SciELO - Scientific Electronic Library Online

vol.30 número4Insegurança alimentar e fatores sociodemográficos em crianças de São José dos Pinhais, Paraná, 2017: estudo transversalCobertura de equipes de saúde bucal na Estratégia Saúde da Família e uso de serviços odontológicos em adolescentes de Mato Grosso do Sul, 2019: estudo transversal índice de autoresíndice de assuntospesquisa de artigos
Home Pagelista alfabética de periódicos  

Serviços Personalizados




  • Não possue artigos citadosCitado por SciELO

Links relacionados

  • Não possue artigos similaresSimilares em SciELO


Epidemiologia e Serviços de Saúde

versão impressa ISSN 1679-4974versão On-line ISSN 2237-9622

Epidemiol. Serv. Saúde vol.30 no.4 Brasília dez. 2021  Epub 24-Nov-2021 

Original Article

Food consumption patterns, overweight and cardiovascular risk: a cross-sectional analysis of the Pró-Saúde Study, Brazil, 2013*

Thalita Fialho da Rocha (orcid: 0000-0002-9284-7780)1  , Cíntia Curioni (orcid: 0000-0002-5160-9567)2  , Eliseu Verly Junior (orcid: 0000-0002-1101-8746)3  , Flávia Bezerra (orcid: 0000-0002-6594-4323)4  , Eduardo Faerstein (orcid: 0000-0002-4027-4896)3 

1Universidade do Estado do Rio de Janeiro, Programa de Pós-Graduação em Alimentação, Nutrição e Saúde, Rio de Janeiro, RJ, Brazil

2Universidade do Estado do Rio de Janeiro, Departamento de Nutrição Social, Rio de Janeiro, RJ, Brazil

3Universidade do Estado do Rio de Janeiro, Departamento de Epidemiologia, Rio de Janeiro, RJ, Brazil

4Universidade do Estado do Rio de Janeiro, Departamento de Nutrição Básica e Experimental, Rio de Janeiro, RJ, Brazil



To identify food consumption patterns and association between overweight and risk of cardiovascular disease.


This was a cross-sectional study with staff of the Universidade do Estado do Rio de Janeiro, Brazil, who took part in the Pró-Saúde Study. Food consumption was investigated using a food frequency questionnaire. Association between dietary patterns (exposure) and overweight and cardiovascular risk (outcomes) was estimated using linear regression.


Among the 520 staff assessed, four dietary patterns were found: ‘ultra-processed’, ‘healthy’, ‘meat’ and ‘traditional’. After adjustment, the ‘meat’ pattern was inversely associated with waist circumference (ß=-1.52 - 95%CI -2.66;-0.39), body mass index (ß=-0.56 - 95%CI -1.01;-0.11), and the Framingham Risk Score (ß=-0.36 - 95%CI -0.64;-0.09).


In view of excess weight, risk of cardiovascular disease and inverse association between the ‘meat’ food consumption pattern and BMI, it is important to conduct further investigations, with non-working groups, with the aim of gaining greater understanding of the health-disease process related to food consumption.

Keywords: Eating; Cardiovascular Diseases; Risk Factors; Overweight


Obesity is growing in Brazil and worldwide, and is considered a global epidemic, a great challenge for public health.1 According to the Brazilian National Health Survey published in 2019, the proportion of obesity among adult individuals had reached 26.8%. Regarding overweight, the prevalence found in adults was 63.3% for women and 60.0% for men.2

A larger reserve of fat in the body, specifically in the trunk, predisposes the occurrence of conditions such as glucose intolerance, plasma lipid profile alterations, and in particular hypertension, all of which are considered risk factors for cardiovascular disease (CVD).3

Diagnostic imaging methods, considered the gold standard for assessing adiposity, are expensive and difficult to implement in epidemiological studies. Among the most adopted anthropometric indicators, the World Health Organization (WHO) recommends the body mass index (BMI) for defining general obesity. However, BMI does not provide information about fat distribution in the body. Waist circumference provides more accurate estimates of intra-abdominal visceral fat. Furthermore, as it is a simple and practical measure it has become widely used in determining abdominal obesity and is capable of predicting cardiovascular risk.3

Because of the need to quantify the effect of each risk factor and to estimate the severity of cardiovascular-related disease, risk scores and guidelines have been developed for CVD prevention and treatment. The Framingham Risk Score (FRS), for example, is widely used to assess an individual's risk of having a coronary event within the next ten years.4

Hypertension, smoking and hypercholesterolemia, risk factors used to calculate this score, are frequent in the Brazilian population, and have prevention as their primary focus, aiming to achieve improved quality of life, change in eating habits, and physical activity. A diet based on fresh and minimally processed foods, as well as regular physical exercise, play this important preventive role, especially in adults with CVD risk factors.5

On the other hand, a diet high in total fat, sugars, synthetic oils, sodium, additives and emulsifiers, and low in fiber, omega-3 and antioxidant nutrients such as vitamin A, C, D and E is associated with higher risk of CVD.6 Studies highlight the relationship between higher carbohydrate intake and increased risk of death, while higher saturated fat intake is associated with decreased occurrence of strokes.7 Indeed, isolated investigations of nutrients underestimate the possibility of interaction between nutrients and other food intake components.8

