Participação crescente de produtos ultraprocessados na dieta brasileira ( 1987-2009 ) Increased contribution of ultra-processed food products in the Brazilian diet ( 1987-2009 )

MÉTODOS: Os dados analisados são provenientes de Pesquisa de Orçamentos Familiares realizadas no Brasil em 1987-1988, 1995-1996, 2002-2003 e 2008-2009. Foram analisadas amostras probabilísticas dos domicílios das áreas metropolitanas em todos os períodos mencionados e, nas duas amostras mais recentes, a abrangência foi nacional. As unidades de análise foram registros de aquisições de agregados de domicílios. Os itens alimentícios foram divididos segundo extensão e propósito de seu processamento industrial em: alimentos in natura ou minimamente processados, ingredientes culinários processados e produtos alimentícios prontos para consumo, processados ou ultraprocessados. A quantidade adquirida de cada item foi convertida em energia. Estimaram-se a disponibilidade diária total per capita de calorias e a contribuição dos grupos de alimentos em cada pesquisa. Calcularam-se estimativas por quintos de renda para as pesquisas nacionais. Variações temporais foram testadas por teste de diferença de médias e modelos de regressão linear.

Increased production and consumption of processed foods and drinks is one of the principal causes of the current obesity pandemic and of non-communicable diseases. 33Studies have shown the link between excessive consumption of these processed products, such as soft drinks, sweets and processed meat, and excessive weight gain. 18,34Longitudinal follow ups of more than ten years in length have shown association between the habit of eating processed food in fast food restaurants and increase in body mass index and in resistance to insulin. 3,23e industrial processing of food is practically ignored in studies on populations' food intake and in dietary recommendations.This is perhaps due to the lack of a classification system which takes into account the type of food processing before being purchased and consumed. 9,14,30classification system for food items was drawn up 9 based on the extent and purpose of the food processing.This classification grouped the food items into three groups: 'in natura' or minimally processed foods; processed culinary ingredients; and ready-to-consume, processed and ultra-processed products.Ultra-processed products are essentially industrial formulations, mainly or totally made from ingredients and containing little or no whole food, whereas processed products are whole foods preserved in salt, sugar or oil. 9,10The relevance of this classification for dealing with obesity and other chronic diseases is reported in the literature. 13,14,30udies in different countries have shown that ready-to-consume, processed and ultra-processed products are more energy dense, have higher sugar, sodium, total and saturated fat content with lower protein and fiber content compared to in natura or minimally processed foods combined with culinary ingredients. 13,14Ultra-processed products have special characteristics which encourage excessive energy intake, such as their being generally served in large portions, their hyper-palatability, their long shelf-life and the ease with which they can be carried around, facilitating the habit of eating between meals and snacking, as well as their aggressive promotion through persuasive marketing strategies. 9,14udies using data from three household budget surveys in metropolitan areas of Brazil between 1987-1988 and  2002-2003 have documented continuous increases in the caloric share of these ready-to-consume products in the diet, together with a decrease in the share of in INTRODUCTION natura or minimally processed foods and of processed culinary ingredients. 12,13A new household budget survey was carried out in Brazil in 2008-2009.This survey offered the possibility of updating the trend of the share of ready-to-consume products in household food purchasing in metropolitan areas of Brazil and, for the first time, to document this trend throughout the country, as the surveys were of a national scope.
The aim of this study was to estimate trends in household consumption of food items in Brazil, taking into account the extent and purpose of their industrial processing.All of the surveys used complex cluster sampling, involving the geographical and socioeconomic stratification of all the census tracts in the country, followed by random selection of tracts in the first stage, and of households in the second.The data collection in all of the surveys was carried out over a 12 month period, in a uniform way in the strata, guaranteeing the representativeness of all four seasons of the year.A detailed description of the sampling process in the four surveys is available elsewhere.a,b,c,d The basic data analyzed in this study refer to the purchase of food items for domestic consumption by the unit of consumption (the household) over seven consecutive days, recorded daily in a notebook by the members of the household or by the IBGE interviewer.

