Introduction
The types of food offered in the first 1000 days of life, especially breast milk, have an influence on food choices throughout life and protection against the development of cardiometabolic disorders, Reference Eshriqui, Folchetti and Valente1 a phenomenon known as metabolic programming. Reference Wang, Shu and Sheng2 This is because early exposure to certain environmental factors influences gene expression, thus affecting the emergence of chronic diseases in adulthood. Reference Paes, Gonçalves and Terra3 Human milk enables the child’s first contact with real food and ensures that their nutrition is adequate to their needs in early life. 4 Its composition, in addition to nutrients essential for the baby, includes hormones, especially leptin, which plays an essential role in controlling food intake and consequent regulation of nutritional status. Reference Chan, Goruk and Becker5,Reference Palou, Picó and Palou6
Infant formulas have a greater amount of protein, which may favor the deposition of fat in the first years of life due to the increase in insulin and Insulin Growth Factor 1 (IGF-1) secretion, unlike breastfeeding, as breast milk has a lower protein content, which contributes to optimal weight gain and fat deposition in childhood. Reference Melnik7 Its effective practice also restricts the intake of age-inappropriate foods, which are rich in sugars, sodium, and saturated fats. Moreover, breastfeeding allows for better control of the portions ingested according to the baby’s own hunger and satiety signals due to hormones present in breast milk, which shapes long-term physiological processes responsible for maintaining energy balance; these hormones are absent in infant formulas. Reference Gonsalez, Retondario and Bricarello8,Reference Rito, Buoncristiano and Spinelli9
The recommendations of the World Health Organization (WHO) on the practice of exclusive breastfeeding (EBF) during the first 6 months and continued breastfeeding up to 2 years of age or over coincide with this critical period. 10 The global epidemiological analysis indicates that breastfeeding decreases as the child grows and its prevalence is higher in less developed regions, justified by the lower socioeconomic level and resulting restricted access to other types of food, making human milk the main food source in early life. Reference Flores, Nunes and Neves11 In contrast, babies from wealthier families have a higher consumption of infant formulas and other types of milk than breast milk, possibly due to their marketing at the national level. Reference Neves, Barros and Baker12 In 2019, the global rates of EBF and continued breastfeeding was 42% and 65%, respectively, considering that there is a modest improvement for EBF, with an increase of only 7% points, but has remained unchanged in certain geographical locations, such as Latin America, since 2005. 13 In the same year, following the trend of breastfeeding rates similar to those in Latin America, only 52.7% of Brazilian children under 6 months of age were on EBF and 53.1% of Brazilian children between 6 and 24 months, on complemented breastfeeding. 14
The duration of breastfeeding is associated with food consumption throughout life. Reference Hamulka, Zielinska and Jeruszka-Bielak15 For instance, in a study including 403 children aged 4–7 years, EBF for <4 months was associated with a lower consumption of fruits and vegetables, classified in the category of unprocessed and minimally processed foods. This was linked to the detrimental greater inclusion of ultra-processed foods; Reference Fonseca, Ribeiro and Andreoli16 both food groups were organized according to the NOVA classification, which considers the level of food processing for its categorization. Reference Monteiro, Cannon and Levy17
To date, limited studies have evaluated the association between early childhood eating practices and food consumption in adults, especially according to the NOVA classification, due to the need for longitudinal studies. Therefore, a cohort of young nutritionists and nutrition undergraduates can be advantageous since they are (or are training to become) health professionals who deal with dietary assessment on a daily basis and could apply their knowledge and expertise to improve memory biases of their mothers, consequently enhancing information precision on their early childhood nutrition. This information precision is essential for understanding how physiological processes can influence food consumption throughout life and help prevent associated diseases.
We hypothesized that events occurring in the first 1000 days of life have an influence on adequate food consumption during adulthood. Based on this, the present study aimed to evaluate the relationship between childhood eating practices and current consumption of unprocessed and minimally processed as well as processed and ultra-processed foods in young adults using data from the Nutritionists’ Health Study (NutriHS) cohort.
Methods
Study design and population
The cross-sectional study used baseline data from the NutriHS conducted in Fortaleza, Ceará; the data were collected from 2018 to 2019. Volunteers signed the Free and Informed Consent Form, and the study was approved by the Ethics Committee of State University of Ceará (CAAE n° 95402618.3.0000.5534).