Several studies have been conducted in order to assess the relationship between food consumption patterns and the risks of developing CVD.8 It is recognized that a dietary pattern rich in fruit, vegetables, fish, and whole grains (defined as the 'healthy' pattern), DASH diet (Dietary Approaches to Stopping Hypertension)9 and the so-called ‘Mediterranean diet’10 reduce the risk of DCV. In turn, patterns of food consumption characterized by high fat and sugar content, or a meat-based diet, especially processed meat, have harmful effects related to increased risk of obesity, type 2 diabetes mellitus and CVD.11

In Brazil, few studies have investigated the relationship between dietary pattern, adiposity measured by dual-energy X-ray absorptiometry and cardiovascular risk factors.12 Inverse association has been found between diabetes mellitus and a dietary pattern characterized by dairy products, as well as the DASH diet associated with metabolic syndrome.12 The typically Brazilian traditional pattern of food consumption has been associated with higher BMI and waist circumference, higher levels of total cholesterol and its LDL-cholesterol and HDL-cholesterol fractions, as well as higher systemic blood pressure.13 Another study found that three out of four patterns, referred to as 'traditional', 'cafeteria' and 'atherogenic', were associated with CVD risk factors.14

Considering that food consumption patterns can be a determining factor in the development of overweight and chronic diseases, the present study aimed to verify association between food consumption patterns, overweight and risk of CVD.


Study design

This was a cross-sectional study within the larger context of a cohort study called the Pró-Saúde Study.


The Pró-Saúde Study is conducted with technical-administrative staff of the Universidade do Estado do Rio de Janeiro (UERJ), and its main objective is to investigate the role of markers of social position and other dimensions related to health in Brazilian adult populations.15


The Pró-Saúde Study began in 1999 and in 2012-2013 a fourth wave of data collection was conducted with a sample of 520 study participants, this being about 1/5 of the study’s baseline population. This sample was randomly selected from within strata relating to sex, age (in years: <50; ≥50), and schooling (<high school education; ≥high school education). The size of the strata corresponded to the proportions observed among baseline participants, while the sample size obeyed the project’s budgetary constraints. Administrative technicians who had retired and those who had moved out of Rio de Janeiro state were excluded. The data were investigated through standardization, and data collection took place between July 2012 and July 2013, carried out by interviewers and technicians who had been trained beforehand.



  • a) Adiposity, assessed according to total body mass, height and BMI, as well as total fat mass, android and gynoid fat mass;

  • b) Cardiovascular risk, measured by waist circumference and the Framingham Risk Score (FRS).

The covariables investigated were socioeconomic profile, represented by age, schooling, and per capita family income.

Data source and measurement

Dietary assessment was based on a validated semi-quantitative food frequency questionnaire (FFQ),16 covering 82 food items or food groups, in previously stipulated quantities and based on household measures or units of food items. Participants were asked to report the frequency of food consumption in the previous 6 months, divided into eight categories, ranging from ≥3 times/day to never or almost never; and the mean amount, which was variable according to each type of food.

Total body fat, total fat mass and android and gynoid region fat mass measurements were taken by means of Dual-Energy X-ray Absorptiometry (DXA), using iDXA Lunar equipment (GE, Health Care) and EnCoRe 2008 software, version 12.20. The scans were performed by the same professional and the equipment was calibrated daily, according to the manufacturer's instructions. During the session, participants were instructed to wear light clothing with no metallic components.

The android region was defined by the upper limit above the cut-off point of the pelvis, at 20% of the distance between the pelvis and the lap cut-off points. The gynoid region was defined by the upper limit below the pelvis line, with a cut-off point 1.5 times the height of the android region.

Android fat mass (AFM) and gynoid fat mass (GFM) were expressed as a percentage of total fat mass (TFM%), obtained via DXA. The android fat mass/gynoid fat mass ratio was expressed in terms of percentage android fat mass divided by percentage gynoid fat mass: AFM%/GFM%.

Waist circumference (WC) was measured in centimeters. Assessment of risk of metabolic alterations associated with obesity followed the criteria proposed by the WHO, based on WC measurement according to sex,17 classified as follows: high risk, where WC is ≥94cm in men and ≥80cm in women; and very high risk, where WC is ≥102cm in men and ≥88cm in women.

Height was measured in meters, using a stadiometer, on which individuals stand barefoot, in an erect position, with their heads positioned according to the Frankfurt plane. Height was measured taking a 90 degree angle to the scale. Total body mass (in kg) was measured using digital weighing scales (Filizola brand). BMI was calculated by the ratio between total body mass (in kg) and height squared (in meters), adopting the cut-off points proposed for adults and for those over 60 years old.

The FRS was used to assess cardiovascular risk. It is based on an algorithm, associating multiple factors in order to estimate the risk of cardiovascular events happening within ten years among individuals not already diagnosed as having coronary disease.18

The risk factors adopted to estimate the SRF were age, sex, presence of dyslipidemia (total cholesterol and HDL-cholesterol), systemic blood pressure, and smoking. Each variable received a rating, and the final sum represented low (less than 10%), medium (10 to 20%), or high (more than 20%) cardiovascular risk. Individuals with a history of diabetes mellitus and CVD were classified as having high risk (more than 20%).

The presence of diabetes mellitus, CVD, and cigarette consumption were assessed by the answers to the questionnaire used in wave 4 of the present study.

Blood pressure was measured using an OMRON® monitor, duly calibrated by the manufacturer. All blood pressure measurements were taken on the right arm, after a five-minute rest, using an armband; they were taken three times, with a two-minute interval between each one, taking the mean of the three measurements.

In order to determine total cholesterol and HDL-cholesterol levels, blood samples were subsequently collected from participating UERJ employees who had fasted for 12 hours. The samples were collected by a trained professional, who used vacutainer® (Becton, Dickinson & Company do Brasil) tubes containing EDTA-Na as an anticoagulant, and tubes without anticoagulant to obtain serum. The sample processing and biochemical analyses were performed at the Laboratório de Lípides (LabLip). Cut-off values established for total cholesterol (<190 mg/dl) and HDL-cholesterol (>40 mg/dl) levels were used.


The following covariables were analyzed with the aim of minimizing possible biases:

  • a) Age categorized in years: under 35; 35-44; 45-49; 50-54; 55-59; 60 or over;

  • b) Sex (female; male);

  • c) Schooling categorized into: up to complete elementary education (including incomplete junior and middle school); complete high school education (including incomplete higher education); complete higher education or above;

  • d)Per capita family income calculated from the midpoint of the reported net income category, divided by those dependent on the income; this quotient was divided by the minimum wage in effect at the time (BRL 622.00) and the resulting value was stratified into 3 categories: less than 3 minimum wages; 3 to 6 minimum wages; more than 6 minimum wages.

Statistical methods

Food consumption pattern

Dietary patterns were identified using principal component analysis (PCA). The food items listed in the FFQ were grouped together according to nutritional similarities or culinary preparations (Box 1). The premises for applying PCA were tested using the Kaiser-Meyer-Olkin (KMO) sample adequacy test, resulting in a value of 0.752 (>0.6 is considered adequate), and the Bartlett sphericity test. The null hypothesis was rejected when the latter test was performed (p<0.001), indicating that the data were adequate for the use of the method. Varimax orthogonal rotation was adopted, and factor loadings equal to or greater than 0.30 were considered as belonging to the patterns. After performing PCA, a standardized variable was predicted (continuous, symmetrical), indicating how much each individual adhered to each pattern identified.

Box 1 Food groups according to nutritional similarity, used to identify dietary patterns, Pró-Saúde Study, Rio de Janeiro, 2012-2013 

Food groups Food items
Rice Rice
Pasta Spaghetti, lasagna, gnocchi or ravioli
Bread and salted biscuits Mini baguette or loaf of bread and salted biscuits
Fruit Orange or tangerine, banana, papaya, apple, water melon, melon, pineapple, mango, lime, passion fruit, grapes
Fresh greens and vegetables Lettuce, kale, cabbage, cauliflower or broccolis, tomato, cucumber, choko, zucchini, pumpkin, carrot, beetroot, okra, French beans, garlic, onion and sweet pepper
Conserved vegetables Palm hearts, olives, maize
Beef and offal Beef, tripe, liver, gizzard, heart and barbecue
Pork Pork
Processed meat and fish Meat or fish conserved in salt, bacon or pork crackling, tinned sardines or tuna
Ultra-processed meat Hamburger, hot dog sausage, sausage, mortadella, luncheon meat, ham and salami
Fresh fowl and fish Fresh fish and chicken
Milk and milk products Cheese, milk, yoghurt and cream cheese
Eggs Eggs
Beans Beans
Fat Butter or margarine and mayonnaise
Confectionary Cake, filled biscuit, sweet biscuit, ice cream, candies, milk-based confectionary, fruit-based confectionary, bar of chocolate or bonbon, powdered chocolate and sugar
Savories and pizzas Pizza, rissole-type savories, deep fried savories
Coffee and infusions Coffee, tea and matte
Soda drinks Cola-based soda drinks and other soda drinks
Natural fruit juice Fruit juice or pulp
Alcoholic drinks Beer, wine and other alcoholic drinks
Snacks French fries, potato crisps or straws, packet snacks, packet peanuts
Pulses Lentils, peas, chickpeas
Cereals Polenta or corn mash
Root vegetables and products Potato or mashed potato, cassava and cassava flour

The continuous variables were expressed in means and standard deviations (SD); while the categorical variables were expressed in percentages.

Among the variables, only TFM% and AFM% reached a 5% significance level when the Shapiro Wilk normality test was applied. However, since distribution was similar to the normal histogram and the sample size was large, linear models were considered for all variables. A linear regression model was estimated for each response variable (TFM%; AFM%; GFM%; AFM%/GFM%; WC; BMI; FRS), in relation to the crude and adjusted explanatory variables (dietary patterns: ultra-processed; healthy; meat; traditional). The explanatory variables were adjusted by the sociodemographic variables (sex; age; schooling; per capita income) and total energy. The model parameters were estimated using the least square model. 95% confidence intervals (95%CI) were used.

