School food policy in secondary schools in England and its impact on adolescents' diets and dental health: the FUEL multiple-methods study

Exemptions to national school food standards legislation

The way in which the SFS legislation in England has been applied means that a group of state-funded schools are exempt from the statutory requirement to comply with the SFS. Academies and free schools (state-funded schools that are independent of local authorities) that were established between 2010 and 2014 are not required by law to comply with the SFS. 45 However, as part of the SFP, and following the introduction of the revised SFS in 2015, these exempt schools were encouraged to voluntarily sign up to the SFS. 46,47 By March 2016, only around one-third of the exempt schools had done so. 48 Although the legislation has not been changed to address this exemption of schools, in the last 18 months of this study there has been a shift in the expectation of the government, whose position is now that all schools, regardless of their exemption status, should be complying with the SFS. 49

The current national school food standards

The SFS currently in place are food-based (rather than nutrient based) but underpinned by a nutrient framework. Before their introduction, the standards were pilot tested to ensure that they led to the same or improved nutrient content of school meals compared with the 2006 SFS. 50

The SFS specify six groups of foods and drinks: (1) starchy foods; (2) F&V; (3) milk and dairy; (4) meat, fish, eggs, beans and other non-dairy sources of protein; (5) foods high in fat, sugar and salt (HFSS); and (6) healthier drinks. There are rules for portion, variety and frequency of provision for each group. The standards are divided into three groups: those applying to school lunch, those applying to foods provided in school other than at lunchtime, and standards that apply across the whole school day. 51 Standardised checklists are available to assist schools and caterers in checking their compliance with the SFS. 51

Despite their legislative status, there are no formal national arrangements for monitoring schools’ compliance with the SFS. 52 Responsibility for ensuring the SFS are met is placed with the governing bodies of schools, but in England there is no requirement for school governors to report on their school’s SFS compliance. 49 As part of their inspection visits to schools, Office for Standards in Education,Children’s Services and Skills (Ofsted, the English schools inspection body) observe the school canteen food and environment and the effect of this on pupils’ behaviour, but they do not review SFS compliance. 53

The School Food Plan

The SFP established a set of wide-ranging recommendations for actions by the government, schools and headteachers to promote a ‘whole-school’ approach to a healthy-eating culture and ethos in schools. 43 One of the central aims of the SFP was to increase school meal uptake, both for economic reasons (higher demand drives better-quality food provision at lower cost) and because there is evidence to suggest that packed lunches are of poorer nutritional quality than school meals. 54,55 The other key aim was to provide practical support, advice and information for headteachers and others involved in school leadership in the form of a set of non-statutory recommendations for action to help improve the quality and uptake of school food and incorporate healthy eating into all aspects of school life and the wider school community. 43

Broadly, the recommendations in the SFP for schools include actions relating to leadership, vision and a whole-school approach; school food monitoring and accountability; school food provision and affordability; the physical and social eating environment; food education encompassing healthy-eating knowledge and skills; linking healthy eating to wider health and well-being; and involving school children and the wider school community in school food and healthy eating.

A checklist of actions for headteachers and guidance for school leadership teams and governors were developed to assist schools in implementing the SFP recommendations and a whole-school approach to creating a healthy-eating culture and ethos. 56–58 The Department for Education (DfE) continues to encourage the implementation of SFP recommendations and the use of these checklists and guidance in schools. 59

Evaluation of English national school food policy to date

Evaluation of the impact of the SFS on nutritional intake following the introduction of the 2006 food- and nutrient-based standards has been conducted among younger children in primary and middle schools. Repeated cross-sectional studies were conducted pre and post introduction of the SFS, collecting nutritional intake data from children aged 4–7 years in primary schools (n = 12)60,61 and children aged 11–12 years in middle schools (n = 6)62 to determine the effect of the SFS on lunchtime and total dietary intake. Following the introduction of the SFS, improvements were seen in the overall nutritional intake of children aged 4–7 years, with the greatest improvements seen in those receiving a school-provided lunch (vs. having a packed lunch). 60,61 By contrast, in the 11–12 years age group, there was limited evidence of a positive impact on the nutritional intake of children following introduction of the SFS, and levels of free sugar intake remained well above the recommended guidelines. 62 Another study examined how food provision and food choices had changed following the introduction of the 2006 standards in a sample of 136 primary schools across England, and reported that lunchtime food provision was healthier, with more fruit, vegetable and salad provision, and less starchy foods cooked in oil, desserts, crisps and confectionery. In line with this, improvements were seen in children’s F&V consumption. 63

Evidence of impact of the SFS on the dietary intake of secondary school pupils is more limited. One study in 80 secondary schools compared school food provision and consumption of foods and drinks by pupils aged 11–18 years at lunchtime in 2011 with that in 2004. 64 They reported an improvement in the nutritional content of school food, with the greatest impact on the availability of confectionery. Small improvements were observed in pupils’ lunchtime food consumption, such as decreased consumption of starchy foods cooked in oil and slight increases in F&V intake. Total dietary intake was not examined.

Since the introduction of the revised SFS in 2015, there have been no large-scale evaluations of their impact on nutritional intake52 and no comprehensive evaluations of the implementation of the SFP actions in secondary schools. In 2017, the Jamie Oliver Food Foundation published A Report on the Food Education Learning Landscape. 65 This reported findings from surveys, interviews and focus groups with school leaders, governors, teachers, children and parents, and focused on food education and the food culture in schools. Key findings relating to secondary schools were cheap and unhealthy (SFS non-compliant) foods on offer throughout the school day; noisy dining environments, often with long queues; inconsistent messaging with unhealthy foods used for fundraising and as rewards; and patchy provision of food education. More recently, in 2020, Guy’s and St Thomas’ Charity (Impact on Urban Health) published a report66 that outlined the findings of a study of primary and secondary schools across London, in which the food on offer, pupils’ food choices and the wider dining and food environments were observed. Only a small number of secondary schools participated, but the key findings in these schools were poor awareness and implementation of school food policies, food offers that were non-compliant with the SFS (particularly at breakfast and breaktimes), and pupils often choosing less healthy options, such as ‘grab and go’ food, even when a more nutritionally balanced meal was on offer. Both reports highlighted the lack of monitoring and enforcement of the SFS in secondary schools. In addition, similar issues were highlighted in the 2019 State of the Nation: Children’s Food in England report by Food for Life, Soil Association. Through consultation with 30 caterers, they estimated that around 60% of secondary schools may not be compliant with the SFS and recommended better monitoring of SFS compliance. 67 The need for robust and independent monitoring of food standards compliance in schools was also identified by Dame Sally Davies in the report she published as outgoing Chief Medical Officer for England in 2019, Time to Solve Childhood Obesity. 68

Two papers, one published in 201369 and one published in 2022,52 provide a comprehensive summary of the introduction and changes to the SFS over the last 20 years across all UK nations, including England. The first reported that in England, since the introduction of compulsory SFS in 2006, school lunch uptake had increased, and there was some evidence to suggest nutritional benefit from the standards, particularly in younger children. 69 However, the 2022 paper identified that no studies had examined the impact of the updated food-based SFS on children’s nutritional intakes, and highlighted that this lack of studies, together with the lack of monitoring of compliance of the SFS, made it difficult to determine the impact of the SFS policy on children’s nutritional intake. 52

School food policy in the context of wider government strategy

The SFS legislation remains an important national strategy for tackling childhood obesity and is a key component of the government’s report Childhood Obesity: A Plan for Action. 70,71 In this, the government states plans to update the SFS, with a particular focus on reducing the sugar content of foods. 71 However, this was put on hold during the COVID-19 pandemic, and the focus has moved to ensuring compliance with the current SFS. 72 Also as part of the their childhood obesity action plan, the government’s DfE introduced the healthy schools rating scheme in 2019. 73 This is a school self-assessment scheme to achieve bronze, silver or gold awards based on a number of factors related to healthy eating and physical activity, which include compliance with the national SFS and food education provision.

More recently, in 2021, Henry Dimbleby led the development of an independent national food strategy report74 in which he makes broad-ranging recommendations for the food system in England. In this are a suite of recommendations relating to food education and a recommendation to extend free school meal (FSM) eligibility. Following this independent report, the government launched two White Papers in 2022: Levelling Up the United Kingdom and the Government Food Strategy. 75,76 While not all of Dimbleby’s recommendations were taken up, in these White Papers the government set out plans to provide training to school governors on a whole-school approach to food, and to invest up to £5M in the food curriculum with the aim of children leaving secondary school with the knowledge of how to cook six basic recipes to support healthy eating. It also set out the intention to test approaches to ensuring transparency of school food arrangements and compliance with the SFS.

Schools

To generate our sampling frame, we identified all state-funded secondary school academies and free schools in the 22 local authority areas outlined above using DfE routine data. As we were comparing two groups of schools, those mandated and those not mandated to comply with the SFS, we restricted the sampling frame to include only academies and free schools. The reason for this was that these are the only types of state-funded secondary schools that can potentially have exemption from the SFS (i.e. if they were established between 2010 and 2014). We excluded other secondary school types as they are all legally required to comply with the SFS and so could only be recruited to the SFS-mandated group, and we wanted to enable comparability in terms of other school characteristics across the two school groups. The excluded school types significantly differ from academies and free schools in their governance structures (local authority-maintained schools) or because they are specialist/alternative educational providers, and therefore their inclusion in the SFS-mandated group would reduce comparability across the two school groups. In 2019, at the start of this study, academies and free schools were providing education to 72% of secondary school pupils in England,81 and by the end of the study this figure had risen to 80%. 82

We used stratified sampling, based on propensity score methods,83 to increase the comparability of schools and reduce the influence of confounding. Propensity scores were generated using the following school characteristics:

• local authority area

• establishment type (academy sponsor-led, academy convertor or free school)

• rural or urban categorisation

• number of pupils

• percentage of female pupils

• proportion of pupils from ethnic minorities groups

• proportion of pupils with English as an additional language (EAL)

• proportion of pupils with special educational needs (SEN)

• proportion of pupils eligible for FSM

• Income Deprivation Affecting Children Index [IDACI, a component of the Index of Multiple Deprivation (IMD) representing the proportion of children aged 0–15 years living in income-deprived families, measured for the lower-layer super output area in which the school is situated]

• presence or absence of a sixth form (education provision for 16- to 18-year-olds)

• selective or non-selective admission policy

• religious status: faith school or secular.

The propensity score was derived from a logistic regression model, fitted using these variables to predict whether SFS compliance was mandated. The schools were split into four strata using the propensity score quartile cut-offs, and within each strata schools were divided into SFS-mandated and SFS-non-mandated schools, generating eight sampling groups within the sampling frame. Schools in each group were randomly ordered and invited sequentially, the intention being to recruit schools from across all eight groups and achieve an even split of SFS-mandated and SFS-non-mandated schools.

Pupils

We sampled pupils attending participating schools. We identified form groups or other class groupings [such as personal, social, health and economic education (PSHE) classes], attempting to select class groupings that were representative of the general year group and avoiding groupings that were streamed according to academic ability or related to a non-core, selected subject. From these class groupings, one class from each of years 7 (age 11–12 years), 9 (age 13–14 years) and 10 (age 14–15 years) were selected to ensure that a range of ages was included in the sample. Pupils in these three selected classes were invited to take part in data collection.

School staff and governors

We invited a selection of staff and governors from participating schools to take part in the study, aiming for at least four representatives from each school (see School-level data capture for details of the staff groups recruited).

Sample size

We conducted a power analysis to estimate the required sample size, based on the comparison of free sugar intake in pupils attending SFS-mandated and SFS-non-mandated schools. We used data on free sugar intake pre and post implementation of the 2006 legislation from a study conducted in English middle schools84 to inform the analysis. The study reported a reduction in free sugar intake at school lunch of 6 g [standard deviation (SD) 11 g] in pupils consuming school-provided food, and 2 g (SD 13 g) in those consuming packed lunches; therefore, for this study we based our power analysis on the ability to detect a 4-g difference between groups. Assuming a SD of 11 g, an intracluster correlation coefficient (ICC) of 0.1 (a conservative estimate85) and balanced cluster sizes, we estimated that to detect this difference with at least 90% power and at 5% significance, we would require 990 evaluable participants and 22 clusters (schools) in each group (total schools, n = 44; total participants, n = 1980). Within each year group cluster, we anticipated that there would be a minimum of 15 students, with a total of at least 45 students for each school (average cluster size, n = 45).

