The development of school food programs in high-income countries has followed three phases, the third of which is just taking hold today [16]. Globally, school food programs were often established primarily to reduce hunger [16]. In the 1970s, in some parts of Europe and in the 1990s and the 2000s in the United States of America and the United Kingdom, a shift toward improving food quality began, creating the second phase of school food programs. This second phase shifted the focus toward dietary guidelines for school food programs to improve the nutritional quality of food served. The third phase, in its infancy in most countries (Canada included), is a response to increased rates of childhood obesity and nutrition-related chronic diseases, a larger societal context of challenges in the food system, climate change, and environmental degradation. This phase incorporates food systems and societal issues into food programs and policies and generally integrates them more closely with curricula and the school environment as a whole [16,24]. Various researchers argue that this approach is the future of school food programs around the world and is consistent with the principles outlined in the CFG [25,26]. We recently examined international literature on school food programs in high-income countries and outlined key characteristics needed in a national school food program for Canada, including universality, curriculum integration, adaptation to local contexts, and sustainable funding [21].

International research on the health and dietary behavior impacts of school food programs in high-income countries has found modest positive effects overall, including higher vitamin intake and increased vegetable and fruit consumption, especially in younger children [27-30]. Studies from various countries that compare the nutritional quality of food consumed at school that was brought from home versus food acquired through school food programs has found that school food programs provide healthier food overall (regardless of the SES of child participants) [31-38].

Only a small proportion of Canadian children meet the Canadian eating recommendations, with low vegetable and fruit consumption of particular concern [39,40]. Research on overall dietary patterns in Canada highlights the very large proportions of ultraprocessed foods high in salt, sugar, and some fats being consumed [41,42]. Families struggle to introduce minimally processed healthy foods for a variety of reasons [3-5].

Provision of healthy school lunches is challenging in the context of parents working long hours [43,44], and families are struggling to adopt healthy food behaviors [44]. Parents may rely on highly processed foods, low in key nutrients but high in salt, sugar, and fat, to cope with demands on time [45,46]. The diets of Canadian children across the socioeconomic spectrum are poor [47]. There is some evidence from the United States that school food programs can reduce disparities in vegetable and fruit consumption between children from higher versus lower SES households and limit the intake of minimally nutritious foods among higher SES households [48]. Introduction of healthy foods in a universal school food program could provide all children with greater opportunities to learn about and eat healthy foods in a way that is not stigmatizing [49].

Our team’s previous observational research on school food in and around Saskatoon, Saskatchewan, Canada, characterized lunches using a school food checklist and digital photography in randomly selected urban schools with a meal program (n=3), urban schools without a meal program (n=3), and rural schools without a meal program (n=3) [50]. The number of servings of each CFG food group was determined and compared with one-third of the recommended daily intake. The Healthy Eating Index (HEI) scores were calculated. In results similar to Tugault-Lafleur et al [14] using the national Canadian Community Health Survey data for the school day, just over half of the students who brought lunches from home met the recommendations for grain products and meat and alternatives, less than one-third met recommendations for vegetables and fruits, under one-fourth met the recommendations for whole grains, and even fewer met the recommendations for milk and alternatives. The HEI scores of students in meal programs were greater than those of students who brought lunches from home because they included more whole grains, met meat and alternatives recommendations, and contained fewer calories from minimally nutritious foods (high-fat and high-sugar foods such as candies, chocolate, and sauces). Children not participating in meal programs brought about one-third of calories as minimally nutritious food, about double that of meal program student lunches.

There are numerous reports of school staff and other stakeholders expressing a desire to offer universal lunch programs rather than the more common targeted programs and expressing concern about the quality of food currently consumed by Canadian children at school outside of lunch programs [51,52]. Drawing on international evidence, an evaluation was conducted of a 2-year pilot of free school meals in 3 local authorities in the United Kingdom [53]. Two local authorities made free school meals universal to all primary school children, whereas the third extended free school meal entitlements to a larger number of students but was not universally free. In the extended entitlement (not universal) authority, there were no impacts found on children’s eating habits, whereas in the universal entitlement authorities, there were reductions in the consumption of potato chips and soft drinks and an increase in vegetables consumed. In the universal pilot areas, parents’ perceptions of school meals for health were more positive and they thought that their children were more willing to try new foods.

