Literature search strategy and selection criteria

The search criteria were developed based on a standardised Population, Exposure, Outcome (PEO) framework to enable transparency and repeatability [31]. Population was searched using overarching terms of child and any extended variations, the institutional setting being school or education and the element being primary and secondary with alternatives of junior, infant, elementary and high. The exposure of interest (independent variable) was kept broad to capture the full range of procurement contract types for the provision of school food. Overarching terminology of purchase, procure, contract, and buy, along with the key components of catering and meal provision being lunch, dinner, and menu was used. Outcomes of interest (dependent variable) were based on factors which could link school food provision with child health and academic results. These were related to food, nutrition, diet, healthy, BMI, quality, uptake, absence, attainment, and results. Quality is often defined in different ways, for example, nutrient composition or meeting nutritional requirements, guidelines or standards. For the purpose of this study a sensitive approach was taken to include papers based on their own definition of quality. The review also sought to establish if further guidance was given in the literature for the quality of food in the context of school food provision contract types.

The following databases, being the preeminent databases for health and education research, were selected [24]: Medline; Web of Science; Scopus; Education Resources Information Center (ERIC). These were searched for peer reviewed articles and the reference lists of these articles were scanned for further applicable evidence. Google was used to search for grey literature. A new google profile was set up and location services were disabled to prevent results being influenced by previous searches or location [32]. Search terms (Supplementary Table 1) were inputted into the advanced search option using Boolean logic. The first 50 Google hits, being the 50 most relevant according to the search engine relevancy ranking, were reviewed for eligibility as is typical in systematic reviews [33]. Web pages for organisations involved in school food or public sector procurement are also a useful source of information relevant to this study; [34] therefore, the following websites were included in the search: Professional associations: LACA (the school food catering association) [35], Association for Public Service Excellence [36]; Government websites and agencies: GOV.UK [37], World Health Organisation [38]; Charities: The Food Foundation [39], School Food Plan [40]. The literature search was completed on 10 September 2023.

Criteria from the PEO model relevant to the review question and objective were used to determine eligibility of the studies [24,41–43]. All study designs, e.g., qualitative reviews, trials, and cohort studies were eligible for inclusion and scientific papers or letters, articles, reviews, book chapters or grey literature reports from reputable agencies and government departments were eligible. Conference abstracts, student dissertations or theses and protocol papers were not included. Documents dated from the devolution of school lunch procurement in 1988 to date of the search were included, as this is the point where the provision model changed in the UK. Eligibility was expanded to international studies to capture relevant findings from other countries however, only sources written in English were included, due to a lack of resources for translation. Any studies not relating to populations of primary or secondary school children were excluded (e.g., preschool children, colleges or universities, or adults). Any studies with exposures not relating to school lunch provision comprising of a full meal for the child or procurement contract type for the provision of that school lunch were excluded (e.g., purchasing by the student, home or family, packed lunches, beverages, and snacks, vending machines and tuck shops, marketing or advertising, food related education, physical activity, after school and breakfast clubs, childcare settings, or nutrition/ education interventions or studies). Any studies with outcomes not related to child health or academic achievement were excluded (e.g., addictive substances, child abuse, food allergies, COVID, food borne or other non-diet related diseases, eating behaviours or disorders, unrelated standard compliance, policies or guidance). Papers focused on global sustainability and food systems were excluded unless they also included research focused on areas that may relate to food quality. Operationally, this included those with data on food waste and organic/ local sourcing that also discussed food quality [44]. An iterative approach was used in case further applicable criteria became apparent during the review [45].

Data extraction and analysis

Covidence systematic review software was used to ensure reliability, whereby the eligibility criteria at both the title and abstract stage were considered by a primary (NN) and experienced second reviewer (LP) to simultaneously review 10% of studies, which were selected using a random number generator. This process was repeated at the full text stage. At both stages, an inter-rater reliability score of 75% [26] was required to determine whether further papers needed to be reviewed or a third reviewer utilised.

Papers deemed potentially eligible after the title and abstract screen were read in full and data relevant to the research question were extracted and analysed using basic descriptive qualitative methods [42]. Whilst scoping reviews do not usually seek to synthesise the results of the existing evidence [28,42,43] a narrative synthesis was used to identify themes bringing homogeneity to the presentation of the results and to highlight the differences and similarities in the findings of the studies reviewed [46]. This approach has been particularly useful for synthesising different types of evidence in other studies [46–48] and was used to develop inductive descriptive themes in a methodologically rigorous way with reference to data returned alongside the study objectives and the research question [30,42]. Narrative synthesis as proposed by Popay et al., [30] was used to bring together findings and highlight key points.

Literature search

The databases, citation review and google search returned 1,288 documents. After removing 433 duplicates, the titles and abstracts of 854 documents were screened for eligibility. This excluded 713 documents and the full texts of the remaining 141 documents were assessed for eligibility, identifying 30 papers and reports which fulfilled the eligibility criteria. Reasons for excluding documents in the full text screen are included in the PRISMA diagram (Fig 1). Over half of the papers (58%) were excluded for not meeting the exposure criteria, mainly due to focusing on procurement (purchases) by the student, or of food other than school lunch, or for examining education or nutrition interventions which were unrelated to school lunch provision or procurement contract type. A further 19% were excluded for non-eligible outcomes, such as eating behaviours or disorders or policies, standards or guidance notes unrelated to the specific outcomes of child health or academic achievement.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flow chart of the literature selection process.

PRISMA diagram showing the search and selection process for the scoping review.

https://doi.org/10.1371/journal.pone.0305685.g001

Overview of included articles

The remaining 30 documents comprised 16 academic papers, 11 grey literature reports and 3 books (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of documents included in review.

https://doi.org/10.1371/journal.pone.0305685.t001

There was international coverage in the papers, and books from South America (n = 2) [20,49], Europe (n = 19) [3,19,20,23,50–64], North America (n = 9) [20,64–71], Asia (n = 3) [20,72,73] and Australia (n = 1) [74]. However, the grey literature related exclusively to the USA, UK, and Ireland.

The dates of the academic papers covered 23 years from 2000 to 2023 with the majority (75%, n = 12) [49–56,65,66,73,74] in the last decade. The grey literature and books were dated between 1996 and 2022 with half (50%, n = 7) in the last decade [3,20,60,62,63,70,71].

Sustainability, relating to organic procurement and waste, was a key focus of seven of the papers [50–54,56,67], ten were mainly concerned with nutrition or guideline compliance [50,53,55–58,65,67,72,74], two with body weight and disease [49,73] and one with academic achievement [66]. Different types of food provision contract types were described (see Table 1): outsourced vs in-house (n = 5) [50,53,55,73,74]; on premises meal preparation vs off premises (n = 1) [49]; organic vs non organic (n = 2) [51,52]; contractor vs local community (n = 1) [72]; healthy vendor vs not (n = 1) [66]; low cost vs local sourcing (n = 1) [56]. Five papers gave background information only and did not compare outcomes by provision contract type [54,57,58,65,67]. The academic papers were not necessarily focused on the comparison of health outcomes by procurement contract type, though the outcomes and procurement contract type were described and therefore the papers were included as eligible. One paper did focus on a comparison but framed it as low-cost vs local sourcing rather than provider contract type [56]. Including these papers captured important characteristics of each contract type providing context and understanding of how the contract type links with food quality and outcomes.

The 11 grey literature reports were mainly commissioned and funded by government departments or advocacy organisations to evaluate or inform policy. Therefore, they were more concerned with providing overall economic, planetary and population health related insight, as opposed to comparing specific outcomes by provider types. Only one report made comparisons between outsourced and in-house provision [68]. Two looked at food service management companies [69,70] and one at local authorities [59]. Two of the three books were mainly concerned with sustainable school meal provision and did not compare different provision contract types [20,64]; one specifically looked at outsourcing the provision and made comparisons to in-house provision [71].

School meal provision contract types

Documents revealed a variety of school food provision contract types between and within different geographical areas. These fell into broad overall categories of government provision, in house catering and outsourced commercial provision. This varied by school type, geographical area, or political orientation [19,49,53,58,64,66,69,71]. In most European countries, the school was responsible for choosing the food provision contract type [58]; however, European school food systems were found to be commonly managed at a higher level than the school which could be at a national or regional level of government [51,54,56]. Countries such as England, Finland, France, Scotland, Italy and Sweden provided a hot lunchtime meal, either through the school, local government or an outsourced catering contract [54,56,58,64]. School meals were prepared in different locations, with some cooked on site and others prepared in central kitchens for transportation to the schools which could be by outsourced catering teams or in house staff or volunteers [54,56,72].

