From preschool to high school, 65·6 % of French children aged between 3 and 17 years have lunch at school at least once a week, 71·5 % of children in lower secondary schools and 64·4 % in upper secondary schools(Reference Dubuisson, Lioret and Dufour1), and 4·2 % have dinner (i.e. internal students)(2). In addition, the last French National Dietary Survey (INCA 2) showed that children aged 3–17 years consumed 33·6 and 31·1 % of their daily energy intake at lunch and dinner, respectively(3). Thus, school meals contribute widely to the dietary and nutritional intakes of schoolchildren in France. Furthermore, both childhood and adolescence are the periods during which food habits and preferences are acquired and settled(Reference Birch4–Reference Nicklaus7). In this context, it is necessary that schools offer balanced meals that help fulfil the nutritional requirements. Schools have to set an example, especially given the growing preoccupation with obesity among the general population and the need to teach children to adopt more healthy dietary behaviour(Reference Lioret, Touvier and Dubuisson8, Reference Lioret, Touvier and Lafay9). In addition, when given the choice, students often prefer fatty and sweetened foods over fruit and vegetables(Reference Kramer-Atwood, Dwyer and Hoelscher10). Consequently, several countries have agreed on the importance of applying food policies at schools to improve the nutritional quality of food and beverages available on school premises (in cafeterias, vending machines or other convenience stores)(Reference Briefel, Crepinsek and Cabili11, Reference Briefel, Wilson and Gleason12). Indeed, several studies have already pointed out the beneficial effect of such policies (restriction of competitive food items) on children's food consumption and choices at school(Reference Cullen, Watson and Zakeri13, Reference Vereecken, Bobelijn and Maes14), even though they were sometimes not effective enough to guarantee balanced diets in school meals(Reference Nelson, Lowes and Hwang15).
In France, several reports on dietary intake in school meals drew attention to the issue of unbalanced nutrition and diet on offer(Reference Czernichow and Martin16–19). In 2007, the study group on catering and nutrition markets (Groupe d'Etude des Marchés de Restauration Collective et Nutrition, GEMRCN) revised its recommendations(20), first published in 1999(21), to help caterers improve the nutritional quality of meals served at school. This revision added two new nutritional objectives (restrict total sugar intake, especially in desserts, and achieve a more balanced intake of fatty acids) to the original ones (reduce fat intakes and increase fibre, Ca and Fe intakes). To reach these nutritional objectives, the GEMRCN proposed fifteen food-group frequency guidelines (FFG) based on twenty meals (either on twenty successive lunches or twenty successive dinners), which were better described and generally more restrictive than the 1999 guidelines (see Table S1 of the supplementary material, available online http://www.journals.cambridge.org/bjn).
Further to the second National Nutrition and Health Programme(Reference Hercberg, Chat-Yung and Chauliac22), these FFG are about to become compulsory instead of simply being recommended. With the advent of this new legislation on school meal composition, it is important to ensure that these guidelines do actually lead to healthier meals by bringing the nutritional amounts served in line with the national nutritional requirements.
Given this background, this article (1) describes both the compliance of nutrients on offer and FFG in secondary school meals and (2) assesses FFG with respect to their effectiveness and relevance to get a balanced nutritional offering.
The present study is based on a cross-sectional survey carried out on a representative sample of 1440 French lower and upper secondary state schools (i.e. 10- to 18-year-old children) from November 2005 to April 2006. The methodology has been described elsewhere(Reference Dubuisson, Lioret and Calamassi-Tran18). Briefly, the schools were selected from two independent samples. The first sample included 1200 lower and upper secondary schools from France's mainland and overseas departments, randomly selected from among the 7000 secondary schools administrated by the Ministry of Education (ES) and having a cafeteria. The second sample included the 240 upper secondary schools administrated by the Ministry of Agriculture (AS). To ensure national representativeness of the sample, we randomly selected the 1200 ES schools by a balanced sampling design, based on four controlled variables (geographical location, school size and type, size of urban area and location in a deprived area or not).
After validation of questionnaire and verification of a minimum of fifteen meals returned, 707 schools were finally retained for statistical analysis. In this sample, 570 schools were administrated by the ES and 137 schools were administrated by the AS. Furthermore, of the final sample of 707 schools, 270 had boarding facilities.
