Hostname: page-component-78c5997874-4rdpn Total loading time: 0 Render date: 2024-11-04T19:11:03.950Z Has data issue: false hasContentIssue false

Determinants of the consumption of ultra-processed foods in the Brazilian population

Published online by Cambridge University Press:  28 October 2024

V. N. C. Silveira*
Affiliation:
Postgraduate program in Public Health, Federal University of Maranhão, São Luís, Brazil
A. M. dos Santos
Affiliation:
Postgraduate program in Public Health, Federal University of Maranhão, São Luís, Brazil
A. K. T. C. França
Affiliation:
Postgraduate program in Public Health, Federal University of Maranhão, São Luís, Brazil Physiological Sciences Department, Federal University of Maranhão, São Luís, Brazil
*
*Corresponding author: V. N. C. Silveira, email [email protected]
Rights & Permissions [Opens in a new window]

Abstract

This article aims to evaluate the sociodemographic determinants of ultra-processed foods (UPF) consumption in the Brazilian population ≥ 10 years of age. The study used data from the personal and resident food consumption module of the Family Budget Surveys, grouping foods according to the NOVA classification of food processing. The classification and regression tree (CART) was used to identify the factors determining the lowest to highest percentage participation of UPF in the Brazilian population. UPF accounted for 37·0 % of energy content in 2017–2018. In the end, eight nodes of UPF consumption were identified, with household situation, education in years, age in years and per capita family income being the determining factors identified in the CART. The lowest consumption of UPF occurred among individuals living in rural areas with less than 4 years of education (23·78 %), while the highest consumption occurred among individuals living in urban areas, < 30 years of age and with per capita income ≥ US$257 (46·27 %). The determining factors identified in CART expose the diverse pattern of UPF consumption in the Brazilian population, especially conditions directly associated with access to these products, such as penetration in urban/rural regions. Through the results of this study, it may be possible to identify focal points for action in policies and actions to mitigate UPF consumption.

Type
Research Article
Copyright
© The Author(s), 2024. Published by Cambridge University Press on behalf of The Nutrition Society

Ultra-processed foods (UPF) are industrialised food products originating from fresh or minimally processed foods with the addition of chemical additives such as flavourings, colourings, sweeteners and other compounds that enhance their sensorial properties and increase their shelf life(Reference dos Santos Costa, de Faria and Gabe1Reference Monteiro, Cannon and Levy3). These products have overt marketing, usually aimed at children and adolescents, and are often associated with practical consumption(Reference Ferreira, Barbosa and Vasconcelos4,Reference da Costa Louzada, dos Santos Costa and Souza5) .

Studies show that the consumption of UPF is also associated with the emergence of non-communicable diseases and conditions such as obesity and cardiometabolic events(Reference da Costa Louzada, Baraldi and Steele2,Reference da Costa Louzada, dos Santos Costa and Souza5Reference Srour, Fezeu and Kesse-Guyot7) . Furthermore, due to water use and waste disposal, the impacts involved in the production and consumption of UPF on the environment are also highlighted(Reference dos Santos Costa, de Faria and Gabe1,Reference Chen, Zhang and Yang8) .

Since 2008, Brazil has conducted a national survey on sociodemographic information and food consumption of the Brazilian population aged ≥ 10 years, the Family Budget Survey (POF)(9). Based on estimates obtained by representative Brazilian surveys, it was observed that the consumption of UPF by the Brazilian population in 2023 was 17·7 %(10), 19·7 % in 2017–2018 and 18·68 % in 2008–2009(Reference da Costa Louzada, da Cruz and Silva11). However, studies that evaluate UPF consumption usually carry out cross-sectional investigations, which cannot indicate the existence of determinants of consumption, as well as interactions between different factors that may indicate different nuances in the consumption of these products. This allows us to identify the sociodemographic determinants of the consumption of UPF in the Brazilian population, a public health issue that negatively impacts the population’s health, as well as the costs of the health sector.