An additional power analysis was conducted in November 2021, as it became apparent that we were unlikely to achieve the original target school sample size (n = 44) due to ongoing interruptions resulting from the COVID-19 pandemic. At this point, based on data collected until that time, we calculated that the average number of students participating from each school was 68, and we took note that it was unlikely that we would be able to recruit equal numbers of schools to the SFS-mandated and SFS-non-mandated groups (due to a lower number of SFS-mandated schools in the sampling frame). We therefore used a cluster size of 68 and accounted for a likely imbalance across the school groups in this additional power analysis. With all the other parameters kept the same, we estimated that 14 schools in the SFS-mandated and 20 schools in the SFS-non-mandated groups would give 87% power, and 17 schools in the SFS-mandated and 23 schools in the SFS-non-mandated groups would give 92% power.

Schools

Postal and e-mail invitations, including a study information leaflet, were sent to the headteachers of schools in a staggered manner, inviting several schools from each of the eight sampling groups at a time. Invitations were followed by telephone calls to schools, with additional e-mails sent as required. For schools expressing an interest, we arranged face-to-face or remote (telephone call or video) meetings to discuss the study and explain what taking part would involve, and to provide an opportunity for the school to ask questions. Once a school agreed to participate, they were asked to complete and sign a school agreement outlining the expectations of the school and the research team. Schools received a £300 payment for participating in phase I and a tailored report detailing the school’s compliance with the SFS and implementation of SFP actions. Schools were asked to provide the study team with a specific study liaison member of staff.

Staff/governors

Relevant staff and governors were identified by the liaison person and invited to take part by e-mail, which included a participant information leaflet. Written consent was obtained. Invited staff were those identified within the school to have roles relating to food provision, the eating environment, the food curriculum or SFS/SFP implementation (including headteachers, catering staff, PSHE leads, teachers with responsibility for the food/cooking curriculum and relevant representatives from the governing body).

Pupils

All pupils in selected classes were invited to participate. Parents were given written detailed information about the study at least 1 week prior to the first data collection session. Schools assisted in the distribution of this information to parents in different formats (e.g. e-mail, post, website). Parents were not asked for active consent but were given the opportunity to complete and return a form to opt their child out of participating in the study. This approach to parental consent was used to ensure that there was no socioeconomic bias in our sample, as previous research has shown that active consent letters distributed by schools are less likely to be returned by more socioeconomically disadvantaged parents. 86 Participant information leaflets were distributed to pupils at least 1 week prior to data collection. Researchers also gave a verbal overview of the study on the day of the first data collection session and pupils were invited to ask questions. Written assent was be obtained from pupils whose parents had not opted them out of participating. Based on the advice we received from our youth public advisory group (see Public involvement methods), pupils were given a £5 shopping voucher as a thank-you for taking part.

Theoretical considerations guiding data collection

We developed a logic model to outline the processes by which the SFS and SFP are assumed to produce health gains (Figure 1). Through this model we articulated that health gain is generated:

FIGURE 1.

Logic model for the influence of SFS and SFP actions on children’s dietary intake and health outcomes.

1. directly via a change in school food consumption

2. indirectly through healthy food environments, and curricular and other activities that are designed to change pupils’ dietary knowledge, attitudes and beliefs, which impact on their food and drink consumption in and out of school.

The logic model also outlines that the extent to which the SFS and SFP positively impact on dietary intake and health depends on their implementation within a school, which is influenced by contextual factors (e.g. school leadership, the wider school culture, the physical environment).

The logic model, along with the principles outlined in the UK Medical Research Council framework for process evaluation,87 guided our data collection strategy. We planned to capture data aligned with each of the boxes in the logic model, enabling us to assess schools’ compliance with the SFS and implementation of the SFP actions. We also planned to assess the wider school context and the extent to which the two national school food policies (SFS and SFP) were adopted by and embedded within schools. Broadly, we structured our data collection to encompass:

• food provision

• the school food culture and environment

• costs relating to school food provision, culture and environment

• school contextual data

• embedding of school food policy within schools

• school meal uptake

• school food sales

• pupil outcomes (nutritional intake and dental caries experience; see Pupil outcomes).

Data collection was undertaken at the school and the individual pupil levels and is outlined below. Recruitment and data collection processes are presented in Figure 2.

FIGURE 2.

Recruitment and data collection flow chart. CHU-9D, Child Health Utility 9-Dimensions; PIL, participant information leaflet.

School-level data capture

Pupil outcomes

We aimed to compare dietary intake and indicators of dental caries in pupils in the SFS-mandated and SFS-non-mandated secondary school groups. As one of the main outcomes of interest was free sugar intake, and for the purposes of our sample size calculation, we defined our primary outcome as intake of free sugars (in grams). We assessed free sugar intake during school day lunch, while present at school, and during the full 24-hour period of the same school day.

Secondary nutritional outcomes were:

1. percentage of dietary energy intake from free sugars

2. total energy intake (TEI; kcal)

3. total fat intake (g)

4. fibre intake (g)

5. fruit and vegetable portions consumed

6. number of SSBs consumed

7. number of sugar and chocolate confectionery items consumed

8. number of HFSS foods consumed (defined according to the Nutrient Profiling model90)

9. free sugar intake providing ˃ 5% of 24-hour TEI

10. consumption of five or more portions of F&V during a 24-hour period

11. number of eating/drinking occasions (excluding plain water) during a 24-hour period.

As with the primary outcome, the secondary nutritional outcomes numbered 1–8 above were assessed during school day lunch, while present at school, and during the full 24-hour period of the same school day.

Dental outcomes were:

1. the presence of any symptoms indicating dental caries in the past 3 months

2. the number of dental caries symptoms in the past 3 months

3. past dental caries treatment.

Pupil data capture

Pupils were invited to two computer-based sessions, completed during timetabled lessons, and asked to complete (1) an online survey (using REDCap software;91 available on the study web page at www.fundingawards.nihr.ac.uk/award/17/92/39) hosted at the University of Birmingham and (2) an online 24-hour dietary recall. For the first session, at least one researcher was present to facilitate the session. This included providing a verbal overview of the participant information leaflet, answering any questions related to participation and providing assistance throughout the session to pupils who required it. The second session was facilitated by the class teacher approximately 2–4 weeks later. Paper copies of the surveys were available if there were any issues with computer/internet access. In some situations, pupils were asked to complete the survey or 24-hour dietary recall at home, for example if a pupil was late to the session, if a pupil had issues with computer/internet access or if the school could not find a timetabled session to complete the second data collection session.

Aim

Through our public involvement activities, we aimed to consult with young people, parents and school staff/governors on the design and conduct of the study, including recruitment, data collection, data presentation and dissemination.

Recruitment and membership

Throughout the study, we had a teacher in a school senior leadership role as part of our research team (SW; see Acknowledgements), who provided advice on a range of aspects of the study, including engaging schools and pupils, access to school-owned food sales data, considerations around school management and governance systems, and planning data collection as part of timetabled school sessions.

We convened three public advisory groups: (1) young people, (2) parents and (3) school staff and governors. Members of the young people’s group were originally recruited through two schools with whom we had established relationships. In each school we met face to face with two groups of six pupils (year 7, age 11–12 years; and year 10, age 14–15 years), with a total of 24 pupils involved. As the study progressed, we linked with the established Young People’s Advisory Group for the National Institute for Health and Care Research (NIHR) West Midlands Clinical Research Network. We met with this group via video call on two occasions, with 7–9 young people in attendance.

We initially recruited a group of four parents of secondary school pupils via personal contacts and networks within the University of Birmingham. This group met face to face on one occasion, with further interaction over e-mail due to the national lockdown in response to the COVID-19 pandemic (March 2020). A new parent group was established towards the end of the study (July 2022) using social media and participating school contacts. This group comprised eight members and met remotely on one occasion to advise on study dissemination.

The school staff/governor group members were recruited through personal contacts and networks within the University of Birmingham. This group met on four occasions, both face to face and remotely, with additional interaction over e-mail during the national lockdown. Membership of this group fluctuated between three and six members and included teachers, senior leadership team (SLT) representatives and governors. Members of all public advisory groups received reimbursement for time and expenses.

We also had three public representatives on our Study Steering Committee (SSC): SB (teacher with a senior leadership role in a multi-academy trust; SSC member March 2019–January 2020), AH (SSC member January 2020–September 2022) and CM (retired headteacher and chief executive officer of multi-academy trust; SSC member March–September 2022). These public representatives provided advice and oversight throughout the study.

Methods of engagement

We developed terms of reference for the parent and staff advisory groups. In the initial meetings with all the public advisory groups, we presented a summary of the study and outlined the role of the group. During group meetings, we used an interactive approach to engaging with our public advisors. This involved working in small groups to review study materials with questions to prompt discussion, and using voting buttons, emoticons, the chat function (if online) and audio–visual materials to engage group members. We took structured notes and presented the thoughts and views expressed in the meetings to the Study Management Team and the SSC for consideration.

Assessment of school food standards adherence

We assessed compliance with the SFS based on (1) observation of mealtimes across a full school day and (2) review of weekly menus. The purpose of completing an assessment of SFS compliance on the 1-day observational data was to capture the off-menu foods on offer in secondary schools such as breakfast and breaktime offers, drinks, condiments, snacks and sandwiches. The observational data also enabled us to gain further information on menu items, as menus often lacked detail such as the types of fruit/vegetables or cakes/biscuits offered. The menu review was required in addition to the 1-day observational data to enable assessment of standards that apply to schools on a weekly (e.g. starchy foods cooked in fat or oil) or 3-weekly (e.g. oily fish) basis.

School food standards compliance criteria were guided by the UK statutory instrument The Requirements for School Food Regulations 201444 and School Food Standards Practical Guidance. 51 A SFS question-and-answer document was also used to check the interpretation and definitions of food categories. 105

Some additional definitions of food categories beyond those given in the SFS documentation were required to categorise all foods and drinks offered, for example desserts, confectionery and snacks (available on the study web page at www.fundingawards.nihr.ac.uk/award/17/92/39). We did not collect recipes, so we developed and applied rules for some of the food types described in the SFS; for example, for ‘fruit-based desserts’ (with a content of at least 50% fruit measured by volume of raw ingredients), we included desserts such as fruit pie, fruit crumble, fruit cobbler and jelly made with fruit. For other dessert items, the research team searched the internet and reviewed a standard recipe (e.g. British Broadcasting Corporation Good Food) for the dessert or a similar item to make a judgement on whether it met the criteria for a fruit-based dessert. Drinks and prepackaged snacks, biscuits and cakes were reviewed for compliance using the ingredients lists detailed on brand or shopping websites.

All foods and drinks recorded in the observation tool and menus were extracted into Microsoft Excel^® (Microsoft Corporation, Redmond, WA, USA) spreadsheets and coded based on the food/drink type. One researcher carried out data extraction, and two researchers familiar with the SFS independently coded foods and drinks and judged SFS compliance for the two sets of data (observation and menu), entering their judgements into a data entry form. The judgements of the two researchers were compared, and, where discrepancies arose, the research team met to agree on a final judgement. The 1-day observation data judgements and menu judgements were combined into an overall SFS compliance judgement for each standard. Details on the standards assessed and how judgements were combined across 1-day and menu assessments are available on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39).

As a final quality check, the coded food/drink data were back-checked against menus and observations for any missing foods/drinks or errors in data entry, and final SFS judgements for all schools were checked for consistency in application (MM). Any judgements that were deemed inaccurate were reviewed and corrected by two researchers (MM and AD). Final judgement options were yes (meets standard), no (does not meet standard), can’t tell (it was not possible to assess compliance due to lack of information) or not applicable (standard did not apply). Not applicable (N/A) was used for standards relating to drinks, condiments, dried fruit and crackers/breadsticks when these items were not sold/not available in the school. We calculated the proportion of standards met for each school using the number of assessable standards as the denominator (i.e. excluding ‘can’t tell’ judgements).

Assessment of School Food Plan actions

Actions included in the SFP were identified following a comprehensive review of relevant SFP resources: the checklist for headteachers, the creating a culture and ethos of healthy-eating guidance, and the guidance for school governors. 56–58 Some actions were unique to each document, whereas others were repeated or similar across documents. Those that were repeated or similar were merged to form single actions, retaining as much of the original wording as possible. We identified a total of 69 actions, which we grouped into 9 ‘themes’ based on the headings used in the SFP resources. More detail about the specific actions within each theme is provided in Appendix 2, Table 32.

We assessed the implementation of SFP actions using a variety of data sources, including researcher observation, survey responses from schools, staff and pupils, and the review of school documents, for example curriculum documents, catering contract and meeting minutes. Further detail on the source of data for specific actions and data points/questions used for assessment is provided in Appendix 2, Tables 32 and 33.