METHODS
The records of food items purchased by households clusters (strata) were used as units of study due to the short reference period for collecting data on the purchase of food items in each household.
The strata of households included in the surveys' sample design were used to study the national evolution of food intake.These strata, homogenous as regards the geographic location of the households and the socio-economic level of the families, added up to 443 in 2002-2003  In the 1987-1988 survey, only a description of the food purchased and the amount spent was recorded, meaning that the quantity purchased (in kg or l) needed to be defined indirectly (based on mean price for the food item recorded by the IBGE in the week and the metropolitan area in question).In the other surveys, the quantity of each food item purchased was recorded directly by members of the household.Food purchased for consumption outside of the home was not recorded with sufficient detail (a simple description of type and cost are available) to be included in this study.
Data on household income and spending -and other information characterizing the residence and its inhabitants -were obtained by trained interviewers using standardized questionnaires.
The total quantities of each food item purchased, after excluding inedible portions, were converted so as to express daily intake values.The total daily purchase of each food item was converted into energy using the Tabela Brasileira de Composição de Alimentos (TACO -Brazilian Table of Food Composition); f if the food item was not included in the table, that of the United States Department of Agriculture (USDA) was used.g The items were grouped according to the abovementioned classification, based on the extent and purpose of industrial processing. 10They were grouped into the three following groups: in natura or minimally processed foods, including rice and beans, meat, milk and eggs, fruit and vegetables, roots and tubers among others; processed culinary ingredients, which are substances extracted from other foods, including oils and fats, flours and sugar; and ready-to-consume products, which could be either processed or ultra-processed.
Processed ready-to-consume products are produced from adding substances such as salt, sugar and oil to whole foods.They include vegetables or legumes preserved in brine, fruits in syrup, fish preserved in oil, salted or smoked fish or meat, and cheeses with added salt.Ultra-processed ready-to-consume products are produced predominantly or entirely from industrial ingredients and have little or no food in their composition.In general, they contain preservatives, cosmetic additives and often added synthetic vitamins and minerals.They include bread, cereal bars, cookies, savory snacks, cakes and baked goods, ice-cream, soft drinks, ready-meals, pizzas, sausages, nuggets, canned and dried soup and infant formula.
We estimated the mean value of available calories per capita for each period studied and the proportion of calories acquired for each of the three groups and their subgroups.The estimates were obtained for all Brazilian households, for household strata according to quintiles of per capita monthly income and for the set of households belonging to the 11 metropolitan areas of the country.
The test of means for independent samples (t-test) was used for each quintile of income in the temporal trend analysis of the national estimates.Statistical significance was tested for the variations using linear regression models for the analysis of temporal trend of estimates restricted to the metropolitan areas.We considered the descriptive level to be 0.05 for statistical significance in all tests.
Weighting factors were used in the analyses to enable the results to be extrapolated to all Brazilian households and for their metropolitan areas.We used the Stata v. 12.1 statistical package for all analyses.

RESULTS
The total daily per capita quantity of energy from the purchased food items corresponded to 1,791 kcal (standard error (se): The caloric share of ready-to-consume products increased significantly (from 23.0 to 27.8%) in the six years between the two surveys, above all among ultra-processed products (from 20.8% to 25.4%).There were significant decreases in the caloric share of in natura or minimally processed foods (1.6 percentage points -pp) and of culinary ingredients (3.2 pp) (Table 1).
The caloric share of all the ultra-processed ready-toconsume products increased, above all for sausages, sugary drinks, sweets, chocolates and ice-cream and for ready meals which tripled their caloric share in the period.Among the processed products there was a small, though significant, growth in the caloric share of cheese and preserved fruit and vegetables.The caloric share of beans (1.0 pp) and milk (0.5 pp) decreased significantly among the in natura or minimally processed foods.The greatest decreases among culinary ingredients were for refined sugar (1.0 pp) and for wheat flour (0.7 pp) (Table 1).
The increase in the caloric share of ready-to-consume products was found in all economic strata, above all with regards ultra-processed products, and tended to be greater among those on lower incomes (Table 2).The caloric share of ready-to-consume products increased in a uniform and significant manner (from 20.3% to 32.1%) over a period of a little more than 20 years (1987 to 2009).There was an increase in ultra-processed products, from 18.7% to 29.6%, and a reduction in the caloric share of in natura or minimally processed foods (5.1 pp) and of processed culinary ingredients (6.8 pp) (Table 3).

