Update of the acquisition parameters of the Brazilian National School Feeding Program based on the Dietary Guidelines for the Brazilian Population

The Dietary Guidelines for the Brazilian Population is acknowledged as a powerful inducer of public food and nutrition policies. In this perspective, this article presents the methodological path and evidence that supported the elaboration of the new parameters of food acquisition of the Brazilian National School Feeding Program (PNAE). This elaboration involved the analyses of: (1) participation of federal resources used to purchase food, grouped according to the NOVA classification, used in Dietary Guidelines for the Brazilian Population , by the set of Brazilian municipalities and according to the classification of the execution (positive or negative); (2) monthly reference menus that were prepared following Dietary Guidelines for the Brazilian Population recommendations; (3) analysis of food acquisition by the sam-pling of 525 municipalities, involving the relative participation of food groups (according to NOVA) in total expenditures and energy and nutritional quality of purchased foods; and (4) analysis of ultra-processed foods that should not be offered in the school environment. We proposed the adoption of the following parameters for the participation of food groups in relation to the total federal resources used in the purchase of food: ≥ 75% of resources for fresh or minimally processed foods; < 20% for processed or ultra-processed foods and < 5% for processed culinary ingredients, as well as the expansion of the list of foods whose acquisition with federal resources from PNAE The possibility of monitoring these parameters by society is a way of complying with one of the principles of the National System of Food and Nutritional Security ( Law n. 11,346/2006 ), a measure of transparency also encouraged in international documents 26 . According to Siqueira et al, in recent years, the PNAE has advanced in its legal framework towards the realization of the human right to food, but it is still necessary to advance in the active involvement of civil society organizations 27 . On the other hand, a low participation of ultra-processed foods in the total of purchased foods was not enough to promote a micronutrient supply or a greater diversity and variety of fresh or minimally processed foods in the list of acquired foods when compared to a greater participation of ultra-processed foods. This second finding was corroborated by the analysis of the menus prepared by CECANE: even though prepared according to the Dietary Guidelines for the Brazilian Population , more than half of them did not reach an “adequate” classification for diversity/variety. The results referring to the supply of micronutrients differ from those found in studies that investigated the nutritional quality of the diet according to the participation of ultra-processed foods in individual food consumption databases. In them, an inverse association was observed between consumption of ultra-processed foods and intake of micronutrients such as B12, iron, zinc, among others 28,29,30,31 . A possible expla-nation for this divergence is the fact that, in our study, the estimate of the micronutrient supply was made based on data of acquisition and not of food consumption. This method may not have been sensitive enough to capture this association. On the other hand, our findings corroborate those of other studies, which show low diversity/variety of the Brazilian diet, regardless of the participation of ultra-processed foods 32,33 . This low diversity/variety also seems to be occurring in the process of acquiring food from PNAE 34 . In view of these findings regarding the supply of micronutrients and food diversity/variety, the criteria for food acquisition must be complemented by supply criteria that contribute to ensuring the quality of the meals offered.