Differences also existed within provision contract types. For example, one US study found several variations of outsourcing school meal provision such as, private company providing the meals, but the school employs the staff; private company providing the meals and the staff; private company provides consultation services only; and on rare occasions (1%) another public school district provides the meals [66]. There was no peer reviewed scientific research on the provision contract types in England. The School Food Plan [19] provided the most comprehensive picture of provision; however, this report is now over a decade old. The report noted the complexity of school meal provision. This could be contracted by the local authority but provided in house through local authority catering services or through a private caterer. Alternatively, it could be contracted through the school which may use the local authority, a private caterer, or cater in house.

Identified papers indicated that good practice existed across the different provision contract types regardless of whether provision was executed by a private caterer, in-house staff, or a local authority catering service. The synthesis of findings highlighted a number of key themes linking the school food provision contract type with food quality and outcomes for children including; contractual relationships, economic factors, nutritional quality, disease, culture, academic achievement, policies and regulation and child acceptability.

Contractual relationships

The importance of the contractual relationship and having binding, monitored, and evaluated specifications was noted to help to manage a complex situation where quality and outcome priorities may differ by procurement contract type. For example, there could be conflicting priorities of profitability for private contractors compared to social, economic, and environmental principles for authorities [20,54]. A lack of precision in contractual terms such as ‘best value’ had allowed interpretation by caterers to suit their business goals rather than the stated aim of a high quality nutritious catering provision [62]. Substandard contracts without legally binding detailed specifications were thought to impact school food take up, profitability, food quality and prices and expensive specialist legal knowledge would be required to renegotiate or exit the contract [19]. Tender design and contract negotiation were identified as key specialist skills which public sector officials and school leaders may not have despite their responsibility for the contractual relationships with large public sector suppliers [64]. This could affect quality through outsourced contracts being awarded at financially unviable rates, or too much (or too little) market competition driving flawed costing models [53,71]. Differing views on contract lengths were presented, with long term contracts thought to facilitate contractor complacency and profit prioritisation over quality [60] and conversely to facilitate capital investment in facilities which is not financially viable over shorter contract terms [63].

Economic factors

The low levels of available spend for ingredients and the impact financial viability had on nutritional quality was highlighted in several reports [19,23,53,58–60,64,67,72]. Nutritional quality of food was related to the available budget, with convenience and price sensitivity driving the increased use of cheaper processed fast food over more expensive fresh food and the expense of reverting to an infrastructure supporting scratch cooking [53,67,72,74]. Examples of spending varied widely and ranged between 32p and £1.10 per meal served [19,23,59,64,67]. Hacking [60] reported that caterers, from all provision contract types, claimed that rising costs were impacting quality and menus as caterers reduced costs to compensate. This included reducing labour costs by prioritising time saving food over nutritious food, more regular use of cheaper processed food and packet mixes and reducing the variety of food offered. In addition, food costs were cut by reducing portion sizes and limiting more expensive healthy food such as protein and fish. Overcoming a lack of funding included a need to be creative by mixing high and low-cost menu items [72].

There was a large per meal price differential between the high-quality meals in Italy, which were between €5 and €6.18, and prices in the rest of Europe. Studies found prices ranged from €1.02 in Serbia to €2.28 in the UK [56]. The price charged for English school lunches was found to be insufficient to cover production costs, particularly affecting smaller schools unable to meet a breakeven rate of serving 100 meals per day [19]. Documents highlighted that the expectation and pressure for school food to generate profit in the UK and US influences policy decisions to prioritise cost savings over food quality [59,60,67]. The soil association [23] reported that successful local authority catering became unviable as profits were redistributed to education budgets.

Cost reduction was discussed as a key reason for outsourcing the school meal provision in several documents and there was an expectation that outsourcing provision would reduce costs [69–71]. However, in practice, the perceived cost effectiveness, efficiencies and economies of scale were deemed to be outweighed by higher costs for contract tendering, management, and monitoring [53,71]. One study found mixed perceptions of whether costs reduced with outsourcing, noting that ‘contractors did not always keep their promises’ [71]. Conversely, Robinson [69] found that costs and deficits were reduced after contracting out services, but the reduction was only enough to bring them in line with food authorities not using contractors. A 2013 report also stated that the provision contract type did not determine the cost of providing the meal, as schools with an in-house provision had a similar overall cost to local authorities, despite catering for much lower volumes [19]. However, there was no comparable data for outsourced private caterers [19]. RSM Ireland [63] noted that excess profits could be made by corporate providers at the expense of the public purse, however providers needed an adequate economic return to ensure quality was maintained for the lifetime of the contract.