STATA version 13 was used in all stages of the analysis.

Ethical aspects

The Research Ethics Committee of the Instituto de Medicina Social, Universidade do Estado do Rio de Janeiro approved the study project in its opinion on Certificate of Submission for Ethical Appraisal No. 0041.0.259.000-11. Its opinion was issued on October 18th 2011. The UERJ staff participated voluntarily. All participants signed a free and informed consent form.


There was no loss of participants during the data collection stages. Of the total individuals studied (n=520), 51.9% were female, 24.6% were between 50 and 54 years old, 54.3% reported having complete higher education or above, and 70.0% had a per capita income of less than three minimum wages (Table 1).

Table 1 General characteristics of technical and administrative staff (n=520) at a university in the state of Rio de Janeiro, Pró-Saúde Study, Rio de Janeiro, Brazil, 2012-2013 

Variables n %
Female 270 51.9
Male 250 48.1
Age (years)
<35 29 5.6
35-44 71 13.7
45-49 115 22.1
50-54 128 24.6
55-59 82 15.8
≥60 95 18.2
Up to complete elementary education 49 9.4
Complete high school education 184 35.3
Complete higher education or above 283 54.3
Per capita income (minimum wages)
<3 364 70.0
3-6 128 24.6
>6 28 5.4
Waist circumference
No risk 114 21.1
At risk 406 77.9
Underweight 20 3.8
Normal 154 29.6
Overweight 346 66.6
Framingham Risk Score (rating)
High 51 9.8
Intermediate 113 21.7
Low 356 68.5

a) BMI: body mass index.

Table 2 shows variables related to adiposity. Among the participants, mean BMI was 27.8 kg/m2 (SD=5.0), TFM was 36.2% (SD=8.3), and the AFM%/GFM% ratio was 0.6 (SD=0.2).

Table 2 General description of the study population (n=520), according to adiposity, Pró-Saúde Study, Rio de Janeiro, Brazil, 2012-2013 

Variables Mean Standard deviation
Total mass (kg) 77.6 15.3
Total fat mass (%) 36.2 8.3
Android fat mass (%) 9.3 1.8
Gynoid fat mass (%) 16.9 3.1
AFM%/GFM% ratioa 0.6 0.2
BMI (kg/m2)b 27.8 5.0
Waist circumference (cm) 97.2 12.4
Framingham Risk Score (rating) 6.5 7.0

a) AFM%/GFM% ratio: android fat mass %/gynoid fat mass %; b) BMI: body mass index.

Risk of cardiometabolic diseases, according to WC, was identified for 77.9% of the study population, while 66.6% were overweight or obese. However, according to the FRS, 68.5% of participants had low risk and 31.5% had intermediate or high risk of developing CVD within 10 years.

Four food consumption patterns were found and referred to as ‘ultra-processed’, ‘healthy’, ‘meat’ and ‘traditional’. The latter pattern included characteristics of a traditional Brazilian diet: ultra-processed (pasta; bread and salted biscuits; fat; confectionary; savories and pizzas; soda drinks; snacks; ultra-processed meat); healthy (fruit; fresh greens and vegetables; fresh fowl and fish; milk and milk products; coffee and infusions; natural fruit juice; pulses); meat (conserved vegetables; beef and offal; pork; processed meat and fish; alcoholic drinks); and traditional (rice; beans; cereals; roots, root vegetables and products). Total variance explained by the four patterns was 37.3% (Table 3).

Table 3 Distribution of food consumption pattern factor loadings (n=520), Pró-Saúde Study, Rio de Janeiro, Brazil, 2012-2013 

Food groups Factors (dietary patterns) identified
Ultra-processed Healthy Meat Traditional
Pasta 0.57 0.03 0.05 0.08
Bread and salted biscuits 0.42 0.25 -0.23 0.25
Fat 0.46 0.06 0.15 0.31
Confectionary 0.64 0.09 -0.14 0.04
Savories and pizzas 0.69 -0.09 0.18 -0.06
Soda drinks 0.54 -0.26 0.14 0.07
Snacks 0.58 0.03 0.21 -0.01
Ultra-processed meat 0.58 -0.07 0.35 0.04
Fruit -0.09 0.69 0.10 0.16
Fresh greens and vegetables -0.16 0.70 0.24 0.17
Fresh fowl and fish 0.08 0.32 0.04 0.08
Milk and milk products 0.17 0.62 -0.22 0.16
Coffee and infusions 0.14 0.45 0.29 0.00
Natural fruit juice 0.04 0.32 -0.20 0.17
Pulses 0.05 0.48 0.16 -0.15
Conserved vegetables 0.13 0.27 0.44 -0.15
Beef and offal 0.18 -0.03 0.56 0.25
Pork 0.14 -0.08 0.64 0.08
Processed meat and fish 0.21 0.15 0.48 0.13
Alcoholic drinks 0.01 0.07 0.49 -0.10
Rice 0.01 -0.09 0.02 0.74
Beans -0.45 -0.01 -0.01 0.78
Cereals 0.15 0.21 -0.01 0.38
Roots, root vegetables and products 0.26 0.18 0.10 0.41
Eigenvalues 2.99 2.41 1.99 1.93
Explained variance (%) 11.95 9.63 7.97 7.73
Total explained variance (%) 37.3
KMOa 0.752
Bartlett sphericity 0.001

a) KMO: Kaiser-Meyer-Olkin Measure of Sampling Adequacy.

