study was to assess, compared with usual care, the effectiveness of a Web-based menu planning intervention in increasing the mean number of food groups on childcare service menus that comply with dietary guidelines. Secondary aims include assessment of the impact of the intervention on the proportion of service menus compliant with (1) all food groups; (2) individual food groups; and (3) the mean servings of individual food groups. Childcare service use and acceptability of the Web-based program were also assessed.

Ethical approval was obtained from the Hunter New England (approval no: 16/02/17/4.05) and the University of Newcastle (approval H-2016-0111) Human Research Ethics Committees. The trial was prospectively registered with the Australian New Zealand Clinical Trials Registry (ACTRN12616000974404). Other registered secondary outcomes will be reported elsewhere. The reporting of this study adheres to the Consolidated Standards of Reporting Trials guidelines [33]. All subjects in this research study provided consent to participate.

As previously described in the study protocol [34], a parallel-group randomized controlled trial (RCT) was undertaken with 54 long day care services in NSW, Australia. The 252 potentially eligible childcare services in NSW that were current clients of a single specific childcare management software (CCMS) provider, and that provided foods to children, served as the study sampling frame. In order for families to receive financial reimbursement from the Australian government to assist with the costs of childcare [35], services are mandated by Federal legislation to use a government-approved CCMS. The Web-based intervention, titled feedAustralia, was developed by Hubcare Innovation, for Healthy Australia and in collaboration with HubHello, and was linked to one such software package used by approximately 20% of childcare services in NSW [36].

Eligible childcare services were required to (1) be open for ≥8 hours each weekday; (2) prepare and provide at least 1 main meal and 2 snacks to children onsite each weekday; (3) have service staff make menu planning decisions; and (4) have a menu planner with sufficient English to engage with the intervention. Services that outsourced menu planning, did not cater for children aged 3-6 years, catered exclusively for special needs children, or were run by the NSW Department of Education were excluded because of differing administrative characteristics.

All services in the sampling frame were posted an invitation letter and information statements about the study in random order, approximately 2 weeks before receiving a call from a research assistant to assess eligibility and obtain service consent to participate (August-December 2016). Recruitment of services was conducted in random order as a subsample of services also participated in a nested evaluation [34]. The CCMS provider also displayed an invitation for services to participate in the trial via their Web access portal. Following provision of consent, nominated supervisors and menu planners were contacted to complete a computer-assisted telephone interview (CATI) to assess baseline service and menu planner characteristics and were asked to provide a 1-week-long menu from their current menu cycle for assessment.

Following the completion of baseline data collection, services were allocated to the intervention or control group in a 1:1 ratio, stratified by service area socioeconomic status (as determined by service postcode) [37] by an independent statistician using a random number function in Microsoft Excel 2010. All outcome data assessors were blind to group allocation; however, owing to the nature of the trial, childcare staff and health promotion officers delivering the intervention were aware of group allocation.

Services received a 12-month implementation intervention consisting of access to a Web-based menu planning program (feedAustralia), in addition to training and support to use the program (Multimedia Appendix 1 [15,28,34,38-44]). The menu planning program was not embedded within the CCMS platform already used by the childcare services as originally planned because of changes in national regulatory requirements for CCMS. Rather, the menu planning program was developed as a stand-alone program, allowing childcare services to access the program outside of CCMS. The program was linked to the Web-based CCMS platform to allow communication between the 2 systems. The intervention was codeveloped and overseen by an experienced multidisciplinary expert advisory group consisting of health promotion practitioners, implementation and behavioral scientists, policy makers, and public health nutritionists with experience working in the setting. To ensure uptake and to enhance use of the Web-based program, the menu planning program was developed using the Technology Acceptance Model [45], with implementation support strategies identified through a barriers assessment using the Theoretical Domains Framework [46]. Further details regarding the theoretical underpinnings and development of the intervention are reported elsewhere [34].

Services randomly allocated to the control group did not receive access to the Web-based menu planning program or other implementation support strategies.

Baseline data were collected during October 2016 to April 2017, with the 12-month follow-up conducted during October 2017 to March 2018.