Nutritional quality and guidelines

Most of the papers (n = 10) [50,53,55–58,65,67,72,74], discussed nutritional quality or guidelines; however, there was inconsistency in the markers of nutritional quality used. Measures included compliance with standards and guidelines, which differed by country for items such as content/ servings of fruit and vegetables, fish, whole grains, high fat meat and dairy, sodium content, energy, protein, fat, vitamins and minerals, fried dishes, red meat content, processed foods, pulses and starches, pastries and cakes, confectionary, desserts and added sugars. Two papers [66,67] specifically noted the difficulty in defining and accurately measuring nutritional quality within the complexity and practical difficulties of school food provision.

Factors influencing the nutritional quality of meals in the different meal provision contract types were addressed in some studies, with in-house providers noted to be affected by a lack of resources, nutritional menu planning expertise and administrative flexibility and private contractors noted to be affected by contractual ambiguity, unsuitable selection processes and management burden [53,71]. However, the results relating to quality outcomes were also inconsistent. Three studies found no significant evidence linking the provision contract type with the markers of nutritional quality tested [50,56,74]. Three studies found that in-house facilities performed better than outsourced catering. One related to reduced sodium levels through replacing ‘heat and serve’ models with scratch cooking [65]. Two found that outsourced caterers had higher levels of operating, contract management, and overhead costs and an overall focus on reduced costs which led to including less of the expensive foods such as fruit and vegetables [53,72].

Included reports indicated that the quality of food from outsourced catering was impacted through offering cheaper, more palatable, but less healthy alternatives in competition with the more expensive healthy options to influence consumption patterns prioritising profit over health [23]. Conversely, one study found that when considering the four most important nutritional quality criteria, as defined in national guidelines and relating to fruit vegetable and meat content, outsourced canteens performed better [53] which could indicate a more commercial ability to prioritise contractual compliance. Moreover, Italian schools, which mainly outsourced their provision, were more likely to serve nutritionally calculated meals and recommend healthier choices than Finnish schools which had lower levels of outsourcing [51].

The use of commercial market forces and practices to influence choice in New York, such as making healthy food resemble fast food, creating a brand for school food and marketing the food to the children as ‘customers’ was said to raise the nutritional value of the food [64]. However, more commercialised food and a higher prevalence of Ultra Processed Food (UPF) in privately catered schools compared to state catered schools was an indicator of reduced nutritional quality elsewhere [49].

Two studies found that higher levels of organic produce were an indicator of better nutritional quality due to the increase in fruit and vegetable content and additional training needed for creativity in menus to accommodate the more expensive ingredients [50,53]. However, there was no robust evidence that organic procurement policies led to healthy eating or compliance with nutritional recommendations in the two studies addressing this topic [51,52].

Disease

In terms of chronic disease, the papers only addressed obesity, overweight and hypertension and findings were inconsistent. Two studies found that the characteristics of the school food environment, including the provision type, were associated with overweight and obesity with lower levels present when meals were prepared in-house [49,73]. However, one study found that hypertension was higher in those schools preparing meals in house [49]. Conversely, Anderson et al., [66] found no evidence of obesity rates being impacted by provider contract type. The authors noted that this could be because all models in the study (healthy, standard, in house) were subject to the same calorie requirements. One study noted that the association between increased body weight in children attending schools that outsourced meals may be due to competition between outsourced caterers to make their meals more palatable to children compared to school nutritionists prioritising health over taste [73].

Academic achievement

Only one study measured the impact of healthy school meal provision on academic results, finding that schools contracting with a healthy meal vendor achieved modestly better academic results than schools with in-house catering and the impact was larger for disadvantaged children [66]. Some reports and studies noted that schools prioritised academic results over quality food provision, for example, through reducing the length of lunch breaks, reducing funding or using school meal budgets to subsidise education budget shortfalls or outsourcing the provision to concentrate resource on the core education provision [19,23,67,68,71].