Introduction
Changes in the epidemiological and nutritional profile of the population and in the way of understanding its determinants require the adequacy and updating of public policies. In this perspective, throughout its decades of existence, the Brazilian National School Feeding Program (PNAE) has been improving its nutritional guidelines. From the publication by the Brazilian Ministry of Health of the first Dietary Guidelines for the Brazilian Population, in 2006 1 , the PNAE adopted parameters so that food purchased with federal resources was aligned with its guidelines. For this, in 2009, it prohibited the acquisition of soft drinks and artificial juices, limited the acquisition of canned foods, processed meats, sweets, compound foods, semi-ready or ready-to-eat food items with high amounts of sodium or saturated fat, and also established a parameter for the minimum supply of fruits and vegetables 2 .
In response to the increase in obesity, chronic diseases and consumption of ultra-processed foods, a new version of the Dietary Guidelines for the Brazilian Population was published in 2014 3 . It adopted as a structuring approach of its guidelines the NOVA classification, which groups food according to the extent and purpose of its industrial processing 4,5 . In view of these new dietary guidelines of the Brazilian Ministry of Health, the Brazilian National Fund for the Development of Education (FNDE) establishes two working groups with a view to updating the nutritional parameters of the PNAE for the acquisition and supply of food. For this, they should propose parameters that would contribute to the achievement of the program's objectives related to the health and food and nutritional safety of schoolchildren and, at the same time, were auditable by their social control institutions.
The working groups were composed of representatives of the FNDE, the Brazilian Ministry of Health, the National Council for Food and Nutrition, the Collaborating Centers on School Feeding and Nutrition (CECANE) 6 , the Federal Council of Nutritionists, the Pan American Health Organization, nutritionists responsible for the PNAE of state and municipal education departments, and specialists from different higher education institutions. The work process was supported by technicians from the Brazilian National Regulatory Agency (ANVISA), the Food Acquisition Program Management Group and the PNAE Advisory Group. The first working group operated from April to November 2017 (hereinafter Phase 1) and the second, from March to September 2018 (Phase 2).
This article aims to present the methodological path and the evidence that supported the work of these two groups regarding the elaboration of new parameters of food acquisition in the PNAE, as well as the proposal resulting from this collective effort. The record of the path, the evidence and the proposed parameters related to the supply of food goes beyond the scope of this article.

• Update of PNAE parameters for food acquisition restriction
In this phase, three groups of analyses were carried out: the first, for the total number of PNAEexecuting entities (municipalities, states, and federal schools); the second, for a set of municipalities according to the level of their performance (positive execution and negative execution) and, the third, for menus prepared by CECANE, as detailed below.
For the three analysis groups, each food was allocated in the NOVA 5 classification groups. Such allocation was made based on the list of ingredients declared on the food label obtained through internet searches on websites of three major supermarket chains. To perform this search, the food nomenclature of the Accountability Management System (SIGPC) was used, for example, chocolate powder. Since SIGPC does not inform the brand of the purchased products, we used the list of ingredients of the food that presented the lowest price among three different brands surveyed, considering that this is the procedure practiced in the bidding processes for the acquisition of food items for schools. To make this allocation, a consensus proposal was sent through an electronic form to the members of the nine entities and bodies represented in the working group, in addition to the CECANE, totaling Cad. Saúde Pública 2021; 37 Sup 1:e00151420 24 specialists consulted. In the event of disagreement, a consensus proposal and its justification were presented to the experts, analyzed and resolved among members of the group.

a) First group of analyses
In the previous legislation 2 , the acquisition of canned foods, processed meats, sweets, compound foods, semi-ready or ready-made preparations and foods with a high amount of sodium or saturated fat was restricted to up to 30% of the federal resources used in the purchase of food for the Program. In order to support the updating of these legislation, an ecological study was developed with secondary data obtained from the SIGPC to identify the participation of each food group, according to the extent and purpose of its industrial processing, in the total federal resources used in the purchase of food for the program. The data refer to the annual values of acquisition of each food by the executing entities of the country between the years 2013 to 2015, the last period available at the time.

b) Second group of analyses
With the SIGPC database, an analysis of the participation of each food group was performed for a set of municipalities according to their level of performance in the program. For this, a municipality was selected by region with the highest number of the following positive implementation criteria: acquisition of foodstuffs from family farming above the minimum percentage provided for by law (30%); accountability approved without reservation by the School Feeding Council (CAE) of the municipality; absence of balance in the specific account of the PNAE beyond what is allowed by the legislation at the end of the year; registration of at least three actions of food and nutritional education in the exercise of 2014; use of own resources for school feeding in percentage equivalent to 50% or more of the amounts transferred by the FNDE; and achieving the target for the initial years of elementary school in the Basic Education Development Index. A municipality was also selected by region of the country with the highest number of the following criteria of negative execution: non-acquisition of food from family farming; accountability not approved by the CAE; complaint regarding the execution of the program from the FNDE ombudsman's office or control body, reprogramming of balance above the allowed value (30%); and registration, in the Food and Nutritional Surveillance System, of prevalence of 50% or more of excess weight in children aged 5 to 10 years. In 2015, 30 municipalities were monitored by the FNDE and characterized according to these criteria. The municipalities included in this analysis were characterized according to number of inhabitants 7 , Brazilian Municipal Human Development Index 8 and number of students enrolled 9 .