Educational achievement is associated with health across life span [54]. Studies on school food programs and academic achievement, attendance, tardiness, and dropout rates point to important impacts of school food programs. Attendance and tardiness appear to be the educational outcomes with the most available evidence, but some studies have found improvements in academic achievement with the introduction of school food programs [55-65]. School food programs can also contribute to teaching culinary heritage, social norms around food, and environmental sustainability [26,66,67]. We are conducting a separate study in partnership with educational research colleagues on educational outcome changes as a result of this intervention.

Various countries, including Canada most recently, have adopted nutritional recommendations that recognize the connections between nutrition and environmental sustainability [1,68]. According to Cullen et al [69], food literacy is “the ability to make decisions to support the achievement of personal health and a sustainable food system considering environmental, social, economic, cultural, and political components.” School food programs focused on the provision of healthy, sustainable foods (including vegetables and fruit, plant proteins, and locally produced foods) along with the promotion of sustainable food behaviors such as school gardening and learning how to reduce food waste, may work together to change food behaviors [26], which may spill into life away from school [70]. Stone [71] and Weaver-Hightower [72] explain how children can be involved in age-appropriate ways to grow and prepare food, along with learning how the food system works and its environmental and social challenges, to integrate learning with a meal program. This integration enables appreciation for food and a greater willingness to try new foods [73,74].

The evidence for school food interventions that include both meals and curriculum is getting stronger internationally, but not much Canadian evidence is available [19,20,62,75]. These interventions include introducing healthy foods, integrated with the curriculum, and with parent involvement [28,76]. They take an education-integrated approach that involves children in growing and preparing food, teaching about food system sustainability, and healthy behaviors [26]. Internationally, Oostindjer et al [26] juxtapose the Swedish and British school food programs and the Japanese, French, Italian, and Finnish school food programs on this front. The Finnish universal lunch program, which is a key aspect of their education system, is integrated with the curriculum, including learning objectives related to social relationships and eating norms that are lacking in Sweden and the United Kingdom [77-79]. According to a review of international literature, this emerging integration of school meals with classroom curricula aligned with food cultural learning and establishing an optimal food and social environment may facilitate learning of healthy and sustainable food behaviors [26].

We have ethical approval in place (University of Saskatchewan BEH-509) and a memorandum of understanding prepared between principal investigator RES and the SPSD. In 2018, we collected baseline data on school food environments in the 2 intervention schools to inform study preparation. Intervention planning was supposed to be conducted from March to May 2020; however, because of the COVID-19 pandemic, it was put off and conducted between January and May 2021. The COVID-19 pandemic-related prohibitions on field data collection also affected this study, and permission to enter schools was granted in mid-May 2021.

RQ2: What is the difference in diet quality, food group, and key nutrient consumption among children in intervention schools after the implementation of a comprehensive, curriculum-integrated universal school lunch program as compared with children in control schools?

A digital photography–enhanced plate-waste study will assess food eaten for lunch by students in intervention and control schools (n=148 minimum each for a total of 296; 4 schools) on 2 days at baseline (preintervention) and then on 2 days at the end point (in the last 2 months of the intervention). Each participating child will have four measurements of plate-waste data (two baseline and two end point data). We will follow the protocol developed by coinvestigator HV, which has been extensively used in studies of school-aged children [85-87]. The procedures for data collection involve weighing the served food and leftovers, using digital photography and a food scale during the lunchtime with an app installed on tablets. This method is considered the most precise measurement of dietary intake compared with other dietary assessment methods that are typically used, such as food records [88,89]. The literature indicates no significant impact of the method on eating behaviors in children [88,89]. Digital photography of plate-waste will be conducted by 2 research assistants in half the school classrooms on each data collection day, for a total of 4 data collection days at both baseline and at end point in each school. The practice of collecting data from half the classrooms in a single school on each data collection day will ensure minimal disruption of regular school days.