A judgement of red, amber or green (RAG) was made on each action for each data source, with red indicating that the action had not been implemented, amber indicating that it had been partially implemented and green indicating full implementation. If data were missing (e.g. incomplete survey responses or documents not provided by schools), a judgement could not be made, indicated by a ‘no judgement’ rating. A final judgement of high, medium or low implementation was made for each action point based on RAG ratings across all data sources, with rules for combining these judgements agreed in advance by the research team (available in Appendix 2, Table 34). The assessment was conducted by one researcher. Uncertainties were discussed with the research team, who then agreed on a final judgement.

The proportion of actions judged as ‘high’, ‘medium’ or ‘low’ was calculated using the number of assessable standards as the denominator (i.e. excluding those actions for which a judgement could not be made).

Assessment of the embedding of national school food policy within schools

Questions were developed to assess the embedding of the SFS and SFP within schools. These were adapted from the NoMAD instrument88,89 (based on NPT106) with statements mapping onto the relevant NoMAD constructs and their constituent components in relation to coherence (individual and collective sense-making), cognitive participation (relational work, building a community of practice), collective action (operational work, enacting a new set of practices) and reflexive monitoring (appraisal, understanding how a new set of practices affects individuals and others). Details of relevant questions and their relation to NPT constructs are available on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39). Data from these questions were extracted from the staff and governor survey data sets and responses presented in tables and graphs to explore variation in responses across the SFS-mandated and SFS-non-mandated school groups and across staff/governor groupings [see Embedding of the school food standards and School Food Plan national school food policies within schools (research objectives 1a and 1b)].

Development of typology schools in relation to school food provision and support for healthy eating

School-level data were used to identify different school ‘types’ in a two-stage process. In the first stage, we used SFS compliance data to generate school types. In the second stage, we inspected SFP implementation data to identify subtypes within the types generated in stage 1.

School food uptake and sales data analysis

Analyses of pupil nutritional intake and dental health data

The study statistical analysis plan is available on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39).

Approach to the economic evaluation

The study health economic analysis plan is available on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39).

The economic analyses took the form of a micro-costing analysis and a cost–consequences analysis. For the micro-costing analysis, the costs of implementing and delivering the SFS and SFP plus the costs of school food provision were calculated for each school using data provided by the school-reported costing survey. For the cost–consequences analysis, the costs of implementing and delivering the SFS and the SFP, the costs of school food provision, and selected consequences (i.e. pupil outcomes) were summarised in the form of a balance sheet for the two groups of schools being compared (SFS-mandated and SFS-non-mandated). A societal sector perspective was adopted for the cost–consequences analysis, meaning that the distribution of costs and consequences borne by schools, catering companies, volunteers, and families and pupils were considered and highlighted. The time horizon for the analyses was 1 year. No discount rate was applied as the data covered a period of < 1 year. All costs were converted to Great British pounds 2021 based on the UK Consumer Price Index as published by Office for National Statistics. 108 Statistical analyses were conducted using Stata.

Costs

All schools were asked for data to capture any one-off or occasional costs associated with the implementation and delivery of the SFS and the SFP, ongoing annual costs associated with food provision and costs associated with other activities to support healthy eating, for example staff training (in addition to those outlined in the SFP). The survey was initially developed by the FUEL study team as a paper-based questionnaire and tailored to individual schools based on the observed school characteristics. The questions reflected each of the elements of the SFS and the SFP. However, after a low response rate and poor completion, the FUEL study team adapted the survey by shortening it to focus on what was deemed the most important resource use items and converted it to an online format using REDCap. The survey was sent to liaison persons at each participating school. The schools were offered help with completing the survey. The paper-based and online versions of the survey are available on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39).

Unfortunately, the school resource-use data (from both paper-based and online surveys) had large numbers of missing data, so these were supplemented with secondary data on catering expenditure by schools. 109 Legally, all schools in England are required to report expenditure data under predetermined budget headings, and this information is publicly available on an annual basis for both catering staff and catering supplies (including the costs of providing FSM and milk). However, it does not include expenditures associated with mealtime assistance and midday supervisors, any training for catering staff or costs related to maintenance and improvement to the canteen. Further detail on what is included in this data capture is provided on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39). For all schools, the catering expenditure for the year 2018–9 was used to reflect catering before the COVID-19 pandemic, apart from four schools for which all other study data were collected during the pandemic (2020–1 academic year), and therefore the same year was used for the catering expenditure to reflect spending during the pandemic.

The costs for pupils and families were collected in the pupil survey (see Sociodemographic and other data and the study web page at www.fundingawards.nihr.ac.uk/award/17/92/39). The pupils were asked to indicate how much they typically spent on food per day during the school day, purchased both at school and at food outlets outside school. The answer options were given in ranges (e.g. £1–2.99). The average spending per pupil per day was calculated by using the mid-point value for each category and summed to calculate the total spending per pupil.

All resource use was valued in monetary terms using appropriate UK unit costs (see Appendix 5, Table 38) or participant valuations estimated at the time of data collection.

Outcomes

The outcome measures for the health economic analysis included:

• dietary intake

• dental health outcomes

• educational outcomes

• quality-adjusted life-years (QALYs) derived from the CHU-9D utility scores.

The data were collected using the study-specific questionnaires for the pupils as described in Sociodemographic and other data. QALYs were measured using the CHU-9D instrument, which was included in the second pupil survey. The CHU-9D has previously been used to measure QALYs in general population samples of children and adolescents and has been shown to have good psychometric performance. 110 Utility scores were derived from responses to the CHU-9D using preference weights obtained from a sample of the UK adult general population. 103

The dietary intake outcomes, detailed in Generation of pupil nutritional intake outcomes, were included in the cost–consequences analysis. The outcomes corresponding to the dietary intake during the school day were selected. These outcomes were calculated only for the pupils whose TEI during school day was above zero. Additional outcomes related to dietary intake during 24 hours and included in the cost–consequences analysis were:

• > 5% of 24-hour TEI from free sugar

• consumption of five or more portions of F&V per day

• number of eating/drinking occasions (excluding plain water).

All dental health outcomes, detailed in Generation of pupil dental outcomes, were included in the cost–consequences analysis.

Educational outcomes data were collected from publicly available school performance tables111 for the year 2019 to reflect school performance before the COVID-19 pandemic. Relevant indicators were identified by the FUEL study team in consultation with the SSC and school public advisors. The following indicators were included in the cost–consequences analysis:

• absenteeism –

• overall rate of absence

• persistent absence.

• secondary performance –

• progress 8 score (a measure of the progress children make between the end of primary school and the end of secondary school relative to the national average, based on attainment 8 scores)

• attainment 8 score (a measure of the average academic performance of a secondary school in eight defined subjects)

• staying in education or entering employment

• grade 5 or above in English and Maths General Certificate of Secondary Education (GCSEs; a subject-specific academic qualification taken in England, Wales and Northern Ireland).

Micro-costing

All schools that completed the costing survey were included in the micro-costing analyses. The resource use was summarised for each school separately, multiplied by the relevant unit costs, and summed to calculate the total ongoing costs of implementing and delivering the SFS and SFP. The wider costs of food provision and creating a healthy food culture, including the costs for schools, catering providers and volunteer time, were included. To calculate the per-pupil costs of food provision/creating a healthy-food culture, these total costs were divided by the number of pupils in the relevant school for the year when the cost data were collected. Given that the main costs related to food provision, we accounted for school meal uptake by also calculating the costs per pupil based on the number of pupils reported to have school meals. Where data from the schools on the number of pupils having school meals were not available, school meal uptake was estimated from pupil participant data (Intake24). As there was variation in how well the schools completed the survey, the percentage of missing cost data was calculated for each school.

Several assumptions had to be made beyond those outlined in Appendix 5, Table 38. These assumptions were validated internally by the FUEL study team. To calculate the resource use associated with mealtime supervision, we assumed that breakfast time lasted 15 minutes, breaktime lasted 15 minutes and lunchtime lasted 1 hour, in the absence of other data to support this. For some schools, data on the duration of mealtimes were available, and these were used instead. To calculate the cost of classroom cover during staff training, it was assumed that one training session lasted 2 hours and corresponded to the unit cost of teaching staff time (i.e. £20 per hour). The school gardening club sessions were assumed to last 1 hour with one staff member leading the session.

Cost–consequences analysis

All schools participating in the FUEL study were included in the cost–consequences analysis, where costs and consequences were listed for each school in the form of a balance sheet. Furthermore, the average values were calculated for the SFS-mandated and SFS-non-mandated schools. The following costs were included in the cost–consequences analysis: the costs of food provision calculated from the micro-costing analysis (both from a school and a public sector perspective); catering expenditure; and the costs for pupils/families, including the cost of buying food during the school day and the cost of buying the ingredients for the cooking lessons. The total costs, the costs per pupil based on the total number of pupils in each school, and the costs per pupil based on the school meal uptake are presented. The following outcomes were included as described in Outcomes: QALYs, educational outcomes, dietary intake outcomes and dental health outcomes.

Aim and objectives

The central aim was to explore with school staff and governors in more depth the way in which the SFS, SFP and local school food policy or initiatives were introduced, embedded and sustained in the schools, and their perceived influence on the dietary intake of pupils.

In relation to pupils, we sought to explore:

• views on school food and the school food environment and contextual factors influencing these views

• how pupils interact with the school food environment and the wider outside school environment in terms of their eating behaviour

• perceived negative impacts of SFS or SFP and regulation of foods provided.

Sampling approach

In our modified school sampling approach, we aimed to ensure representation of schools with a high proportion of pupils eligible for FSM and a mix of schools based on SFS status and catering provision (external and in-house). We aimed to recruit four case sites from the 36 participating schools.

Our participant sampling approach was to identify key school management, teaching and catering staff, along with governors, in each of the case sites. We invited these identified staff/governors to participate in an interview. Interviews were planned with flexibility of setting – either face to face or via video/telephone call – to reflect that research was conducted during the COVID-19 pandemic. It was anticipated that, depending on the size of the school, four to six staff/governor participants per school would be interviewed. Our sampling approach in relation to pupils was to sample pupils from year groups 7, 9 and 10, aligning with the main study. It was anticipated that two or three focus groups would be conducted in each of the case sites, involving approximately 16–24 pupils in each school.

Recruitment and consent

Data collection

Data collection was conducted during the 2021–2 academic year. Each school was visited on one occasion by different members of the research team (MM, RD, AD, LM, RA) who conducted interviews and focus groups. On one occasion, our public representative on the SSC (CM) also attended to observe.

Focus groups were conducted face to face in a quiet space and during the timetabled school day. Interviews with staff were individual and face to face, with the option of remote administration if required, and were also conducted in a quiet space during the timetabled school day. Semi-structured topic guides were used to guide the discussion (see the study web page at www.fundingawards.nihr.ac.uk/award/17/92/39).

During the fieldwork, two to four researchers typically visited each case site, with one to two researchers per interview/focus group. In the focus groups, the ‘lead’ researcher focused on facilitating the focus group, while the second researcher’s role oriented around recording observational notes, operating the audio-recorder and collecting demographic data from pupils and staff.

All interviews and focus groups were recorded using a digital audio-recorder. Verbal permission was sought before recording, in addition to the written consent/assent provided by participants.

Demographic data were also collected from participants via a paper/online questionnaire. For pupils, this included questions about their ethnicity, home postcode and FSM take-up and whether they bought food from school. For staff, questions related to their specific role in the school and the number of years they had been in post.

Participants were assigned participant IDs and their demographic records were pseudonymised. Audio-recordings were transcribed by an external transcription service, which was asked to remove identifiable data such as names, organisational names and geographical areas. Researchers could refer to the original audio file during the analysis phase if required. For staff interviews, transcripts were labelled using the participant ID number. For focus groups, the transcript was treated as one unit, and excerpts were not assigned to individual participants.

Analysis

A thematic analysis of data gathered from the case study was conducted and was informed by the Framework analytical approach. 117

Audio-recordings from both interviews and focus groups were transcribed verbatim by an external transcription service (stage 1) and underwent quality checking by members of the research team (stage 2) to enable early familiarisation with this textual data. With the aid of NVivo v11 (QSR International, Warrington, UK), five members of the research team (MM, RA, CR, MP, PA) were involved in the initial exploratory coding (stage 3) of a small number of transcripts selected from researchers’ memory as being particularly interesting. Here, an inductive approach was utilised to generate codes and group them into categories.

Following initial coding, a working analytical framework (codebook) was developed (stage 4) (the codebook is available on the study web page at www.fundingawards.nihr.ac.uk/award/17/92/39). This consisted of seven parent codes (categories) and a large number of subcodes, each of which had positive and negative values that corresponded to whether or not that particular aspect of school food was present. We also used two a priori codes relating to views on the SFS and implementation, and awareness and implementation of the SFP. Subsequently, three members of the team (MM, RA, CR) were each allocated a single case site, with a fourth case site split between two researchers (RA, CR), and the framework was applied to the coding of all transcripts related to each site (stage 5). After generating framework matrices in NVivo (stage 6), with interviewees/focus groups in rows and codes in columns, each member of the analysis team wrote coding summaries for their allocated sites. Summaries were reviewed collaboratively (MM, RA, CR, MP, PA) to support identification and discussion of interpreted themes. Thematic summaries were initially compiled for each individual case site and then combined by comparing sites and exploring common and contrasting themes across sites, (stage 7), in line with other health services research using Framework analysis in a case study approach. 116 The thematic overview was sense-checked by the wider team.

Implementation of the school food standards: normalization process theory constructs

Responses to the questions on implementation of the SFS, relating to the four NPT constructs, are presented in Figure 3.

FIGURE 3.

Staff responses to questions relating to NPT implementation constructs for the SFS. Note: number of respondents for each statement ranges from 112 to 119; x-axis indicates the percentage of respondents within each agreement category.

Implementation of the School Food Plan: normalization process theory constructs

Responses to the questions on implementation of the SFP actions, relating to the four NPT constructs, are presented in Figure 4. A detailed table of responses is provided in Report Supplementary Material 1, Table 2.

FIGURE 4.

Staff responses to questions relating to NPT implementation constructs for the SFP. Note: number of respondents for each statement ranges from 54 to 56; x-axis indicates the percentage of respondents within each agreement category.

Uptake of school lunches

School lunch uptake data were reported by 20 schools (56% of sample), but there was a large difference in the proportion of SFS-mandated and SFS-non-mandated schools providing this information (23% of SFS-mandated schools vs. 74% of SFS-non-mandated schools). On average, schools reported that 54% of pupils purchased school lunch, with a similar range of values for SFS-mandated and SFS-non-mandated schools. The SFS-mandated group includes one school that had a policy of no packed lunches, so it includes a school with 100% school meal uptake (reported by the school). We also calculated the proportion of pupil participants having school-provided food for lunch in each participating school from Intake24 data. We observed similar levels of school lunch uptake using this pupil-reported data (Table 11).

School food sales

Thirteen of 36 schools (36%) provided monthly itemised food sales data. Eleven of these schools provided data for more than one time period. The number of food and drink items sold in 1 month ranged from 8821 (ID6 – a small school that reported 54% school meal uptake) to 62,592 (ID17 – a large school reporting 70% school meal uptake). The average number of food/drink items sold per pupil attending the school for each month ranged from 12 (ID39 and ID45, June 2020) to 36 (ID14, November 2019). Sales data were not requested for June 2020 due to restricted school openings during the first national lockdown in response to the COVID-19 pandemic.

For schools providing data for both June and November of any year (n = 9), sales per pupil were higher in November than in June. For schools providing data before the COVID-19 pandemic and during the pandemic (n = 4), total sales were substantially lower during the pandemic (i.e. November 2019 vs. November 2020–November 2021). Comparisons between SFS-mandated and SFS-non-mandated schools were limited due to the small number of SFS-mandated schools providing data, with only one school providing data over multiple time points. However, when the mean for all schools and all time points was taken for the SFS-mandated and SFS-non-mandated groups, sales per pupil were similar (Table 12).

Sales of indicator foods by school food standards-mandated/school food standards-non-mandated status

Of the three SFS-mandated schools providing food sales data, two provided November data only, and one included pre-pandemic data (2019). Of the 10 SFS-non-mandated schools, 2 provided November data only, and 6 included pre-pandemic data.

Figure 5 provides details of sales of indicator items as a percentage of total sales per month by SFS group. Sales of confectionery/chocolate items (which are not permitted within the SFS) and pastry items were higher in SFS-non-mandated schools than in SFS-mandated schools (1.5% vs. 0% and 6% vs. 4% of total sales, respectively). Sales of desserts, cakes and biscuits were slightly higher in SFS-non-mandated schools (17% vs. 15.5%). Sales of starchy foods cooked in fat/oil and other non-compliant snacks were slightly higher in SFS-mandated schools (4% vs. 2% and 4% vs. 1% of total sales, respectively). Sales of fruit were the same across both groups, while sales of vegetables and salads were higher in SFS-non-mandated schools (0.5% vs. 0.1%). Sales of non-SFS-compliant drinks were higher in SFS-mandated schools than in SFS-non-mandated schools (10% vs. 6%), but sales of SFS-compliant drinks were also higher (12% vs. 9%), while sales of plain water were lower (3% vs. 6%).

FIGURE 5.

Sales of indicator items as a percentage of total sales per school per month, by SFS-mandated and SFS-non-mandated schools.

There are key limitations to note with these data. Only three SFS-mandated schools provided data, so outliers (e.g. high values in single schools) are likely to have a larger impact on the mean values for this group, and only one-third of the school sample provided sales data overall, which limits the representativeness of the data and the usefulness of comparisons across the two groups.

Full details of mean monthly sales of indicator items by school are provided in Report Supplementary Material 2, Table 1.

Nutritional intake

Of the 2273 participants providing nutritional intake data, 1046 (46%) had two 24-hour dietary records, with the remaining 54% having one record. The proportion of participants with two 24-hour dietary intake records was lower in the SFS-mandated school group than in the SFS-non-mandated group (37% vs. 51%).

Pupils’ mean nutrient intakes at lunch, across the school day and over 24 hours are shown by SFS-mandated and SFS-non-mandated schools in Table 15. There was some variation across SFS-mandated and non-SFS mandated schools, with pupils from SFS-mandated schools reporting less free sugar intake at lunch, but more during the school day and over 24 hours, and a higher percentage TEI from free sugar across the whole day (24 hours). Pupils in the SFS-mandated schools had lower energy, fat, fibre and F&V intake across all time periods, and a lower proportion of pupils in the SFS-mandated schools consumed five or more F&V portions in 24 hours.

Nutritional intake: adjusted analyses using multilevel models

Dental outcomes

Dental outcomes and other factors related to dental health are presented for participants in the SFS-mandated and SFS-non-mandated school groups in Table 19. Some marginal variations were present across the two school groups, with a lower proportion of pupils in the SFS-mandated group having dental caries symptoms, and a slightly higher proportion in this group receiving dental caries treatment in the past. Home fluoridation levels were higher, but the proportion of pupils brushing their teeth at least twice a day was lower in the SFS-mandated school group.

Dental outcomes: adjusted analyses using multilevel models

Multilevel models to compare the presence and number of dental caries symptoms in the previous 3 months and the past receipt of treatment for dental caries between SFS-mandated and SFS-non-mandated schools are presented in Table 20. There were no significant differences in the presence or the number of dental caries symptoms or in the likelihood of dental caries treatment between the SFS-mandated and SFS-non-mandated school groups. Male participants had lower odds of caries [odds ratio (OR) 0.76, 95% CI 0.62 to 0.94], less extensive symptoms and lower odds of dental caries treatment, when adjusting for other variables. There were no significant differences across the individual and school-level deprivation indicators (IMD quintile group, school % FSM, school IDACI) in the odds of having any dental caries symptoms, but there were differences in the number of symptoms experienced in the previous 3 months and receipt of caries treatment. Compared with the most deprived IMD quintile group, other groups had less caries symptoms and lower odds of caries treatment, although there was no gradient present across the groups, and the rate/ORs were only significant for IMD quintile groups 2 and 3. At the school level, a higher percentage of pupils with FSM eligibility was marginally associated with more caries symptoms but a higher school IDACI score (indicating higher income deprivation in families with children in the area surrounding the school) was associated with marginally lower caries symptoms. Toothbrushing at least twice a day was associated with lower odds of dental caries symptoms and treatment and a smaller number of caries symptoms.

Missing deprivation (Index of Multiple Deprivation) data and age data

The percentage of pupils with primary outcome data, but missing data for IMD were 11% and 8% for participants in the school SFS-mandated and SFS-non-mandated groups, respectively. Therefore, we re-ran our primary analysis to include imputed IMD values. Where a pupil had missing IMD data, we used the median IMD rank for pupils in their school and assigned them to the appropriate IMD quintile group. The coefficients in the complete-case and the IMD-imputed models were similar, with no changes in their significance or direction (see Report Supplementary Material 8, Table 1). The significant difference in free sugar intake at lunch between SFS-mandated and SFS-non-mandated school groups remained in the IMD-imputed model [mean difference (MD) –2.03 g, 95% CI –3.76 to –0.30 g; p = 0.02].

Although data on age were missing for only 2.6% of the sample with primary outcome data, we also explored the impact of missing age data on this primary analysis by re-running our model to include imputed age values. Where pupils had missing age data, we used the average age for their year group in their school. The coefficients for the complete-case and the age-imputed models were again similar (see Report Supplementary Material 8, Table 2), with the significant difference in free sugar intake at lunch remaining between school SFS-mandated and SFS-non-mandated groups (–2.88 g, 95% CI –4.75 to –1.02 g; p = 0.003).

Exclusion of participants reporting 24-hour total energy intake of > 4000 or < 400 kcal

To explore the impact of implausible dietary intake reporting, we re-ran our primary analysis excluding participants who reported a 24-hour TEI of < 400 or > 4000 kcal. The adjusted model coefficients were similar for the outcome of free sugar intake at lunch (see Report Supplementary Material 8, Table 3), with a significant difference between SFS-mandated and SFS-non-mandated school groups remaining (MD –2.29 g, 95% CI –4.07 to –0.50 g; p = 0.01). The significance and direction of other coefficients in the model remained the same.

Comparison of pupil nutritional outcomes across the four-group school typology based on school food standards compliance

We explored differences in pupil outcomes across the four identified school types based on levels of implementation of (1) standards relating to obesity and dental health; and (2) standards relating to dietary variety, adjusting for individual-level and school-level factors, as in the primary analyses. The nutritional outcome model regression coefficients for the school types are presented in Table 21. The full models are provided in Report Supplementary Material 9, Tables 1–9.

Compared with the type 1 schools (low implementation of obesity/dental health-related SFS and high implementation of dietary variety-related SFS), pupils in the type 2 school (medium implementation of both sets of SFS) had a lower percentage of TEI from free sugar at lunch (–8.12%, 95% CI –14.48% to –1.76%). The same trend was seen for the school day, but the difference was non-significant, and there was little difference over 24 hours.

Compared with type 1 schools, there were some differences in consumption of sugar and chocolate confectionery items in the types 2 and 4 schools. The type 2 school (medium implementation of obesity/dental health-related SFS and dietary variety-related SFS) had a higher consumption of confectionery items during the school day (IRR 2.70, 95% CI 1.30 to 5.59), but a lower consumption of these items at lunch (non-significant). Of note here is that the type 2 school had a very high percentage (94%) of pupils consuming a school-provided lunch. Pupils in the type 4 school (high implementation of obesity/dental health-related SFS and dietary variety-related SFS) had a higher consumption of confectionery items at lunch (IRR 2.15, 95% CI 1.16 to 3.98) and over 24 hours (IRR 1.46, 95% CI 1.02 to 2.10) than those in type 1 schools.

Level of school compliance with school food standards related to obesity/dental health and dietary variety and pupil nutritional outcomes

Due to the limitations of only having one school each in types 2 and 4, we constructed additional models to explore the relationship between school SFS compliance and pupil nutritional outcomes. We developed two variables based on the data used to inform the development of the four school types: (1) the percentage of the 12 standards related to obesity and dental health met by the school and (2) the percentage of the 15 standards related to dietary variety met by the school. We ran models, including both SFS-related variables and individual and school-level covariates, to explore the association between these variables and pupil nutritional outcomes. The nutritional outcome model regression coefficients for these two SFS-related variables are presented in Table 22. The full models are provided in Report Supplementary Material 10, Tables 1–9.

There were some associations between nutritional intake and the percentage of dietary variety standards met by the school. This variable was positively associated with the percentage of TEI from free sugar at lunch (0.14%, 95% CI 0.07% to 0.22%), with the same positive association across the school day. There were small but significant inverse associations between the percentage of dietary variety standards met, and lunch TEI (–2.14 kcal, 95% CI –3.88 to –0.41 kcal) and lunch fat intake (–0.12 g, 95% CI –0.19 to –0.04 g). The direction of association was the same for the school day and 24-hour models, but the coefficients were not significant. There was also a small negative association between the percentage of dietary variety standards met and fibre intake at lunch (–0.03 g, 95% CI –0.05 to –0.01 g) and across the school day, but no association between this variable and F&V intake.

There was a very small positive association with the percentage of obesity/dental health-related standards met and free sugar intake, but only in the 24-hour intake model (0.29 g, 95% CI 0.03 to 0.56 g). Consistent with this was an association between the percentage of these standards met and consumption of sugar and chocolate confectionery items in the school day and 24-hour models, with a higher percentage of standards met associated with a higher consumption (school day model IRR 1.01, 95% CI 1.00 to 1.02). The same direction of association was seen in the lunch model, but it was not significant. This finding is, to an extent, consistent with the finding of higher confectionery consumption in the types 2 and 4 schools (see models presented in Comparison of pupil nutritional outcomes across the four-group school typology based on school food standards compliance), as these schools had medium and high compliance with the obesity/dental-health-related standards, respectively, and higher consumption of confectionery items during the school day/at lunch, compared with type 1 schools, which had low compliance with the obesity/dental-health-related standards.

Comparison of pupil nutritional outcomes across the six-group school typology based on school food standards compliance and School Food Plan implementation

We also explored pupil nutritional outcomes across the six school types that we identified by subdividing the initial SFS-compliance-based types according to their level of implementation of SFP actions (described in Development of typology schools in relation to school food provision and support for healthy eating). The nutritional outcome model regression coefficients for these school types are presented in Appendix 8, Table 43, and the full models are provided in Report Supplementary Material 11, Tables 1–9. Overall, there were few additional differences across the six school types. There were significant differences in consumption of HFSS items, with a marginally higher consumption of these items across the school day and over 24 hours in school types 1B and 3B (i.e. types with higher implementation of SFP actions than the reference type 1A). A similar, but non-significant, association was seen for school lunch consumption.

Comparison of pupil dental outcomes across school types

We constructed adjusted models to explore differences in dental outcomes, using the SFS-based four-group school typology and the SFS- and SFP-based six-group school typology. There were no significant differences in the presence or number of caries symptoms in the previous 3 months, or in caries treatment received in the past across either the four- or six-group typologies. The full models are provided in Report Supplementary Material 12, Tables 1 and 2.

Micro-costing

Twenty-two schools completed the costing survey and were included in the micro-costing analysis. The condensed summary of micro-costing is presented in Table 23. The micro-costing is available on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39). The results are presented separately for the schools that completed the paper-based version (n = 8) and the online version (n = 14). Of these 22 schools, 6 were mandated and 16 were not mandated to comply with the SFS. Thirteen schools had external caterers and nine schools had in-house catering.

The annual ongoing costs associated with implementing and delivering the SFS and the SFP, and of food provision for schools ranged from £8500 to £861,950, with a mean of £182,732. The equivalent per-pupil cost ranged from £10 to £869, with a mean of £173 (based on the total number of pupils), and from £16 to £2311, with a mean £388 (based on the school meal uptake rate).

Annual ongoing societal costs of food provision included costs to schools, catering providers and volunteer time and ranged from £8500 to £974,563 with a mean of £207,094. Annual ongoing societal costs per pupil ranged from £9 to £982, with a mean of £195 (based on the total number of pupils), and from £16 to £2613, with a mean of £426 (based on school meal uptake rate). Eleven schools did not report any societal costs in addition to the school costs, and hence for these schools the total estimate from the societal perspective was the same as from the school perspective.

Staff costs, including catering staff, teaching staff involved in meal supervision, and teaching staff involved in delivering cooking lessons, were the three largest cost categories; they were also in the top five of the best completed categories with the smallest number of missing data. The costs associated with the employment of catering staff ranged from £37,166 to £191,100, with a mean of £108,719. The costs associated with teaching staff supervising meals ranged from £13,974 to £191,100 with a mean of £72,488. The costs of providing cooking lessons varied across the schools, and for all schools that completed the paper-based version, these costs were consistently lower than the costs for the schools that completed the online version. For the paper-based version, the costs ranged from £2400 to £9640 (mean £5733), and for the online version, the costs ranged from £23,400 to £102,960 (mean £106,519).

Six schools with external caterers reported the cost of their catering contract. Four schools reported having zero cost contracts, potentially indicating a profit-sharing arrangement with the caterers. One school reported a cost of £9003 and one school reported a cost of £200,000.

All schools provided data on the availability of vending machines. Five out of 22 schools reported having vending machines and the costs of their maintenance ranged from £1794 to £7176 per year depending on the number of vending machines available in each school. All schools that completed the paper-based survey reported the number of water fountains. The annual maintenance costs for these ranged from £34 to £102 (mean £55). Fifty per cent of the schools that completed the online survey reported the cost of providing water to pupils; these costs ranged from £200 to £4000 per year (mean £1700).

The schools that completed the paper-based survey were asked to indicate the costs associated with setting up and maintaining the school food payment system. One school reported a one-off cost of £15,000 associated with setting up the system and three schools reported annual ongoing costs ranging from £2000 to £2648.

Five schools reported spending between £0 and £3000 per year on maintaining the dining environment. One school reported spending £60,000 per year on ‘other’ maintenance. Six schools reported spending £682 on average per year, ranging between £250 and £2199, on food rewards for the pupils. Seven schools reported spending between £200 and £5000 per year on school food (e.g. breakfast) clubs.

Four schools reported having a food growing area and spending between £600 and £8200 on its maintenance per year. This is the only food-related activity for which the involvement of volunteers (e.g. parents) was mentioned by any of the schools. Three schools reported organising gardening clubs with annual costs ranging from £600 to £1400. Three schools reported having school cooking clubs with annual cost ranging from £1200 to £2002. Six schools reported costs associated with staff well-being training that ranged from £100 to £1500 per year. Only one school reported the use of independent food monitoring by third parties, at an annual cost of £2000. Two schools reported spending £500 and £1000 per year, respectively, on promoting healthy eating by, for example, organising healthy-eating events. Two schools reported the costs of providing subsidies to the staff associated with food provision: one reported an annual cost of £81,900 and the other reported an annual cost of £10,000.

Six schools provided information on the costs of purchasing ingredients for cooking lessons incurred by pupils and/or families. These costs ranged from £1913 to £35,827 per year for all pupils or from £15 to £107 per year per pupil. This question was only asked in the paper-based version of the survey.

In general, the surveys were not well completed. The percentage of missing data varied from 0% to 89%, with an average of 21% missing in the schools that completed the paper-based version, and 39% in the online version. The cost categories with the highest percentage of missing data were other ongoing maintenance costs (86% of missing data), annual costs of subsidies for the staff (71% of missing data) and staff well-being training (59% of missing data).

Cost–consequences analysis

Thirty-six schools (SFS-mandated, n = 13; SFS-non-mandated, n = 23) and 2543 pupils were included in the cost–consequences analysis (2273 providing nutritional outcome data, 2268 providing dental outcome data and 1495 providing HRQoL data). The summary of the analysis is presented in Table 24 and the full analysis is available on the study web page (www.fundingawards.nihr.ac.uk/award/17/92/39). Compared with the schools that were not mandated, SFS-mandated schools had a higher IDACI, a higher proportion of pupils eligible for FSM (23% vs. 18%), a lower proportion of pupils from black and ethnic minority groups (24% vs. 32%) and a lower proportion of pupils whose first language is other than English (11% vs. 16%).

Sample description

In total, 16 out of 36 schools were invited to take part, and 4 progressed to data collection (25% response rate). Table 25 provides details of the characteristics of case sites using 2021–2 school census data.

In total, 21 staff members and governors participated across the 4 case sites and represented members of school leadership (38%), catering (19%), teaching (29%) and the governing bodies (14%) (Table 26).

In total, 137 (male, n = 67; female, n = 68) pupils participated across the 4 school sites, with roughly equal proportions of year 7 (39%), year 9 (28%) and year 10 (33%) pupils. There was wide variation across case sites in terms of other participant characteristics, with FSM eligibility ranging from 9% to 53% and the proportion of ethnic minorities participants ranging from 25% to 97% (Table 27).

Between three and seven interviews were conducted with school staff at each of the four case sites. The average duration of interviews was 35 minutes, with a range of 16–80 minutes. Three pupil focus groups were conducted at each school, with an average length of 34 minutes, ranging from 15 to 46 minutes.

Description of case sites

Table 28 provides further details about the characteristics of case site obtained from researcher observations, the key information survey and pupil data from the main study (phase I).

All schools offered breakfast, with half offering this for free to all pupils (both of these schools had a high proportion of pupils eligible for FSM). School 1 did not offer breaktime food provision. This site had the highest proportion of pupils having school food, as the school had a ‘family dining service’ and did not allow packed lunches except in exceptional cases. In the remaining schools, school meal take-up ranged from 44% to 66%. Schools ranged in their compliance with the SFS from 53% to 70% and in their implementation of the SFP from 40% to 53%. Schools from three out of the four SFS types (1–3) and four of six SFP subtypes were represented in the sample.

Summary of themes interpreted from the data

Table 29 provides an overview of six themes interpreted from the data. These were mainly derived from site-specific subthemes; however, some did not appear in all schools (e.g. perception of the importance of food in school; wider context of school food). We also identified three interwoven themes that appeared to underlie the narratives described in some of the themes, which are summarised in Interwoven themes. Descriptions on individual site-specific subthemes are provided in Appendix 10, Table 45. Each theme is described below with supporting quotations provided. Similarities and differences across case sites are discussed within each theme.

Healthy eating at school

This theme related to perceptions of the importance of healthy eating at school, if/how the school supports healthy eating, and the barriers to healthy eating.

Pupils and staff felt it was important that schools supported healthy eating. They understood that healthy eating prevented ill health later in life. Some staff and pupils also linked healthy eating to learning:

I think it’s important because if you haven’t eaten your brain won’t work and you go to sleep in lessons and you’re not prepared to learn.

Year 7

Pupils knew the importance of eating healthily but perceived that the healthy options offered at school were not filling and/or tasty and/or value for money. Healthy eating was not generally at the forefront when pupils were making decisions about what to eat:

I don’t think the pot with the fruit is really that healthy.

You get a tiny amount of that for a very high price.

Year 10

We don’t go there and think I am eating something really healthy, it’s just that it tastes nice when we’re eating it.

Year 7

Pupils’ opinions were mixed about whether school food was healthy. In some schools (school 3 and 4), the use of Fridays as a ‘treat’ day reinforced the idea that the majority of school food was healthy and enhanced the desirability of the ‘Friday foods’:

The food is mainly healthy, because on Friday is the end of the week we can have a treat like pizza or burgers […] but mainly throughout the week it’s mainly healthy.

Year 7

At some sites, thought had been invested into how to make healthy food affordable and appealing, for example adding cheaper fruit or waste fruit to fruit salads and cooking vegetables in more interesting ways:

If they like corn on the cob, roasted veg, squash, sweet potato, which they would never normally have, but if we roast it with garlic then they will see it and they will choose it. But if it’s just a general carrots on the menu [they won’t eat it].

Catering

Pupils and staff spoke about the lack of consistent messaging around healthy eating, in that healthy-eating teaching is not reinforced in the school food provision and not encouraged by staff, and this also applied to foods and drinks brought in from outside:

There’s no point telling us to eat healthy, then in the canteen they are not serving healthy meals.

Year 10

Even if you’re buying loads of bad stuff they won’t say anything to you. To be honest they probably don’t really care.

Year 9

However, there were occasions where staff tried to redirect pupils to healthier choices. For example, staff at one school did monitor food choices at the point of payment and would intervene if they felt unhealthy choices had been made (although this was largely targeted at pupils having FSM):

The canteen staff are really good, because when it comes to the free school meal, youngsters they will say to them if they come with all biscuits they are like ‘no’, or they will call me on the radio and I am like, ‘No you can’t have two waffles’. We will have conversations with them and make them get something healthy.

SLT

Pupils also reported policies aimed at promoting healthy eating. In particular, in school 1, pupils’ awareness of the schools’ policy on drinks and snacks brought into school suggested that it was indeed strictly monitored:

You’re only allowed water, only water.

So you’re not allowed to bring in other drinks?

No fizzy drinks, no juice, no squash.

No fizzy drinks or energy drinks, no canned drinks, nothing, just water, or flavoured.

Year 10

In some schools, the pricing of ‘healthy’ foods also gave the appearance to pupils that healthy eating messages were not supported in practice:

You get about six grapes and it costs you about 80p … When I go outside and buy the grapes from there it’s £1 but they are really big and stuff, so they have more quantity, but in school it’s really small and expensive.

Year 9

Other factors that were seen to contrast with the idea that schools were supportive of healthy eating in some case study sites included the lack of choice at the end of service, the negativity of the lunchtime experience, the perception of poor quality of food, a lack of operational water fountains and poor provision for special diets, for example vegetarian, vegan and those with food allergies.

There was a staff perception that pupils had strong preferences for unhealthy food:

I know that if given the choice children tend to choose the unhealthy, the quick option.

SLT

To some extent, this contrasted with the pupil perspectives described in the ‘Difficulties in school catering’ theme: healthy eating was just one of a number of factors that pupils considered when making their choices. However, some pupils also acknowledged a preference for unhealthy food such as waffles and chips.