On the basis of pilot data (unpublished) with a standard deviation of 1.23, a sample of 27 services in the intervention and 27 services in the control would enable detection of a 0.96 (approximately 1) change in the mean number of food groups compliant between intervention and control groups at the 12-month follow-up (primary outcome) with 80% power and a 2-sided alpha of .05. From a population health perspective, increasing compliance with just 1 food group may contribute to important improvements in public health nutrition. For example, based on current data regarding food provision by childcare services in Australia [49], achieving compliance with guideline recommendations for vegetables would be equivalent to an increase of 60 grams (0.8 servings) per child, while compliance with discretionary foods would be equivalent to a decrease of 360 kilojoules (0.6 servings) per child [14]. Such improvements have been associated with important child health outcomes and reductions in disease risk [50,51].

All statistical analysis was undertaken using SAS 9.3 (SAS Institute Inc) [52] by a statistician blinded to group allocation. All statistical analyses were 2-tailed with an alpha value of .05. Service postcodes ranked in the top 50% of NSW according to the 2016 Socioeconomic Indices for Areas were classified as higher socioeconomic status [37]. Geographical characteristics of service locality were classified as either urban or rural according to the Australian Statistical Geography Standard [53]. Chi-square and t test analyses were used to compare characteristics of consenters and nonconsenters, and service and menu planner characteristics between intervention and control groups at baseline. The primary (mean number of food groups compliant with guidelines) and secondary menu outcomes (individual and all food group compliance with guidelines, and mean daily servings of individual food groups) were analyzed with generalized linear mixed models to account for repeated measures at the service level, as well as potential service level clustering effects for the mean daily servings of food groups analysis. All models included a random effect for service, as well as a group by time interaction to assess intervention effectiveness over the 3 time points (summarized as relative mean difference for the continuous measures and relative odds ratio [OR] for the categorical outcomes). All models assessed the relative difference in menu outcomes between the 2 groups from baseline to 3 months, as well as the relative difference from baseline to 12 months. For the primary and secondary outcomes, under an intention-to-treat framework, a complete case analysis was performed using all available data based on group allocation (without imputation), in addition to analysis using multiple imputation for missing data at follow-up undertaken using the MI procedure in SAS.

Of the 252 long day care services, who were current clients of a single specific CCMS provider in the study region, 54 services declined to participate in the study before eligibility assessment. A total of 198 services were assessed for eligibility, with 42.4% (84/198) deemed ineligible, most commonly because of the inability of service staff to make menu planning decisions (28/84, 33%), and not providing meals and snacks to children (24/84, 29%); (Figure 1). Of the remaining 114 eligible services, 47.4% (54/114) provided consent to participate in the study. There were no significant differences in service area socioeconomic status or service geographic location between consenters and nonconsenters.

A total of 27 services were randomized to the intervention and 27 services to the control. Two intervention services withdrew from the study before the 12-month follow-up; 1 no longer prepared and provided meals and the other no longer wished to participate. Services in the control arm had a significantly higher proportion of menu planners with a university qualification (5/27, 19%) compared with services in the intervention (0/27, 0%; P=.02; Table 1).

This study is the first RCT measuring the effectiveness of a Web-based menu planning program, linked to a CCMS system, in improving childcare service compliance with dietary guidelines. The study found that, despite being considered acceptable by childcare service staff, the intervention did not significantly improve childcare service menu or food group compliance with dietary guidelines compared with the control. However, significant increases in the servings of fruit, vegetables, dairy, and meat on the menu, and a significant reduction in the number of times discretionary foods were provided were observed at 3 months. At 12 months, a significant increase in servings of fruit and a significant reduction in the provision of discretionary foods was found. Such findings suggest that despite increases in the quantity of some healthy foods and a decrease in unhealthy (discretionary) foods provided on the menu, the Web-based intervention was not sufficiently effective to ensure children are provided with servings of food groups consistent with dietary guidelines for the setting. As foods provided in the home and other settings often fail to align with dietary guidelines [54], such findings are of concern.

The lack of a significant effect of the intervention on menu compliance with all food groups is similar to previous Australian studies in the childcare setting [17,19]. This suggests the achievement of a fully compliant menu in accordance with the current dietary guidelines for the setting is a sizeable challenge [55], and perhaps an unachievable goal for many childcare services at present. To be fully compliant with guidelines, services are required to provide adequate servings of each of the AGHE foods groups, and no discretionary foods, for every child in attendance on every single day. Reviews of public health program implementation more broadly suggest that implementation of more than 80% of recommended program elements is rarely achieved across a range of settings [56]. As such, continuous, incremental changes to practice may be more manageable, and over time may result in greater improvements in the provision of healthy food in childcare.

