Services on Demand
- Cited by SciELO
- Access statistics
- Similars in SciELO
Print version ISSN 1415-790X
Rev. bras. epidemiol. vol.14 suppl.1 São Paulo Sep. 2011
Amanda de Moura SouzaI; Ilana Nogueira BezerraII; Diana Barbosa CunhaI; Rosely SichieriI
IDepartment of Epidemiology -
Instituto de Medicina Social - Universidade do Estado do Rio de Janeiro
(UERJ) - Rio de Janeiro (RJ), Brazil
IIPost-Graduation program in Clinical and Experimental Physiopathology - Faculdade de Medicina of Universidade do Estado do Rio de Janeiro (UERJ) - Rio de Janeiro (RJ), Brazil
OBJECTIVE: To evaluate markers of food
intake of the telephone-based risk factor surveillance system for chronic diseases
(VIGITEL) and the trend of these markers.
METHODS: A total of 135,249 subjects from 27 Brazilian cities interviewed in the 2007 - 2009 surveys were evaluated. Eating habits were evaluated based on the frequency of intake of fruit, vegetables, beans, whole and skim milk, regular and diet/light soft drinks and visible fat in meat and poultry. These items were used to create a diet quality score and to identify dietary patterns in a cluster analysis.
RESULTS: Time trends indicated statistically significant increase in the frequency of intake of beans, whole milk and regular soft drinks and decline in vegetables and skim milk. There was an increase in the frequency of individuals who reported consuming beans daily, from 11 to 13%. Beans are considered as a protective factor and the prevalence of usual intake is still low. Over the past three years, less than 15% of the studied population reported eating the Brazilian recommended number of 3 servings of fruits and 3 servings of vegetables per day. As to the vegetable intake, a decrease from 5 to 3% was reported. The consumption of regular soft drinks had the highest increase, ranging from 60 to 67%. The assessed items showed a weak correlation and did not represent a sole healthy eating construct.
CONCLUSION: The diet quality of the Brazilians has gotten worst and eating markers that are associated with high risk of chronic diseases should be better qualified.
Keywords: food habits; feeding behavior; data collection; Brazil.
Chronic Non-Communicable Diseases (CNCD) contribute with a lot of obits in the country, and many risk factors are associated with such diseases1,2. Among these factors, inadequate eating habits, physical inactivity, smoking and obesity are important to determine CNCD2.
The insufficient intake of fruit and vegetables and the high intake of sugar and saturated fat are habits related to the risk factors of CNCD. These habits are included in the Dietary Guidelines for the Brazilian Population3 established by the Ministry of Health in 2006, and in the Telephone-based Surveillance of Risk and Protective Factors for Chronic Diseases (VIGITEL)4.
In the first VIGITEL survey, in 2006, 7% of the participants reported the adequate intake of fruit and vegetables according to the Brazilian Guidelines; less than 50% of the interviewees reported the regular consumption of these foods5. In this study, we analyzed the consumption of foods from 2007 to 2009, from the seven intake-related questions (beans, fruits, vegetables, whole milk, skim milk, regular soft drinks, diet or light soft drinks and consumption of saturated fat), we evaluated the ability to identify food intake patterns. Therefore, we applied some techniques that are commonly used to assess eating dietary patterns, summarized by Hu6 in two general formats. The first one consists of indexes created from guidelines, like the Index of Food Quality (IFQ)7,8; and the second is comprised of statistical procedures to analyze the covariation structure of different variables (usually foods or food groups) in order to show a restricted number of eating patterns in the assessed group6,9. These strategies of analysis have been widely known as a priori and a posteriori, even though these terms are not properly used, once the former does not concern knowledge based only on theory (a priori), and the latter consists of exploratory analyses, and not on experimental knowledge (a posteriori).