This theme highlights a mismatch between staff and pupil experiences and perceptions related to healthy eating. From the staff perspective, the pupils were making poor/unhealthy choices, which might signal the need for increased regulation. However, from a pupil perspective, the healthy food offer did not align with what they wanted from school food, and in some cases they sensed contradictions in the messaging around healthy eating.

Interwoven themes

These interwoven themes underpin many of the narratives brought out in the descriptions above.

Implementation of the school food standards and School Food Plan

Data were also examined to understand implementation of the SFS and SFP, which naturally focuses on the staff perspective.

Implementation of the school food standards

A key finding of this study is that no schools fully complied with the SFS. On average, schools met 64% of the standards. Standards applying to school lunchtime had the highest compliance (average 81% met), followed by standards for food other than lunch (average 64% met), and finally standards that apply across the whole day (average 43% met). Most secondary schools have extensive provision of food that includes breakfast and breaktime offers (30 and 34 schools in this study provided breakfast and breaktime food, respectively), and so the low compliance of standards across the whole day and for food other than lunch is a concern when considering the overall food offer of secondary schools.

There was large variation in compliance across the 36 schools. However, there was only a marginal difference in average levels of compliance when comparing SFS-mandated and SFS-non-mandated schools. Some larger variations were seen across the two groups for individual standards. Given that during the period of this study there has been increasing government encouragement and expectation of schools that are exempt from SFS-legislation to comply with it,49 this is an unsurprising finding. In our SFS-non-mandated school sample, 48% reported that they had voluntarily signed up to the SFS.

On further examination of compliance with the individual standards across all schools, the lunchtime standards relating to ensuring a good balance of food and nutrition daily and throughout the week were generally well complied with, but standards relating to the restriction of unhealthy food items (mostly standards relating to foods other than lunch and the whole day) were much less well met. For example, standards relating to the restriction of cakes, biscuits, confectionery, pastry, fried foods, processed meat products and SSBs were not complied with in the majority of participating schools. Although we had limited data from schools on food sales, and there were issues with comparisons across schools, the data we obtained provided further evidence of low compliance with these standards, as a substantial proportion of food sales comprised these restricted or prohibited items. We found that, overall, sales of desserts, cakes, biscuits, confectionery, chocolate, pastries, starchy foods cooked in fat or oil and SFS-non-compliant drinks/snacks accounted for 34% of all sales.

Our case study data help with further interpretation of these findings. There was a sense from both school leadership and catering staff that there needs to be flexibility around meeting the SFS, and an acknowledgement by some that the standards were not always adhered to, especially at breakfast and breaktimes. The driving factors in this appeared to be related to staff perceptions of what the pupils wanted (based on items that appeared to sell well), thus ensuring uptake of school food, which is important for catering service viability and profit. There was a perception among staff and caterers that healthier alternatives to the more traditional breakfast and breaktime foods (bacon, sausages, pastry products, etc.) were not purchased by pupils, but they also acknowledged that some pupils buy their lunch at breaktime, and so low compliance with the SFS at breaktime is an issue for this reason.

Other factors influencing compliance with certain standards that we identified in our case study data related to ease of implementation. Some were perceived as easier to implement (e.g. compliant drinks) and others much more difficult, for example providing oily fish, which resulted in concerns about low uptake and wastage, which in turn has an impact on profit. Overall, levels of compliance with the SFS, together with the understanding we have gained from the case study phase, provide some evidence that even though the updated standards introduced in 2015 were simplified to make it easier for schools to comply, many implementation challenges remain in the secondary school context.

In addition to assessing compliance with the SFS, we explored how well the SFS had been embedded as a policy in participating schools through surveys with school leaders, teachers with responsibility for/interest in food and nutrition, caterers and governors, informed by NPT. 106 The majority of participants were aware of the SFS and perceived that the policy had been implemented in their schools, with minimal difference between schools mandated and schools not mandated to comply with the SFS. Of note is that no school governors who reported implementation of the SFS in their schools perceived that they had a role in the SFS policy implementation, despite the government expectation that it the school governing body is responsible for ensuring that the SFS are met. 49 Overall, the majority of participants who were aware of the SFS were engaged with the policy and perceived that there was collective action in their schools to implement it. However, only one-third of respondents felt that there was a shared understanding of the purpose of the SFS policy across their schools. There were some differences across respondents from SFS-mandated and SFS-non-mandated schools, with a higher proportion from mandated schools having an understanding of the benefits of the SFS and what is required to deliver them (coherence), a perception that everyone worked well together to implement the SFS (collective action) and a belief that the SFS were worthwhile (reflexive monitoring). It is possible that these differences were due to the differences in statutory obligation to comply with the SFS across the two groups.

In line with our staff survey findings, the case study phase suggested generally high levels of support for the SFS from staff and governor interviewees, but the governors did not recognise oversight and monitoring of the SFS as part of their role. There was a sense among school senior leaders and governors in schools with external catering companies that the responsibility of SFS implementation rested with the catering company, implying that the SFS as a policy was not centrally embedded in the schools. There was also a perception that use of specific ‘approved’ suppliers (e.g. those designated by the catering company) provided assurance that pre-packaged items such as drinks and snacks were SFS-compliant; however, we found that in many schools some of these drinks and snacks were not compliant.

In the case study data, there was a disconnect between the school leadership perceptions and the catering provider perceptions in terms of the extent to which the schools were compliant with the SFS, with examples of those in leadership positions assuming full school compliance as part of the contractual obligations of their external catering company. By contrast, school caterers (both external and in-house) had developed a more pragmatic and flexible approach to the implementation of the SFS, balancing conflicting priorities. These included cost/profit, customer (pupil) preferences (driven by a number of factors, including convenience, taste and cost) and competition from outside (i.e. pupils can choose to buy preferred foods/drinks from food outlets before or after school or bring them from home if schools are not offering these products).

Implementation of the School Food Plan

All schools in our study were implementing some of the actions recommended in the SFP, even though evidence from our survey of school senior leadership, teachers, caterers and governors showed that fewer than half of respondents were aware of the SFP. Again, there was large variation in implementation across the schools, with the percentage of actions implemented ranging from 25% to 63%. There were no substantial differences in the level of implementation across SFS-mandated and SFS-non-mandated schools, which was expected, as the SFP is non-statutory national guidance that applies to all schools.

We explored implementation of the actions across our nine identified themes within the SFP but were limited in doing so to a certain degree by a large number of missing data from schools on some of the actions. Nevertheless, we were able to identify actions within themes that were either well or poorly implemented. Of note, no actions within the themes of leadership and oversight on school food, engagement (with pupils, parents and the community) and catering practices (catering staff encouraging healthy behaviours and nutritional balance) were implemented to a high level in the majority of schools. Within all other themes, some actions were well implemented and some not well implemented by the majority of schools.

Actions relating to the lunchtime experience presented a very mixed picture in terms of level of implementation, with some contrasting findings. Although over three-quarters of schools had implemented strategies to reduce queuing times and over 60% were assessed (using multiple participant perspectives) as providing pupils with enough time to eat lunch, in the majority of schools pupils reported feeling rushed when purchasing school food. Also of note was that many schools (58%) did not ask for feedback from their pupils on the lunchtime experience. This lack of opportunity for feedback is supported by our finding on the related SFP action of providing opportunities for pupils to contribute views on school food (within the theme of leadership and oversight on school food), which was also not well implemented, with only just over one-quarter of schools rated as having a high level of implementation, and nearly half of schools rated as having a low level of implementation (see Appendix 6). There was also a mixed picture for food education. Curricular education relating to food and healthy eating was assessed to be well implemented in most schools, but education around growing and extracurricular food education was not implemented in most schools.

Particular SFP actions to note that were poorly implemented in the majority of schools included governor leadership and engagement with school food (low implementation in 64% of schools), independent checking of compliance of the school with the SFS by either governors or an external service (low implementation in 79% of schools) and governor review of school food uptake data each school term (low implementation in 74% of schools). These findings are coherent with the governor survey findings, which provided evidence that governors do not identify they have a role in implementation and oversight of the SFS. This perception of governors also surfaced in the case study, but they did perceive that they had a role relating to school food. Governors in schools in socioeconomically deprived areas were concerned about undernutrition and were focused on making sure those in need had adequate energy intake and nutrition rather than being concerned about compliance with the SFS. This was also the case for senior leadership staff in these schools; they were concerned about a lack of food, as well as unhealthy and imbalanced diets in their pupils. Also of note here is that no schools in this study had participated in the government’s healthy schools rating scheme, which includes self-assessment of SFS compliance. 73 This gives a further indication of the minimal oversight on healthy school food and SFS compliance among school SLTs and governors in secondary schools.

Our case study data helped us to further interpret our findings in relation to the implementation of SFP actions. Both staff and pupils in some of the case sites perceived that consistent messages around healthy eating were not present, and some pupils perceived that there was a lack of action from staff when they saw pupils consuming unhealthy foods and drinks. This resonates with our findings that some actions related to a whole-school approach, including the promotion of healthy-eating messages across the school, were not well implemented in most schools. There was also a strong negative perception of the lunchtime experience by pupils from some of the case sites, including a lack of time to purchase and eat their food, queuing, fear of food running out and crowded dining environments. These negative perceptions were not particularly echoed by staff in these case sites, which, in the light of our findings that some strategies to improve the lunchtime experience are well implemented in most schools, may suggest that even though these strategies might be in place, they are not achieving the desired effect. In one case site, the lunchtime experience was perceived positively by both staff and pupils. This school had a policy of all pupils having a school-provided lunch, and family-style dining was in place, with staff and pupils sitting together at tables and eating the same food. This school was the only school in our sample to adopt this method at mealtimes, and the key motivation for doing so related to supporting the social and holistic development of the pupils, rather than aiming to improve nutritional intake per se. In general, the family-style lunch was positively viewed by staff, governors and pupils, and pupils valued the opportunity to sit and eat with staff. This finding was corroborated in our assessment of SFP implementation, as this was the only school in which pupils did not feel rushed to buy their food (information is available on the study web page at www.fundingawards.nihr.ac.uk/award/17/92/39).

Our case study data highlighted differences among school staff and pupils’ perceptions of the opportunities for pupils to feedback about school food and the eating experience. From the staff perspective, there were opportunities for pupils to give their feedback, but often pupils either were not aware of systems of feedback or felt that they were inadequate. Some pupils felt that no action was taken in response to the feedback that they gave.

Our case study also provided some insights around food education in schools, which is a key component of the SFP. Overall, food education was viewed as important by school staff and was perceived as sitting within broader holistic ‘life learning’. However, staff acknowledged that it was difficult to fit it in with other educational priorities, and some staff felt that food and healthy eating were not high on the agenda in their schools. In line with this view, some pupils perceived that food education outside of food preparation and nutrition lessons was scant and ad hoc. There was also a perception among some pupils that practical education on cooking often involved preparing desserts and so was less useful in developing the skills needed for future independent living.

In terms of the survey data on how well the SFP was embedded within schools as a policy, from school senior leadership staff, teachers with responsibility for/interest in food, caterers and governors, among those aware of the SFP (i.e. under half of respondents), around 60% perceived that the SFP was being implemented in their schools, and catering staff reported being most involved in implementation. Fewer than half of the respondents who were aware of the SFP felt that there was a shared understanding of it in their schools (coherence) and that staff worked well together to implement the SFP (collective action), although there were higher levels of engagement (cognitive participation). The SFP was perceived by most as worthwhile. As with the SFS, governors did not feel that they had a role in the implementation of the SFP in their schools.

Within our case study, there was variation in the awareness of the SFP as a national policy, but all four schools had initiatives and actions to support healthy eating, which were in line with some of the SFP actions. There was some tension between school teaching/leadership staff and caterers as to where the responsibility for SFP implementation lay, but this may in part be due to the limited awareness and understanding of the SFP among some participants. In one case site, there had been explicit implementation of the SFP as a policy, and, as a result, the whole school and community were involved in school food and healthy eating. A key driver of implementing the SFP in this school had been to increase school meal uptake, and this had been achieved following implementation of a variety of SFP actions. This SFP implementation had been driven by a particular staff member (a food preparation and nutrition teacher) who had received support through an external programme. However, this person no longer held a teaching post at the school, and the current food preparation and nutrition teacher had little awareness of the SFP.

Another perception among school staff articulated in the case sites was the sense that the school was limited in its influence on pupil food choices, with staff acknowledging the wider socioeconomic, demographic and environmental influences driving these dietary behaviours both in and out of school. They felt that if pupils chose to bring unhealthy foods into school or chose the unhealthier options in school, there was little they could do about it. The case site that used the family dining model was an exception. In this school there was a policy prohibiting certain drinks and snacks, which staff felt was important in reinforcing education around healthy eating.