On measures of individual food group compliance, the ORs reported in this study at any time point (0.37-4.33) were generally smaller than those found in a previous randomized trial (1.19-17.83) which, using the same measure, found statistically significant improvements in compliance for fruit, meat, dairy, and discretionary foods [19]. In that 6-month face-to-face intervention, support for childcare service staff included securing executive support, 2 rounds of staff training and ongoing telephone support from an implementation support officer, provision of resources, and 2 rounds of audit and feedback from a dietitian. The findings may reflect a greater capacity of the more intensive face-to-face implementation support offered in the trial by Seward and colleagues to address a broader range of barriers to implementation (eg, environmental context). Such findings suggest the inclusion of additional implementation support strategies as an adjunct to the Web-based program, may be required in order for larger improvements in guideline implementation to be achieved. Future research testing this hypothesis is warranted.

Notwithstanding the lack of statistical significance between group effects on these measures, increases in compliance for all food groups and individual food groups for the control group were observed and were similar to those found in the intervention group. A possible explanation for this could be an increased awareness of the importance of healthy food provision in childcare in the external environment, other secular trends, or changes to childcare service accreditation requirements during the study period [57]. Alternatively, this may be the evidence of measurement reactivity or Hawthorne effect [58], in that the act of evaluating childcare service menus by external dietitians on multiple occasions within a 12-month period may lead to an increase in menu compliance with guidelines. To reduce the impact of any research reactivity effects, future studies should investigate alternate methods of measuring guideline implementation.

The exploratory analysis identified a statistically significant increase in the mean daily servings of food groups, in particular fruit, and a reduction in discretionary foods at both 3 months and 12 months among the intervention group, compared with the control. As the program focused on supporting services to make incremental changes to the quantities of healthy food groups provided on the menu via recipe substitution and modification, such improvements to servings are not surprising. In addition, the mean number of daily servings for some food groups (eg, fruit, breads and cereals, and dairy) was higher than the required minimum servings to be considered compliant with the guidelines, suggesting it is likely that services were compliant on some, but not all days of the week (as required for menu compliance). Assessments of any adverse impacts of the provision of foods above the recommended minimum on child-level outcomes or service outcomes (eg, increased waste) warrants investigation.

Among intervention services, there were high levels of acceptability and variable levels of use of the Web-based program (as evidenced by the large SDs and IQRs in program use data). Previous research has identified engagement with Web-based interventions to be associated with a range of health behaviors [59,60]. As such, research exploring perceived barriers and enablers to use of the program and identification of strategies to best support end-user engagement with the Web-based program is warranted.

The study had notable strengths including the design (RCT), rigorous evaluation approaches, and inclusion of theory-driven and evidence-based intervention and implementation support strategies. Limitations, however, were also present. Similar to previous trials within childcare services [61], the study yielded a moderate consent rate (47.4%). Although there were no significant differences in service area socioeconomic status or geographic location for consenters and nonconsenters, given the study was conducted within 1 state in Australia (NSW) with few Indigenous services, it is unclear whether these findings are generalizable nationally or internationally. Furthermore, despite randomization, services in the control arm had a significantly higher proportion of menu planners with a university qualification compared with the intervention services. It is possible that this may account for the improvement in menu compliance observed in the control arm. The findings report the overall effects of the intervention, which may mask differences in outcomes at the subgroup level. Future exploratory studies reporting findings from the trial will describe any differential effects by subgroups based on service locality (eg, service area socioeconomic status and geographic location), service characteristics (eg, size), or other contextual factors. Although the menu planning program was linked to a CCMS platform to increase uptake and integration into daily routines, the program was not viewable on the main child enrollments page that is accessed on a daily basis by childcare service staff. Integrating the Web-based menu planning program into the main CCMS platform of the software may reduce variability in service use of the program. Finally, the outcome relating to servings of individual food groups provided on the menu was not prospectively registered and should be interpreted with caution.

The study is the first RCT measuring the effectiveness of a Web-based menu planning program to improve childcare service compliance with dietary guidelines in NSW, Australia. Findings indicate that the Web-based program was not effective in increasing the mean number of food groups compliant with dietary guidelines, nor the proportion of service menus compliant with dietary guidelines for all food groups and individual food groups. Despite this, significant improvements in the mean number of servings of healthy food groups and a reduction in the provision of discretionary foods provided on the menu were found. Future research should aim to reduce potential measurement reactivity or Hawthorne effects. Exploration of differing strategies in supporting uniform use of the Web-based program, and the implementation of dietary guidelines, among childcare services is warranted to ensure potential public health benefits are achieved.