As to the second approach, cluster analysis (in groups)10,11 and factorial analysis, or principal components analysis12-14, have been more frequently reported in literature. The main difference among them is that the cluster analysis gathers individuals, while the factorial analysis gathers variables. The former is adequate for three situations: to explore eating patterns when it is suspected that the sample is not homogeneous; when there are no statistical properties necessary for the factorial analysis; or when the intention is to keep all the proposed food items in the instrument. Thus, the method aims at assigning individuals to different subgroups (the clusters). Food intake of individuals in the same cluster is expected to be reasonably homogeneous, showing more differences among clusters15-17.
This article aims to assess food intake markers of the VIGITEL system and its temporal evolution.
Study population and design
This study used data from VIGITEL, conducted in 2007, 2008 and 2009. This survey is a probability sampling of adults (>18 years) who own a telephone line in 26 Brazilian state capitals and the Federal District. The sampling process was based on the draw of 5,000 households per city, followed by the selection of one inhabitant per household. At least 2,000 interviews were conducted in each city. More details on sampling procedures may be found in Moura et al.18.
Individuals aged between 20 and 65 years were included in this study, and the women who were pregnant at the time of the survey were excluded. Due to some changes in the questions related to eating habits, the year 2006 was not included in the analyses.
From 2007 to 2009, 162,971 individuals were interviewed for VIGITEL. After the exclusion of 26,430 individuals who were not in the analyzed age group and 1,292 pregnant women, 135,249 adults were assessed.
Description of the variables
The assessed sociodemographic variables were: age group, schooling and gender. Data regarding reported weight and height were used to calculate the Body Mass Index (BMI) - body mass (kg) divided by height in squared meters (m2). Nutritional status was assessed according to the guidelines by the World Health Organizations19, that is, excess weight: BMI>25 kg/m2 and obesity: BMI>30 kg/m2.
Concerning food intake, the survey included a question about the consumption of beans: "On how many days of the week do you usually eat beans?"; the responses were: never/almost never, once or twice a week, three or four times a week, five or six times a week, or every day of the week.
The questionnaire had three questions on the weekly frequency of fruit, salad, raw salad and cooked vegetables intake. Possible responses were: never/almost never, once or twice a week, three or four times a week, five or six times a week, or every day of the week. Another question enabled the analysis of the frequency of fruit intake on a regular day, with the following possible responses: once, twice a day, or three or more times a day. Individuals who reported eating fruits at least once or twice a week, but did not respond as to the frequency of fruit intake on a regular day were classified in the low intake category (once a day). Both questions were matched in order to estimate the daily frequency of fruit intake. The frequency of salad, raw salad and cooked vegetables intake on a regular day was assessed from the following characteristics: for lunch (considered as once a day), for dinner (considered as once a day), for lunch and for dinner (considered as twice a day). Individuals who reported eating salad, raw salad and cooked vegetables at least once or twice a week, but did not respond as to the frequency of salad, raw salad and cooked vegetables consumption on a regular day were classified in the low intake category (once a day). The daily consumption of vegetables was estimated by the daily intake of salad, raw salad and cooked vegetables.
The consumption of soft drinks (or artificial juice drinks, such as Tampico) was analyzed from the weekly intake frequency (six possible responses, varying from never to every day of the week); type (regular, diet/light, or both) and number of consumed glasses/cans (possible responses ranging from 1 to 6 or more). For those who could not estimate the amount of consumed soft drinks, the average established number was one glass.
The intake of saturated fat was assessed with two questions related to meat consumption: "when you eat red meat, do you usually: remove the visible fat, eat the meat and the fat, or do not eat meat with a lot of fat", and "when you eat chicken, do you usually: remove the skin, eat the skin or do not eat pieces of chicken with skin". The answers were dichotomized into: consumption of visible fat in the meat or not, and the consumption of the chicken skin or not, respectively.
Two questions regarding milk consumption also showed the usual consumption of saturated fat: weekly frequency of milk consumption (ranging from never to every day of the week) and type of milk (whole, skim or both). From these variables, the daily consumption of whole milk was estimated, including those who consumed both types of milk, as well as the daily consumption of skim milk.