Table 4 shows the crude and adjusted coefficients to find association between the food consumption patterns and each of the outcomes. In the bivariate analysis, inverse association was found between the 'ultra-processed' pattern and the Framingham Risk Score (ß=-0.52 - 95%CI -1.01;-0.02), and between the 'meat' pattern and waist circumference (ß=-1.33 - 95%CI -2.42;-0.24) and body mass index (ß=-0.44 - 95%CI -0.87;-0.01). After adjustment, the 'meat' pattern was inversely associated with waist circumference (ß=-1.52 - 95%CI -2.65;-0.38), body mass index (ß=-0.56 - 95%CI -1.00;-0.10) and Framingham Risk Score (ß=-0.36 - 95%CI -0.63;-0.09). As for the other patterns, no association was identified with any of the outcomes assessed.

Table 4 Association between dietary patterns and each of the outcomes assessed (n=520), Pró-Saúde Study, Rio de Janeiro, Brazil, 2012-2013 

Variables Non-adjusted β 95%CIa p-value Adjusted βb 95%CIa p-value
Total fat mass (%)
‘Ultra-processed’ pattern 0.64 (-0.08;1.37) 0.082 0.06 (-0.84;0.96) 0.890
‘Healthy’ pattern 0.38 (-0.34;1.11) 0.303 0.06 (-0.70;0.83) 0.871
‘Meat’ pattern -0.22 (-0.95;0.50) 0.551 -0.42 (-1.17;0.33) 0.275
‘Traditional’ pattern 0.70 (-0.02;1.43) 0.055 0.23 (-0.61;1.07) 0.586
Android fat mass (%)
‘Ultra-processed’ pattern 0.06 (-0.09;0.22) 0.439 -0.02 (-0.21;0.17) 0.833
‘Healthy’ pattern 0.08 (-0.06;0.24) 0.271 0.06 (-0.10;0.23) 0.461
‘Meat’ pattern -0.10 (-0.26;0.04) 0.173 -0.14 (-0.31;0.01) 0.083
‘Traditional’ pattern -0.10 (-0.26;0.04) 0.177 -0.15 (-0.33;0.03) 0.101
Gynoid fat mass (%)
‘Ultra-processed’ pattern -0.11 (-0.38;0.15) 0.415 -0.22 (-0.56;0.10) 0.183
‘Healthy’ pattern 0.11 (-0.15;0.38) 0.394 0.02 (-0.25;0.31) 0.848
‘Meat’ pattern 0.17 (-0.08;0.44) 0.191 0.15 (-0.12;0.43) 0.279
‘Traditional’ pattern 0.22 (-0.04;0.49) 0.104 0.04 (-0.26;0.36) 0.761
AFM%/GFM% ratioc
‘Ultra-processed’ pattern 0.01 (-0.07;0.02) 0.281 0.01 (-0.01;0.02) 0.496
‘Healthy’ pattern 0.01 (-0.01;0.01) 0.792 0.01 (-0.01;0.20) 0.733
‘Meat’ pattern -0.01 (-0.02;0.02) 0.169 -0.01 (-0.02;0.01) 0.132
‘Traditional’ pattern -0.01 (-0.03;0.02) 0.071 -0.01 (-0.03;0.01) 0.162
Waist circumference
‘Ultra-processed’ pattern 0.43 (-0.66;1.52) 0.430 0.06 (-1.29;1.42) 0.924
‘Healthy’ pattern 0.79 (-0.29;1.88) 0.151 0.69 (-0.46;1.85) 0.235
‘Meat’ pattern -1.33 (-2.42;-0.24) 0.012 -1.52 (-2.65;-0.38) 0.009
‘Traditional’ pattern -0.25 (-1.34;0.84) 0.656 -0.41 (-1.67;0.85) 0.522
‘Ultra-processed’ pattern 0.20 (-0.22;0.63) 0.351 0.10 (-0.43;0.64) 0.709
‘Healthy’ pattern 0.29 (-0.13;0.72) 0.182 0.23 (-0.22;0.69) 0.308
‘Meat’ pattern -0.44 (-0.87;-0.01) 0.043 -0.56 (-1.00;-0.10) 0.015
‘Traditional’ pattern -0.10 (-0.54;0.32) 0.638 -0.20 (-0.70;0.29) 0.427
Framingham Risk Score
‘Ultra-processed’ pattern -0.52 (-1.01;-0.02) 0.040 0.11 (-0.21;0.44) 0.483
‘Healthy’ pattern -0.24 (-0.73;0.25) 0.341 0.25 (-0.02;0.52) 0.072
‘Meat’ pattern -0.00 (-0.50;0.49) 0.982 -0.36 (-0.63;-0.09) 0.009
‘Traditional’ pattern -0.09 (-0.58;0.40) 0.724 0.16 (-0.13;0.46) 0.280

a) 95%CI: 95% confidence interval; b) Least square linear regression, adjusted for sex, schooling, per capita income, age and energy (total calorie intake); c) AFM%/GFM% ratio: android fat mass %/gynoid fat mass %; d) BMI: body mass index.


The analyses revealed four food consumption patterns, referred to as ‘ultra-processed’, ‘healthy’, ‘meat’ and ‘traditional’, as well as association between the ‘meat’ consumption pattern and overweight, and risk of developing CVDs according to the FRS.

Although food consumption is influenced by regional and cultural factors, similar food consumption patterns to those we identified have been described by other authors in France,19 China,20 the United States21 and also in Brazil.22 In general, patterns usually recognized as traditional reflect those foods that are most consumed by the population that is being studied and usually cost less.22

The 'meat' pattern, comprised of beef and pork, processed meat and fish, conserved vegetables and alcoholic beverages, was inversely associated with WC, BMI and FRS. This finding contrasts with the literature, which indicates that so-called healthy or prudent patterns are protective against obesity. However, it has been found that the protein content of the diet results in a reduction of carbohydrates, generating a calorie deficit of approximately 6%, which is favorable to weight loss or maintenance.23

Some potential explanatory mechanisms are related to the fact that the consumption of protein-rich foods promotes a negative fat balance, and consequently a reduction in body fat stores. One of these mechanisms is related to the thermal effect of protein, which is greater than that of carbohydrate or fat. In addition, protein generally exerts a greater satiety effect than other macronutrients, partially mediated by a synergistic effect of the sociogenic hormones (glucagon-like peptide-1 [GLP-1] and peptide YY [PYY]) released in the small intestine. Other mechanisms involve reduction of orexigenic hormone secretion (ghrelin) and protein-induced changes in gluconeogenesis, improving glucose homeostasis.24

We found that alcoholic beverage consumption was present in the 'meat' pattern and this is worthy of note. The literature shows that habitual light to moderate alcohol consumption (1 to 2 doses/day) is associated with lower risk of total mortality from diabetes mellitus or CVD. These findings may be related to the action of ethanol in reducing platelet aggregation and its effect in increasing high density lipoproteins (HDL-c), besides the antioxidant action of some beverages.25 However, excessive alcohol consumption (>4 doses/day) is associated with a higher risk of death and presence of CVD.25

The effect of alcohol consumption in relation to body composition is controversial. In a systematic review of studies on the subject, some studies reported that individuals with low to medium alcohol consumption had lower weight, compared to those who never drank and also to those with heavy alcohol consumption, while other studies showed a positive effect of alcohol on BMI.26

An investigation carried out earlier with the same population taking part in the present study showed low alcohol consumption: 41% reported not consuming any type of alcoholic drinks, 41% reported consuming up to one dose per day and 18% reported consuming more than one dose per day.27 Those results corroborate the understanding of the associations we found related to the increase in BMI.

Dietary patterns identified as healthy or prudent include foods such as fruit, vegetables, fresh greens and vegetables, and cereals, associated with better health outcomes. In turn, the so-called processed, or 'western', patterns reflect higher consumption of sugars, fat, sodium and energy, which are inversely associated with health outcomes.28

Studies with the Brazilian population have found a relationship between a traditional dietary pattern (rice and beans) and metabolic complications,22,29 although the present study did not detect this association. A survey of baseline participants of the 2008-2010 Brazilian Longitudinal Study of Adult Health (Estudo Longitudinal de Saúde do Adulto - ELSA-Brasil), identified that the traditional Brazilian dietary pattern (beans, white rice, processed and fresh meat, and beer) was related to greater diagnosis of diabetes mellitus and metabolic syndrome. It is noteworthy that that survey also identified four food patterns, similar to those identified in this study.29 However, authors have observed an inverse association between this traditional pattern and overweight/obesity in women.22 Traditional dietary patterns represent particular characteristics of the populations studied, although the 'traditional' pattern is mostly represented by the Western pattern associated with the prudent or healthy pattern.

This study identified a 'healthy' pattern, consisting of foods such as fruit, fresh greens and vegetables, fresh fowl and fish, milk and milk products, coffee and infusions, natural fruit juice and pulses. Other studies have revealed an inverse association of the healthy pattern with cardiovascular events.8-10 Among Chinese men, an intake pattern characterized by fruit and milk was inversely associated with hypertension. The same pattern was associated with lower blood pressure levels, diabetes mellitus, and lower CVD risk among adults, and with a lower risk of hypertension for women on a DASH diet.30

Quality of food and an active lifestyle are two fundamental conditions for preventing and controlling modifiable risks of CVD and obesity.1 Therefore, knowing the main dietary patterns related to protection or risk of these morbidities is of utmost importance in public health strategies.

The main limitation of this study is its cross-sectional design, which weakens inferences, of a causal nature, regarding associations between food intake patterns, overweight and cardiovascular risk. The participants of the Pró-Saúde Study are civil servants, which makes the occurrence of healthy worker bias possible, given the better conditions of this population for maintaining their health and keeping to a diet, compared to the average population. It should be noted that the instrument used to evaluate food consumption is subject to errors, since it is a retrospective instrument, dependent on the memory of the interviewee, on individual choices, in addition to the quantification and classification of foods, even though it is a validated instrument.16

Although more precise methods may be used to assess abdominal fat, waist circumference should be considered to be a good predictor of cardiometabolic risk, in view of it being a practical and low cost measure, in addition to the use of the FRS for predicting cardiovascular risk.18

This was the first study in Brazil to use a food intake pattern and which, by adopting a food frequency questionnaire related to body composition data, such as total fat mass, android and gynoid fat mass, measured via DXA, the gold standard for anthropometric measurement, overcame limitations of previous studies that used double indirect anthropometry. Observational analyses were performed between consumption patterns and sex; however, the directions of association were unchanged.

Given the high prevalence of overweight and risk of CVD, as well as the inverse association found related to ‘meat’ dietary pattern and BMI, it is important to conduct further prospective investigations to help understand the role of diet in the development of cardiovascular disease.


1. World Health Organization. Healthy diet. Fact Sheet. 2017;(394). [ Links ]

2. Instituto Brasileiro de Geografia e Estatística. Pesquisa nacional de saúde: 2019: atenção primária à saúde e informações antropométricas: Rio de Janeiro: IBGE; 2020. [ Links ]

3. Darsini D, Hamidah H, Notobroto HB, Cahyono EA. Health risks associated with high waist circumference: a systematic review. J Public Health Res. 2020 Jul 3;9(2):1811. doi: [ Links ]

4. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive Summary of the third report of the national cholesterol education program. expert panel on detection, evaluation, and treatment of high blood cholesterol in adults: adult treatment panel III. JAMA. 2001;285(19):2486-97. doi: [ Links ]

5. World Health Organization. Prevention of cardiovascular disease: guidelines for assessment and management of cardiovascular risk. Geneva: WHO; 2007. [ Links ]

6. Miller V, Mente A, Dehghan M, Rangarajan S, Zhang X, Swaminathan S, et al. Fruit, vegetable, and legume intake, and cardiovascular disease and deaths in 18 countries (PURE): a prospective cohort study. Lancet. 2017;390(10107):2037-49. doi: [ Links ]

7. Kang Z-Q, Yang Y, Xiao B. Dietary saturated fat intake and risk of stroke: systematic review and dose-response meta-analysis of prospective cohort studies. Nutr Metab Cardiovasc Dis. 2020 Feb 10;30(2):179-89. doi: [ Links ]

8. Wang D, He Y, Li Y, Luan D, Yang X, Zhai F, et al. Dietary patterns and hypertension among Chinese adults: a nationally representative cross-sectional study. BMC Public Health. 2011 Dec 14;11:925. doi: [ Links ]

9. Fung TT, Chiuve SE, McCullough ML, Rexrode KM, Logroscino G, Hu FB. Adherence to a DASH-style diet and risk of coronary heart disease and stroke in women. Arch Intern Med. 2008 Apr 14;168:713-20. doi: [ Links ]

10. Rees K, Harty L, Flowers N, Clarke A, Hooper L, Thorogood M, et al. 'Mediterranean' dietary pattern for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev. 2013 Aug 12;(8):CD009825. doi: [ Links ]

11. Na L, Han T, Zhang W, Wu X, Na G, Du S, et al. A snack dietary pattern increases the risk of hypercholesterolemia in northern Chinese adults: a prospective cohort study. PLoS One 2015 Aug 5;10(8):e0134294. doi: [ Links ]

12. Drehmer M, Odegaard AO, Schmidt MI, Duncan BB, Cardoso LO, Matos SMA, et al. Brazilian dietary patterns and the dietary approaches to stop hypertension (DASH) diet-relationship with metabolic syndrome and newly diagnosed diabetes in the ELSA-Brasil study. Diabetol Metab Syndr 2017 Feb 13;9:13. doi: [ Links ]

13. Olinto MTA, Gigante DP, Horta B, Silveira V, Oliveira I, Willett W. Major dietary patterns and cardiovascular risk factors among young Brazilian adults. Eur J Nutr 2012;51(3):281-91. doi: [ Links ]

14. Neumann AICP, Martins IS, Marcopito LF, Araujo EAC. Dietary patterns associated with risk factors for cardiovascular disease in a Brazilian city. Rev Panam Salud Publica. 2007;22(5):329-39. doi: Portuguese. [ Links ]

15. Faerstein E, Chor D, Lopes CDS, Werneck GL. Estudo pró-saúde: características gerais e aspectos metodológicos. Rev Bras Epidemiol. 2005;8(4):454-66. doi: [ Links ]