The NutriHS is a longitudinal study conducted with a cohort of undergraduates in nutrition and nutritionists graduated from three universities campi in Brazil (the University of São Paulo, in Sao Paulo city, the State University of Campinas, in Limeira city, and the University of Fortaleza/State University of Ceara, in Fortaleza city). Data on general health was collected through electronic and self-applied questionnaires (including sociodemographic data, current health, early life factors, childhood and current food consumption, and morbid personal and family history) via an online platform (http://www.fsp.usp.br/nutrihs). Reference Folchetti, Silva and Almeida-Pititto18
The minimum sample size for the NutriHS was previously estimated considering the detection of a correlation coefficient of ≥0.20 (alpha: 5% and beta: 20%), with an addition of 20% to cover for follow-up losses. Thus, a minimum of 233 volunteers were required for each center. Reference Hulley, Cummings and Browner19,Reference Valente, De Almeida-Pititto and Archanjo20 For the present study, the same parameters were considered for sample size estimation.
The recruitment for potential subjects was carried out by placing advertisements around campus and inviting students and alumni by email messages. From all the recruited subjects, it was estimated that 61% accessed the website and at least partially filled the self-applied questionnaire. All undergraduates in nutrition and newly graduated nutritionists, ageing 18 years or more at baseline were included, and those that were pregnant at baseline, with missing breastfeeding and FFQ data, or who presented a dietary energy intake above 5,000 kcal/day were excluded (Fig. 1), comprising 392 participants.
Flowchart of selection of participants included in the present study, Fortaleza, Brazil.
Health data
Volunteers who met the eligibility criteria were asked to complete questionnaires using the e-NutriHS system (http://www.fsp.usp.br/nutrihs), which were presented in blocks of questions. Sociodemographic data (age, sex, skin color, family income, and smoking status), events that occurred in the first 2 years of life, current food consumption and nutritional status through self-reported body weight and height for the calculation of body mass index (BMI), calculated by dividing weight (in kilograms) by height squared (in meters), were investigated and classified according to the World Health Organization cutoff values. 21 Previous studies indicate that self-reported weight and height are accurate for calculating BMI in adults. Reference Olfert, Barr and Charlier22,Reference Hodge, Shah and McCullough23
Past events data collection: first 2 years of life
Participants were asked to fill part of the e-NutriHS questionnaire with aspects concerning their first 2 years of life. To obtain the information in a more reliable way, the volunteers were instructed to consult their mothers or their birth cards. In particular, the following information was collected: breastfeeding practice and duration, introduction of infant formula or other types of milk, and consumption of fruits and vegetables during childhood.
The practice of breastfeeding regardless of duration was categorized as no or yes, and the duration of breastfeeding was categorized as <6 months of life, 6–24 months of life, and ≥24 months of life. Type of feeding during the first 6 months of life (main exposure) was classified as not breastfed, introduction of solid, semi-solid or soft foods before 6 months, and EBF. The introduction of infant formula or other types of milk was categorized as follow: <1 month of life of life, 1–3 months of life of life, 3–5 months of life of life, and ≥6 months of life. The habit of eating fruits and vegetables during childhood was categorized as yes or no.
Current food consumption data
Current food consumption, the main outcome variable, was assessed using a previously validated quantitative food frequency questionnaire (FFQ). Reference Selem, Carvalho and Verly-Junior24 The quantitative FFQ contemplates the eating habits of individuals over a period of 1 year, identifying the usual individual portion of 136 food items, with options for frequencies (0–10 times), time units (day, week, month, or year) and portions (described in homemade measures, grams, and millimeters). The data were processed in the Nutrition Data System for Research (Nutrition Coordinating Center, University of Minnesota, Minneapolis). 25
Food consumption was assessed according to the NOVA classification, which categorizes foods into four groups according to their level of processing and their processing purpose – unprocessed and minimally processed foods, processed culinary ingredients, processed foods, and ultra-processed foods. Reference Monteiro, Cannon and Moubarac26
According to the FFQ, food items were classified into 74 unprocessed or minimally processed foods, 3 processed culinary ingredients, 19 processed foods, and 40 ultra-processed foods.