Assessment of healthy eating scores
A healthy eating score with the following variables was established: beans, fruits, raw salad, cooked vegetables, milk and soft drinks, in four levels of weekly frequency, as described in Chart 1. The score was calculated from the sum of the punctuation of the items, ranging from 0 to 24 points.
The internal consistency of the score was assessed by the correlation of the items and with the Cronbach's α coefficient, which was 0.24. Food intake patterns were also assessed by the cluster analysis, with the following variables: intake of fruits, vegetables, whole and skim milk, regular and diet/light soft drinks, consumption of meat and chicken fat. For the consumption of fruits and vegetables, daily frequency was dichotomized into less than three times a day and three or more times a day, according to the Dietary Guidelines for the Brazilian Population3. The consumption of soft drinks was assessed as to daily frequency (glasses/day).
Descriptive data analysis was conducted by the calculation of the distribution of variables frequency and their respective 95% confidence intervals; concerning continuous variables, the calculation consisted of the mean and the standard error, according to year of monitoring.
Simple linear regression models were used to assess temporal variation of the continuous variables in the years of monitoring. The chi-square test was used to assess the temporal variation of the distribution of frequencies per year of monitoring. Weighted analyses were conducted using the Statistical Analysis System (SAS), version 9.1.
For the cluster analysis, the 16.0 version of the SPSS k-means procedure was used, which consists of a non-hierarchical cluster technique that classifies individuals in a pre-defined number of clusters by the Euclidean distance, in a way to minimize the distances between observations inside a cluster in relation to the distances between clusters. This requires the previous definition of a number of clusters to be used in the analysis16,20.
The analysis was performed separately for each year of monitoring, with two hypothetical clusters: a healthy cluster and an unhealthy cluster. In this analysis, F-statistics values identify the food items that contribute the most with the solution of clusters. Variables with high F values present a larger separation between clusters.
VIGITEL was approved by the National Committee of Ethics in Research with Human Beings, of the Ministry of Health. Because it was a telephone survey, the verbal consent was obtained instead of the informed consent form.
According to the year of monitoring, there was no difference related to sociodemographic characteristics (age, gender and schooling). The prevalence of excess weight showed statistically significant increase over the years (Table 1).
There was statistically significant increase in the frequencies of bean, whole milk and regular soft drink intake; meanwhile, the consumption of vegetables and skim milk decreased throughout the years (Table 1). The recommended consumption of three servings of fruit and three servings of vegetables per day was reported by less than 15% of the population in all analyzed years, with a decrease from 5 to 3% for vegetables.
The average healthy eating index score did not change throughout the years (12.2 in 2006, 12.4 in 2007 and 12.4 in 2008). Correlation values between score items ranged from -0.05 to 0.17.
Due to the low internal consistency of the score (Cronbach's α=0.24), it was not considered a good marker of healthy eating. If the beans were removed, they would be the only item that could increase Cronbach's α coefficient to 0.29 (data not shown).
The cluster analysis identified the same patterns in the three evaluated years: cluster I, characterized by the intake of vegetables, fruits, diet/light soft drinks and skim milk, and cluster II, comprised of whole milk, regular soft drinks, visible fat of the meat and chicken skin; however, some items such as fruits and milk did not properly discriminate both patterns. According to the F values obtained by the ANOVA table, the variable which presented the largest separation among clusters were the regular soft drinks. In 2007, 90.7% of the individuals were part of cluster I; in 2008, this percentage decreased to 86.4% and, in 2009, it was 87.5% (Table 2).
The results in this study show that a low percentage of individuals analyzed in VIGITEL met the Dietary Guidelines for the Brazilian Population regarding the intake of fruit, vegetables and beans. We observed four types of evolution: negative (for vegetables, skim milk and diet soft drinks), positive (for beans, whole milk and regular soft drinks), unstable (for fruit) and with no changes regarding the action of removing the fat from the meat or chicken skin.
In relation to the intake of fruit and vegetables, our findings are similar to those demonstrated by the Household Budget Survey (Pesquisa de Orçamentos Familiares - POF) carried out by the Brazilian Institute of Geography and Statistics (IBGE) from 1970 to 2003. The presence of fruit, salads and vegetables in the diet remained relatively constant (from 3 to 4%) and below the recommendation of consuming 9 to 12% of the total calories from this food group. The low intake of fruit and vegetables has also been observed in the United States21, and some analyses show that even though healthy eating costs more than industrialized products, it is possible to choose healthy items at a lower cost22. Thus, it is not only the costs that determine the eating choices. De Bem Lignani et al.23 also identified that the higher purchasing power of the population addressed by Brazilian income transfer programs (Bolsa Família) increased the intake of unhealthy foods.
Concerning the consumption of beans, POF identified a 30% reduction in the household availability of this item in the same period. However, although our data show that the percentage of individuals who consume the adequate amount of beans increased throughout the years, this percentage remained low.
Milk consumption is recommended as a good source of riboflavin and the main source of calcium in the diet. Nowadays, recommendations addressed to the adult population related to milk consumption are restricted to low fat sources in order to reduce the intake of saturated fat3. Our data meet the recommendations, once they showed opposite trends for the consumption of whole milk and skim milk in the period, with increasing values for the former and decreasing values for the latter. Another marker of saturated fat intake is the consumption of visible fat of meat and chicken. Even though these markers did not present statistically significant changes between surveys, about 70% of the participants reported removing the fat of the meat and the chicken skin.
The intake of regular soft drinks, which is a marker of unhealthy eating habits, increased through the years, just as reported in data from POF. In 30 years, the household availability of soft drinks increased 400%24.
The food items analyzed did not allow to establish a global indicator of healthy eating. Methods to assess food intake, such as the construction of indexes and scores, have been proposed to analyze the general quality of the diet and, although most of these methods are able to predict health-related outcomes, the associations have low magnitude25. Besides, the choice of the variables, cut off values, and the exact method to calculate a score are the main difficulties to develop such instrument25. In this study, a healthy eating score was established including the following variables: beans, fruits, raw salad, cooked vegetables, milk and soft drinks. The internal consistency of this score was low, with Cronbach's α coefficient of 0.24, like that obtained in the healthy eating score evaluation used in the United States, the Healthy Eating Index (HEI) - 20057; in this case, Cronbach's α coefficient was 0.43, which is higher than the value found in this study, but it is still a moderate consistency value among the items.
Cluster analysis was also unsatisfactory. This method was chosen due to the possibility to produce mutually exclusive clusters and to identify a healthy pattern. This procedure is also justified by the heterogeneity of the sample and by the initial intention of keeping all the food items that were proposed, as well as having the advantage to classify individuals in only one pattern15. Despite the fact that this method led to the identification of a cluster comprised of foods considered healthy or frequently consumed by individuals who are more concerned about their health, and of another cluster comprised by items that are considered unhealthy, there were high percentages (around 90%) of adherence to cluster I, which is closer to healthy eating habits, in all investigated years. This clearly does not represent the eating habits of the Brazilian population24. Besides, the mean consumption of the food items did not significantly vary between the identified clusters. This may be explained by the low percentage of individuals who consume in accordance with the Ministry of Health guidelines related to fruits and vegetables25. However, the employed procedures did not lead to a sole healthy eating construct, indicating that the seven questions in VIGITEL should be reviewed as to the constructs that will be assessed and their real ability to measure healthy eating habits. It is acceptable that if the quality of the diet is a multidimensional construct, we could not recover a global measure from the assessed items. Thus, those items represent different consumption dimensions considered adequate by the guidelines discussed for HEI 2005-US4,7.
It is worth to emphasize that the three analyzed years may not be sufficient to detect great changes regarding the consumption of the assessed items. Besides, the changes observed in this short period may reflect alterations in price, distribution or access to the products.
Among the analyzed items, the regular soft drinks represented the largest separation between clusters, which demonstrates that this item is the one that discriminates food intake the most. Thus, this item is essential to investigate food intake of populations.
Monteiro et al.26 demonstrated good reproducibility and adequate validation for most of the questions in the eating analysis of VIGITEL; thus, the non recovery of a sole healthy eating construct is not a result of information bias.
Factors that could explain the difficulty of questions present in VIGITEL to define healthy eating habits included: 1) the absence of questions regarding the intake of unhealthy foods in Brazil, like crackers - which increased 400% in 30 years24, according to data from POF -, pastry and fast food; 2) the low percentage of individuals who presented inadequate consumption of healthy items, such as fruits and vegetables; 3) the complexity and the multidimensionality of this construct.
The inclusion of new variables in the eating analysis of VIGITEL and the identification of items that do not characterize health consumption may contribute with the construction of eating patterns that discriminate more and identify individuals with eating habits considered risky for chronic non-communicable diseases.
1. Brasil. Sistema de informações sobre mortalidade. Brasília: Ministério da Saúde, 2010. [cited 2010 Dez 06]. Available from: http://tabnet.datasus.gov.br/cgi/tabcgi.exe?sim/cnv/obt10uf.def [ Links ]
2. World Health Organization. Reducing risks, promoting healthy life. Geneva: WHO; 2002. [ Links ]
3. Ministério da Saúde. Secretaria de Atenção à Saude. Coordenação-Geral da Política de Alimentação e Nutrição. Guia alimentar para a população brasileira: promovendo a alimentação saudável. Brasília: Ministério da Saúde; 2006. [ Links ]
4. Estados Unidos. Department of Agriculture, Agricultural Research Service. Center for Nutrition Policy and Promotion. Report of the Dietary Guidelines Advisory Committee on the Dietary Guidelines for American, 2010. USDA; 2010. [cited 2011 Jun 15]. Available from: http://www.cnpp.usda.gov/Publications/DietaryGuidelines/2010/DGAC/Report/2010DGACReport-camera-ready-Jan11-11.pdf. [ Links ]
5. Jaime PC, Figueiredo IC, Moura EC, Malta DC. Factors associated with fruit and vegetable consumption in Brazil, 2006. Rev Saúde Pública2009; 43 Suppl 2:57-64. [ Links ]
6. Hu FB. Dietary pattern analysis: a new direction in nutritional epidemiology. Curr Opin Lipidol. 2002;13(1):3-9. [ Links ]
7. Guenther P, Reedy J, Krebs-Smith S, Reeve B, Basiotis P. Development and Evaluation of the Healthy Eating Index - 2005: Techinical Report.: Center for Nutrition Policy and Promotion. United States. Department of Agriculture - 2007. [cited 2010 Dez 06]. Available from: http://www.cnpp.usda.gov/Publications/HEI/HEI-2005/HEI-2005TechnicalReport.pdf8. [ Links ]
8. Fisberg RM, Slater B, Barros RR, Lima FD, Cesar CLG, Carandina L, et al. Índice de Qualidade da Dieta: avaliação da adaptação e aplicabilidade. Rev. de Nutr. 2004;17(3):301-8. [ Links ]
9. Panagiotakos D. α-priori versus α-posterior methods in dietary pattern analysis: a review in nutrition epidemiology. Nutr Bull. 2008;33(4):311-15. [ Links ]
10. James DC. Cluster analysis defines distinct dietary patterns for African-American men and women. J Am Diet Assoc. 2009;109(2):255-62. [ Links ]
11. Newby PK, Muller D, Hallfrisch J, Qiao N, Andres R, Tucker KL. Dietary patterns and changes in body mass index and waist circumference in adults. Am J Clin Nutr. 2003;77(6):1417-25. [ Links ]
12. Cunha DB, de Almeida RM, Sichieri R, Pereira RA. Association of dietary patterns with BMI and waist circumference in a low-income neighbourhood in Brazil. Br J Nutr. 2010;104(6):908-13. [ Links ]
13. Carrera PM, Gao X, Tucker KL. A study of dietary patterns in the Mexican-American population and their association with obesity. J Am Diet Assoc. 2007 107(10):1735-42. [ Links ]
14. Sichieri R, Castro JF, Moura AS. Fatores associados ao padrão de consumo alimentar da população brasileira urbana. Cad Saúde Pública. 2003;19 Suppl 1:S47-53. [ Links ]
15. Hearty AP, Gibney MJ. Comparison of cluster and principal component analysis techniques to derive dietary patterns in Irish adults. Br J Nutr2009;101(4):598-608. [ Links ]
16. Jinlin F, Binyou W, Terry C. A new approach to the study of diet and risk of type 2 diabetes. J Postgrad Med. 2007;53(2):139-43. [ Links ]
17. Olinto MTA. Padrões alimentares: análise de componentes principais. In: Kac G SR, Gigante DP (editor). Epidemiologia Nutricional. Rio de Janeiro: Fiocruz; 2007. p. 213-25. [ Links ]
18. Moura EC, Neto OLM, Malta DC, Moura L, Silva NN, Bernal R, et al. Vigilância de fatores de risco para doenças crônicas por inquérito telefônico nas capitais dos 26 estados brasileiros e no Distrito Federal (2006). Rev Bras Epidemiol. 2008;11 Suppl 1:20-37. [ Links ]
19. World Health Organization. Obesity: preventing and managing the global epidemic. Geneva: WHO; 1997. [ Links ]
20. Schulze MB, Hu FB. Dietary patterns and risk of hypertension, type 2 diabetes mellitus, and coronary heart disease. Curr Atheroscler Rep. 2002;4(6):462-7. [ Links ]
21. Li R, Serdula M, Bland S, Mokdad A, Bowman B, Nelson D. Trends in fruit and vegetable consumption among adults in 16 US states: Behavioral Risk Factor Surveillance System, 1990 - 1996. Am J Public Health. 2000; 90(5):777-81. [ Links ]
22. Bernstein AM, Bloom DE, Rosner BA, Franz M, Willett WC. Relation of food cost to healthfulness of diet among US women. Am J Clin Nutr. 2010;92(5):1197-203. [ Links ]
23. de Bem Lignani J, Sichieri R, Burlandy L, Salles-Costa R. Changes in food consumption among the Programa Bolsa Família participant families in Brazil. Public Health Nutr. 2011;14(5):785-92. [ Links ]
24. Levy-Costa RB, Sichieri R, Pontes Ndos S, Monteiro CA. Disponibilidade domiciliary de alimentos no Brasil: distribuição e evolução (1974 - 2003). Rev Saúde Pública. 2005;39(4):530-40. [ Links ]
25. Waijers PM, Feskens EJ, Ocké MC. A critical review of predefined diet quality scores. Br J Nutr. 2007;97(2): 219-31. [ Links ]
26. Monteiro CA, Moura EC, Jaime PC, Claro RM. Validade de indicadores do consumo de alimentos e bebidas obtidos por inquérito telefônico. Rev Saúde Pública. 2008;42(4):582-9. [ Links ]
Correspondence to: Study carried out at the Department of Epidemiology
- Instituto de Medicina Social of Universidade do Estado do Rio de Janeiro (UERJ) - Rio de Janeiro (RJ), Brazil
Amanda de Moura Souza
Department of Epidemiology - Instituto de Medicina Social at Universidade do Estado do Rio de Janeiro
Rua São Francisco Xavier, 524, 7º floor, bloco E
CEP: 20550-900 - Rio de Janeiro (RJ), Brazil
Financial Support: National Council for Scientific and Technological Development (CNPq); Research Support Foundation of the State of Rio de Janeiro (FAPERJ) and Coordination for Improvement of Higher Education Personnel (CAPES).
Conflict of interest: nothing to declare.
Submitted on: 05/01/2011
Final version presented on: 06/04/2011
Accepted on: 29/04/2011
Study carried out at the Department of Epidemiology
- Instituto de Medicina Social of Universidade do Estado do Rio de Janeiro (UERJ) - Rio de Janeiro (RJ), Brazil