ABSTRACT OBJECTIVE To study the association between ultra-processed food consumption and carbon and water footprints of the Brazilian diet. METHODS Cross-sectional analysis on data collected in 2008–2009 on a probabilistic sample of the Brazilian population aged ≥ 10 years (n = 32,886). Individual food intake was assessed using two 24-hour food records, on non-consecutive days. The environmental impact of individual diets was calculated by multiplying the amount of each food by coefficients that quantify the atmospheric emissions of greenhouse gases in grams of carbon dioxide equivalent (carbon footprint) and freshwater use in liters (water footprint), both per gram or milliliter of food. The two coefficients consider the food life cycle ‘from farm to fork.’ Crude and adjusted linear regression models and tests for linear trends assessed the association between the ultra-processed food contribution to total energy intake (quintiles) and the diet carbon and water footprints. Potential confounders included age, sex, education, income, and region. Total energy intake was assessed as a potential mediation variable. RESULTS In the crude models, the dietary contribution of ultra-processed foods was linearly associated with the carbon and water footprints of the Brazilian diet. After adjustment for potential confounders, the association remained significant only regarding the diet water footprint, which increased by 10.1% between the lowest and highest quintile of the contribution of ultra-processed foods. Additional adjustment for total energy intake eliminated this association indicating that the dietary contribution of ultra-processed foods increases the diet water footprint by increasing energy intake. CONCLUSIONS The negative impact of ultra-processed foods on the diet water footprint, shown for the first time in this study, adds to the negative impacts of these foods, already demonstrated regarding dietary nutrient profiles and the risk for several chronic non-communicable diseases. This reinforces the recommendation to avoid ultra-processed foods made in the official Brazilian Dietary Guidelines and increasingly in dietary guidelines of other countries.

ABSTRACT OBJECTIVE To study the association between ultra-processed food consumption and carbon and water footprints of the Brazilian diet. METHODS Cross-sectional analysis on data collected in 2008–2009 on a probabilistic sample of the Brazilian population aged ≥ 10 years (n = 32,886). Individual food intake was assessed using two 24-hour food records, on non-consecutive days. The environmental impact of individual diets was calculated by multiplying the amount of each food by coefficients that quantify the atmospheric emissions of greenhouse gases in grams of carbon dioxide equivalent (carbon footprint) and freshwater use in liters (water footprint), both per gram or milliliter of food. The two coefficients consider the food life cycle ‘from farm to fork.’ Crude and adjusted linear regression models and tests for linear trends assessed the association between the ultra-processed food contribution to total energy intake (quintiles) and the diet carbon and water footprints. Potential confounders included age, sex, education, income, and region. Total energy intake was assessed as a potential mediation variable. RESULTS In the crude models, the dietary contribution of ultra-processed foods was linearly associated with the carbon and water footprints of the Brazilian diet. After adjustment for potential confounders, the association remained significant only regarding the diet water footprint, which increased by 10.1% between the lowest and highest quintile of the contribution of ultra-processed foods. Additional adjustment for total energy intake eliminated this association indicating that the dietary contribution of ultra-processed foods increases the diet water footprint by increasing energy intake. CONCLUSIONS The negative impact of ultra-processed foods on the diet water footprint, shown for the first time in this study, adds to the negative impacts of these foods, already demonstrated regarding dietary nutrient profiles and the risk for several chronic non-communicable diseases. This reinforces the recommendation to avoid ultra-processed foods made in the official Brazilian Dietary Guidelines and increasingly in dietary guidelines of other countries.

RESUMO OBJETIVO: Estimar a pegada de carbono da dieta brasileira e de estratos sociodemográficos dessa população. MÉTODOS: A pegada de carbono da dieta foi estimada com base nos dados de dois registros alimentares de 24 horas, obtidos em 2008 e 2009, de uma amostra probabilística da população brasileira com 10 ou mais anos de idade (n = 34.003) e em coeficientes de impacto ambiental de alimentos e preparações culinárias consumidos no Brasil (gCO2e/kg). Médias com intervalos de confiança de 95% do consumo alimentar (kcal/pessoa/dia) e da pegada de carbono da dieta (gCO2e/pessoa/dia e em gCO2e/2.000kcal) foram calculadas para o conjunto da população e para estratos segundo sexo, idade, renda, escolaridade, macrorregiões e Unidade Federativa. Modelos de regressão linear foram utilizados para identificar diferenças significativas (p < 0,05) na pegada de carbono da dieta de diferentes estratos sociodemográficos. RESULTADOS: A pegada média de carbono da dieta brasileira foi de 4.489gCO2e/pessoa/dia. Foi maior para o sexo masculino, para a faixa etária de 20 a 49 anos e para as regiões Norte e Centro-Oeste, e tendeu a aumentar com a renda e a escolaridade. O padrão de associação da pegada a variáveis sociodemográficas não se alterou substancialmente com o ajuste para diferenças na quantidade consumida de alimentos, exceto por uma redução no excesso relativo da pegada entre homens e pelo aumento no excesso relativo da pegada na região Centro-Oeste. CONCLUSÃO: A pegada de carbono da dieta brasileira excede em cerca de 30% a pegada da dieta humana que poderia atender, simultaneamente, os requisitos nutricionais de uma dieta saudável e a meta global de contenção do aumento da temperatura média do planeta. O padrão de associação dessa pegada às variáveis sociodemográficas pode auxiliar na identificação de alvos prioritários para ações públicas que visem a reduzir os impactos ambientais do consumo alimentar no Brasil.