Every school was asked to fill out a questionnaire on its catering service management and the implementation of the guidelines, and was requested to enclose all menus (lunches and dinners) proposed during the last twenty school days. The menus consisted of a list of the foods and dishes served by courses (starters, main dishes, side dishes, dairy products and dessert) during twenty successive days. However, information about the amounts of foodstuffs provided was not collected. Thereby, we used the adequate serving size recommended by the GEMRCN, according to the food characteristics and courses that were established to help caterers with their menu elaboration to match the recommendations.
First, the traditional French courses (i.e. starter, main dish, side dish, dairy product, dessert) were counted and identified for all the 22 409 menus received (16 893 lunches and 5516 dinners). Then, the food items recorded were allocated a nutritional composition in accordance with the French food composition database: CIQUAL(Reference Favier, Ireland-Ripert and Toque23). When food items met the specifications for meal courses and nutritional composition, they were assigned to one or more of the fifteen food groups defined by the guidelines (see Table S1 of the supplementary material, available online http://www.journals.cambridge.org/bjn). Non-described ‘cheese’ and ‘yogurts’ (69·9 % at lunch and 73·5 % at dinner) were randomly assigned, or not, to a particular food group after taking into account the statistical distribution observed in the remaining and clearly described cheese and yogurts. The number of menus sent could differ from one school to another (sixty-four schools sent fifteen to twenty meal sets and 643 schools sent twenty or more meal sets), and most of the menus were free-choice menus. Thus, to enable comparisons with the guidelines, the FFG observed in school meals were related to a set of twenty meals, according to the following formula:
Calculation example: Estimation of a school adequacy to the FFG ‘fried products ≥ 15 % of lipids’.
This particular FFG is to apply to main courses and side dishes (two possible courses).
If a school returned twenty single menus including twenty starters, twenty main courses, twenty side dishes, twenty desserts, in which ten menus proposed fried products at the main course and ten menus proposed fried products at the side dish, the adequacy to the FFG ‘fried products ≥ 15 % of lipids’ would be equal to: ((10+10)/(20+20)) × (20 × 2) = 20. The school would have served twenty times products from the FFG ‘fried products ≥ 15 % of lipids’ on twenty meals.
If the school returned fifteen single menus only, in which fried products were also served ten times at the main course and ten times at the side dish, the adequacy to the FFG ‘fried products ≥ 15 % of lipids’ would be equal to (10+10)/(15+15)) × (20 × 2) = 26·7.
The school would have served 26·7 times products from the FFG ‘fried products ≥ 15 % of lipids’ on twenty meals.
The number of FFG met by each school was calculated as the sum of each FFG, followed or not (0 or 1), considering the meal type (lunch or dinner) separately. The schools' compliance regarding the recommendations was estimated both quantitatively (on a scale from 0 to 15 for the FFG met) and qualitatively through a three-class variable: low compliance with the recommendations (six or less out of fifteen FFG followed), medium (from seven to eight out of fifteen) or high (nine or more out of fifteen).
Nutrient amounts and macronutrient contributions to energy were obtained by the addition of nutritional amounts in each course of the meal: starter+main dish+side dish (+dairy products)+desserts. Dairy products were considered to be a single course in the case of a five-course menu, and were included in the desserts in the case of four-course menu. When free-choice menus were offered to the students, a mean of nutrient amounts was first estimated by course and then added to the means of other courses of the meal.
Finally, a mean of nutritional amounts was calculated per school, considering lunch and dinner separately, and compared to the national nutritional requirements when they were available. On average, a lunch meal must provide between 11 and 15 % of energy from proteins, 50 and 55 % from carbohydrates, 30 and 35 % from lipids, at least 300–400 mg Ca and 4–7 mg Fe(Reference Hercberg, Chat-Yung and Chauliac22). No information on the amounts of vitamin C required has been provided for lunches. The requirements were set at 35–50 mg as the requirements for 1 d are estimated at 110 mg for teenagers.
All analyses were carried out using the Statistical Analysis System, version 9.1 (SAS Institute, Inc., Cary, NC, USA). The representativeness of the final samples was tested according to the controlled variables used for the balanced sampling design: geographical location, school size and type, size of urban area and location in a deprived area or not. Only a little deviation regarding the geographical location was observed. Therefore, a weighting coefficient was calculated by the ratio method of the CALMAR macro(Reference Sautory24) to ensure the national representativeness of the final sample.