The quantity of daily
An increase in the caloric share of ultra-processed products in the diet was observed; for sausages, ready meals, sweets, soft drinks and other sugary drinks, this share more than doubled.The share of in natura or minimally processed foods significantly decreased, except fruit and fish, whose share remained practically constant.Among the processed culinary ingredients, except pasta, the caloric share decreased.Cassava flour and wheat flour had the greatest reductions (1.2 pp and 0.8 pp respectively) (Table 3).

DISCUSSION
The results of this study confirmed the trend of increase in the share of ready-to-consume products in household food purchases in Brazilian municipalities.For the first time, the same trend was documented in the country as a whole and for all income groups.Household budget surveys using the same classification as that used in this study have shown ready-to-consume products with higher shares than that in Brazil: 61.7% of total calories in Canada in 2001 and 63.4% in the United Kingdom in 2008. 14,15The share of in natura or minimally processed foods and of culinary ingredients in the total calories was around half the value observed in Brazil in the two analyses.
Trends similar to those found in this study have been reported in other countries for products classified as ultra-processed.Studies in Europe report increases in the share of ready meals and soft drinks in the diet, and reductions in the share of olive oil, fruit, vegetables and milk. 19,32In Taiwan, an increase in consumption of sweets and sugary drinks was observed between 1993-1996 and 2005-2008. 22There was increased consumption of soft drinks and other sugary drinks, products which are typically ultra-processed, in Mexico City 1 and in the USA, h as well as among Chilean women who tripled the caloric share of sugary drinks in their diet between 1999 and 2006. 26creased share of ready-to-consume products in the diet is not limited to the Brazilian population.The substitution of in natura or minimally processed foods and culinary ingredients by ready-to-consume products is a 20 th century phenomenon in developed countries.In Canada, the share of ready-to-consume products increased from 28.7% in 1938-1939 to 61.7% in 2001; ultra-processed products increased from 24.4% to 54.9% 14,16 The same phenomenon can be seen in developing countries from the 1980s onwards. 11e rapid expansion of the share of ready-to-consume products in Brazil and other developing countries, such as China and Mexico, can be explained by changes in the food systems of these countries, due above all to national economic growth and transnational food industries (Big Food) penetrating the local markets. 11,28he increased supply of products such as soft drinks and snacks in the globalized world, accompanied by a reduction in the relative price of these products, caused the gradual replacement of traditional diets based on foods with diets composed of processed and ultra-processed products.The increased share of ready-to-consume products, even among the poorest levels of the Brazilian population may be explained by real growth in household income, especially for families on low income.j Ready-to-consume products typically have high energy density and are rich in sugar and fat and poor in fiber. 13,14hese characteristics are considered to be risk factors for obesity, 24,29,33,k and the increased consumption of these products may be one of the explanations for the growing trend of the prevalence of being overweight in the Brazilian population.The percentage of overweight adults in Brazil increased from 24% in 1975-1975 to 49% in 2008-2009, while the prevalence of obesity in adults tripled in the same period.l In addition to their unfavorable nutritional profile, ready-to-consume products have characteristics   , 1987-1988, 1995-1996, 2002-2003 and 2008-2009.