Our plate-waste approach allows us to infer the amount of food consumed by calculating the difference between the amount served and the amount left over [88,90]. At the end point, the information about the menus (recipe level data) on the days of data collection will be entered in the app on tablets before lunch. A label with the identification number of the child is placed on the tableware to correctly identify each child’s meal and leftovers and scanned at the time of data collection for each child. A photograph (62 cm away from the food, at a 45° angle [91]) is taken of student’s tableware (ie, each container in a lunch box, plates, or other) to visually capture the size and general appearance of items to subtract them from the food consumed. Each entry contains the child’s identification number, the photo forms to record the weight of the food, and whether the photo was taken before or after the child’s meal. Leftover food is weighed and photographed again at the end of the meal. By subtracting the weighed leftovers from the weighed food served, the food consumed will be obtained. The information pertaining to energy, macronutrients, and micronutrients consumed will be derived using Food Processor nutrition analysis software (The Food Processor, Esha Research version 11.3.285). The final report will be exported to Microsoft Excel where food groups based on the new CFG will be calculated (new recommendations on the amount and type of food taken by age group are not available as of June 2021 but will be used once released by Health Canada).

The collated baseline and end point data will go through data cleaning, processing, and quality control using protocols developed in the Vatanparast Nutritional Epidemiology laboratory. After creating the master data file including original and derived variables of interest, such as the Nutrient Rich Food index (9.3) [92], HEI [14] based on the new CFG, sex assigned at birth, self-identified gender, age, grade, and neighborhood of residence as a proxy for SES, the descriptive analyses will provide the initial information on these variables. Categorical variables (ie, HEI categories) will be summarized in frequency tables. Regression models will examine changes from baseline to the end point across key outcome variables including the Nutrient Rich Food index scores, HEI scores, CFG servings, and key nutrients (calcium, sodium, vitamin D, folate, and calories from minimally nutritious food) and whether changes differ between the control and intervention schools. Final models will adjust for potentially confounding factors including sex assigned at birth, self-identified gender, age, grade, neighborhood of residence, and baseline dietary intake (for more detail on identifying a priori potential confounders and controlling for them, see Phase 5: Final Analysis and Dissemination section below). We will also examine potential interaction effects between the intervention and control groups and age, sex, and gender (as detailed above in the sample description) to assess whether the effect of the intervention on dietary outcomes was modified by key student-level characteristics. Clustering of students within the classroom and schools (nonindependence) will be taken into account by applying random-effects multilevel modeling approaches to data.

RQ3: What is the difference in food-related KAP among child participants in intervention schools after the implementation of a comprehensive, curriculum-integrated universal school lunch program as compared with children in control schools?

At baseline and at the end point in both intervention and control schools, we will administer a survey of food-related KAP adapted from the Individual Eating Assessment Tool (I-EAT) to all participating students in grades 4-6 in the intervention and control schools. According to the KAP model, knowledge (K) accumulates and attitudes (A) change and these changes promote changes in practices (P) over time. We cannot administer the survey to children younger than grade 4 because of inconsistent literacy levels before that age and older than grade 6 because they will no longer be at the school at the end point. Given our expected 80% response rate, we will have a sufficiently large sample in these three grades only to carry out these analyses. This survey was originally developed by Black et al [47,93-95] and we have adapted and piloted it with 100 children. We conducted a factor analysis of the pilot results for validation and then made some additional minor changes following the analysis, specifically removing four questions that did not add relevant data. We included I-EAT questions on food, cooking and nutrition knowledge and beliefs, confidence in participating in various food-related activities, attitudes toward whole grains, vegetable and fruit consumption, participation in learning about food, nutrition and cooking, confidence in engaging in food-related activities, engagement in food-related sustainability practices, and demographics.[47,93,94] We will analyze I-EAT derived outcome data and compare the differences at end point between the intervention and control student data.

Analyses of plate-waste data and survey data will be conducted using the intention-to-treat principle [96]. To assess the effect of the intervention, we will use mixed-effect models using time of measurement (baseline or end point), group (intervention or control), and an interaction between time and group as fixed effects (base model). The neighborhood of residence (proxy for SES) will be used to determine whether SES influences the response to the intervention. To account for clustering related to repeated measures and because of sampling of students within classes and schools, variables representing participants and schools will be included as random effects in all models. Additional models will account for potentially confounding variables identified using directed acyclic graphs for each outcome [97,98]. Directed acyclic graphs are used to identify potential confounders using prespecified conceptual diagrams and testing these using the data to increase the precision of estimates and causal paths based on regression models. Analyses will be conducted using the MIXED procedure in SAS version 9.4 (SAS Institute, Inc).