Other factors influencing quality and outcomes

Other factors which were not dependent on provision contract type were also noted to influence the quality and outcomes from school meals. These included leadership attitude to food provision [19], inability to recruit and retain experienced staff [59,71], and administrative burden [69,70].

The impact of culture was highlighted in two studies [52,58] and one book [64]. Despite extensive public funding and resources devoted to school meals in Copenhagen, only 7% of pupils bought a school lunch due to Danish cultural preferences for packed lunches [52]. Similarly, a cultural prioritisation of education over school food in the UK has led to a reduced focus on quality [64]. The stronger food culture in Italy was felt to underpin the quality of the school food provision, which also incorporated environmental credentials, curriculum compatibility, local cultures and traditions, freshness & organic provision [64].

Several documents addressed the complexity of policies and regulation for the provision of school food of an appropriate nutritional quality, which were found to be a challenge for providers and, consequently, led to a lack of implementation and compliance. Key issues included resource constraints, lack of understanding, and a lack of adequate monitoring or consequences for non-compliance [50,53,55,60,64,67,70]. Some studies found that the existence of policies, regulations, and guidelines per se did not influence outcomes [19,52,53]. One study found that the provision contract type did not impact compliance with state menu guidelines which was low for all provision contract types tested [74]. Concerns over the ability to understand regulations and demonstrate compliance was a key reason for outsourcing provision and a robust standards regime was a priority for ensuring nutritional quality [56,70,71]. The monitoring of compliance was found to vary considerably. Italy maintains its high-quality standards with strict compliance monitoring in every school [54,64]. In contrast, England, Wales, Australia, and USA were found to have no compulsory compliance monitoring in place [59,70,74].

The difficulty of increasing child acceptability of healthy meals in a food environment where children prefer fast food to the compliant reduced salt, sugar, and fat healthy food was noted regardless of the provision contract type [71]. Moreover, Tregear et al., [56] found that the best nutritional quality Italian menus also generated the most plate waste (38%) and the Croatian menus with the highest level of noncompliance with nutrient standards generated the least plate waste (12%) indicating that the nutritionally balanced meals may not be appealing to the children. One report noted that healthier menus reduced the number of children taking a school meal in secondary schools but increased the number of children taking a school meal in primary school [59].

Recommendations

More research is needed to inform robust regulation, monitoring and standards to improve outcomes from school food. Furthermore, research is needed to establish whether a particular provision contract type is more suited to leverage the magnitude of public sector spend on school meals to simultaneously impact environmental, social, and economic objectives. Additionally, many shortcomings in areas of good practice were identified which would benefit the design of future policy and practice relating to the provision of school meals. For example, school leaders and public sector procurement officials should have access to specialist commercial support to design, tender, negotiate and manage detailed and binding school food contracts. However, to optimise the benefit of this, there needs to be a clear, current and mandatory definition of school food quality to include in the specifications.

Additionally, work to fully understand the current cost of providing a good quality, nutritious school meal is needed which does not rely on the use of cheaper ultra processed food for financial viability. An adequate economic return for providers which is transparently monitored and reported and does not result in excessive profits at the expense of the public purse is needed. The model should adapt to differing sizes and locations of school which affect profitability and ‘ring fence’ funding for meal provision to prevent use for other budget areas. Further, the current model of voluntary participation of children in school meals in the UK should be considered. This contrasts with some other European countries which require all children to have a school meal or provide universal free school meals to better influence dietary health by enabling caterers to focus on providing health supporting meals rather than competing for business against heavily marketed, more appealing alternatives in an obesogenic food environment.

Strengths and limitations

The robust and systematic nature of the search strategy provides comprehensive and reproducible data on what is known about contract types for the provision of school meals and their impact on outcomes for children. The inclusion of a narrative synthesis to explore the relationships will better inform policy and practice. However, it should be noted that most studies were not specifically designed to address the differences in outcomes by provision contract type and therefore may not be ideally suited to address the research question. Less than half of the documents were academic peer reviewed studies. The grey literature reports written for advocacy organisations have been written with a specific purpose in mind, which may or may not include all available information. The findings in some studies may not be generalisable, as some geographic areas may not translate to the English cultural system. In line with usual scoping review procedure [24], the quality of the included studies was not evaluated, however, it should be noted that some studies were very small [55,56,65,71], others were limited by missing and incomplete data [51,52] and others used subjective measurement methods [66].