The number of meal courses (four or five) and type of menu (free choice, guided choice, no choice (unique menu)) were considered as possible confounding variables, as they have been shown to be associated with the level of implementation of frequency guidelines and the nutritional amounts in the meals(Reference Bertin, Lafay and Calamassi-Tran25). Indeed, five components and free-meal choice (often associated with five components) were found to be related to an increase in nutritional amounts (results not shown). The mean nutrient amounts provided by school meals were compared between ES and AS schools and lunch and dinner, using a t test. The relationships between nutrient amounts (i.e. lipids, proteins, total sugars, Ca, Fe and vitamin C and energy) and the compliance with their corresponding FFG (i.e. the FFG that aimed at the nutrient reduction or increase) and the overall recommendations adequacy were tested using ANOVA after controlling for the two adjustment variables. The level of significance used was P < 0·05. ES and AS schools were studied separately because of their different sampling design.
Compliance of meal food composition with the recommendations
Table 1 shows the proportion of schools meeting each FFG and their distribution according to the three levels of compliance defined.
ES, general and vocational teaching; AS, agricultural teaching.
* Reading example: 60·9 % of the ES schools and 40·7 % of the AS schools met the guideline limiting starters with 15 % of fat or more at lunch; 76·1 and 63·1 % of ES and AS schools, respectively, met this guideline at dinner.
Recommendations were less followed at dinnertime (P < 0·05) for both ES and AS schools, with 44·8 and 60·2 % of ES and AS, respectively, having a low compliance level, 47·4 and 33·9 % an intermediate one, and 7·7 and 5·9 % a high level. ES schools followed more guidelines at dinner than AS schools, whereas the contrary was observed for lunches (P < 0·05). According to the three levels of compliance defined, 32·9 and 28·2 % of the ES and AS schools, respectively, had a low level for lunches, 52·1 and 45·5 % an intermediate level, and 15·0 and 26·3 % a high level. Food groups consisting of high-fat starters, fried products, raw fruits, vegetables, dairy products and starchy foods were well offered whereas food groups consisting of highly sweetened desserts, main courses, preparations, cheese, meat and fish were insufficiently offered in school meals.
Nutrient amount offered in school lunches and dinners
The mean meal nutrients on offer and macronutrient contribution to energy are shown in Table 2. A significant effect of the type of meals (lunch or dinner) and of the types of school was observed for several nutrient offerings. Lipid, protein and carbohydrate contributed respectively to 42·8, 21·5 and 35·7 % of energy offering at lunch in ES schools. Similar proportions were noted for dinners in ES schools and for lunches and dinners in AS schools. Average offerings of vitamin C, Fe and Ca matched the estimated average requirements with 42·2 and 40·6 mg vitamin C, 5·4 and 5·6 mg Fe, and 338·3 and 355·9 mg Ca, respectively, in ES and AS schools at lunch.
ES, general and vocational teaching; AS, agricultural teaching.
For a given meal type (lunch or dinner), nutrient mean values were significantly different between ES and AS schools (t test): * P < 0·05; ** P < 0·01; *** P < 0·001.
† Reading example: a meal served at lunch provided on average 3139·7 kJ in ES schools and 3266·5 kJ in AS schools.
‡ t test used to compare differences in different nutrient offers at lunch v. dinner.
Impact of food-group frequency guidelines on nutrient offering
The impact of each FFG on its respective nutrient offering is shown in Table 3. In general, the observations confirmed that the FFG had been effective in adjusting the nutrient amount offered.
ES, general and vocational teaching; AS, agricultural teaching.
Mean values were significantly different according to compliance or not with specific food-group frequency guidelines, tested using ANOVA adjusted on the type of menu and number of components: * P < 0·05; ** P < 0·01; *** P < 0·001.
† Reading example: at lunch a meal served in ES schools provided on average 34·3 g lipids if they met the FFG ‘Starter with >15 % lipids’ and provided 37·8 g lipids if not.
‡ A relation between a particular nutrient and a specific FFG was studied if the nutrient was concerned by an FFG. For example: the amounts in Ca were analysed in relation to the FFG that were proposed to improved amounts of Ca the schools meals that are the FFG: cheese >150 mg Ca, cheese with 100 to 150 mg Ca and dairy products.
The impact of the FFG on nutrient offering was generally greater at lunch and for ES schools. No significant impact was observed for the food groups that were either fully followed (fried products, desserts with >15 % of lipids) or poorly followed (desserts < 15 % of lipids). However, they did decrease the amounts of lipids and total sugars (desserts) offered in the meals.