16. Sichieri R, Everhart JE. Validity of a Brazilian food frequency questionnaire against dietary recalls and estimated energy intake. Nutr Res. 1998;18(10):1649-59. doi: [ Links ]

17. World Health Organization. Obesity: preventing and managing the global epidemic: report of a WHO expert consultation on obesity. Geneva: WHO; 1998. [ Links ]

18. Framingham Heart Study. FHS: bibliography, 2017. Framingham: FHS; 2017 [acesso 13 set. 2021]. Disponi´vel on-line em: [ Links ]

19. Bertin M, Touvier M, Dubuisson C, Dufour A, Havard S, Lafay L, et al. Dietary patterns of French adults: associations with demographic, socio-economic and behavioural factors. J Hum Nutr Diet. 2016;29(2):241-54. doi: [ Links ]

20. Zheng P-F, Shu L, Zhang X-Y, Si C-J, Yu X-L, Gao W, et al. Association between dietary patterns and the risk of hypertension among Chinese: a cross-sectional study. Nutrients. 2016 Apr 23;8(4):239. doi: [ Links ]

21. Kim WK, Shin D, Song WO. Are dietary patterns associated with depression in U.S. adults?. J Med Food. 2016;19(11):1074-84. doi: [ Links ]

22. Cunha DB, Almeida RMVR, Pereira RA. A comparison of three statistical methods applied in the identification of eating patterns. Cad Saude Publica 2010;26(11):2138-48. doi: [ Links ]

23. Astrup A, Raben A, Geiker N. The role of higher protein diets in weight control and obesity-related comorbidities. Int J Obes (Lond). 2015;39(5):721-6. doi: [ Links ]

24. Schmidt JB, Gregersen NT, Pedersen SD, Arentoft JL, Ritz C, Schwartz TW, et al. Effects of PYY3-36 and GLP-1 on energy intake, energy expenditure, and appetite in overweight men. Am J Physiol Endocrinol Metab. 2014 Jun 1;306(11):E1248-56. doi: [ Links ]

25. Pesta DH, Samuel VT. A high-protein diet for reducing body fat: mechanisms and possible caveats. Nutr Metab (Lond). 2014 Nov 19;11(1):53. doi: [ Links ]

26. Sayon-Orea C, Martinez-Gonzalez MA, Bes-Rastrollo M. Alcohol-consumption and body weight: a systematic review. Nutr Rev. 2011;69(8):419-31. doi: [ Links ]

27. Rocha TF, Hasselmann MH, Curioni CC, Bezerra FF, Faerstein E. Alcohol consumption is associated with DXA measurement of adiposity: the Pró-Saúde Study, Brazil. Eur J Nutr 2017;56(5):1983-91. doi: [ Links ]

28. Bowley CI, Blundell LL. Dietary patterns and sociodemographic factors: considerations for nutrition research. Public Health Nutr. 2016;19(16):3055-56. doi: [ Links ]

29. Cardoso LO, Carvalho MS, Cruz OG, Melere C, Luft VC, Molina MC, Faria CP, et al. Eating patterns in the Brazilian longitudinal study of adult health (ELSA-Brasil): an exploratory analysis. Cad Saude Publica. 2016;32(5):e00066215. doi: [ Links ]

30. Schulze MB, Hoffmann K, Kroke A, Boeing H. Risk of hypertension among women in the EPIC-potsdam study: comparison of relative risk estimates for exploratory and hypothesis-oriented dietary patterns. Am J Epidemiol. 2003 Aug 15;158(4):365-73. doi: [ Links ]

*Article derived from the Ph.D. thesis entitled ‘Dietary patterns and cardiovascular risk: The Pró-Saúde Study’, defended by Thalita Fialho da Rocha at the Postgraduate Program in Food, Nutrition and Health, Universidade do Estado do Rio de Janeiro, in 2017. FAPERJ Project - ‘Obesity epidemic, beyond the energy balance: contextual and prolonged factors in Rio de Janeiro’ - Process No. E-26/110.119/2013.

Received: February 11, 2021; Accepted: August 11, 2021

Correspondence: Thalita Fialho da Rocha - Rua São Francisco Xavier, nº 524, Pavilhão João Lyra Filho, 12º andar, Bloco D, Rio de Janeiro, RJ, Brazil. CEP: 20559-900. E-mail:

Associate Editor

Doroteia Aparecida Höfelmann - 0000-0003-1046-3319

Authors´ contributions

Rocha TF contributed to the concept and design of the study, analysis and interpretation of the results, drafting and critically reviewing the contents of the manuscript. Curioni C contributed to the study design, analysis and interpretation of the results, drafting and critically reviewing the contents of the manuscript. Verly-Junior E, Bezerra F and Faerstein E contributed to data analysis and interpretation, drafting and critically reviewing the contents of the manuscript. All the authors have approved the final version of the article and are responsible for all aspects thereof, including the guarantee of its accuracy and integrity.

Creative Commons License Este é um artigo publicado em acesso aberto sob uma licença Creative Commons