c) Third group of analyses
The relative participation of each food group in the total cost of 22 monthly menus prepared by CECANE from the five regions of the country was described according to the following Dietary Guidelines for the Brazilian Population 3 (p. 26) recommendation -"make fresh or minimally processed foods the basis of their food" -that were feasible from the point of view of the complexity of preparing meals in the school environment. Menus were prepared to meet 20%, 30% and 70% of the daily nutritional needs of students from daycare, preschool, elementary and high school. The relative participation of each food group was compared with the national average of its participation in the number of federal resources used in the purchase of food.
The relative participation of food groups in the total expenditure on food of the monthly menus was calculated as follows: (1) price survey carried out on the website of the supply center (CEASA) of the state referring to CECANE or the nearest state, in 2017; (2) for food items in which no price was found on the CEASA website, the lowest search price performed on the Internet was used, as explained earlier. The list of food prices has been validated by the respective CECANE.
The quality analysis of the menus prepared by CECANE was also performed, applying the Quality Index of the Coordination of Food and Nutrition Security (IQ COSAN) 10 to evaluating characteristics such as the food diversity of the menus, presence of regional foods and sociobiodiversity and absence of prohibited foods, incorporated aspects of Dietary Guidelines for the Brazilian Population into Cad. Saúde Pública 2021; 37 Sup 1:e00151420 its evaluation matrix, attributing positive scores to the absence of processed or ultra-processed foods in place of the positive score for the absence of restricted foods. The IQ COSAN score ranges from 0 to 95 points and classifies menus as: inadequate (0-45 points), needs improvements (46-75 points) and adequate (76-95 points).
Based on this set of analyses, a proposal was formulated to update the parameters for restriction of the acquisition of foods that establish maximum percentage values for the acquisition of processed culinary ingredients and processed and ultra-processed foods and minimum percentages for the acquisition of fresh or minimally processed foods.

• Update of PNAE parameters for the prohibition of the purchase of certain foods
In order to update the list of foods whose acquisition with federal resources from the PNAE was prohibited in the resolution in force at the time 2 , the ultra-processed foods purchased in 2015 by the executing entities (municipalities, states and federal schools) of the country, obtained from the SIGPC, were analyzed based on the guidelines of the PNAE, the precepts of Dietary Guidelines for the Brazilian Population and the promotion of adequate healthy food 8 , according to which junk food, sugary drinks and others should not be part of the children's eating routine and, above all, should not be offered in the school environment.
In addition, these foods were analyzed based on the Pan American Health Organization Nutrient Profile Model 11 . Then, we identified those who had excessive sodium content (≥ 1g per 1Kcal), free sugar (≥ 10% of the total energy value -TEV), total fats (≥ 30% of the TEV), saturated fat (≥ 10% of the TEV), trans fat (≥ 1% of the TEV) and also those that contained sweeteners. As the declaration of the number of total sugars is not mandatory on food labels in Brazil, the difference was calculated between the amount of total carbohydrates of the food analyzed with sugar and the sugar-free food available in the 4th edition of the Brazilian  14 .
Based on this set of analyses, ultra-processed foods that should have prohibited acquisition were listed.

Phase 2
The objectives of this phase were: (a) to analyze the relative participation of the food purchased in the PNAE, according to the extent and purpose of its industrial processing, in the total federal financial resources used in its acquisition and in the total energy purchased and (b) to identify the relationship between the relative participation of food groups in these financial resources and the quality of the food purchased. For this, an ecological study was developed with secondary data from SIGPC, referring to the year 2016, the last year available when the analyses were performed. Data were used regarding the quantities of each food and the paid amounts (BRL) for 525 municipalities. For the selection of municipalities, the first draw was made for a state of each region (Acre, Alagoas, São Paulo, Santa Catarina and Goiás) and then, considering the set of municipalities of these states (n = 1,310), a simple random sample of 40% of the municipalities was performed.
The quantity of each food, recorded by the municipalities in different units of measure (e.g.: kg, unit, can, pack), was converted into grams 15 and the inedible fraction was excluded, according to the corresponding correction factors 11 . The nutritional composition of the purchased foods was calculated using TACO 11 and, for foods not available in it, the official table of American nutritional composition was used 13 .
All food items purchased were classified according to extension and purpose of industrial processing, as proposed in Dietary Guidelines for the Brazilian Population, according to the NOVA 3,4,5 classification, as described in Phase 1.
Additionally, for the evaluation of the quality of food provided by the municipalities, the following data were analyzed: the percentage distribution of energy from each macronutrient (carbohydrates, Cad. Saúde Pública 2021; 37 Sup 1:e00151420 proteins, and lipids) in the total energy acquired, the density of micronutrients (amount of micronutrient/total energy acquired by the municipality) (vitamin C, vitamin A, iron and calcium) and the number of fresh or minimally processed foods acquired. The latter variable was constructed from the counting of different types of foods, according to the concepts of diversity 16 and variety 3 , which, to some extent, express the contribution of micronutrients from the diet.
For the analysis of the Phase 2 data, in addition to the mean values obtained for the set of municipalities analyzed, these were divided into quarters according to relative participation of energy from ultra-processed foods in the total number of foods purchased, with the first quarter (Q1) being the one with the lowest participation.
We estimated the relative mean participation (%) of energy and financial resources used for each of the four food groups for the set of municipalities analyzed, and for each quarter of the participation of ultra-processed foods. The percentage distribution of macronutrients, the density of micronutrients and the average number of fresh or minimally processed foods acquired for the set of municipalities analyzed for each quarter of the participation of ultra-processed foods were also estimated. In addition, for the set of municipalities that met the parameters of relative participation of the different food groups in the total federal resources used for the acquisition of foods proposed in Phase 1 (spending parameters), the percentage distribution of macronutrients and the average number of fresh or minimally processed foods were estimated.

Results
Phase 1

• Update of PNAE parameters for food acquisition restriction
In the period analyzed (2013-2015), the group of fresh or minimally processed foods had a greater participation in the federal resources used in the purchase of total food for the program (67.6%). The average participation of the sum of the groups of ultra-processed and processed foods corresponded to almost one third (29.3%) federal resources spent (Table 1). When compared to the national average participation of each food group (Table 1), the municipalities with positive PNAE execution presented higher values for the participation of fresh or minimally processed foods and lower for processed and ultra-processed foods; the opposite is observed for municipalities with negative execution. The average participation of processed culinary ingredients in both groups of municipalities was similar to the national average (Table 2). We can also observe that, in the first group of municipalities, the average participation of fresh or minimally processed foods was higher than 75%; that the participation processed and ultra-processed, added, was less than 20%; and that that of processed culinary ingredients was less than 5%.

Table 1
Participation (%) of each food group in relation to the total federal resources used in the purchase of food. Brazilian National School Feeding Program    The average participation of the group of fresh or minimally processed foods in the cost of the 22 monthly menus prepared by CECANE (88%) ( Table 3) was higher than the average participation of this group in the total number of federal resources used in the purchase of food for the program ( Table 1). The average participation of processed culinary ingredients (6.7%) also exceeded the national average participation of this group and the sum of the average participation of processed and ultra-processed foods (5.3%) was lower than the national average participation of these groups. Despite the expressive relative participation of fresh or minimally processed foods, of the 22 monthly menus prepared by CECANE, 10 were adequate, 11 needed improvements and one was inadequate with regard to its quality according to the IQ COSAN (Table 3). All menus provided fruits and culinary preparations that could be produced in the school environment, such as rice and beans, baião de dois, arroz carreteiro, feijão tropeiro, maxixada, canjica and curau (data not shown in the table).
Based on this set of results, the working group proposed that the following parameters be adopted for the participation of food groups in relation to the total federal resources used in the purchase of food: ≥ 75% of resources for fresh or minimally processed food; < 20% for processed or ultraprocessed foods and < 5% for processed culinary ingredients. Starting from the average distribution obtained in the studied period, the proposed parameters sought to establish a goal of participation of different food groups that would promote an improvement in the profile of food acquisition within the scope of the PNAE.