Therefore, robust statistical methods, such as classification and regression trees (CART), appear as an analysis option because they consider the interaction of distinct sociodemographic factors on UPF consumption in the general population. Through this analysis, it is possible to identify consumption patterns across different population groups, making it possible to target guidelines and policies more specifically and understand the relationship between individuals and their diets.

Methods

Ethical aspects

The data in this study come from an open-access information system, therefore eliminating the obligation of prior requests to government bodies or institutions and approval by the Research Ethics Committee.

Design and sample

This is a cross-sectional study with data from the personal food consumption module of the National Food Survey (INA) of the POF, a nationally representative survey that took place between July 2017 and June 2018 in Brazil. It is considered the most complete survey to date in Brazil, and the POF is representative of the Brazilian population and investigated living conditions based on the analysis of their household budgets, food availability and food consumption(9).

The data collection used a complex sampling plan by conglomerates in two stages, with the drawing of census tracts in the first stage and households in the second. The census sectors come from the master sample of the Brazilian Institute of Geography and Statistics (IBGE), grouped into strata of households with high geographic homogeneity in the sector. Data collection occurred throughout the years of 2017 and 2018 divided into four quarters to consider dietary variability and foods from different seasons(9).

The POF involved 46 164 residents aged ≥ 10 years in Brazil. The sample of households was randomly selected, and all individuals within the target age range were invited to participate. With the expansion of the sampling plan, information was obtained from 52,906,759 Brazilian individuals aged ≥ 10 years(9).

Food consumption

The individuals’ food consumption was assessed using two food records applied on two non-consecutive days using the Automated Multiple Steps Method(Reference Steinfeldt, Anand and Murayi12). In various stages, information was collected on all foods consumed the day before the application, their amounts in household measures, preparation method and, for some predetermined foods, information was requested on the addition of ingredients such as sugars, sweeteners and oils.

Foods with amounts considered unlikely or absent were imputed using the similarity matrix method(Reference Andridge and Little13) from variables correlated with the possible amount consumed. The foods were combined with the food codes present in the Brazilian Food Composition Table (TBCA)(14), while the preparations were disaggregated considering the TBCA standardised recipes. Finally, the reported/imputed amount of each food was converted into kJ using the TBCA information.

Subsequently, the foods were classified according to the NOVA classification criteria(Reference Monteiro, Cannon and Levy3) in in natura or minimally processed foods (INMP), culinary ingredients, processed foods and UPF. The classification of the UPF followed the concept that they are industrial formulations obtained through the fractionation of foods from INMP foods(Reference da Costa Louzada, dos Santos Costa and Souza5). Dyes, flavourings, emulsifiers, thickeners and other additives are often added, giving the formulations high palatability and extended shelf life(Reference da Costa Louzada, Baraldi and Steele2).

For this work, the outcome was the percentage share in energy provided by UPF, which was obtained through the equation:

$${\rm{\% UPF}} = {{{{\rm{kJ\;from\;UPF}} \times 100}}\over{{\rm{total\;dietetic\;kJ}}}}$$

Sociodemographic variables

Sociodemographic information was collected using standardised resident information questionnaires. The following variables were used: sex (male/female), age (in years), years of education (in years), home situation (urban/rural) and per capita family income (converted to US dollars at the 2018 exchange rate for comparative purposes). These variables were selected because they are the sociodemographic conditions assessed in the resident information module.

Data analysis

Initially, categorical data were described in absolute (n) and relative (%) frequencies. Continuous variables had their assumptions of normality tested through the asymptotic one-sample Kolmogorov–Smirnov test, which were rejected (P < 0·05), therefore being described in medians and interquartile ranges (IQR). Due to this study using complex sampling, the sample weights were considered in all analyses.

The CART was used to evaluate the determinants of UPF consumption. The CART is a method that divides the data into segments that are as homogeneous as possible relative to the outcome variable (percentage of energy participation of UPF in the individual’s diet)(Reference Carrizosa, Molero-Río and Romero Morales15,Reference Breiman, Friedman and Olshen16) . A homogeneous node is considered one in which all cases have the same value for the outcome, therefore being a terminal node(Reference Breiman, Friedman and Olshen16).