ABSTRACT OBJECTIVE: To estimate the carbon footprint of the Brazilian diet and of sociodemographic strata of this population. METHODS: Carbon footprint of the diet was estimated based on data from two 24-hour diet records, obtained in 2008 and 2009, from a probabilistic sample of the Brazilian population aged 10 years and over (n = 34,003) and on environmental impact coefficients of food and culinary preparations consumed in Brazil (gCO2e/kg). Means with 95% confidence intervals of food consumption (kcal/person/day) and the carbon footprint of the diet (gCO2e/person/day and in gCO2e/2,000kcal) were calculated for the population as a whole and for strata according to sex, age, income, education, macro-regions and Federative Unit. Linear regression models were used to identify significant differences (p < 0.05) in the dietary carbon footprint of different sociodemographic strata. RESULTS: The average carbon footprint of the Brazilian diet was 4,489gCO2e/person/day. It was higher for males, for the age group from 20 to 49 years and for the North and Midwest regions, and tended to increase with income and education. The pattern of association of footprint with sociodemographic variables did not change substantially with adjustment for differences in the amount of food consumed, except for a reduction in the relative excess of the footprint among males and an increase in the relative excess of the footprint in the Midwest region. CONCLUSION: The carbon footprint of the Brazilian diet exceeds by about 30% the footprint of the human diet, which could simultaneously meet the nutritional requirements of a healthy diet and the global goal of containing the increase in the planet's average temperature. The pattern of association of this footprint with sociodemographic variables can help identify priority targets for public actions aimed at reducing the environmental impacts of food consumption in Brazil.

RESUMO OBJETIVO: Estimar a pegada de carbono da dieta brasileira e de estratos sociodemográficos dessa população. MÉTODOS: A pegada de carbono da dieta foi estimada com base nos dados de dois registros alimentares de 24 horas, obtidos em 2008 e 2009, de uma amostra probabilística da população brasileira com 10 ou mais anos de idade (n = 34.003) e em coeficientes de impacto ambiental de alimentos e preparações culinárias consumidos no Brasil (gCO2e/kg). Médias com intervalos de confiança de 95% do consumo alimentar (kcal/pessoa/dia) e da pegada de carbono da dieta (gCO2e/pessoa/dia e em gCO2e/2.000kcal) foram calculadas para o conjunto da população e para estratos segundo sexo, idade, renda, escolaridade, macrorregiões e Unidade Federativa. Modelos de regressão linear foram utilizados para identificar diferenças significativas (p < 0,05) na pegada de carbono da dieta de diferentes estratos sociodemográficos. RESULTADOS: A pegada média de carbono da dieta brasileira foi de 4.489gCO2e/pessoa/dia. Foi maior para o sexo masculino, para a faixa etária de 20 a 49 anos e para as regiões Norte e Centro-Oeste, e tendeu a aumentar com a renda e a escolaridade. O padrão de associação da pegada a variáveis sociodemográficas não se alterou substancialmente com o ajuste para diferenças na quantidade consumida de alimentos, exceto por uma redução no excesso relativo da pegada entre homens e pelo aumento no excesso relativo da pegada na região Centro-Oeste. CONCLUSÃO: A pegada de carbono da dieta brasileira excede em cerca de 30% a pegada da dieta humana que poderia atender, simultaneamente, os requisitos nutricionais de uma dieta saudável e a meta global de contenção do aumento da temperatura média do planeta. O padrão de associação dessa pegada às variáveis sociodemográficas pode auxiliar na identificação de alvos prioritários para ações públicas que visem a reduzir os impactos ambientais do consumo alimentar no Brasil.

ABSTRACT OBJECTIVE: To estimate the carbon footprint of the Brazilian diet and of sociodemographic strata of this population. METHODS: Carbon footprint of the diet was estimated based on data from two 24-hour diet records, obtained in 2008 and 2009, from a probabilistic sample of the Brazilian population aged 10 years and over (n = 34,003) and on environmental impact coefficients of food and culinary preparations consumed in Brazil (gCO2e/kg). Means with 95% confidence intervals of food consumption (kcal/person/day) and the carbon footprint of the diet (gCO2e/person/day and in gCO2e/2,000kcal) were calculated for the population as a whole and for strata according to sex, age, income, education, macro-regions and Federative Unit. Linear regression models were used to identify significant differences (p < 0.05) in the dietary carbon footprint of different sociodemographic strata. RESULTS: The average carbon footprint of the Brazilian diet was 4,489gCO2e/person/day. It was higher for males, for the age group from 20 to 49 years and for the North and Midwest regions, and tended to increase with income and education. The pattern of association of footprint with sociodemographic variables did not change substantially with adjustment for differences in the amount of food consumed, except for a reduction in the relative excess of the footprint among males and an increase in the relative excess of the footprint in the Midwest region. CONCLUSION: The carbon footprint of the Brazilian diet exceeds by about 30% the footprint of the human diet, which could simultaneously meet the nutritional requirements of a healthy diet and the global goal of containing the increase in the planet's average temperature. The pattern of association of this footprint with sociodemographic variables can help identify priority targets for public actions aimed at reducing the environmental impacts of food consumption in Brazil.