The lipid offering was significantly lower when the starters with >15 % of lipids were limited by the schools, as suggested in the recommendations. Following FFG for the main course with a protein:fat ratio < 1 and desserts with >15 % lipids food groups also lowered the lipid amounts offered, even though the differences were not significant for dinner and AS schools.
Furthermore, a proper offering in red meats and preparations including more than 70 % of eggs, meat or fish increased the protein amounts in the meals served. Amounts of Ca, as well as amounts of lipids, were significantly higher when the guidelines related to cheese were met. However, suitable servings of dairy products decreased the offering in Ca. In addition, a significant increase in vitamin C was noted when starters with raw vegetables and desserts with fruit were proposed according to the guidelines. This improvement was also observed with a greater offering in cooked vegetables in the side course, but only at dinner. Finally, a better offering in Fe was found when the FFG regarding red meat was followed, but the difference was significant only for ES lunches. The respect of the preparations including more than 70 % eggs, meat or fish FFG showed to enhance Fe offering for ES schools only, at both lunches and dinners, whereas following FFG on fish did not increase the dietary Fe offering.
Compliance with the overall food-group frequency guidelines on nutritional offering
Table 4 presents the impact of the overall FFG on the different nutrients and energy offerings in school meals. Higher compliance regarding the overall recommendations is associated with a ‘lower energy and lipids offering’ (with a significant difference at lunch for both ES and AS schools). It was also shown that greater compliance with the FFG made it possible to increase the ‘vitamin C and reduce the carbohydrate offering’.
ES, general and vocational teaching; AS, agricultural teaching; FFG, food-group frequency guidelines.
Mean values were significantly different between the three levels of compliance with FFG, tested using ANOVA adjusted on the type of menu and number of components: * P < 0·05; ** P < 0·01; *** P < 0·001.
† Reading example: in ES schools with low adequacy to the FFG, a meal provided on average 3225·9 kJ at lunch.
The main objective of the study was to identify the effectiveness of the dietary recommendations in enabling schools to achieve an adequate nutritional offering in school meals. The present study used data from the only national survey on school meals and catering in France. The large sample and national representativeness of the survey on school catering and meals served offered many advantages for estimating the effectiveness of the dietary recommendations in adjusting the nutritional offering of school meals.
Nevertheless, there are several limitations to the present study. First, the study design itself, which included only state schools. This limitation remained acceptable, as 78 % of French pupils go to state schools. However, data on meals served in private schools should be collected in the future as they receive nearly 20 % of the schoolchildren that eat at school. Moreover, the menus were collected only in autumn/winter, which did not take into account seasonal availability in some foods such as fruits and vegetables. Nevertheless, this does not invalidate the study results as the FFG were set up to be relevant throughout the year. Apart from that, after two reminders during the school year, the final response rate remained at 49 %, which matched the range of response rates observed in studies based on a similar design(Reference Vereecken, Bobelijn and Maes26, Reference French, Story and Fulkerson27). However, representativeness of the sample was verified and information was collected on 22 409 meals (16 893 lunches and 5516 dinners), which strengthens the present findings. The present study is based on the menus returned by the schools, which implies that the quality of the data relied on the quality of the menu description. Indeed, a large number of cheese and yogurt items in the menu were not precisely identified and were randomly assigned to an amount of Ca (using the amounts observed for other cheeses or yogurts), which might differ from reality. Furthermore, to ensure the representativeness of the sample, schools with only fifteen menus returned or with free-meal choice were included in the study. Nevertheless, the application of the formula enables the management of this aspect and allows a correct comparison between the offering and the frequency guidelines.
In addition, theoretical portions were used to estimate nutrient amounts in the meals, which might differ from the portions actually served and then eaten by schoolchildren(28). To avoid this issue, in some surveys, food offered and consumption are collected at the same time on school premises(28). This method would enable a better accuracy in the description of the food offered and portion size served but would also require more materials and involve more people to collect the data. It has mainly been used in studies with a small sample of schools, and thus making it difficult to adapt to our large study. Finally, since the recommendations are first to be targeted to the offer, the present study concerned the dietary offer of school meals and did not provide information on the foods chosen by children.