RQ1: What are the benefits and challenges faced in establishing and administering a universal, curriculum-integrated healthy school lunch program?

RQ4: What aspects of curriculum integration show promise in enhancing food-related learning?

We will collect qualitative data from the 2 intervention schools. In the case study research, the investigator explores a bounded system or case (a school) over time through detailed data collection, drawing on multiple sources of information [99,100]. We have already conducted preintervention research in the intervention schools examining the school food environment using a tool called the School Food Environment Assessment Tool (SF-EAT) [101]. SF-EAT assesses the integration of healthy and sustainable food strategies in the school food environment, including food availability, food preparation facilities, gardening, composting, integration of food issues into the curriculum, and availability of healthy and environmentally sustainable food [101]. The administration of the SF-EAT tool was complemented with key informant interviews on barriers and facilitators for adopting the curriculum-integrated lunch program.

We will use semiparticipant observation and key informant interviews with school staff to evaluate the process of implementing the curriculum-integrated school lunch programs in the 2 intervention schools. We will study 3-5 individual classrooms in different grades to detail the aspects of each case. Observation will occur during (1) meetings when school staff are planning for the establishment of the school lunch program; (2) meetings when school staff are discussing curricular integration; (3) during lunch preparation, service, and clean-up; (4) when children are involved in preparation; and (5) during other curriculum integration activities such as during lessons involving growing vegetables. Observations will occur at minimum for 1 school day each month in each school and at school staff meetings as needed during the intervention period, with key informant interviews occurring throughout. Observations will be recorded using voice memos and field notebooks, which will be transcribed or detailed upon leaving the school at the end of the day. We will focus our data collection on how the intervention is being carried out, including challenges experienced and its fidelity to the original intervention design. We will interview caregivers on their perceptions of the program and behavioral changes they identify, including caregivers of children who have never participated in a school food program and those who have, across grades.

The qualitative data will document how schools establish a universal lunch and parent perceptions of the program. Our observations and interview questions will focus on the perceived benefits and challenges of establishing and administering a universal, curriculum-integrated healthy school lunch program and on the aspects of curriculum integration that appear to show promise in terms of enhancing food-related learning. The cases will inform the establishment of programs on a larger scale and in other contexts.

Interviews will be audio-recorded, transcribed verbatim, and analyzed using NVivo software (QSR International) for coding and theme development. Semiparticipant observation data will be recorded as field notes during observation as well as immediately after. Qualitative data will be analyzed initially in an open coding process to allow for emerging themes, followed by analysis using a priori generated code lists derived from the literature and our RQs. Our analytical approach will build concepts and themes inductively, testing and refining them with participants [102]. This hybrid inductive and deductive thematic analysis has been described by Fereday and Muir-Cochrane [103]. A regular debriefing with the research team and key school informants will assist in the refinement of concepts and themes. These steps will strengthen the credibility and dependability of the results [104,105]. Our contextualized interpretation and meaning making with stakeholders is consistent with Moss and Dahlberg [106] for research in educational contexts.

We will enter, clean, and analyze postintervention data. This phase will include end-of-study knowledge translation (KT), although because of the existing collaborative relationship between the academic and nonacademic partners, KT will be ongoing throughout the entire study. Although this study is about filling gaps in the academic literature, it is also about ensuring practice relevance. Everything from conceiving the study to the development of RQs and practices has been and will be conducted closely with the school division, the CHSF, and other partners.

We have experience with integrated community-university KT strategies, which have generated practice and policy change [107-110], and we will apply what we have learned [111,112]. The inclusion of knowledge users or decision makers as members of the research team will facilitate evidence-informed decision-making and integrated knowledge transfer and exchange. This collaborative approach informs and directs our research and sharing of results and other outputs and is consistent with the Canadian Institutes of Health Research Guidelines for Research with Aboriginal People [113]. Although the focus of this research is not Indigenous people, we believe that these guidelines are relevant for all community-based research. We will work closely with our partners to conduct KT as needed and requested. Materials will be developed and designed with the goal of their use in future programming.