The algorithms usually used to build trees work from top to bottom by grouping independent variables, which allows complex interactions to be established between variables and the outcome without prior specification. Also, the CART algorithm itself determines the ideal cut-off point for identifying risk or protection groups through interaction with one or more variables(Reference Breiman, Friedman and Olshen16).

To analyse the adjustment of the CART, two metrics were used: root mean squared error, which measures the average prediction error committed by the model in predicting the result of observation, and the mean absolute error, an alternative to root mean squared error that is less sensitive to outliers and corresponds to the absolute average difference between observed and predicted results. These measurements were calculated on different subsets of data by the k-fold cross-validation method (k = 10). For each subset, the root mean squared error and mean absolute error were obtained. Namely, the mean mean absolute error was 15·09 with a sd of 0·15 and the mean root mean squared error was 18·61 with a sd of 0·16. These values reinforce a low variability in the ten repetitions performed, therefore confirming the reproducibility, reliability and generalisation of the results obtained.

Analyses were performed in the open-access statistical program R (R Core Team, 2023)(17). The CART was created using the rpart package(Reference Therneau and Atkinson18), and the sample weights were included in this method for sampling expansion. The survey (Reference Lumley19) and gtsummary (Reference Daniel, Whiting and Curry20) packages were also used to describe the population.

Results

We present the sociodemographic and food intake characteristics of the 52,906,759 Brazilian individuals interviewed in Table 1. Most individuals were female (54·1 %) and lived in urban areas (85·0 %) in the southeast region of Brazil (43·2 %). The median energy intake was 6789·6 kJ (1616·4 kcal), and the median contribution from UPF was 37 %.

Table 1. Sociodemographic and food intake characteristics of the Brazilian population ≥ 10 years from the National Food Survey, 2017–2018

IQR, interquartile ranges; UPF, ultra-processed foods.

* Median (IQR); n (%).

The results of the CART are presented in Fig. 1. The home situation, years of education, age (in years) and per capita family income (in US dollars) were selected by the CART to group the mean percentage participation of UPF in the individual’s diet. For the continuous variables, the best cut-off points were determined by the CART algorithm.

Fig. 1. Classification and regression tree of the determinants of ultra-processed food consumption in the Brazilian population ≥ 10 years. From left to right, the colour gradient represents a progressive increase in the percentage contribution of ultra-processed foods. Green tones correspond to lower values, yellow tones indicate intermediate values and red tones signify higher values. UPF, UPF, ultra-processed foods.

The CART algorithm identified eight terminal nodes. The home situation, age and per capita family income were considered the most important predictors, since the people who lived in urban areas were < 30 years and had per capita income ≥ 257 US$ had the highest UPF participation in the diet. The lowest UPF participation was determined by living in rural Brazil and having less education (< 4 years).

The nodes obtained show a growth behaviour in energy values from UPF and their percentage contributions as observed in CART. However, although the values of total energy ingested according to the nodes also show a similar growth behaviour, we can observe that node 3 (lowest proportional UPF consumption) presented a higher and similar total energy consumption when compared with nodes 8 and 9. Furthermore, node 4 (third lowest proportional consumption of UPF) presented energy consumption similar to that observed by individuals from nodes 12 and 14, who had a significantly higher consumption of UPF (Table 2).

Table 2. Total and from UPF energy intake

UPF, ultra-processed foods; IQR, interquartile ranges.

Discussion

Through the results of this work, we can admit that the percentage participation of UPF in the diet of the Brazilian population is more dependent on the individuals’ residence conditions, since the root node that divided the sample was the household situation. It is important to emphasise that Brazilian individuals residing in rural areas still show a large variation in their UPF consumption since years of study ≥ 4 years (node 4) were responsible for an increase of 7·29 p.p.