The analysis of meal composition showed that the FFG were well followed for high-fat foods, plant products (raw or cooked vegetables, fruits and starchy foods) and dairy products but not for food groups aimed to improve the nutritional quality of main courses, to promote the consumption of cheese or to limit the consumption of high-sugar desserts. On the whole, the schools appeared to have complied less with the 2007 FFG than the one released in 1999(21). This decrease was found for both types of school (ES and AS) and meals (lunch and dinner). Indeed, the 2007 FFG focused on more specific food groups and made stricter demands than the 1999 FFG(21). The differences in compliance between the two versions of the guidelines are mainly explained by the reduction of the maximum FFG defined for food groups to be limited in the earlier guidelines (e.g. high-fat starters), and by the revision of the foods allowed in food groups (e.g. red meat, preparations, fishes, main course with protein:fat ratio). Regarding other food group guidelines, no direct comparison can be made because of the changes and supplementary recommendations made in the 2007 guidelines. However, the new recommendations revealed that the guidelines for fruits in dessert were followed more than those for raw vegetables in starters, and that dairy products were more often proposed than cheeses. The new constraint on total sugar content of fatty desserts in the 2007 guidelines is less restrictive than the previous one and allows more fatty desserts in meals. Thereby, some adjustments are still to be made to improve the recommendations. Following the present study findings, the constraint on sugar content of fatty desserts will be removed in the upcoming regulations.
The nutritional composition analysis confirmed the beneficial effect of FFG on specific nutrient offers, especially when the food group concerned a particular nutrient that should be limited or increased. Thus, food groups aimed at reducing the lipids and total sugar offered favourably influenced the offering for such nutrients. Similarly, the food groups established specifically to increase the offering for Ca, vitamin C, Fe and animal proteins were found to be effective on the whole. The decrease in Ca offered when more dairy products are served may be explained by a substitution effect with cheese, which contains greater amounts of Ca than other dairy products.
Furthermore, the difference in the nutrient offering was robust to adjustment for the number of courses and type of menu, suggesting that they were directly related to compliance with the recommendations rather than reflecting the meal structure.
Differences between lunch and dinner and ES and AS schools were noted. The recommendation impact would appear to be greater for lunch and for ES schools. This difference might be due to the fact that the ES schools and lunch samples were bigger than those for the AS schools and dinner ones.
A comparison between the nutritional offering and general compliance with the recommendations confirmed that high observance of the guidelines tended to lead to a more appropriate global nutrient offering, especially in limiting the energy and fat offered and promoting the vitamin C offered in school meals. The results confirm the influence of the FFG on the global nutritional offering of school catering. They also suggest that it is necessary to follow all the recommendations to get a balanced diet. Even though no significant evidence of compliance with the general recommendations was observed for dinners, similar trends were found.
Similar benefits of schools meal recommendations have been reported in other studies. Gould et al. (Reference Gould, Russell and Barker29) have shown that English schools with food provisions closest to the guidelines (mandatory nutritional standards) enabled pupils to benefit from superior nutrient intakes. Two others studies have concluded that the school meal guidelines have the same beneficial effect on the nutritional offering(Reference Cullen, Watson and Zakeri13–Reference Nelson, Lowes and Hwang15). However, one study observed a decrease in nutritional intake improvement for the second year. This reversal was explained by students compensating by consuming foods from vending machines rather than (or in addition to) school cafeteria meals during the second year(Reference Cullen, Watson and Zakeri13, Reference Neumark-Sztainer, Story and Perry30).
