Alignment Of US School Lunches With The EAT-Lancet Healthy Reference Diet’s Standards For Planetary Health

By feeding more than thirty million schoolchildren daily, the National School Lunch Program’s food offerings substantially influence a contributing factor to climate change: US food production methods. Modifying school lunch nutrition standards could be a policy strategy to reduce climate change while promoting human health. To estimate how school lunches could be adjusted to encourage both outcomes, we compared the composition of school lunches with the EAT-Lancet Commission’s healthy reference diet science-based benchmarks for reducing food system impacts on climate change. Analyzing more than 5,000 lunches served during school year 2014–15, we found that they exceeded EAT-Lancet targets for dairy, fruit, refined grains, red meats, and starchy vegetables while containing insufficient whole grains, legumes, vegetables, and nuts. In addition, estimated food costs were higher for school lunches than for lunches meeting EAT-Lancet targets. Our findings suggest redesigning school lunches could provide high-quality nutrition while benefitting the environment and reducing food costs. Food systems are among the primary to climate change and account for up to 30 percent of global greenhouse gas emissions. 1 Current methods of food production also to land interfering with 2 Most systems direct result of practices used for food land fertilizer

associated with better long-term health outcomes including reduced risk for cardiovascular disease and mortality. 7 Shifting a population's diet, however, is complex and requires massive collective action from consumers, producers, and policy makers. 8 In the US, federal food assistance programs play a major role in shaping food systems. 9 The National School Lunch Program, which serves more than thirty million US youth and accounts for 30-55 percent of their daily food intake, 10 is particularly influential. In addition to providing a nutritious diet for millions of children and alleviating food insecurity, 11 the National School Lunch Program influences the food and agricultural industries by way of its substantial impact on market share. For example, in response to the Healthy, Hunger-Free Kids Act of 2010 mandating stronger nutrition standards for school meals and snacks, 12 many food manufacturers reformulated products specifically for sale in schools. 13 The National School Lunch Program additionally affects the agricultural industry by creating demand for and increasing the value of US Department of Agriculture (USDA) commodities (foods the USDA purchases from US farmers and subsequently donates to food programs), which are required to account for 12 percent of funding per school year. 14 For the average cost of $3.81 per National School Lunch Program lunch during the 2014-2015 school year, $0.27 (7 percent) is from USDA commodities. 14 Hence, modifications to the National School Lunch Program nutrition standards could affect the food system at multiple levels.
Understanding how well food programs such as the National School Lunch Program currently align with best practices for minimizing environmental harm is critical for understanding what kinds of policy action may be needed to meet climate and environmental goals. When considering the human and environmental impacts of food choices, the EAT-Lancet Commission's healthy reference diet serves as a useful set of benchmarks. 2 The healthy reference diet was developed by the EAT-Lancet Commission, a comprehensive, global team of more than thirty-seven leading experts in nutrition, climate, and agricultural sciences. Combining comprehensive reviews of existing science on nutrition and health with existing science on the effects of food production on greenhouse gas emissions, land use, energy use, acidification potential, and eutrophication potential, the commission developed benchmarks for average consumption of key food categories that would promote health while staying within a "safe operating space" for environmental systems. The healthy reference diet encourages the consumption of fruits and vegetables, whole grains, unsaturated fats, and plant-based proteins. Seafood and poultry are recommended in moderation, and added sugars, starchy vegetables, red meats, refined grains, and saturated fats are discouraged (online appendix exhibit A). 15 The National School Lunch Program's own nutrition standards were revised by the Healthy, Hunger-Free Kids Act of 2010 to incorporate more fruits, vegetables, and whole grains and less saturated fat, sodium, and added sugar. 12 Since the Healthy, Hunger-Free Kids Act's enactment, researchers have documented both higher nutritional quality of meals 16,17 and higher student intake of fruits, vegetables, and whole grains 18 for meals served to National School Lunch Program participants. The implementation of the Healthy, Hunger-Free Kids Act is also associated with reductions in the risk for childhood obesity for participating students. 19 Despite these positive impacts on nutritional health, little is known about the potential environmental impacts of lunches served through National School Lunch Program.
Because the National School Lunch Program standards allow flexibility for the types of foods served within food categories (for example, types of vegetables), it is unclear, by looking at the National School Lunch Program standards, how substantially National School Lunch Program lunches would have to change to align with the healthy reference diet's safe operating space benchmarks for environmental health.
The present study aims to use the healthy reference diet as a framework for understanding how the National School Lunch Program could build on its substantial successes in improving nutritional health by also protecting the environment. In this study, we quantify the amounts of food, according to the healthy reference diet food categories, found in average lunches served nationwide to elementary, middle, and high school students and compare those amounts with healthy reference diet targets to identify excessive and insufficient quantities for environmental health. To accomplish this, we obtained lunch menu data from a nationally representative sample of National School Lunch Programparticipating schools for the 2014-15 school year. Because cost is often cited as a barrier to change, we also conducted an exploratory analysis to estimate the relative difference in food costs between healthy reference diet and National School Lunch Program lunches.