One of the central aims of the introduction of the SFP was to increase the uptake of school food. However, in our participating schools, on average, only just over half of school pupils were having school lunches. There was no difference across SFS-mandated and SFS-non-mandated school groups, but large variation was seen across schools, with uptake ranging from one-third to three-quarters of pupils (excluding the case site that had a family-style lunch approach and a policy of compulsory school lunches). In the case study, pupils articulated views that may contribute to their decisions to have school-provided food. There were mixed views on the quality of school food, and other issues that arose were the lack of diverse cuisines, lack of catering for special dietary needs, and mistrust over school food pricing and value for money of foods. Crucially, pupils felt they could get more for their money from food outlets outside school. The lunchtime experience was also a factor for pupils when making decisions on purchasing school food, with school food being perceived as a risk because of food running out and limited choice if they were at the back of the queue.

School types

As a way of trying to capture and articulate the variation across schools in relation to school food provision and the food and healthy-eating culture and environment, we attempted to develop a typology of schools in relation to the SFS and the SFP. Using an approach we developed to type schools based on their SFS compliance, we categorised the majority of schools into two types: both had high compliance with SFS relating to increasing dietary variety, but one had low compliance with standards relating to obesity and dental health (type 1) and the other had medium compliance with this latter set of standards (type 3).

There were two schools in our sample that were assigned as different types. The school identified as type 2 had a more restrictive food offer, with lower compliance with standards related to dietary variety than the other three school types and medium compliance with the standards related to obesity and dental health. This school was also the case site (school 1) with a family-style dining approach and limited meal choices both at breakfast and lunch, which may explain a lower compliance with standards relating to dietary variety. However, with the more controlled menu and limited choice available in this school, one might think that there would be opportunity to be highly compliant with the standards related to obesity and dental health, yet this was not the case. In fact, in the qualitative data from this school, catering staff reported that the need to cater according to the family dining approach was one of the factors making it more challenging to comply with the SFS (related to the need to ensure that all pupils were happy with the limited choices on offer).

The school identified as type 4 was different from other schools because it was the only school that had high compliance with the standards relating to obesity and dental health and high compliance with the dietary variety-related standards. Thus, the four identified types provided us with a range of levels of compliance with the SFS targeting different aspects of achieving healthy nutritional intake, which enabled us to explore nutritional intake in pupils across the different school types. Our subtyping based on implementation of the SFP actions was less clear-cut, and we were hampered by a substantial number of missing data from schools. There were no patterns or clusters of actions that were implemented in some schools and not in others. Therefore, we based our subtypes on the proportion of SFP actions implemented.

Pupil nutritional intake and dental health, and school food policy

Costs and economic impacts of school food policy

The economic evaluation was designed as a cost–consequences analysis due to the range of outcomes potentially impacted by the SFS and SFP, and the wide range of costs to be included. A comprehensive micro-costing approach was essential for this study to fully estimate the societal-sector cost of the SFS and the SFP. However, encouraging the schools to complete the costings survey was challenging, mainly due to the difficulty of identifying the relevant staff member who had the knowledge to complete the survey. Halfway through the study, this led to the survey being adapted and shortened to an online version to try to boost completion rates, but completion remained a problem. The data that were collected produced a wide range of costs reported, and, in some cost categories, these costs ranged substantially between schools.

The cost–consequences analysis suggests that the costs in the SFS-mandated schools are lower than those in the SFS-non-mandated schools, and this direction of difference remained the same across all the different methods for measuring costs. Given that the costing analysis was fraught with methodological difficulties, we cannot be certain of the exact magnitude of this difference; however, it did appear to show SFS-mandated schools having consistently lower costs for items relevant to the SFS and the SFP. However, the data on the implementation of the SFS and the SFP do not provide an explanation for this. The initial plan for the economic evaluation included an exploratory cost–utility analysis. 77 Due to the lack of baseline data, a large proportion of missing cost data, and the uncertainty in the cost and outcomes data, this analysis was no longer feasible.

Overall, this study has highlighted the methodological challenges of collecting cost data in school settings, and, despite the substantial study resources devoted to the data collection, this remained a challenge. We supplemented our analysis with the publicly available catering expenditure data reported by the schools. It should be noted that these data do not include all costs of food provision incurred by the schools. Therefore, some important differences across the schools and the two main school groups compared in this study might not have been captured when comparing the schools based on their catering expenditure. Furthermore, for the majority of the schools, the actual expenditure for the year 2021–2 would be expected to be higher than in 2018–9, even after adjusting for the Consumer Price Index, due to the increase in the proportion of pupils receiving FSM. 118

It is important to note that the socioeconomic differences across the two school groups might have influenced both the costs and the pupil outcomes. For example, the SFS-mandated schools had on average a higher IDACI score than the SFS-non-mandated schools; this indicates higher deprivation, and it might be associated with poorer educational performance, lower quality of life and poorer dietary intake independently from the food provision in these schools. Furthermore, the SFS-mandated schools had a higher proportion of pupils eligible to receive FSM than did the SFS-non-mandated schools. This might be reflected in the funding that is available to support the eligible pupils.

Implementation of the school food standards and School Food Plan

Our findings of incomplete compliance with the national SFS in our study sample is consistent with the 2020 Guy’s and St Thomas’ Charity report Serving Up Children’s Health. 66 This London-based study only included a small number of secondary schools, but also highlighted low compliance with the SFS, especially with foods served for breakfast and at breaktimes. Our study builds on this previous work by identifying the standards with particularly low compliance (i.e. the standards relating to the restriction of unhealthy foods, which often apply to food served across the whole day and outside of lunchtime) and provides additional insight into the reasons why these standards have low compliance (i.e. the need to balance pupil preferences, school food uptake and financial viability with SFS compliance). The low compliance with the SFS concerned with restricting unhealthy foods and drinks is also corroborated by evidence on nutritional intake from school food in UK secondary school children. High consumption of sweet and savoury snacks119 and high levels of ultra-processed foods and drinks120 have been reported in the secondary school age group consuming school food at lunch.

The lack of monitoring and oversight of SFS compliance by school governing bodies that we have found has been previously emphasised. 52,65,66 This study sheds further light on this issue in terms of the misconception around roles and responsibilities in implementing the SFS, with governing bodies and school senior leadership representatives often seeing this as the responsibility of the catering provider, and a lack of recognition from governors of their role in overseeing SFS compliance. However, particularly in schools in areas with high deprivation, we found that school governors, senior leaders and teachers focus on food in the context of food insecurity, with their major concern being school pupils going hungry.

In our exploration of implementation of the SFP, we identified similar issues to those presented by Hart and Page, who examined qualitative data on school food education and culture from pupils attending schools in England, along with survey data from school leaders, food teachers and parents. 121 They found that in secondary schools there was lack of a whole-school approach, with mixed messages around healthy eating and foods high in fat, salt and sugar given as rewards. They also reported views of overcrowding in dining areas, and pupils perceiving that healthier foods were expensive and poor value for money. In line with these findings, we found that the SFP actions around consistent healthy-eating messaging across the school, and sweets not used as rewards, were generally not well implemented in schools. Our case study data highlighted the issue of crowding and poor dining environments and pupils’ perceptions of healthy food being poor value for money.

Another key finding of the Hart and Page study was the variable provision of food education and practical food preparation skills and the low status of food education in secondary schools. 121 Our SFP assessment suggests that curricular food and healthy eating is generally delivered well in schools, but our case study data is more in line with this previous study, with healthy food/eating education being perceived as low priority, patchy and ad hoc from a pupil perspective, and a lack of focus on healthy food when teaching practical cooking skills.

A consistent theme throughout our findings in relation to both the SFS and the SFP was that these were not priorities for most schools’ SLTs and governing bodies, although they did identify that SFS compliance and support for healthy eating were important and part of the school’s role. This resonates with findings of a mixed-methods study with English secondary school headteachers and chairpersons of governing bodies. 122 The authors explored participants’ perspectives on adolescent obesity prevention in school settings and reported that, while headteachers and governors recognised that schools have a role in supporting healthy eating (and other obesity-related behaviours) and saw it as part of holistic adolescent development, they identified a range of factors that prevented them from prioritising actions to support this, including financial pressures and the requirement from national government to focus on academic achievement. Financial issues relating to school food provision and other educational priorities competing with the need to provide food and healthy-eating education also featured in our case study data.

School food and pupil nutritional intake

No existing studies have explored the influence of the current SFS (introduced in 2015) on secondary school pupil nutritional intake, but the limited evidence from two studies examining pupil nutritional intake before and after the introduction of the 2006 SFS in secondary-school-aged pupils suggested only a marginal positive impact on dietary intake at lunch64 and very limited impact on total dietary intake. 62 In our exploratory findings on the level of SFS implementation and pupil nutritional intake, although we did not detect any beneficial associations, our analyses suggest that there may be an undesired association between SFS implementation and pupil nutritional intake. We found that higher compliance with standards related to dietary variety and higher compliance with standards related to obesity and dental health were associated with higher intake of sugar/sugary foods, with these associations mostly seen over the school day or over 24 hours.

Consideration of factors influencing adolescents’ food choices may help to explain this suggestion in our data of an unfavourable association between SFS compliance and sugar intake. Although adolescents have an awareness of the need for and benefits of a healthy diet, the main drivers for their food choices are desirability, value for money and convenience, with peer and social influences also playing a part. 25,123 Our case study data surfaced the issue around school food’s perceived poor value for money, and also the low desirability of the food on offer in some instances. As adolescents have increasing autonomy over their food choices, these drivers may lead them to purchase foods from out of school or obtain foods from home, especially if the foods that are desirable to them are restricted in the school food offer. The perception in our case study of schools and school food having limited influence on pupils’ food choices, outside the canteen offer, is coherent with this potential explanation.

Another factor that may contribute to poorer-quality dietary intake during school time, even when the school food offer is relatively healthy overall (i.e. with relatively high compliance with the SFS), is the way in which secondary school food sales are organised. Pupils are often able to purchase food at multiple times throughout the day from multiple service points, with no requirement to purchase a set meal. As highlighted in Hart and Page’s study,121 this freedom to purchase single items can lead to a pupil making multiple purchases of unhealthier foods (e.g. high in sugar and/or fat) throughout the day, even if there is a range of healthy foods on offer. Again, this resonates with our case study data, in which school staff perceived pupils as opting for the unhealthy options available in schools.

This perceived propensity of pupils to choose less healthy foods and purchase the more convenient ‘grab and go’ options (avoiding the main meal options that are more nutritionally balanced) was also found in a qualitative study with catering staff in three Welsh middle schools (pupils aged 11–14 years). 124 The caterers perceived that a variety of influences led to these pupil choices, including taste preferences (i.e. desirability), pricing and convenience. Consistent with our findings about pupil lunchtime experiences, the caterers commented on the pupils having to queue and being rushed at lunchtime, which prompted them to purchase less nutritionally balanced food options (e.g. sandwiches, pasta pots, pizza slices and cakes) that are portable and more easily consumed without sitting down. Also in line with our qualitative findings, caterers in this study acknowledged that they sold unhealthy foods that were not compliant with the SFS that are in place, but that omission of these items would significantly affect food sales and thus threaten financial viability.

Strengths

A key strength of this study is our use of a multiple-methods approach, incorporating both quantitative and qualitative data collection. This has enabled us to gain a detailed understanding of how the SFS and SFP are implemented in secondary schools. Our school sampling approach enabled us to include schools with a range of different characteristics.

We collected extensive data from schools, achieving a comprehensive understanding of SFS compliance through menu analysis and researcher observation, and SFP implementation through multiple data sources. Our tools for collecting data relating to school food provision and SFP actions were developed using existing online resources and legislative documents, which also supported interpretation of the SFS and SFP actions. Use of multiple data sources for SFP assessment also allowed for different perspectives to be captured and enhanced the credibility of the assessments. Our systematic approach to SFS and SFP assessment, with rules agreed on how multiple data sets were combined, enhanced the transparency and reliability of the assessments.

The study included a large sample of school pupils, which is broadly representative of the national population in terms of ethnicity (with 70% of participants in our sample from white ethnic groups compared to 71% across state secondary schools in England in 2022) and deprivation (26% of our sample were in the most deprived quintile). We had a high response rate (99%) and collected a range of outcome data using validated tools. This included our online dietary assessment tool (Intake24), which we adapted for use in an ethnically diverse population. In our pupil outcome analyses, we used multilevel modelling to account for potential clustering and explored the effect of missing data and misreporting of dietary intake on the findings through our sensitivity analyses.

We conducted a qualitative case study with four school sites, sampled to provide some representation of our main study sample, to better understand the wide variation in SFS and SFP implementation across secondary schools. We used topic guides with broad opening questions to allow participants to guide the discussion and our inductive analysis approach enabled our interpretation to reflect the most relevant issues for participants. A team of five researchers were involved in analysing the qualitative data, allowing for a range of interpretations to be considered and providing an opportunity to challenge assumptions and preconceptions. We used a structured method of data analysis,117 and have presented verbatim quotations from participants in our results, providing transparency in the development of our interpretative themes.

Limitations

There are several limitations within this study. To maximise study power, we aimed to recruit an even number of schools in the SFS-mandated and SFS-non-mandated groups but were unable to do so. However, the proportion of participating schools in each group reflected that of the wider sampling frame (64% and 60% SFS-non-mandated schools in our sample and the sampling frame, respectively). There were some differences in the characteristics of schools and pupil participants between the SFS-mandated and SFS-non-mandated school groups, but we adjusted for school and pupil characteristics in our pupil outcome analyses to account for these.

There were also some differences in our pupil characteristics from those of the secondary school pupil population across England. For example, the proportion of pupils reporting that they had a FSM in our sample was 15.4%. This compares with 20.9% FSM eligibility in secondary-school-aged pupils nationally. 118 This difference may be explained partly by substantial increases in the FSM eligibility in the population over the duration of the study (eligibility was 15.9% in 2020) and partly by a degree of under-reporting in our sample, as pupils were asked to self-report this information. Not all pupils would be aware of their eligibility for FSM, and some may have chosen not to report that they were eligible.

Regarding data collection, there were substantial disruptions in schools recruited in the 2019–20 academic year due to school closures in response to the COVID-19 pandemic. This resulted in lower complete data collection in these schools, for example non-completion of the second pupil data collection session or non-response to school surveys and document requests. This affected a greater number of SFS-mandated schools, since a higher proportion of SFS-mandated schools were undergoing data collection at that time (38% of SFS-mandated schools vs. 30% of SFS-non-mandated). School food provision and the dining arrangements in schools were also temporarily changed to accommodate COVID-19 measures in schools in the summer term of 2021. We collected data in four schools during this school term and so may not have obtained a representation of their typical food provision and environment. This impacted on SFS-non-mandated schools to a greater extent than SFS-mandated schools (13% of SFS-non-mandated schools vs. 8% of SFS-mandated schools). We adjusted for academic year of data collection in our pupil outcome analyses to account for changes that may have occurred in the 2020–1 year in response to the pandemic.

Some schools had low levels of completion of the second pupil data collection session or school surveys for other reasons, for example if they could not find time to run a second pupil session, the school liaison person was absent, or there was a lack of response to requests for school surveys and documents (these issues arose in both school groups). In addition, the costings survey was complex and difficult to complete, despite our attempts to simplify it, and this may have influenced the response from schools. The sales data proved difficult to obtain from schools, resulting in large numbers of missing data. We also had a low overall response rate to surveys from staff, and particularly from governors. This limited our ability to assess the implementation of SFP actions in some schools. It also reduced our ability to make meaningful comparisons between SFS-mandated and SFS-non-mandated school groups in our analysis of SFS and SFP policy implementation, as there were proportionally fewer staff/governor survey responses in SFS-mandated schools than in SFS-non-mandated schools. Overall, the substantial number of missing data prohibited the full assessment of SFP implementation in some schools.

In addition, the large number of missing cost data impacted the economic evaluation. This was somewhat mitigated by supplementing our analysis using publicly available data on schools’ catering expenditure, but it prevented us from conducting an exploratory cost–utility analysis in line with the study protocol. 77 We explored the impact of missing data on pupil characteristics in our pupil outcome analyses through imputation and sensitivity analyses.

Regarding pupil nutritional outcome data, there was some loss of dietary data due to being unable to match pupil demographic data to Intake24 data, but this affected only 4.7% of Intake24 records (165 of 3488). Another limitation of the pupil dietary data is that although we planned to collect two non-consecutive 24-hour dietary recall records per pupil, 54% had one only record. This only provides one ‘snapshot’ of dietary intake, which may not be representative of typical intake. Misreporting is a common limitation of self-reported dietary intake measures such as 24-hour recall, especially in adolescent populations,125 so there is likely to be some measurement error in the nutritional outcomes. This impacts particularly on reported energy intakes, which are likely to be an underestimation of true intakes. However, we would expect this misreporting to be similar in the two SFS comparison groups. We explored the impact of the inclusion of dietary records reporting extreme values of 24-hour energy intake in our pupil outcome sensitivity analyses. The dental outcome data collected from pupil participants were self-reported and not from objective dental examination, and so may not accurately reflect the dental health of the sample. However, the self-report questions were taken from the National Children’s Dental Health Survey and have been previously tested for use in this age group.

In our pupil outcome analyses, we explored multiple nutritional outcomes, each during three different time periods. Therefore, it is possible that some of the findings that reached statistical significance were as a result of multiple testing, rather than a true difference between groups. We did not apply a correction for multiple testing as this analysis was exploratory, and we have interpreted our findings according to patterns seen (e.g. across the three time periods investigated). Regarding the analysis of dental outcomes, while we adjusted for toothbrushing habits and home water fluoridation status, we did not adjust for eating behaviours and habits, which may impact on dental health.

In terms of the qualitative component, we had planned to sample our case sites using our developed school typology. However, we were unable to do this, as we had to conduct the main and qualitative phases of the study concurrently, rather than sequentially, due to study interruptions relating to the COVID-19 pandemic. This meant that we missed an opportunity to explore the experiences of SFS/SFP implementation in SFS type 4 school, which had the highest level of SFS compliance. Another limitation was that the pupil focus groups were larger than planned, due to a variety of reasons, which may have impacted on the data collected. For example, there may have been fewer opportunities for everyone in the group to voice their opinion and more pupils talking over one another, which may have affected the quality of the audio data.

Implications for policy and practice

In this study we have demonstrated no difference in compliance with the SFS between schools mandated and schools not mandated to follow them, and minimal or no differences among the two school groups in the implementation of actions to support healthy eating (outlined in the SFP), food-related costs, and the nutritional intake and dental health of pupils. This has limited implications for policy for the following reason: since the introduction of the 2015 SFS legislation, exempt schools have been encouraged to voluntarily sign up to the standards, and more recently this has been a firm government expectation. 49 Therefore, in the current context, there is no practical difference between the two school groups that we compared in terms of the expectation for them to comply with the SFS.

Our findings of incomplete implementation of the SFS in secondary schools, regardless of whether they have a statutory requirement to comply, have greater implications for policy and practice. These findings suggest that the standards are challenging to fully implement in a secondary school context. The extensive provision of school food at eating occasions other than school lunch, the sale of single items for purchase (and no obligation for pupils to choose the set menu meals), and the need to provide foods and drinks to meet the pupils’ demands to ensure school food uptake and financial viability, are all factors that contribute to this incomplete compliance. There was also some evidence of difficulty in interpreting and applying the SFS within schools (e.g. in relation to the compliance of pre-packaged snacks). From a policy perspective, this raises a question as to how policy around the provision of healthy school food can be adapted and developed to better suit the secondary school context of food provision, without impacting on the viability of school food catering services. There is also the question of how schools and caterers can better be supported to interpret and apply the standards in the context of a financially viable food provision service. From a school and catering practice perspective, there may be changes to the way in which food is provided and choices are presented to pupils (e.g. nutritionally balanced meals in favour of single items, and portable healthy meal options), which could encourage healthier school food choices. Our finding that schools were not perceived to be offering meaningful opportunities for pupils to feedback on school food is relevant here. Better communication between schools and pupils would enable the multiple drivers of pupils’ decisions relating to school food to be understood in more detail, which would inform the changes needed by schools and caterers to enhance the healthy food offer and increase school meal uptake.

The lack of oversight of healthy nutritionally balanced school food provision and SFS compliance in schools has previously been highlighted and is a prominent finding in this study. The government has started to address the absence of external accountability for school governing bodies in relation to compliance with the SFS through their Levelling Up the United Kingdom White Paper,75 in which they announced a pilot programme involving local authorities assuring school compliance with the SFS. However, given the extensive remit and other priorities of governing bodies and school SLTs, support and training for schools and school governing bodies in terms of negotiating food provision contracts and putting in place internal mechanisms to monitor compliance is likely to be needed alongside a system of external accountability. It is worth noting here that a formal process of external monitoring and inspection of school food is in place in Scotland, which may provide valuable learning for the English context. 52

Overall, the SFP as a policy was not adopted in schools, and although all schools had some curricular and environmental measures in place relating to food and healthy eating, these were, in most cases, not a central priority in the schools. In that sense, the SFP, which was launched nearly a decade ago and set out actions for schools to support them in developing a whole-school approach to create a healthy-eating culture, has not achieved the desired outcomes in the secondary school setting. In line with our findings on food education, in his independent National Food Strategy report,74 Dimbleby reflected that despite the legal requirement to teach cookery and nutrition up to the age of 14 years (introduced following the SFP), food education was still not taken seriously. He set out a recommendation to address this (recommendation 3), which included government actions to reinstate the food A-Level and review other qualifications in food; require schools to work with food and food education accreditation schemes; and include rigorous inspection of food lessons within the national Ofsted inspection framework. Dimbleby also highlighted the need for the government to increase the numbers of specialist food teachers in secondary schools and provide funding for the ingredients used in cookery lessons. He suggested that the implementation of these recommendations be led by a dedicated ‘Eat and Learn’ team situated within the DfE. Dimbleby also reiterated the need for a whole-school approach to food (originally set out in the SFP) and specified that this should be championed at the governmental level. 74 Our findings support the need for this, as we found that some actions related to a whole-school approach, such as the promotion of consistent healthy-eating messaging across the school, were not well implemented in most schools. To date, most of Dimbleby’s recommendations have not been translated into government policy, despite the publication of the Government Food Strategy White Paper in 2022. 76

Finally, acknowledging the limitations of dietary assessment in our study (see Limitations), our findings suggest that even when the SFS are relatively well implemented, this either has no beneficial effect on the nutritional intake of pupils or may even lead to marginally worse nutritional intake. To an extent, addressing the issues of poor SFS implementation and the lack of prioritisation of healthy eating and nutrition should lead to more beneficial effects on pupil dietary intake; however, it is important to recognise that adolescents are interacting in a variety of environments and have increasing autonomy over their food choices. Therefore, schools need to be considered as only one part of the food system that influences adolescents’ dietary intake, and it is important to address other aspects of this system to substantially influence nutritional intake in this age group. 126

To sum up, this study has provided a range of evidence that could inform the future direction of school food policy and practice. Work is now needed to translate this evidence into recommendations for specific actions on school food, and identify which bodies are responsible for their implementation (e.g. national government, local authorities, schools, caterers). We are currently conducting this work, involving multiple stakeholders, in an ongoing NIHR Policy Research Programme-funded study (NIHR204247).

Future research recommendations

Key areas of future research have been identified through this study. At the individual level, there is a need to better understand how school food provision and environments influence the food choices of adolescents. Given our findings, we particularly need to understand how more restrictive food policies and provision in schools may shape adolescent food choices and their decisions on where they obtain the food that they consume in school (i.e. school vs. home vs. external food outlets). Related to this, we need to better understand how healthier food can be provided that matches the preferences of secondary school pupils. For example, healthier school food may be achieved by providing nutritionally balanced ‘sit-down’ meals or offering healthier ‘grab and go’ options. Currently, the SFP has an emphasis on ‘sit-down’ meals and set menus, but a further understanding of whether this is what pupils want would inform future school food provision arrangements.

At the school level, we need to better understand how secondary schools can increase the focus on the school food and healthy-eating agenda so that it is part of the core of daily school business and a whole-school approach to healthy eating is fostered. We particularly need to explore challenges that schools experience in prioritising school food and food education (e.g. managing competing educational priorities) and the support that they require to enable them to overcome these challenges. As part of this, we need to be able to evaluate the economic implications for schools taking actions to improve school food and support healthy eating. In this study, we demonstrated the difficulty of conducting economic evaluations in school settings; therefore, there is an additional need for methodological research to understand how to appropriately capture the economic impact of providing healthy school food and developing a healthy school food culture and environment.

On a societal level, we need to further understand the wider food system influences on adolescent nutritional intake and how these interact with school influences, including how the home and school food environments interact to influence nutritional intake in this age group. Ultimately this will help us to identify a range of interventions, targeting different parts of the food system, that will complement secondary school efforts to positively influence adolescent nutrition. Finally, it is imperative to involve schools, caterers and pupils in the development of further research questions, as this will enhance the relevance of future school food research.