• Update of PNAE parameters for the prohibition of the purchase of certain foods
From a total of 390 ultra-processed foods purchased with federal funds from the program in 2015, 160 were identified that should not be offered in the school environment, which were grouped into 13 categories (Table 4). More than half of the total of these foods are powdered or reconstituted products (52.2%), such as powdered milk drinks, powdered soups, canjica mix, cakes, puddings, risotto, and rice pudding. The second main group, which added 14% of the listed foods, refers to concentrates based on guaraná or currant syrup, sweetened fruit drinks with additives and other similar drinks, such as concentrated juice, fruit nectar and drink syrup. We also analyzed the ban on the purchase of margarine and ultra-processed meat products, such as mechanically separated meat, mortadella, hot dogs, meatballs, ham, canned beef, salami, and sausages. Although they have high levels of critical nutrients, the ban could be difficult to implement considering that they are low-cost products that make up or accompany culinary preparations. To avoid the frequent presence of these products on the menus, parameters have been suggested to limit their offer.
Cad. Saúde Pública 2021; 37 Sup 1:e00151420 Table 4 Food categories that must not be purchased in the Brazilian National School Feeding Program (PNAE) and percentage of foods with a high content of critical nutrients and the presence of sweeteners by food category. Brazil, 2015. Complementarily, based on the Pan American Health Organization Nutrient Profile Model, it was identified that all 160 foods had at least one critical nutrient, of which 62% contained a high content of free sugar; 27.2%, high content of saturated fat; 25.9%, high content of total fats; 25.3%, high sodium content; 2.5%, high content of trans fat; and 6.3% contained sweeteners. The details of these results by food category are shown in Table 4.

Categories
Based on these analyses, we propose that the list of foods whose acquisition would be banned with federal resources would then become: soft drinks and artificial soft drinks, already provided for in the Resolution n. 38 of 2009 2 , drinks or concentrates based on guaraná or currant syrup, readymade teas and other similar drinks, fruit drinks with additives or sweeteners, cereals with additives or sweeteners, candies in general, confectionery, bonbons, chocolates in bars or sprinkles, cookies or filled cookies, cakes with icing or filling, cereal bars with additives or sweeteners, ice creams and variants, gelatins, spices with monosodium glutamate or sodium salts, mayonnaise and powdered or reconstituted foods.