Study Design
We analyzed National School Lunch Program lunch data from the USDA Food and Nutrition Service's School Nutrition and Meal Cost Study, a cross-sectional, nationally representative study of US public schools during the 2014-15 school year. 20 School nutrition managers completed a survey to describe five days of lunch menus during a specific week. Complete data were available for 1,207 (94.1 percent) of the 1,282 schools recruited to participate.

Primary Outcomes
Our primary outcomes were the average amounts of foods and beverages prepared (in grams) per lunch across the two sets of nutritional standards (National School Lunch Program versus healthy reference diet). Using the School Nutrition and Meal Cost Study, we calculated average amounts prepared per National School Lunch Program lunch across eighteen food and beverage categories to compare with the healthy reference diet: four categories of vegetables (dark green, red and orange, other, and tubers and starchy vegetables); fruit; refined grains; whole grains; dairy; beef, lamb, and pork; chicken and other poultry; eggs; seafood; beans, lentils, and peas; soy foods; peanuts and tree nuts; unsaturated oils; solid fats; and added sugars. We chose the average amounts prepared per lunch, rather than the amounts selected by students, to better reflect the quantities planned for by school food service programs, both because this relates to how much food is actually purchased for each category (thus affecting market demand) and because changes to menu planning practices would need to take place at the school food service level. We then calculated healthy reference diet targets 2 for each of these categories for lunch by dividing the daily grams recommended by the healthy reference diet (means, lower bounds, and upper bounds for recommendations with ranges) by three, with the assumption that a lunch should account for approximately one-third of daily intake. To allow for age-appropriate comparisons across elementary, middle, and high schools, these amounts were adjusted to be parallel with USDA target levels of 1,800, 2,000, and 2,400 calories per day (that is, approximately 600, 667, and 800 calories per lunch) for elementary, middle, and high school students, respectively. 21 For each school level, we then calculated the mean difference between the healthy reference diet targets for each food category and the average amounts of each food category actually prepared. See appendix A for more methodological details. 15

Secondary Outcome
To create exploratory estimates of our secondary outcome-the relative difference in cost of food between the National School Lunch Program and healthy reference diet lunches-in the absence of food cost-specific data from the School Nutrition and Meal Cost Study or EAT-Lancet, we estimated the average cost of each type of lunch using the following approach. We first identified the most commonly served food items (excluding mixed dishes and combination items) per food and beverage category in National School Lunch Program lunches (defined by the School Nutrition and Meal Cost Study as those appearing on at least 5 percent of daily menus or by the authors for categories that were offered less often, both of which can be found in appendix exhibit B). 15 We then estimated costs per gram for each of these items (n = 66 items across eighteen categories), using the USDA's Center for Nutrition Policy and Promotion price database, 22 which contains the average national 2003-04 costs of more than 4,000 foods and beverages, adjusting these price estimates to account for the inflation of food prices using the Consumer Price Index for food at employee sites and schools from 2014. 23 For each of the eighteen food categories, we then calculated the average cost per category by averaging the costs of commonly served items in that category. Then, for each school's lunch, we calculated a total cost by multiplying this calculated average price per gram per food category by the grams of that food category prepared for a given lunch and summing these estimated costs across all the foods and beverages contributing to the lunch. We repeated this procedure to estimate food costs for the calculated comparison healthy reference diet lunches (that is, using the same commonly served food items across the eighteen categories and then calculating costs per healthy reference diet lunch based on the combination of servings from these food categories). We then calculated the percentage difference between the estimated costs for the National School Lunch Program lunches versus those for the healthy reference diet lunches. See appendix A for an example calculation. 15