We suggest that individuals with less education, and consequently lower socio-economic status, may be more dependent on their agricultural production, therefore with lower UPF consumption(Reference Baraldi, Steele and Canella21). At the same time, individuals residing in rural areas with more education may experience greater economic growth and prosperity; however, alone, these conditions are not markers of improvements in eating habits and may even reduce the consumption of healthy foods to the detriment of an increase in unhealthy foods, as occurs in low-income countries, especially in Latin America(Reference Muhammad, D’Souza and Meade22). Furthermore, it is important to highlight that Brazil has a vast use of land for agroecological production, which may be associated with greater consumption of INMP in regions adjacent to agricultural centres(Reference Muhammad, D’Souza and Meade22,Reference Imamura, Micha and Khatibzadeh23) .

On the other hand, residing in an urban environment was the ramification that showed the greatest participation of UPF in the diet of the Brazilian population. Possibly, this is due to easier access in large centres and penetration of these products in markets and snack bars(Reference da Costa Louzada, da Cruz and Silva11,Reference Khandpur, Cediel and Obando24) . However, age was the knot that later divided the population into different categories of consumption, since age is positively associated with nutritional knowledge and better dietary choices, as well as scientific evidence reinforces the inverse association between UPF consumption and the advancement of age as a global standard(Reference Juul, Lin and Deierlein25,Reference Mariath, Machado and Ferreira26) .

In the urban environment, even the node with the lowest dietary contribution of UPF (node 8) was 6·28 p.p higher than the lowest contribution observed in the tree (node 3). Also, the participation observed in node 8 was slightly lower than in node 4 (–1·01 p.p). We showed that, even if the individuals lived in an urban area, the population of node 8 was divided according to their age (≥ 30 years) and years of schooling (< 4), factors that have a strong influence on the eating habits of populations(Reference Baraldi, Steele and Canella21). Individuals in this age group usually have better nutritional knowledge and greater resistance to UPF consumption due to the prudence of their age and greater concern about their health status, as well as having less education, therefore lower socio-economic status, and consuming more INMP foods(Reference da Costa Louzada, Baraldi and Steele2,Reference Mariath, Machado and Ferreira26) .

We hypothesise that a portion of the population at node 8 may represent individuals from the rural area who emigrated to large centres in search of better living conditions, but who may be suffering from food deprivation. We support this hypothesis through the differences in energy consumption between individuals from nodes 3 and 8, the latter of which generally consume less food.

Also, we can observe the influence of schooling in the following nodes, since even being in the same age group (≥ 30 years), having only more years of study, but still, low schooling increased the participation of UPF in the diet by 4·01 p.p. of the population (node 9). We suggest that this may be due to the greater possibility of access to jobs and, consequently, a slight increase in income. However, we hypothesise that the jobs available for this level of education are usually made up of strenuous workloads and long hours, in addition to the time spent commuting between home and the workplace(Reference Pagliai, Dinu and Madarena27). Thus, there is an inclination towards the choice of UPF, since they are more practical to acquire and prepare(Reference dos Santos Costa, de Faria and Gabe1,Reference Monteiro, Cannon and Levy3) .

When the age group ≥ 30 years is maintained, but the level of education increases to ≥ 10 years, we can observe the emergence of the effect of per capita family income on UPF consumption. Since this level of education can cover high school, as well as higher education and postgraduate courses, it was possible to distinguish that those with higher per capita income (node 12) had higher UPF consumption daily. We suggest that higher per capita income may represent a sign of economic prosperity and greater susceptibility to ostensive UPF marketing, as well as increased consumption due to practicality(Reference Muhammad, D’Souza and Meade22,Reference Imamura, Micha and Khatibzadeh23,Reference Moubarac, Claro and Baraldi28) .

Differently from the population ≥ 30 years old, individuals under the age of 30 years did not show consumption variation according to their level of education, but only with per capita family income. Younger age is globally associated with higher UPF consumption(Reference dos Santos Costa, de Faria and Gabe1,Reference da Costa Louzada, da Cruz and Silva11) , with a progressive decrease as age advances, thus justifying the higher dietary UPF contributions at this extreme of the CART. Also, in addition to younger age, per capita income affected UPF consumption since those with higher income had the highest national consumption of these products(Reference Muhammad, D’Souza and Meade22,Reference Moubarac, Claro and Baraldi28Reference Claro, Maia and Costa30) .

In high-income countries, UPF prices are lower and, therefore, more accessible to the general population, but this trend is not followed by middle- and low-income countries(Reference Imamura, Micha and Khatibzadeh23,Reference Moubarac, Claro and Baraldi28) . This behaviour was observed by Claro et al. (Reference Claro, Maia and Costa30), in which the average value of UPF was higher than that of other foods; therefore, there was no economic advantage for most of the population. In some countries, a process of reversal of eating habits and nutritional status of their populations has already started, but Brazil and other middle- and low-income countries do not follow this trend(Reference Popkin, Adair and Ng29).

We emphasise that the two nodes with the highest income (nodes 12 and 15) had the highest UPF shares, corroborating the greater access to these products granted by income(Reference Shim, Shim and Cha31). Furthermore, we highlight that individuals from nodes 11, 12, and 14 have median energy consumption lower than those from node 4, but their average UPF consumption is much higher, indicating that the diet of these individuals is mostly constituted of UPF.

This study has limitations arising from inherent biases of the dietary survey methods used, such as underestimation/overestimation of foods and/or groups of foods and preparations. However, to minimise these potential biases, the instruments used were validated and procedures were carried out to certify the quality and validity of the information obtained. Furthermore, scientific evidence indicates that the consumption of UPF can be mediated by different mechanisms and sociodemographic conditions; however, to our knowledge, the effects of interactions of different sociodemographic conditions grouped on the consumption of these products in the Brazilian population have not yet been evaluated, reinforcing the uniqueness of this work, especially in the largest Brazilian food survey. It is important to highlight that existing research(Reference dos Santos Costa, de Faria and Gabe1,Reference da Costa Louzada, Baraldi and Steele2,Reference Baraldi, Steele and Canella21,Reference Muhammad, D’Souza and Meade22,Reference Khandpur, Cediel and Obando24,Reference Pagliai, Dinu and Madarena27,Reference Shim, Shim and Cha31) on the consumption of UPF and their correlated factors is supported by cross-sectional associations with methods that consider only the consumption outcome, such as regressions. The unique feature of this study allows for a homogenisation of the Brazilian population according to different sociodemographic conditions, thus allowing a classification of individuals according to their higher or lower consumption. It was possible to observe that lower levels of consumption were primarily correlated with fewer factors (education and housing situation), while higher levels led to a gradual increase in the conditions necessary for their consumption. This reinforces the multi-motivated behaviour pattern of human eating, as well as filling specific gaps regarding which population groups are more and less subject to the deleterious effects of UPF.

We concluded that the findings of this study indicate that the Brazilian population aged ≥ 10 years has a different pattern of consumption of UPF. We observed consumption averages ranging from 23 to 46 % that were influenced by residence conditions, demographics and income. The biggest divisor of food consumption was the region of residence, with the urban area responsible for accommodating the highest consumption of UPF.

Unlike high-income countries that show a downward trend in UPF consumption, Brazil follows the nutritional transition pattern of middle- and low-income countries, especially its neighbours in Latin America that have high UPF consumption rates and a reduction of dietary participation of INMP foods. We emphasise that Brazil is the protagonist of food and nutrition education campaigns, as well as the production of educational materials such as the Dietary Guide for the Brazilian Population, but the recommendations are general and aimed at the average population. Through the results of this work, it may be possible to determine focal points for the action of education and intervention policies that minimise or circumvent the rise in the consumption of these products and mitigate the appearance of health problems in the medium and long term.

Acknowledgements

This research received no specific grant from any funding agency, commercial or not-for-profit sectors.

Substantial contributions to conception and design, data acquisition, analysis and/or interpretation: V. N. C. S., A. M. S. and A. K. T. C. F.; drafting the article or revising it critically for important intellectual content: V. N. C. S., A. M. S. and A. K. T. C. F.; and final approval of the version to be published: A. M. S. and A. K. T. C. F. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved: V. N. C. S., A. M. S. and A. K. T. C. F.

The authors declare none.

References

dos Santos Costa, C, de Faria, FR, Gabe, KT, et al. (2021) Escore Nova de consumo de alimentos ultraprocessados: descrição e avaliação de desempenho no Brasil. Rev Saude Publica 55, 13.CrossRefGoogle Scholar
da Costa Louzada, ML, Baraldi, LG, Steele, EM, et al. (2015) Consumption of ultra-processed foods and obesity in Brazilian adolescents and adults. Prev Med (Baltim) 81, 915.CrossRefGoogle Scholar
Monteiro, CA, Cannon, G, Levy, R, et al. (2016) Classificação dos alimentos. Saúde Pública. NOVA. A estrela brilha. World Nutr 7, 2840.Google Scholar
Ferreira, RC, Barbosa, LB & Vasconcelos, SML (2019) Estudos de avaliação do consumo alimentar segundo método dos escores: uma revisão sistemática. Cien Saude Colet 24, 17771792.CrossRefGoogle Scholar
da Costa Louzada, ML, dos Santos Costa, C, Souza, TN, et al. (2021) Impacto do consumo de alimentos ultraprocessados na saúde de crianças, adolescentes e adultos: revisão de escopo. Cad Saude Publica 37, e00323020.CrossRefGoogle Scholar
Neri, D, Steele, EM, Khandpur, N, et al. (2022) Ultraprocessed food consumption and dietary nutrient profiles associated with obesity: a multicountry study of children and adolescents. Obesity Reviews 23, e13387.CrossRefGoogle ScholarPubMed
Srour, B, Fezeu, LK, Kesse-Guyot, E, et al. (2020) Ultraprocessed food consumption and risk of type 2 diabetes among participants of the NutriNet-Santé prospective cohort. JAMA Intern Med 180, 283291.CrossRefGoogle ScholarPubMed
Chen, X, Zhang, Z, Yang, H, et al. (2020) Consumption of ultra-processed foods and health outcomes: a systematic review of epidemiological studies. Nutr J 19, 110.CrossRefGoogle ScholarPubMed
wIBGE (2020) Pesquisa De Orçamentos Familiares : 2017–2018 : Análise Do Consumo Alimentar Pessoal No Brasil. Rio de Janeiro, RJ: Instituto Brasileiro de Geografi a e Estatística – IBGE.Google Scholar
Brasil (2023) Vigitel Brasil 2023: Vigilância de Fatores de Risco e Proteção Para Doenças Crônicas Por Inquérito Telefônico: Estimativas Sobre Frequência E Distribuição Sociodemográfica de Fatores de Risco E Proteção Para Doenças Crônicas Nas Capitais dos 26 Estados Brasileiros e No Distrito Federal Em 2023. Brasília: Ministério da Saúde.Google Scholar
da Costa Louzada, ML, da Cruz, GL, Silva, KAAN, et al. (2023) Consumo de alimentos ultraprocessados no Brasil: distribuição e evolução temporal 2008–2018. Rev Saude Publica 57, 12.CrossRefGoogle Scholar
Steinfeldt, L, Anand, J & Murayi, T (2013) Food reporting patterns in the USDA automated multiple-pass method. Procedia Food Sci 2, 145156.CrossRefGoogle Scholar
Andridge, RR & Little, RJA (2010) A review of hot deck imputation for survey non-response. Int Stat Rev 78, 4064.CrossRefGoogle ScholarPubMed
Universidade de São Paulo (USP) (2023) Tabela Brasileira de Composição de Alimentos (TBCA). Food Research Center (FoRC). São Paulo: Universidade de São Paulo.Google Scholar
Carrizosa, E, Molero-Río, C & Romero Morales, D (2021) Mathematical optimization in classification and regression trees. Top 29, 533.CrossRefGoogle ScholarPubMed
Breiman, L, Friedman, J, Olshen, R, et al. (1984) Classification and Regression Trees. New York: Chapman and Hall/CRC.Google Scholar
R Core Team (2023) R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing.Google Scholar
Therneau, T & Atkinson, B (2022) rpart: Recursive Partitioning and Regression Trees. https://cran.r-project.org/package=rpart (accessed January 2024).Google Scholar
Lumley, T (2023) Survey: analysis of complex survey samples. J Stat Softw 9, 119.Google Scholar
Daniel, DS, Whiting, K, Curry, M, et al. (2021) Reproducible summary tables with the gtsummary package. R J 13, 570.Google Scholar
Baraldi, LG, Steele, EM, Canella, DS, et al. (2018) Consumption of ultra-processed foods and associated sociodemographic factors in the USA between 2007 and 2012: evidence from a nationally representative cross-sectional study. BMJ Open 8, e020574.CrossRefGoogle ScholarPubMed
Muhammad, A, D’Souza, A, Meade, B, et al. (2017) How income and food prices influence global dietary intakes by age and sex: evidence from 164 countries. BMJ Glob Health 2, e000184.CrossRefGoogle ScholarPubMed
Imamura, F, Micha, R, Khatibzadeh, S, et al. (2015) Dietary quality among men and women in 187 countries in 1990 and 2010: a systematic assessment. Lancet Glob Health 3, e132e142.CrossRefGoogle ScholarPubMed
Khandpur, N, Cediel, G, Obando, DA, et al. (2020) Sociodemographic factors associated with the consumption of ultra-processed foods in Colombia. Rev Saude Publica 54, 19.CrossRefGoogle ScholarPubMed
Juul, F, Lin, Y, Deierlein, AL, et al. (2021) Trends in food consumption by degree of processing and diet quality over 17 years: results from the Framingham offspring study. Br J Nutr 126, 18611871.CrossRefGoogle ScholarPubMed
Mariath, AB, Machado, AD, Ferreira, LNM, et al. (2022) The possible role of increased consumption of ultra-processed food products in the development of frailty: a threat for healthy ageing? Br J Nutr 128, 461466.CrossRefGoogle ScholarPubMed
Pagliai, G, Dinu, M, Madarena, MP, et al. (2021) Consumption of ultra-processed foods and health status: a systematic review and meta-analysis. Br J Nutr 125, 308318.CrossRefGoogle ScholarPubMed
Moubarac, J-C, Claro, RM, Baraldi, LG, et al. (2013) International differences in cost and consumption of ready-to-consume food and drink products: United Kingdom and Brazil, 2008–2009. Glob Public Health 8, 845856.CrossRefGoogle ScholarPubMed
Popkin, BM, Adair, LS & Ng, SW (2012) Global nutrition transition and the pandemic of obesity in developing countries. Nutr Rev 70, 321.CrossRefGoogle ScholarPubMed
Claro, RM, Maia, EG, Costa, BVL, et al. (2016) Preço dos alimentos no Brasil: prefira preparações culinárias a alimentos ultraprocessados. Cad Saude Publica 32, e00104715.CrossRefGoogle Scholar
Shim, J-S, Shim, S-Y, Cha, H-J, et al. (2021) Socioeconomic characteristics and trends in the consumption of ultra-processed foods in Korea from 2010 to 2018. Nutrients 13, 1120.CrossRefGoogle ScholarPubMed
Figure 0

Table 1. Sociodemographic and food intake characteristics of the Brazilian population ≥ 10 years from the National Food Survey, 2017–2018

Figure 1

Fig. 1. Classification and regression tree of the determinants of ultra-processed food consumption in the Brazilian population ≥ 10 years. From left to right, the colour gradient represents a progressive increase in the percentage contribution of ultra-processed foods. Green tones correspond to lower values, yellow tones indicate intermediate values and red tones signify higher values. UPF, UPF, ultra-processed foods.

Figure 2

Table 2. Total and from UPF energy intake