From this classification, the percentages of each group’s contribution to the daily energy intake were calculated, based on previous studies. Reference Berti, Rocha and Curioni27,Reference Louzada, Martins and Canella28 Then, nutritional indicators were used to assess the relationship between current food consumption and the duration of EBF. The nutritional indicators chosen were carbohydrates, dietary fiber, added sugar, lipids, saturated fats, and proteins, based on the United States Department of Agriculture (USDA) National Nutrient Database for Standard Reference 25 , and for which the WHO has established recommendations for the consumption for specific populations, which are independent of sex, age group, and physiological status. 29,30
Data analysis
Descriptive variables are expressed as absolute and relative values or mean and standard deviation. For the association between childhood eating practices (dependent variable) and current food habits (independent variables) a multiple linear regression model was used, selection of co-variates was based on literature well-known confounding factors. The linear trend between food consumption and NOVA classification and EBF was assessed by using linear regression analysis after adjusting for sex, age, and BMI. The beta estimate and 95% confidence interval (95% CI) for the association between current energy consumption from the NOVA classification food groups and breastfeeding practice and eating practices in childhood were evaluated by linear regression analysis using an unadjusted model (crude) and a model adjusted for age (continuous, years), sex (men or women), BMI (continuous, kg/m2), skin color/self-declared race (white or black/brown or others), use of dietary supplement (yes or no), family income (<1 or 1–5 or 6–10 or >10 Brazilian minimum wages – equivalent to 257 USD per month), and smoking habit (yes or no).
For multiple models only participants with no missing data for any of the variables were included.
Statistical analyses were performed using the Stata (version 12.0, 2011, StataCorp LP, College Station, TX). P < 0.05 was considered significant.
Results
Among the 392 participants (average age: 23.9 ± 5.9 years) with complete breastfeeding and FFQ data, 83.2% participants were women, 56.1% participants self-reported having black/brown skin color, and 60.5% participants had family income of 1–5 minimum wages. The mean BMI was 24.2 ± 4.1 kg/m2, and 60.5% participants were classified as eutrophic. Regarding the events that occurred in the first 2 years of life, 87% participants were breastfed, and among them, 46.7% participants received EBF for 6 months (Table 1).
Description of the sample of young adults from the NutriHS, Fortaleza, Brazil
* Value expressed in average.
** Value expressed in standard deviation.
*** Sample reduction due to missing information (Duration of Breastfeeding: n = 367; At how many months was formula or milk introduced: n = 346; As a child, did you have a habit of eating fruits and vegetables?: n = 385).
The current average daily energy intake was 2565.6 ± 898.9 kcal, with 62% intake from unprocessed and minimally processed foods, 3.4% intake from processed culinary ingredients, 19.6% intake from processed foods, and 15.1% intake from ultra-processed foods (Table 1). No significant linear trend was observed between mean food consumption in adulthood among the categories of EBF practice (not breastfed, introduction of solid, semi-solid or soft foods before 6 months, and EBF) (Table 2).
Association between food consumption and type of feeding during the first 6 months of life in young adults from the NutriHS, Fortaleza, Brazil
* Linear regression model adjusted for sex, age, and BMI. Values are expressed as mean and standard deviation.
Participants who did not have the habit of eating fruits and vegetables during their childhood consumed less unprocessed and minimally processed food (β = −3.76; 95% CI = −6.70 to −0.82) (Table 3) and more processed culinary ingredients (β = 0.84; 95% CI = 0.06–1.64) (Table 4) and processed foods (β = 2.62; 95% CI = 0.04–5.20) in adulthood (Table 5) than those who had the habit of eating fruits and vegetables during their childhood.
Association between current energy consumption from unprocessed and minimally processed foods and the practice of breastfeeding and childhood eating experiences in young adults from the NutriHS, Fortaleza, Brazil
* Linear regression model adjusted for age, sex, BMI, skin color/self-declared race, use of dietary supplement, family income, and smoking status.
Values in bold indicate significant coefficients. 95% CI: 95% confidence interval.
Association between energy consumption from culinary ingredient foods and breastfeeding practice and childhood eating experiences in young adults from the NutriHS, Fortaleza, Brazil
* Linear regression model adjusted for age, sex, BMI, skin color/self-declared race, use of dietary supplement, family income, and smoking status.
Values in bold indicate significant coefficients. 95% CI: 95% confidence interval.
Association between current energy consumption from processed foods and breastfeeding practice and childhood eating experiences in young adults from the NutriHS Fortaleza, Brazil
* Linear regression model adjusted for age, sex, BMI, skin color/self-declared race, use of dietary supplement, family income, and smoking status.
Values in bold indicate significant coefficients. 95% CI: 95% confidence interval.
Among the breastfeeding practices evaluated in the present study, adults who were breastfed but started introduction of solid, semi-solid or soft foods before 6 months had a higher consumption of processed foods (β = 4.30; 95% CI = 0.56–8.04) than those who were not breastfed (Table 5). No difference was observed for other breastfeeding practices in relation to current consumption of unprocessed and minimally processed foods (Table 3), processed culinary ingredients (Table 4), and ultra-processed foods (Table 6).
Association between energy consumption from ultra-processed foods and breastfeeding practice and eating experiences in childhood in young adults from the NutriHS Fortaleza, Brazil
* Linear model adjusted for age, sex, BMI, skin color/self-declared race, use of dietary supplement, family income, and smoking status.