Statistical Analysis
We used one-sample t-tests, accounting for the complex sampling design of the School Nutrition and Meal Cost Study, to test whether the amounts prepared of each food category in National School Lunch Program lunches were different from the amounts recommended for healthy reference diet lunches (means, lower bounds, and upper bounds), stratified by school level. Using a Bonferroni correction to our α level for statistical significance to account for multiple comparisons across eighteen food categories and three school levels, results were considered statistically significant when p < 0.0009. Analyses were conducted using SAS, version 9.4 (SAS Institute, Cary, NC).

Limitations
Although this study offers an innovative perspective on National School Lunch Program lunches and sustainability, it has several limitations. Although we were able to estimate differences between healthy reference diet and National School Lunch Program lunches, we did not implement a real-world intervention. We therefore cannot quantify the short-term potential impacts on dietary intake, food waste, and acceptability to stakeholders that would result from a transition to healthy reference diet targets. Future research is needed to understand how the healthy reference diet could be translated, implemented, and evaluated in schools, as well as the facilitators and barriers to implementation including perspectives from stakeholders (for example, students, food service workers, policy makers).
Another limitation is that the healthy reference diet, although based on a comprehensive review and analysis of existing science by international experts from the fields of nutrition, agriculture, climate, and environmental health, was not designed specifically for US children. Rather, the guidelines were designed for adults with daily calorie needs exceeding those of young children (2,500 per day). We addressed this by scaling EAT-Lancet's daily guidance to line up with the daily calorie needs of elementary, middle, and high school students (1,800, 2,000, and 2,400, respectively) and then dividing by three to estimate values for one meal. We found the resulting calories of our estimated healthy reference diet lunches to be consistent with the required calorie ranges for elementary, middle, and high school National School Lunch Program lunches. 21 In addition, the healthy reference diet targets were developed based on overall global agricultural practices; it is possible that in a US context, some agricultural practices may have differential impacts. 24 However, overall, the healthy reference diet benchmarks remain the most comprehensive set of goals predicting optimal health and environmental outcomes. 25 We also faced a limitation when estimating food costs. The School Nutrition and Meal Cost Study data set includes the costs of food, labor, and operating costs per National School Lunch Program lunch reported by schools, but not at the level needed to generate cost estimates for specific food categories used in this study. Our use of Center for Nutrition Policy and Promotion prices of food items likely resulted in higher price estimates per food category than what school districts relying on bulk purchasing and commodities actually pay, which precluded us from presenting absolute cost estimates; however, given that these higher price estimates would be expected to be evenly applied to both current National School Lunch Program lunches and those meeting the EAT-Lancet standards, this should not affect our estimation of the relative differences in costs.

Study Results
Compared with the average healthy reference diet-based standard lunch, which was estimated to have approximately equivalent average calories per lunch as a National School Lunch Program lunch, National School Lunch Program menus across all school levels included significantly smaller amounts, in descending order, of whole grains (−67 g); beans, lentils, and peas (−28 g); dark green vegetables (−13 g); peanuts and tree nuts (−12 g); other vegetables (−8 g); seafood (−7 g); soy foods (−4 g); unsaturated oils (−2 g); and eggs (−2 g) (all p values < 0.0001). National School Lunch Program lunches also included significantly larger amounts, in descending order, of dairy (+233 g); fruit (+102 g); refined grains (+25 g); beef, lamb, and pork (+15 g); tubers and starchy vegetables (+13 g); added sugars (+12 g); red and orange vegetables (+10 g); chicken and other poultry (+7 g); and solid fats (+3 g) (all p values < 0.0001). Even when compared with the lower and upper bounds of food categories for which the healthy reference diet recommends a range of possible intakes, the National School Lunch Program lunches were still significantly different from the healthy reference diet according to the pattern described earlier (except for other or starchy vegetables and poultry). Results can be found in exhibit 1 for elementary schools, exhibit 2 for middle schools, and exhibit 3 for high schools. Appendix exhibits C-N include values for the estimates of each difference and accompanying confidence intervals. 15 There were also differences in estimated average food costs per lunch, with National School Lunch Program estimates exceeding healthy reference diet estimates for elementary, middle, and high school lunches by relative differences of 60 percent, 46 percent, and 32 percent, respectively (data not shown).