These findings suggest that complementary actions are necessary to provide healthy and nutritious meals at school. Government policies or recommendations for school cafeteria dietary offerings do influence the students' choice of food but they have to be implemented for all food environments (through education on nutrition or changes in the whole school environment (including vending machines)) so that students cannot change to a different source of foods, thus getting around any dietary changes(Reference Briefel, Crepinsek and Cabili11, Reference Cullen, Watson and Zakeri13). The third school nutritional dietary assessment in the USA has notably shown a significant reduction in the energy offering from schools that did not have any snack bars or vending machines. The reduction was also observed in schools with unique menus or fewer food choices available in the cafeteria(Reference Briefel, Crepinsek and Cabili11). In France, food vending machines have been prohibited(Reference Martin31). It has also been suggested that fruits should be provided free of charge during morning and afternoon breaks as an alternative, to increase their consumption. Meanwhile, other recommendations for morning snacks have been developed in France(Reference Lafay, Boclé and Kalonji32). Consequently, the whole school environment is taken into account. More than merely taking positive measures to foster the access to healthy food or educate children on nutrition and cooking, the French authorities are considering using negative actions such as restricting the food proposition (through FFG) or source of foods at school. Both methods have been found to be complementary and necessary. Nevertheless, the recommendations are first established to balance dietary offering. Thereby, education and training of the person in charge of school meals design is necessary as a first step to success. An international collaborative study has concluded that there is a real and urgent need for national and international health and nutrition programmes to improve the feeding habits of adolescents and help them choose healthier foods(Reference Vereecken, Bobelijn and Maes14, Reference Nelson, Lowes and Hwang15, Reference Howerton, Bell and Dodd33, Reference Vereecken, De Henauw and Maes34). Designing effective nutrition interventions for children and appropriate educational strategies is a step forward(Reference Hoelscher, Evans and Parcel35–Reference Robinson-O'Brien, Story and Heim38) and limiting the temptation is another. The integration of positive measures such as nutritional educational programmes, cooking or garden-based education has been shown to foster healthier eating behaviour(Reference Martin31, Reference Perez-Rodrigo and Aranceta36–Reference Robinson-O'Brien, Story and Heim38) and is necessary to make the restrictive recommendations easy to understand. As the environment has been shown to influence food choices and intake, improvements will also be needed in the dining room, to make school mealtimes a more attractive, social and enjoyable experience(Reference Briefel, Crepinsek and Cabili11, Reference Gould, Russell and Barker29, Reference Birch39–Reference Story, Neumark-Sztainer and French42).
However, even though the effect of recommendations on the nutrients on offer was confirmed, an unbalanced dietary offering in the macronutrients (excess in proteins, fat and lack of carbohydrates) was observed. On the other hand, adequate dietary offerings in Ca, Fe and vitamin C were shown. These findings suggest that improvements are needed to get schools to comply with the FFG by offering students the nutritional requirements. A US study also found similar results, still with excess in fat but with most of the schools following the nutritional requirements for Ca, Fe and vitamin C as well(Reference Crepinsek, Gordon and McKinney43).
Moreover, in France, in 2002 a study on meals eaten by children at school found similar results concerning the excess of fat and proteins but better amounts of Ca, Fe and vitamin C from schools meals have been found since(Reference Lafay, Volatier and Martin44). This last observation confirms improvements in the dietary offering in school meals over the past few years and the need for compliance with nutritional recommendations. Nevertheless, this last study focused on nutrient intakes whereas our results were based on the nutrients available from the meals proposed. This could also largely explain the improvement in the nutritional offering for Ca, Fe and vitamin C. Hopefully, children's dietary intake in school canteens could be soon estimated from the INCA 2(Reference Birch4), and it will then be possible and relevant to confront offering and consumption. Indeed, when given the choice, students generally choose less healthy options. Consequently, better nutritional results are found when the offering is considered than when the analysis concerns food consumed directly by the students(Reference Nelson, Lowes and Hwang15, Reference Fox, Gordon and Nogales45). Furthermore, the present study focused on nutrients that were identified directly associated to one or several FFG. Therefore, this analysis excluded other micronutrients (vitamins and minerals), fibres and saturated fat that were previously shown to be consumed higher in schools meals as compared to home meals(Reference Lafay, Volatier and Martin44). On the contrary, Na intakes were higher in the meals taken at school. Further analyses on dietary intakes at school based on the INCA 2 survey will manage to consider all the nutrient intakes and will enable the trend in dietary intakes over time in French schools to be watched.
In conclusion, the present study showed that a relatively high proportion of schools followed the FFG. However, improvements are still needed to get a proper and balanced dietary offering for school meals. As these findings confirm the effectiveness and relevance of the recommendations as a way of positively affecting the nutritional offering, full compliance with the FFG will help to improve the dietary offering. Besides, forthcoming legislation based on the last version of FFG drawn up by the GEMRCN in 2007 will compel all caterers to provide healthier meals, which should hopefully increase the compliance of school meals with the recommendations and therefore make their dietary offering healthier.
The present survey was supported by grants from the Ministry of Health within the context of the National Nutrition and Health Programme. It was supervised by a steering committee including the Ministry of Health, Ministry of Agriculture, Ministry of Education, a state high school in Paris and the French Food Safety Agency. There were no conflicts of interest while preparing this paper. M. B. and C. D. designed the study, analysed data and wrote the paper. L. L., J.-L. V. and C. D. designed the survey and contributed to the data collection. G. C.-T. defined the food nomenclature and assigned to the food items collected in the survey to categories in the nomenclature. We are grateful to Ian Margo for revising the manuscript.