Values in bold indicate significant coefficients. 95% CI: 95% confidence interval.
Participants who were fed infant formula or other types of milk between 3 and 5 months of life had a lower current consumption of ultra-processed foods (β = −3.09; 95% CI = −6.12 to −0.06) than those who were fed infant formula or other types of milk before 1 month of life (Table 6).
Discussion
The present study is the first study to assess the relationship between early childhood eating practices and food consumption in adulthood, considering levels of food processing according to the NOVA classification.
We found that breastfed young adults who started introduction of solid, semi-solid or soft foods before 6 months of age had a higher average consumption of processed foods than those who were not breastfed. Participants who did not have the habits of eating fruits and vegetables during childhood consumed less unprocessed and minimally processed foods in adulthood than those who had the habit of eating fruits and vegetables during infancy. Volunteers fed infant formula or other types of milk between 3 and 5 months of age had a lower current consumption of ultra-processed foods than those fed infant formula or other types of milk before 1 month age of life.
In addition, we found that 34.7% energy intake was from processed and ultra-processed foods, with 15.1% intake from ultra-processed foods alone. According to the 2017–2018 Household Budget Survey, the percentage of consumption of ultra-processed foods was 14.8% in Brazil. 31 A comparison of our results with the nationally representative data revealed that the daily intake of ultra-processed foods in the study population was similar that in the Brazilian population. As our study cohort included Nutrition Course undergraduates and recent graduated nutritionists, it was expected that the consumption of ultra-processed foods would be lower as they tended to a follow diet in greater agreement with the Food Guide recommendations for the Brazilian Population Reference da Saúde32 due to their greater knowledge about the health impacts of nutrition. However, the average of 15.1% does not represent a high consumption of ultra-processed foods.
Our results showed that breastfed volunteers who were started introduction of solid, semi-solid or soft food before 6 months of life had a higher consumption of processed foods in young adulthood than those who were not breastfed. In a birth cohort study in Pelotas, located in Brazil’s South, children who started introducing food before 4 months of life consumed more snacks and treats, predominantly coffee and bread, at 6 years of age than those who started food after 4 months of life. Reference Santos, Assunção and Matijasevich33 Acknowledging that the present study chose to work with different reference categories for breastfeeding practices, the Pelotas cohort study findings suggest that introduction of solid, semi-solid or soft foods before recommended starts with the supply of processed foods, such as dairy products and bread, Reference Maciel, Moraes and Soares34,Reference Wang, van Grieken and van der Velde35 even in a region of the country different from ours. This increases the chances for the consumption of these food items throughout life, which could contribute to unhealthy eating habits when in excess.
Participants who did not have the habit of eating fruits and vegetables during childhood had the lowest current consumption of unprocessed and minimally processed foods and the highest consumption of processed foods. This corroborates with the findings of the Avon Longitudinal Study of Parents and Children, conducted in southwest England, which demonstrated that vegetable consumption at an older age was directly related to their smaller intake at age of 13 years. Reference Lauzon-Guillain, Jones and Oliveira36 Our findings are also in accordance with results of a population-based study of young adults, the EAT Project (Food and Activity in Adolescents and Young Adults), conducted in Minnesota, which found that selective feeding in childhood predisposed to lower dietary quality in young adults, characterized by lower intake of fruits, vegetables and whole grains and more frequent consumption of drinks sweetened with sugar and preparations rich in sodium. Reference Pesch, Bauer and Christoph37 These findings lead to the conclusion that, when the introduction of healthy foods occurs in an early, varied, and repeated way, it has positive influences throughout life. Reference Hetherington, Schwartz and Madrelle38
Although breastfeeding practices were not related to minimally processed foods in the present study, it is worthy mentioning the relevance of early weaning over adulthood dietary habits, evidenced by other studies, Reference Eshriqui, Folchetti and Valente1,Reference Fonseca, Ribeiro and Andreoli16 as only 46.7% of the studied sample was exclusively breastfed for a minimum of 6 months. Long-term breastfeeding is known to improve the child’s acceptance of new foods Reference Santos, Assunção and Matijasevich33 due to the ability of human milk to transmit different smells and flavors from the mother’s diet, expanding the healthy food variety and minimizing the presence of foods with the highest level of processing.
The same cannot be achieved by offering infant formula owing to its standard flavors. Reference Beckerman, Slade and Ventura39 In addition, the differences in macronutrient profile between infant formula and human milk and the absence of hormones in the former prevent the self-regulation of satiety. Breastfeeding makes it possible to conduct an autonomous feeding as it allows the baby to actively suck the milk and control the amount of milk consumed, Reference Azad, Vehling and Chan40 which promotes responsive signs of hunger and satiety throughout life.
Among the hormones associated with this ability, leptin controls food intake and regulates nutritional status by interacting with its receptor to activate the anorectic branch through the POMC peptide (proopiomelanocortin), with its alpha-cleavage product MSH (alpha-melanocyte-stimulating hormone) and CART peptide (cocaine-anphetamin-regulated transcript), and inhibit the orexigenic branch by means of NPY (neuropeptide Y) and AgRP (agouti gene-related protein). Reference Chan, Goruk and Becker5,Reference Palou, Picó and Palou6
We also observed that participants fed infant formula or other types of milk between 3 and 5 months of age had the lowest consumption of ultra-processed foods in adulthood than those who were fed with human milk substitutes early, that is, before 1 month of life. Similar to our findings, a study including children participating in the Lactation Support Program in Viçosa, MG, found that breastfeeding for <4 months was associated with a higher intake of ultra-processed foods to the detriment exclusion of unprocessed foods from 4 to 7 years old. Reference Fonseca, Ribeiro and Andreoli16 Similarly, in a cross-sectional analysis of data from the NutriHS São Paulo cohort, women who were breastfed for <6 months as babies were less likely to adhere to a dietary pattern consisting of healthy foods such as fruits, vegetables, whole grains, poultry and fish. Reference Eshriqui, Folchetti and Valente41
However, we found that most of the evaluated breastfeeding practices were not associated with current food consumption, in contrast to findings reported in the literature, which indicates that there is a relationship between greater exposure to breast milk and adherence to the prudent dietary pattern in adulthood, which usually consists of fruits, vegetables, oats, lentils, nuts, brown rice, brown bread, and lean meats. Reference Eshriqui, Folchetti and Valente41,Reference Robinson, Ntani and Simmonds42 This finding can be explained by the early introduction of other foods, especially water and water-based foods, such as juices and teas, preventing the study of the real effects of EBF. Consistent with our results, the birth cohort of Terneuzen, the Netherlands reported no significant association between EBF and eating behavior at later ages. Reference De Kroon, Renders and Buskermolen43 As both are studies related to adulthood, other exposure variables may have confounded their results. In the present study, the homogenous level of education and access to information on nutrition found in our sample, being mainly composed of students and nutritionists, may be weakening the association.
Our study evaluated past information, which may imply memory bias. Data on the practice and duration of breastfeeding and food consumption in childhood were recollected by the participants with the help of their mothers. However, previous research has shown that these variables are reported accurately up to 20 years after birth. Reference Li, Ingol and Smith44,Reference Natland, Andersen and Nilsen45 As the average age of participants was the same, the influence of this type of bias on our results may be small. The study cohort, composed of undergraduate students from the Nutrition Course and recently graduated nutritionists, is a strength of our study because they are aware of the importance of providing reliable information on food intake and are more familiar with the tools used for dietary assessment, ensuring internal validity. Therefore, this homogeneity may not allow generalization of results to other populations, interfering with external validity.
In conclusion, breastfed adults who started introduction of solid, semi-solid or soft foods before 6 months of age had a higher average consumption of processed foods than those who were not breastfed. Participants who did not have the habit of eating fruits and vegetables as children consumed less unprocessed and minimally processed foods in adulthood than those who ate fruits and vegetables during childhood. Participants fed infant formula or other types of milk between 3 and 5 months of age had a lower current consumption of ultra-processed foods than those fed infant formula or other types of milk before 1 month of life. Therefore, feeding practices in the first 1000 days of the child’s life seems to influence food choices in the long-term. Prospective studies, with different and bigger samples are desired to confirm the present results and to clarify literature inconsistences regarding associations between breastfeeding practices and adulthood dietary habits.
Acknowledgments
We thank the research team of the Nutritionists’ Health Study for providing the collected data.
Authors’ contributions
Sthefani da Costa Penha contributes to the data collect, interpretation of data and writing. Marina Maintinguer Norde collaborated with the critical review. Antônio Augusto Ferreira contributed to the design of the project, analysis and interpretation of data and critical review.
Financial support
This work was supported by the National Council for Scientific and Technological Development, Brazil (processes 422721/2018-2) and the Cearense Foundation to Support Scientific and Technological Development, Brazil (processes BMD-0008-01356.01.09/21).
Conflicts of interest
None.
Ethical standards
The study was conducted in accordance with the Declaration of Helsinki of 1975 (revised in 2008) and was approved by the Ethics Committee of State University of Ceará.