Discussion
Because food production contributes significantly to climate change, this study explores a policy-level response to shift the American diet in a more environmentally sustainable direction. Specifically, we assessed the degree of alignment between the EAT-Lancet Commission's healthy reference diet targets and US National School Lunch Program lunches for eighteen food categories. Both the National School Lunch Program nutrition standards and the healthy reference diet were developed with the intention of promoting human health. The healthy reference diet sets targets that also promote environmental health, but our findings suggest National School Lunch Program lunches do not meet these targets, particularly for four categories: dairy, red meat, whole grains, and vegetables. Compared with a healthy reference diet-compliant lunch, a typical National School Lunch Program lunch contained more dairy, red meat, and tubers and starchy vegetables and fewer whole grains, legumes, and dark green vegetables. Thus, our findings suggest that considerable modifications to the standards that guide National School Lunch Program food service practices would be necessary, especially with respect to protein sources, vegetables, and grains, to address students' nutritional needs while remaining within EAT-Lancet's safe operating space for the environment.
Although our study examines differences between National School Lunch Program and healthy reference diet lunches, it does not directly quantify National School Lunch Program greenhouse gas emissions or attempt to estimate potential reductions from moving in the direction of the healthy reference diet. Other studies, however, suggest that calibrating meal standards with guidelines such as the EAT-Lancet's healthy reference diet could have beneficial effects. A study in the United Kingdom, for example, estimated that adopting healthier school meals could save more than 100 million kilograms of carbon dioxide emissions annually. 26 It is possible that modifying National School Lunch Program standards to reflect sustainable dietary patterns similar to those of EAT-Lancet's healthy reference diet could reduce the program's environmental impact while enhancing the Healthy, Hunger-Free Kids Act's nutritional gains. Specifically, the planet and National School Lunch Program participants could most benefit from fewer animal-based foods (for example, red meat, dairy) and more plant-based foods (for example, vegetables, legumes). Although most School Nutrition and Meal Cost Study menus met the National School Lunch Program's daily requirement for meat or meat alternates (for example, plant-based meats, nut butters), 21 schools are serving far more meat than meat alternates. The substitution of animal-based foods with plant-based foods may contribute to a reduction in greenhouse gas emissions [4][5][6] and is also associated with the prevention of cardiovascular disease and all-cause mortality in adulthood. 7 Reducing the amount of dairy in the National School Lunch Program could be another step toward minimizing the carbon footprint of school lunches, particularly given that a high proportion of fluid milk served to schoolchildren is wasted. 27 All School Nutrition and Meal Cost Study menus met the National School Lunch Program daily requirement for milk, 21 but the average amounts of dairy prepared per lunch exceeded the healthy reference diet guidance. Although dairy products can be good sources of calcium and vitamin D, the EAT-Lancet Commission suggests limiting consumption to 250 g per day (about one cup of milk), based on a diet of 2,500 calories daily. 2 This recommendation was informed by scientific literature that demonstrates that dairy has no added health benefits when consumed in excess, 28 and its production significantly contributes to greenhouse gas emissions and land use. 3 Reductions in dairy intake would require schools to ensure that children receive adequate calcium and vitamin D through alternative sources such as leafy greens and soy products, as well as adequate hydration through drinking water.
Importantly, aligning National School Lunch Program standards with sustainable eating patterns could also reduce food costs, which accounted for nearly 45 percent of National School Lunch Program lunch costs during the 2014-15 school year, 14 with substantial savings from meat and dairy reductions. It is unknown, however, whether shifting to guidelines such as the healthy reference diet would affect labor costs, which accounted for another 45 percent of lunch costs. 14 Schools operate with limited funds on cost-neutral budgets, 14 so reducing food costs while maintaining the nutritional value and quality of meals might allow schools to reallocate funds to other National School Lunch Program needs.
Despite improvements in human and environmental health that could result from applying sustainability standards to the National School Lunch Program, modifying National School Lunch Program standards would require time, political will, and collective stakeholder action. Such adjustments may be politically difficult given previous federal resistance to addressing climate change through food policy, as was seen in the Dietary Guidelines for Americans' omission of environmental sustainability as a dietary consideration, despite the Advisory Committee's recommendation to include it. Moreover, since the National School Lunch Program's inception in 1946, its nutrition standards have only been formally revised twice (in 2004 and 2010). 29 The 2010 changes were met with significant political pushback and counter-lobbying from various industry groups, and the law passed ultimately through persistent and substantial work from nutrition and hunger advocates. 29 Recent attempts to roll back multiple standards 30 might make it particularly challenging to address gaps between the healthy reference diet and National School Lunch Program lunches for items such as whole grains, vegetables, and processed meats. Restrictions on items such as dairy and meat also are likely to be met with resistance based on their importance to US agriculture. As the controversy continues for the 2010 National School Lunch Program standards, inclusion of standards for environmental health would add another layer of complexity.
It is also likely that, at least in the short term, changing long-established dietary habits that focus on heavy consumption of meat, dairy, and refined grains through sustainable National School Lunch Program standards would be challenging and may initially result in increased food waste (although studies have found no difference in food waste from before to after the Healthy, Hunger-Free Kids Act changes). 31,32 Incremental modifications to National School Lunch Program standards must be accompanied by resources ensuring meals are appetizing to students while adequately addressing allergen concerns (that is, beans, lentils, and peas may be preferable to soy and peanuts for plant-based protein sources). It is also possible, when considering documented concern about climate change among younger generations in the US 33 and the superior persuasiveness of sustainability messages compared with health messages, 34 that students may be more willing to consume nutritious lunches if they are also environmentally sustainable.
In light of increasing global interests in solutions for climate change mitigation, there is an opportunity for policy makers, advocates, program leaders, and researchers from nutrition and environmental sectors to explore federal and state policies that consider all aspects of the food system: how food is grown, produced, transported, purchased, distributed, consumed, and wasted. Dialogue between these stakeholders could result in actionable steps that programs such as the National School Lunch Program can take to adopt sustainable practices. For example, at the federal level, such partnerships could lead to identification of cost-neutral or cost-saving substitutions for protein sources that are better for the environment. Similarly, states that wish to prioritize environmentally conscious nutrition standards have the option of adopting their own policies that exceed federal National School Lunch Program standards. State and local school wellness policies provide additional opportunities for the incorporation of sustainable strategies.
Beyond the policy setting, there are numerous ways that individual schools can support environmentally sustainable lunches. For example, schools can make concerted efforts to serve more plant-based proteins, promote drinking water, and reduce food waste while still adhering to existing National School Lunch Program standards. Schools can also share their environmentally conscious practices with policy makers to demonstrate their feasibility and reception from students, families, and school nutrition professionals. Such small shifts initiated by schools when adopted across the US could lead to meaningful improvements in food production and consumption with lasting impacts on human and environmental health.
To advance progress on strategic policy making and program implementation for sustainable school lunches, studies specific to the US context examining multiple environmental indicators such as greenhouse gas emissions, land and water use, and loss of biodiversity are needed. The possible cost savings of adopting National School Lunch Program standards that align with guidelines such as the healthy reference diet warrant further investigation using detailed, school-level cost data that can be evaluated in a real-world context. Pilot testing, routine monitoring, and comprehensive evaluation will be critical for ensuring success with food service operations, reception from students, and impacts on nutrition and environmental health.

Conclusion
As the threat of climate change continues to affect the US and the world, federal, state, and local officials can consider incremental changes to meet targets that support human and planetary health. Modifying a wide-reaching program such as the National School Lunch Program presents a powerful opportunity to develop policies that simultaneously address the nutritional needs of children and those of the environment while potentially reducing food costs. Key areas for intervention include replacing animal proteins with plant-based proteins, replacing refined grains with whole grains, and decreasing dairy products while increasing the quantity and variety of vegetables. This shift in approach to food policy making could serve as a model for building sustainable food systems that foster nutritional health.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material.