Abstract

Obesity prevalence is rapidly increasing in developing countries. Existing research investigating social patterning of obesity and its risk factors in Latin American urban contexts has inconsistent findings. This study analyzed a multistage household survey in adults in Belo Horizonte, Minas Gerais State, Brazil. Marginal models were used to examine the association of education and household and neighborhood income with body mass index (BMI), obesity, physical inactivity, and low fruit and vegetable intake after adjusting for age and ethnicity and stratifying by sex. BMI and obesity were inversely associated with education in women. BMI was positively associated with household and neighborhood income in men. Additionally, physical inactivity and low fruit and vegetable intake were inversely associated with education and household income in both men and women, and physical inactivity was inversely associated with neighborhood income in men. Understanding the drivers of these patterns will allow for development of appropriate policy and interventions to reduce cardiovascular disease risk in large cities in Latin America.

Socioeconomic Factors; Body Mass Index; Obesity; Urban Health

Resumo

A prevalência da obesidade está aumentando nos países em desenvolvimento. Estudos investigando os padrões sociais da obesidade e seus fatores de risco em áreas urbanas da América Latina têm resultados inconsistentes. Neste estudo foram analisadas as informações provenientes de inquérito populacional sobre a saúde de adultos residentes em Belo Horizonte, Minas Gerais, Brasil. Foram utilizados modelos marginais para examinar a associação entre a escolaridade, a renda familiar e da vizinhança com o IMC (índice de massa corporal), a obesidade, a atividade física e o consumo de frutas e verduras. O IMC e a obesidade se associaram inversamente à escolaridade entre as mulheres e o IMC positivamente à renda do domicilio e da vizinhança entre os homens. O baixo consumo de frutas e verduras e a inatividade física foram inversamente associados à escolaridade e à renda familiar em ambos os sexos. A inatividade física se associou inversamente à renda da vizinhança entre os homens. Compreender as causas da distribuição social da obesidade é importante para reduzir o risco de doença cardiovascular nos centros urbanos da América Latina.

Fatores Socioeconômicos; Índice de Massa Corporal; Obesidade; Saúde Urbana

Resumen

La prevalencia de la obesidad está aumentando en los países en desarrollo. Estudios sobre la distribución social de la obesidad y sus factores de riesgo en ciudades de América Latina cuentan con resultados inconsistentes. En este estudio se analizó una encuesta de adultos de Belo Horizonte, Minas Gerais, Brasil. Se utilizaron modelos marginales para examinar la asociación entre educación, ingresos del hogar y barrio sí como el índice de masa corporal (IMC), la obesidad, la inactividad física, y el consumo de frutas y verduras. Se observaron relaciones inversas entre el IMC, obesidad y la educación en las mujeres, así como una relación positiva entre el IMC y los ingresos del hogar y barrio en el caso de los hombres. Además, el consumo bajo de frutas y verduras y la inactividad física se asociaron inversamente con la educación y los ingresos del hogar en ambos sexos. La inactividad física se asoció inversamente con los ingresos del barrio en los hombres. Comprender las causas de esta distribución social es importante para reducir el riesgo de enfermedades cardiovasculares en las ciudades de América Latina.

Factores Socioeconómicos; Índice de Masa Corporal; Obesidad; Salud Urbana

Introduction

In many countries a nutrition and physical activity transition has accompanied the epidemiologic shift from infectious to chronic disease ¹1. Popkin BM, Gordon-Larsen P. The nutrition transition: worldwide obesity dynamics and their determinants. Int J Obes Relat Metab Disord 2004; 28 Suppl 3:S2-9.,²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657.. Recognition of the influence of obesity on non-communicable disease rates has led to an increasing focus on prevalence of obesity and its risk factors worldwide. Obesity, defined as a body mass index (BMI) ≥ 30kg/m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657., is a physical activity and nutrition-related risk factor for many chronic diseases including hypertension, diabetes, cardiovascular disease, and several types of cancer ³3. Finucane MM, Stevens GA, Cowan MJ, Danaei G, Lin JK, Paciorek CJ, et al. National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9·1 million participants. Lancet 2011; 377:557-67..

The association of obesity with socioeconomic factors has been well documented in developed countries ⁴4. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993; 88(4 Pt 1):1973-98. where cardiovascular disease risk factors such as obesity are inversely associated with socioeconomic position ⁴4. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993; 88(4 Pt 1):1973-98.,⁵5. McLaren L. Socioeconomic status and obesity. Epidemiol Rev 2007; 29:29-48.,⁶6. Dinsa GD, Goryakin Y, Fumagalli E, Suhrcke M. Obesity and socioeconomic status in developing countries: a systematic review. Obes Rev 2012; 13:1067-79.. Health behaviors such as diet and physical activity may represent possible mediators in the association of socioeconomic position with obesity and have shown inverse trends with socioeconomic position similar to those with obesity in developed nations ⁴4. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993; 88(4 Pt 1):1973-98.. Additionally, studies in industrialized nations have demonstrated that neighborhood factors, such as availability of healthy foods and recreational resources, influence diet and physical activity over and above individual-level characteristics ⁷7. Diez Roux AV. Residential environments and cardiovascular risk. J Urban Health 2003; 80:569-89..

Research demonstrates that rates of chronic diseases and associated risk factors are increasing faster in developing regions than in developed regions and that cardiovascular disease is the greatest cause of mortality in developing nations ⁸8. Magnusson RS. Rethinking global health challenges: towards a “global compact” for reducing the burden of chronic disease. Public Health 2009; 123:265-74.. The largest increase in BMI between 1980 and 2008 occurred in areas of Oceania and Latin America ⁸8. Magnusson RS. Rethinking global health challenges: towards a “global compact” for reducing the burden of chronic disease. Public Health 2009; 123:265-74.. In Brazil, for example, between 1975 and 2008 the percentage of the population that classified as obese increased from 2.7 to 16.0 in men and from 7.4 to 21.4 in women ⁹9. Monteiro CA, Conde WL, Popkin BM. Income-specific trends in obesity in Brazil: 1975-2003. Am J Public Health 2007; 97:1808-12.,¹⁰10. World Health Organization. NCD country profiles: Brazil. Geneva: World Health Organization; 2011..

Historically in developing nations, low socioeconomic position was protective against cardiovascular disease risk factors ⁹9. Monteiro CA, Conde WL, Popkin BM. Income-specific trends in obesity in Brazil: 1975-2003. Am J Public Health 2007; 97:1808-12.,¹¹11. Sobal J, Stunkard AJ. Socioeconomic status and obesity: a review of the literature. Psychol Bull 1989; 105:260-75.. However, a growing body of research in Latin America indicates a shift in the burden of these risk factors to impoverished populations ¹²12. Marins VM, Almeida RM, Pereira RA, Sichieri R. The association between socioeconomic indicators and cardiovascular disease risk factors in Rio de Janeiro, Brazil. J Biosoc Sci 2007; 39:221-9.,¹³13. Boissonnet C, Schargrodsky H, Pellegrini F, Macchia A, Marcet Champagne B, Wilson E, et al. Educational inequalities in obesity, abdominal obesity, and metabolic syndrome in seven Latin American cities: the CARMELA Study. Eur J Cardiovasc Prev Rehabil 2011; 18:550-6.,¹⁴14. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H. Social patterning of chronic disease risk factors in a Latin American city. J Urban Health 2008; 85:923-37.. Additionally, urban populations have been shown to have a higher prevalence of obesity, low levels of physical activity, and poor diet ¹⁴14. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H. Social patterning of chronic disease risk factors in a Latin American city. J Urban Health 2008; 85:923-37.,¹⁵15. Machado-Rodrigues AM, Coelho-e-Silva MJ, Mota J, Cumming SP, Riddoch C, Malina RM. Correlates of aerobic fitness in urban and rural Portuguese adolescents. Ann Hum Biol 2011; 38:479-84.. Given the already high and growing levels of urbanization in many low- and middle-income countries, describing the association of these risk factors with socioeconomic position in urban settings is of special significance. Describing and understanding the association of obesity and associated behavioural risk factors with socioeconomic position in countries such as Brazil, a nation with large urban centers that suffers from great inequity, is fundamental to developing appropriate policies and interventions to stymie the obesity epidemic.

In Brazil in 1989, obesity in adults was positively associated with socioeconomic position ¹⁶16. Monteiro CA, Benicio MH, Conde WL, Popkin BM. Shifting obesity trends in Brazil. Eur J Clin Nutr 2000; 54:342-6.. In 2003, obesity prevalence remained higher in high-income groups than in low-income groups, but the rate of increase was higher in low-income communities ⁹9. Monteiro CA, Conde WL, Popkin BM. Income-specific trends in obesity in Brazil: 1975-2003. Am J Public Health 2007; 97:1808-12.. It is important to continue to characterize the magnitude and direction of the association of obesity, and the associated behaviors of physical inactivity and poor diet, with socioeconomic position as it shifts over time to burden impoverished populations. Utilizing both individual- and contextual-level measures of socioeconomic position can provide greater insight into the specific aspects of socioeconomic position that influence health. The purpose of this study is to investigate the association of both individual-level and neighborhood-level measures of socioeconomic position on risk factors for cardiovascular disease in an urban Brazilian population. Using a unique household survey, we examine the influence of education, household income, and neighborhood income on BMI, obesity, physical activity, and dietary factors in the city of Belo Horizonte, Minas Gerais State, Brazil. Given the shifting trends observed in Brazil over the last several decades, we expect that the present pattern of these outcomes in Belo Horizonte will mirror that of developed nations.

Methods

Study population

From August 2008 to February 2009 the Belo Horizonte Observatory for Urban Health conducted a cross-sectional survey in Belo Horizonte. The survey was conducted in the Barreiro and West districts, two of the city’s nine districts. Each district has approximately 250,000 inhabitants, and the two districts have a total geographical area of 33.16km²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657.. The West and Barreiro districts were chosen due to their geographical proximity and their significant internal heterogeneity of various demographic, socioeconomic, and health indicators. These characteristics were similar to those of other city districts.

Details of the study design have been reported previously ¹⁷17. Friche AA, Diez-Roux AV, César CC, Xavier CC, Proietti FA, Caiaffa WT. Assessing the psychometric and ecometric properties of neighborhood scales in developing countries: Saúde em Beagá Study, Belo Horizonte, Brazil, 2008-2009. J Urban Health 2013; 90:246-61.. In brief, stratified three-stage cluster sampling was used to select the sample; this included selection of census tracts as the first stage, households as the second stage, and individuals as the third stage. The sample strata were defined according to the Health Vulnerability Index ¹⁸18. Turci MA, Lima-Costa MF, Proietti FA, Cesar CC, Macinko J. Intraurban differences in the use of ambulatory health services in a large brazilian city. J Urban Health 2010; 87:994-1006., an index created by combining social, demographic, economic and health indicators from the census for each census tract. Census tracts, defined by the Brazilian Census Bureau, include an average of 1,000 residents. In the first stage of sampling, 149 census tracts were randomly selected from a total of 588 tracts in the sampling frame. A total of 6,493 households made up the 149 tracts. In the second stage of sampling, vacant lots, institutional and commercial buildings, and households that could not be reached after three attempts were deleted. In the final sampling stage, one adult from each household was selected to participate; interviewers randomly selected this participant from a list of resident adults for each selected household. Approximately 75% of the households agreed to participate leaving a total study sample of 4,048 households. The main causes of refusal were inability to contact the selected participant after three consecutive attempts and participants refusing to answer the questionnaire due to its length. Those who refused to participate did not differ from participants who completed the questionnaire by sex or age.

Before the survey was conducted, the instruments were pilot tested, and the 23 interviewers completed centralized training. Interviewers visited each sample household, and if the selected resident was present at the time, the interviewer explained the study aims and procedures and answered participant questions. Participants agreeing to take part in the study signed Informed Consent Forms. The interviewer then administered the face-to-face questionnaire, which lasted approximately 45 minutes, and performed anthropometric measurements.

We originally intended to interview all adults in the household to increase sample size after estimating a total sample size of approximately 3,000 persons over the age of 18 from 1,750 households. These calculations were based on the 2000 Brazilian Census and an 80% response rate ