Parental dietary patterns and social determinants of children ’ s dietary patterns

1 Universidade Federal do Rio de Janeiro, Instituto de Nutrição Josué de Castro, Departamento de Nutrição Social e Aplicada. Av. Carlos Chagas Filho, 373, Edifício do Centro de Ciências da Saúde, Bloco J, 2o andar, Cidade Universitária, 21941-902, Rio de Janeiro, RJ, Brasil. Correspondência para/Correspondence to: R SALLES-COSTA. E-mail: <rosana@nutricao.ufrj.br>. 2 Universidade Federal Fluminense, Hospital Universitário Antônio Pedro, Serviço de Nutrição. Niterói, RJ, Brasil. 3 Universidade do Porto, Instituto de Saúde Pública, Unidade de Pesquisa Epidemiológica. Porto, Portugal. Support: Conselho Nacional de Desenvolvimento Científico e Tecnológico Grant CT-Agrobusiness (MCT/CNPq/MDS-2003), and Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro (Process no E-26 103.296/2011). ORIGINAL | ORIGINAL


I N T R O D U C T I O N
Infant feeding is an important health determinant in the first year of life.Evidence suggests that the food behavior developed in this phase can track into adulthood 1 .Encouraging healthy food habits is a crucial education tool to prevent chronic noncommunicable diseases later on (e.g., type 2 diabetes, obesity, and cardiovascular diseases) as their prevalence has been increasing in the last few decades 2 .
Traditionally, epidemiological studies have tried to establish associations between the intake of nutrients or food groups with health outcomes 3 .However, since people do not consume foods or nutrients separately 4 , and food practices are modulated by social and cultural factors, principal http://dx.doi.org/10.1590/1678-98652016000400004component analyses have proven to be more effective in describing food intake in the form of dietary patterns.These can be defined as the set of foods or food groups consumed by a given population, for which principal component analysis is the most widely used method for their identification 5 .
Few studies have evaluated the dietary patterns of children under the age of three 6,7 .In the United Kingdom a population-based cohort study with 7,052 participants aged 6 months and 5,610 aged 15 months found an association between children's dietary patterns and the family's socioeconomic context, as well as the influence exerted by parental dietary patterns, given that parents were responsible for household food availability and quality 8 .
The identification of dietary patterns in nutritional epidemiology opens a new perspective towards more effective health promotion strategies, with nutritionally, socially, and culturally contextualized practices.This study aimed to identify the dietary patterns of children aged up to thirty months, and to verify whether they are associated with parental dietary patterns, and socioeconomic and demographic characteristics.

M E T H O D S
This is a population-based cross-sectional study with a representative sample of households in Duque de Caxias city, located in the metropolitan area of Rio de Janeiro, Brazil.Details on sample design were previously described by Barroso et al. 9 .Households with at least one child aged six to thirty months were considered eligible for the study.A total of 402 of the 1,085 investigated households were eligible: 366 (91.0%) completed the children's food intake questionnaires, and 365 (90.8%) completed the parental food intake questionnaires (mother or father).
The interviews were carried out between May and December 2005.The team of interviewers consisted of fifteen community residents with complete high school and five dietitians, who collected information about the children's and parental food intake.All interviewers were trained by professionals with experience in the administration of structured questionnaires on socioeconomic and demographic characteristics.The group of dietitians received training to collect food intake data.Both the 24-hour recall (24HR) and the Food Frequency Questionnaire (FFQ) were administered to measure the food intake of the children and parents, respectively.The FFQ was previously validated by Sichieri & Everhart 10 .
The study was approved by the Research Ethics Committee of the Institute of Social Medicine of the Universidade do Estado do Rio de Janeiro (number 02/2004).All participants signed an informed consent form.
Principal component analysis determined the children's dietary patterns, which were grouped by age: 6 to 17 months (n=149) and 18 to 30 months (n=216).Previous pattern analysis was performed by child's sex (not showed).Since the results were similar, the study evaluated the patterns of both sexes together.The Kaiser-Meyer-Olkin test ≥0.60 estimated the applicability and adequacy of the analytical method in relation to the dietary data.Varimax orthogonal rotation was used to improve data interpretation.The number of retained factors was defined by: 1) inflection point of the curve in the graph of scree plot; and 2) eigenvalues higher than 1.0.Foods or food groups with loading factors higher than 0.25 were considered significant for the characterization of each pattern.Negative loading indicated an inverse association, and positive loads, a direct association.Cronbach's alpha verified the internal consistency of each extracted factor.The patterns were named according to the foods or/and food groups they contained.
A structured questionnaire to collect socioeconomic and demographic characteristics was administered to the heads of 366 (91%) households with at least one child: monthly household income per member; number household dwellers (n); age (years) and parental education level (years of schooling).
The parental dietary patterns in the last year were obtained by a FFQ composed of 82 items, with the portions usually consumed and eight intake frequency options, ranging from three times a day to never/almost never.The intake of each food (portion*frequency) was scanned, with subsequent calculation of total daily intake.
The age at complementary feeding introduction was also considered an explanatory variable (in months) reported by the parents.
The children's birth weight (grams) and energy intake (kcal) were considered confounders after univariate analysis (p-value ≤0.20).Total energy intake was estimated from the average of two 24HR, calculated by the software Nutwin (Universidade Federal de São Paulo, São Paulo, Brazil) 11 .Information on foods that were not listed in the software (5% of the study foods) was taken from the Brazilian  13 .

R E S U L T S
The households had a mean of 4.9 (Standard Deviation -SD=1.7)members, and the mean monthly household income per member was 93.9 United States dollars (USD, SD=132.3).The mean children's birth weight was 3,165 grams (SD=576.7),and complementary feeding was introduced at 5 months of age (SD=4.3)(Table1).
Principal component analysis identified, in children aged 6 to 17 months, three dietary patterns that explained 34.7% of the total variation.The pattern named basic-mixed explained 14.5% of the intake variation and included rice, beans, bread, butter and oil, eggs, sausages and fats, and cookies and cake.The second pattern was labeled mixed-plus and explained 10.2% of the intake variation.This pattern included fruits, nonstarchy vegetables, grains and tubers, meats, cheeses and yogurt, candies, chocolate milk and sugary beverages.The third pattern, named milk and flours, explained 10.0% of the variation and was characterized by milk, baby cereals, and sugar, and negatively associated with coffee intake.In children older than 17 months, two dietary patterns were identified: the basic-mixed pattern (rice, beans, eggs, breads, oil and margarine, and sugary drinks) and the mixed-plus pattern (grains and tubers, fruits, non-starchy vegetables, meats, cheeses and yogurt, sausages and fats, biscuits and cakes, and candies), which explained 12.9 and 11.5% of the variation, respectively (Table 3).
In children aged 18 to 30 months, age at complementary feeding introduction (β=0.07;p=0.01) was associated with the basic-mixed pattern.The parental traditional dietary pattern (β=-0.09;p=0.03) was negatively associated with the children's mixed-plus pattern (Table 4).
Multivariate linear regression for children under 18 months, adjusted for energy intake and child birth weight, found a positive association between family income (β=0.002;p≤0.01) and the mixed-plus dietary pattern.Complementary feeding (β=0.108;p=0.004) was positively associated with the basic-mixed dietary pattern.In children up to 18 months, parental education (β=0.368;p≤0.01) was positively associated with the mixed-plus dietary pattern, and age at complementary feeding introduction (β=0.062;p=0.047) was positively associated with the  basic-mixed pattern.The parental traditional dietary pattern was negatively associated with the children's mixed-plus pattern, with β=-0.187(p<0.01) for children aged 6-17 months and β=-0.120(p=0.016) for children aged 18-30 months (Table 5).

D I S C U S S I O N
The study found three different dietary patterns in children less than 18 months of age and an association with the following explanatory variables: family income, parental dietary intake, and complementary feeding.The milk and flours dietary pattern identified in children less than 18 months of age regarded a monotonous diet.On the other hand, children aged 18-30 months presented only two dietary patterns, including the consumption of some food groups that suggested a greater participation in the family's diet, such as the intake of coffee and a possible substitution of milk and baby cereals for sugary drinks.
It is important to highlight that two mixed patterns were found in both groups of children.The first was named basic-mixed and consisted of basic foods, such as rice and beans, and also of processed foods, such as cookies, cakes, butter, and oils.The other dietary pattern, named mixedplus, was characterized by fruits, vegetables, and yogurt, but it also included unhealthy foods (candies, sugary drinks, sausages, and salty products).This pattern was similar to the one found in Brazilian adults by Marchioni et al. 4 and Bezerra & Sichieri 14 , which revealed that the intake of healthy foods does not necessarily exclude the intake of energy-dense or nutrient-sparse foods.
Children under 30 months of age are already susceptible to the early introduction of processed foods and high intake of bread, biscuits, cakes, commercially processed products, and sausages, similar to the diet of adult Brazilians 15 .These results are corroborated by Toloni et al. 16 , who found early introduction of candies, instant noodles, sausages, soft drinks, and drink mixes in children aged four to twenty nine months from public daycare centers in São Paulo/Brazil.
Kudlová & Schneidrová 17 examined the feeding patterns of 1-5-year-olds from Prague and verified that the intake frequency of milk and dairy products, fruits, vegetables, and poultry decreased significantly with age, and that the intake frequency of smoked meat and meat products, sweets, and deep-fried foods increased significantly with age.Also, a study from the United States reported that 46% of the children aged seven to eight months had already consumed some type of candy or sugary drink 18 .Similarly, Rangan et al. 19 found that 24% of the total daily intake of Australian two-and three-year-olds consisted of energy-dense foods, such as soft drinks, fatty and sweet cookies, and potato chips.
The reduction of milk intake and inclusion of sugary drinks (soft drinks and drink mixes) as children age is another aspect to consider.A diet high in sugar and fat is associated with childhood overweight and obesity 20 , as well as diets low in fruits and non-starchy vegetables 21,22 .
Socioeconomic and demographic factors also determined the dietary patterns of the study sample.Monthly household income per member was associated with the mixed-plus dietary pattern in both groups of children.Children from families with higher income had a more varied diet.This result is corroborated by D'Innocenzo et al. 23 , who studied children aged 4 to 11 years from Salvador/Brazil and found that having better socioeconomic level resulted in greater dietary diversity with the presence of health foods, like fruits, green vegetables, legumes, grains, fish, milk, and dairy products, among others.The Family Budget Survery 24 also reported a similar trend: the intake of dairy products, fruits, nonstarchy vegetables, animal fat, and soft drinks increased with family income.
The positive association between parental education level and the intake of fruits and vegetables in children aged 18 to 30 months in the mixed-plus pattern is confirmed by the literature 25,26 .On the other hand, as this pattern was also characterized by foods not indicated for the study age group, having higher education level did not necessarily lead to healthy food intake per se.Parents with a higher education level tend to have higher purchasing power, giving their children access to a more varied diet, which may, therefore, justify this finding. /dx.doi.org/10.1590/1678-98652016000400004 /dx.doi.org/10.1590/1678-98652016000400004

Family income 1 Parental 2
Note: * Multivariate linear model: adjusted for energy intake and birth weight. 1 Monthly household income per member (1.0 Real = 2.0 USD); Dietary pattern of 366 children's parents; 3 Age at introduction of complementary feeding.Rev. Nutr., Campinas, 29(4):483-493, jul./ago., 2016 Revista de Nutrição http://dx.doi.org/10.1590/1678-98652016000400004 Table of Food Composition 12 , of the Universidade Estadual de Campinas, Brazil.Initially, bivariate linear regression evaluated associations with socioeconomic and demographic variables (monthly household income per http://dx.doi.org/10.1590/1678-98652016000400004member; number of household dwellers; parental ages and education levels); parental dietary patterns, age at complementary feeding introduction, and children's dietary patterns.Variables with p<0.20 were included in the multivariate model.The models were adjusted for energy intake and child's birth weight.The significance level was set at 0.05.All the analyses were performed by Stata 13.0 (Stata Corporation, College Station, Texas, United States)

Table 1 .
Socioeconomic and demographic characteristics of the children and parents from 366 households.Duque de Caxias (RJ), Brazil, 2005.
Note: 1 Monthly household income per member (1.0 Real = 2.0 USD); 2 Parental energy intake collected from the food frequency questionnaire; 3 Children's energy intake calculated from the 24-hour recall; 4 Age at introduction of complementary feeding.M: Mean; SE: Standard Error.

Table 2 .
Foods and factor loading for the dietary patterns of 366 children's parents.Duque de Caxias (RJ), Brazil, 2005.

Table 3 .
Foods and factor loading of the dietary patterns of children aged 6 to 17 months (n=149) and 18 to 30 months (n=217).

Table 4 .
Univariate linear regressions of the dietary patterns of children aged 6 to 17 months (n=149) and 18 to 30 months (n=217).