Phase 2
The group of fresh or minimally processed foods had a greater relative share of energy (52.9%) in the total of foods purchased for PNAE in the municipalities studied. The group of ultra-processed foods accounted, on average, for almost 30% of the total energy purchased, ranging from 5.1% (for the 25% of municipalities with less participation) to 64% (for the 25% of municipalities) with greater participation). In addition to processed foods, these accounted for 1/3 of the energy purchased (Table 5).
Cad. Saúde Pública 2021; 37 Sup 1:e00151420 Table 5 Mean relative participation (%) of energy and financial resources executed with each of the groups of the NOVA classification, for 525 Brazilian municipalities and according to quarters of energy participation of ultra-processed foods. Brazil, 2016. The comparison of the results referring to the relative participation of ultra-processed foods in the total energy acquired with the participation of these food groups in the total resources employed indicated that the financial approach was related to the nutritional approach (Table 5). Such coherence was reiterated when the relative participation of food groups in total calories was compared with the proposal, originating from Phase 1, of adopting parameters of relative participation of food groups in total expenditures: it was observed that the municipalities located in the first quarter of energy participation of ultra-processed foods (the most desirable situation for the program) presented, on average, a distribution close to that of the proposed parameters.
Regarding micronutrients, the energy share of ultra-processed foods in the total of purchased food items did not influence the density of vitamin C, vitamin A and calcium. The density of vitamin C was lower, but not significantly, in the upper quarters of participation of ultra-processed foods, while vitamin A and calcium did not show a clear relationship with participation of these foods. As for iron, a positive gradient of its density was observed according to the share of ultra-processed foods in the total calories acquired by the municipalities.
The expenditure parameters proposed in Phase 1 were already practiced by 22.5% (n = 118) of the 525 municipalities studied in this Phase. In this group, the average energy distribution from each macronutrient was as follows: 54% carbohydrates, 17.5% proteins and 28.5% lipids. These values are in accordance with the recommendations for the distribution of macronutrients (respectively 55-75% for carbohydrates, 10-15% for proteins, and 15-30% for lipids) 17,18,19 , despite the fact that they refer to a food day and no to the acquisition of food. Considering the expenditure parameter for all food groups, the average number of types of these foods was 34.9, but no differences were found between the municipalities that reached or did not reach the parameters.
Cad. Saúde Pública 2021; 37 Sup 1:e00151420 The finding that the participation of ultra-processed foods did not influence the number of micronutrients or the number of fresh or minimally processed foods purchased (proxy for variety and diversity) led to four recommendations. The first three, which are mandatory, focused on the supply of micronutrients, namely: inclusion of foods containing heme iron at least four days a week and foods containing vitamin A at least three days a week. In the case of non-heme iron source foods, these must be accompanied by facilitators of their absorption, such as food items that are sources of vitamin C. The fourth recommendation is suggestive and focuses on acquisition: that there should be at least 50 different types of fresh or minimally processed foods acquired annually by the municipalities, with a view to promoting greater variety and diversity in school meals.

Discussion
The proposed parameters for food acquisition contribute to the incorporation of the Dietary Guidelines for the Brazilian Population in the PNAE, since they are based on the extent and purpose of industrial food processing. In addition, the relative participation of the resources used in the acquisition of food groups in the total of executed resources was consistent with the relative participation of energy of these groups in the total of purchased energy. Consequently, the establishment of cut-off points for the participation of food groups in the total federal resources used in their purchase that provide for a majority share of fresh or minimally processed foods and limited of processed and ultra-processed foods converge with the Dietary Guidelines for the Brazilian Population recommendations.
We also observed consistency between the proposed parameters and the relative energy distribution of each macronutrient (carbohydrates, proteins, and fats) in the total energy acquired. The relative shares, in the total amount of energy purchased, both from food groups according to the extent and purpose of industrial processing 5 and from macronutrients are important proxies for food quality 20 . It is worth noting that the observed equivalence between the acquisition parameters and the nutritional parameters was due to the food prices practiced in 2016. Considering the tendency of a relative decrease in the prices of ultra-processed foods in relation to fresh or minimally processed foods 21 , the cut-off points of the proposed parameters should be periodically reviewed.
The proposed acquisition parameters proved to be feasible, since, of the 525 municipalities studied in the second phase, 22.5% practiced them. In addition, the menus developed by CECANE proved to be viable for preparing meals. Feasibility is an indispensable attribute of any proposal to improve a public policy, being just as important as its technical-scientific quality, as it will enable its applicability in different realities.
At the same time, if fully practiced, the proposed acquisition parameters will induce the improvement of the PNAE, since the average percentage of financial resources used in the acquisition of fresh or minimally processed foods will increase from 69% to at least 75%.
In addition, it is important not only to recommend minimum amounts for spending on fresh or minimally processed foods, which could be simpler operationally, but also pointing out the limits for spending on processed culinary ingredients and processed foods and ultra-processed, since these recommendations are complementary and act to favor the quality of the food offered by PNAE.
From the perspective of public policy, it is important to establish parameters that induce municipalities and states to improve the nutritional quality of the menus offered in schools, since the PNAE is one of the strategic policies for ensuring food and nutrition security 20 . In addition, schools can be protective factors for obesity and chronic non-communicable diseases 22,23 contributing to a school food environment that promotes healthy eating practices 24,25 .
The proposed parameters are also auditable by the instances of social control. Municipal managers are obliged to submit detailed information on food purchase expenditures to the Schools Food Council. Based on these and with a list of the classification of foods according to NOVA (which can be made available by the FNDE), the counselors will be able to calculate the relative participation of each group of foods in the total resources. This represents an important advance for social control practices, since, until 2020, part of the nutritional parameters of the PNAE was focused on nutrients and depended on computational tools and specialized professionals to verify compliance.
Cad. Saúde Pública 2021; 37 Sup 1:e00151420 The possibility of monitoring these parameters by society is a way of complying with one of the principles of the National System of Food and Nutritional Security (Law n. 11,346/2006), a measure of transparency also encouraged in international documents 26 . According to Siqueira et al, in recent years, the PNAE has advanced in its legal framework towards the realization of the human right to food, but it is still necessary to advance in the active involvement of civil society organizations 27 .
On the other hand, a low participation of ultra-processed foods in the total of purchased foods was not enough to promote a micronutrient supply or a greater diversity and variety of fresh or minimally processed foods in the list of acquired foods when compared to a greater participation of ultra-processed foods. This second finding was corroborated by the analysis of the menus prepared by CECANE: even though prepared according to the Dietary Guidelines for the Brazilian Population, more than half of them did not reach an "adequate" classification for diversity/variety. The results referring to the supply of micronutrients differ from those found in studies that investigated the nutritional quality of the diet according to the participation of ultra-processed foods in individual food consumption databases. In them, an inverse association was observed between consumption of ultra-processed foods and intake of micronutrients such as B12, iron, zinc, among others 28,29,30,31 . A possible explanation for this divergence is the fact that, in our study, the estimate of the micronutrient supply was made based on data of acquisition and not of food consumption. This method may not have been sensitive enough to capture this association. On the other hand, our findings corroborate those of other studies, which show low diversity/variety of the Brazilian diet, regardless of the participation of ultra-processed foods 32,33 . This low diversity/variety also seems to be occurring in the process of acquiring food from PNAE 34 . In view of these findings regarding the supply of micronutrients and food diversity/variety, the criteria for food acquisition must be complemented by supply criteria that contribute to ensuring the quality of the meals offered.
It is important to comment on some limitations of the study. The SIGPC data refer only to acquisitions made with funds transferred by FNDE, disregarding the counterpart of the municipalities. Thus, acquisitions made with own resources were not counted. Although they do not reflect what was actually consumed by students, purchase estimates are accurate to determine the types of food and the amounts spent on school meals. The lack of information on the value of the counterpart from states and municipalities also did not allow estimating the financial impact of the proposed parameters, an important aspect for assessing their feasibility.
Regarding the application of the Pan American Health Organization Nutrient Profile Model, in view of the non-mandatory declaration of the number of total sugars on food labels in Brazil and the unavailability of product brands in the SIGPC, there was an inaccuracy in the estimates of nutritional composition and free sugars in purchased foods. However, in view of the limitation of the labeling legislation in the country and that the study was based on secondary data, a well-founded and reproducible analytical strategy was outlined.
Another relevant aspect concerns the differences in the registration of information from the municipalities with regard to the amount of food purchased. In some cases, measures such as bundle and unit were found without information regarding weight. Based on the literature, a weight standardization was adopted for these measures, but it is not possible to guarantee the precision of the values. Furthermore, it was not possible to assess the daily per capita contribution of micronutrients, since only annual food purchase data were available, with no information on per capita quantities and number of beneficiaries, for example. However, the analysis of the number of types of food purchased can be considered a proxy for variety and diversity, which are associated with the supply of micronutrients.
It is also important to comment on the strengths of the study. The first was the adoption of the NOVA classification, the Dietary Guidelines for the Brazilian Population and the Pan American Health Organization Nutrient Profile Model, consistent theoretical references on a topic of interest to the study. The second concerns the fact that the proposed parameters are based on an empirical basis derived from the PNAE routine, which allowed working with a large volume of data and producing robust evidence for the intended objectives.
As a lesson learned, the experience of reconciling scientific aspects and methodological rigor with the knowledge accumulated in the management experience, allowed us to arrive at a proposal of