Period
Group 1987-1988 1995-1996 2002-2003  which facilitate the habit of eating between meals and which encourages excess calorie consumption.The sensory characteristics 5,35 of these products, along with their aggressive marketing campaigns contribute to explaining the accelerated growth of these products in Brazil.Their practicality and because they need to or next to no preparation are attractive to a population with a lifestyle dominated by lack of time. 8e data collected for the household budget surveys are not sufficiently detailed on the consumption of food items outside of the home.Another limitation is that food waste was not considered, i.e., the portion of purchased food not actually consumed by the household.Ready-to-consume products tend to be less perishable than in natura or minimally processed foods.It is possible that the share of ready-to-consume products has been underestimated.However, this situation does not substantially change the growth trend identified in this study.
The search for strategies to reduce or slow the expansion of ready-to-consume products is essential.Among them, educational activities on food and nutrition, based on dietary guidelines which emphasize the adoption of eating patterns based on in natura or minimally processed foods are necessary in order to reduce and prevent non-communicable diseases. 17No less important are activities to regulate the relative prices of food items and to regulate the marketing of ready-to-consume products (especially for the children's market). 6,7,21The strategies adopted so far have not proved adequate, such as the inefficient self-regulation by the industry with regards advertising aimed at children. 25 The effects of increasing tax on soft drinks, 2 and energy dense 20 food items, and of those high in saturated fat 31 on reducing the prevalence of excess weight and obesity 27 and reducing calories purchased for consumption 20 have been shown.These studies need to be applied in the form of effective public policies which act in conjunction with other activities in order to modify the food system as a whole.
The presence of ultra-processed products increased exponentially in Brazilian eating habits, raising the need for an in-depth investigation of their impact on the population's health.The majority of data available, both on the trends and on the impact of these products on diet and health are restricted to a limited number of items, but not to processed and ultra-processed products overall, this being a recent method of classifying food items.New studies to investigate the influence of regional, cultural and socio-economic characteristics and culinary practices on eating habits are needed, using the classification of food items based on the extent and purpose of the industrial processing.
The data used in this study are from the surveys carried out by the Brazilian Institute of Geography and Statistics (IBGE) in four periods: March 1987 to February 1988, October 1995 to September 1996, June 2002 to July 2003 and May 2008 to May 2009.Representative samples of a set of Brazilian households located in 11 metropolitan areas (Belem in the North; Fortaleza, Recife and Salvador in the Northeast; Belo Horizonte, Rio de Janeiro and Sao Paulo in the Southeast; Curitiba and Porto Alegre in the South and the Federal District and the municipality of Goiania in the Midwest) were studied in the four periods of the study.The total sample of households in these areas was 13,611 in 1987-1988, 16,014 in 1995-1996, 13,848 in 2002-2003 and 15,399 in 2008-2009.The metropolitan areas together represented around 1/3 of households and of the Brazilian population.a,b,c,dThe two most recent household budget surveys had a national coverage and the total number of households was 48,470 in2002-2003 and 55,970 in 2008-2009.

Table 1 .
Distribution of the share (%) of food, culinary ingredients and ready-to-consume products in the total consumption of calories purchased byBrazilian households, 2002-2003 and 2008-2009.
SE: Standard errorNumbers in bold highlight the three large groups and are not included in the total sum.a p < 0.05 for the t-test for means of independent samples.b Grains (other than rice and beans), nuts and seeds (unsalted), seafood, coffee, tea, condiments and dried eggs.c Other flours, starches, sugars and sweeteners and coconut milk.

Table 3 .
Distribution of the share (%) of food items, culinary ingredients and ready-to-consume products in total consumption of calories purchased by Brazilian households, metropolitan areas of Brazil

21.0 0.8 26.1 0.8 29.6 a 0.7
SE: Standard errorNumbers in bold highlight the three large groups and are not included in the total sum.a p < 0.05 for the t-test for means of independent samples.b Grains (other than rice and beans), nuts and seeds (unsalted), seafood, coffee, tea, condiments and dried eggs.c Other flours, starches, sugars and sweeteners and coconut milk.
Considering the trend for spending on food consumed outside of the home to grow (from 24.5% of total spending in 1987-1988 to 33.1% in 2008-2009 in metropolitan areas, d and from 24.1% in 2002-2003 to 31.1% in 2008-2009 in the country),dit is likely that the total share of calories consumed outside of the home is also growing.Eating ready-to-consume products tends to be greater outside of the home.It is, therefore, possible that the increased consumption of these products in Brazil identified here is actually even higher.