Multi-strategic intervention to enhance implementation of healthy canteen policy: a randomised controlled trial

Nutrition policies governing the availability of foods in schools have been recommended by the World Health Organization to improve child nutrition [1]. In schools, such policies offer an opportunity to ensure that the foods made available to students comply with dietary guidelines. Systematic reviews of trials of the effectiveness of policies restricting the availability of unhealthy foods have consistently reported beneficial effects on child diet [2, 3]. As such, internationally, in jurisdictions where meals provided to children as part of school food services (e.g. UK), or where foods are available for sale to children at kiosks, tuckshops or canteens (e.g. Australia), countries have announced nutrition policies governing school food availability [1, 4‐6].

Despite their popularity, research suggests that school nutrition polices are poorly implemented, limiting their potential public health impact. For example, in the USA, the Healthy, Hunger-Free Kids Act of 2010 (P.L. 111-296) required all schools participating in the National School Lunch Program to adopt and implement nutrition guidelines for all foods and beverages available at school (including competitive foods sold outside the school lunch programme) [7]. However, only 61% of students attend schools in a district with guidelines on competitive foods and less than 10% of middle and high school students were in a district that banned sugar sweetened beverages other than soda [7]. Similarly, in Australia, just 5–35% of schools in most states comply with government healthy canteen policies with marginal improvements in policy implementation occurring over time [8‐10].

To maximise the public health benefits of school nutrition policies strategies are needed to overcome barriers to their implementation [11, 12]. Research investigating strategies to facilitate the implementation of health innovations in schools, however, is limited [13]. An updated Agency for Health Care Policy and Research systematic review identified just one randomised trial of a school-based strategy to improve the implementation of healthy eating policies or practices [14]. The trial reported little improvement in the provision of healthier school lunches between intervention schools who received training, technical assistance, resources and coalition building support to implement school food service guidelines, compared to control schools [15].

In the context of this limited evidence base, the primary aim of this study was to assess the effectiveness of a multi-strategic intervention to increase implementation of a state-wide healthy canteen policy (hereafter referred to as the ‘policy’). We also assessed the impact of policy implementation on the energy, total fat and sodium of children’s canteen purchases and on canteen revenue.

The study was approved and the procedures monitored by the University of Newcastle Human Research Ethics Committee (Approval Number H-2008-0343) and the Hunter New England Human Research Ethics Committee (06/07/26/4.04) and was prospectively registered with Australian New Zealand Clinical Trials Register ACTRN12613000311752. Full details of the study methods are published [16]. The project was also approved by the NSW Department of Education and Communities (DEC) (#2012277).

Of the 124 schools randomly selected, 80 returned menus which were assessed for eligibility, eight were ineligible and two refused participation; 70 provided consent and were randomly allocated to the intervention or control group (Fig. 1). Of the schools who did not return menus, six had recently closed their canteen, 29 principals or school representatives could not be contacted, and nine schools refused to provide a menu.

The baseline characteristics of participating schools were similar between groups (Table 2). At baseline, all schools provided their menu for assessment and 69 principals and 69 canteen managers completed their respective telephone interview. At follow-up, 57 schools provided their menu for assessment and 56 principals and 49 canteen managers completed a telephone interview. There were no significant differences in baseline characteristic among schools that did and did not provide follow-up data (p = 0.19–0.54). Of the 50 schools randomised to participate in the observation of canteen purchases, 20 out of 24 intervention schools and 18 out of 26 controls provided consent. There were no significant differences among schools that did and did not consent to participate in observations in baseline school characteristics (p = 0.23–1.0).

The failure of schools to implement government polices to reduce the availably of unhealthy foods at schools has been documented internationally [7‐10] and undermines the potential public health benefits such policies were intended to deliver. To our knowledge, this is the first randomised trial designed specifically to test the impact of a strategy to improve implementation of a school nutrition policy [14]. The findings of this trial suggest that with multi-strategic support, implementation of healthy canteen policies can be achieved in the vast majority of schools (>70%).

The size of the intervention effect, equivalent to an absolute improvement within the intervention group of 54–67% on the primary trial outcomes, was larger than previous non-randomised trials of interventions which have sought to enhance implementation of nutrition policies or guidelines in school or childcare food service settings [15, 28]. The effect size was also larger than reported for strategies to support implementation of school-based health promotion programmes more broadly [29‐34], which typically increase implementation by between 13 and 45%. Despite such encouraging findings, longer-term follow-up to examine whether improvements in policy implementation are sustained in the absence of intervention support is warranted.

The findings suggest that the intervention was successful in overcoming many of the barriers that reportedly hinder nutrition policy implementation in schools [11, 12]. However, the mechanism by which the intervention support strategies yielded improvements is unclear. A lack of understanding of the mechanism leading to implementation improvements has been previously identified as hindering optimization and the design of more efficient and effective implementation strategies [35, 36]. While the design of the implementation support was informed by considerable formative evaluation and utilised a comprehensive implementation theoretical framework, an examination of the impact of the strategy on hypothesised mediators was not conducted. The development and validation of tools to measure implementation constructs suitable for use in non-clinical settings would facilitate this research in the future [37].

Concerns regarding canteen profitability are frequently cited by canteen managers as an impediment to implementation of health canteen policies given the perceived higher costs, lower profit margin and greater labour required to prepare healthy foods for sale [11, 25, 38, 39]. Increases in the proportion of canteens making a loss have also been suggested in a study by Pettigrew and colleagues following the introduction of the Western Australian Healthy Food and Drink policy [40]. While statistical tests did not identify a significant reduction in profits of intervention canteens relative to control, the magnitude and direction of effect suggests that further investigation into the potential for such adverse effects is warranted.

The study findings should be considered in the context of the trial methods. The trial randomly sampled from the study region to maximise school representativeness and achieved relatively high participation rates. The generalisability of the findings to other jurisdictions or schools systems required to implement alternate school-based nutrition policies, however, is unknown. Risk of bias was reduced by central randomisation to groups, allocation concealment, blinded assessment and low trial attrition. Furthermore, project records suggest a high degree of intervention fidelity for a complex public health intervention with 94% of intervention schools receiving at least six of the nine intervention strategies. Student purchase data was included as a secondary trial outcome as additional resource became available to the researchers, in a subset of schools and at post-intervention only. Student level data were included in the trial register before any post-intervention data were collected but following initiation of the trial. While the inclusion of such data enables assessment of the impact of implementation of an evidence-based policy on the nutritional value of student purchases in the ‘real world’ the findings of the trial on this outcome should be considered in the context of these limitations. Missing data for assessments of canteen revenue, however, was considerable at both baseline and follow-up. As such, assessments of the impact of the intervention on canteen profits are likely to be unreliable. Finally, the trial included a relatively narrow range of potential implementation outcomes. Proctor and colleagues suggest that the inclusion of a broader range of implementation outcomes including acceptability, adoption, appropriateness, feasibility, penetration and sustainability may offer a more in-depth assessment of implementation processes and impact [41]. Consideration of such measures in future trials is therfore warranted.

Without local level implementation support, government school nutrition policies are unlikely to yield improvements in public health nutrition. Notwithstanding the trial limitations, the study provides strong evidence supporting the effectiveness of one approach to support nutrition policy implementation in schools. While effective, providing multi-strategic implementation support to large numbers of schools may represent a considerable challenge for governments. Data regarding the perceived helpfulness of support strategies included in this trial, such as performance feedback, provide some suggestions as to which strategies may be most important. Nonetheless, future research to identify which intervention support strategies are the key drivers of improved policy implementation, and means of providing such support using lower cost delivery modalities (such as via telephone or the web) may be particularly beneficial.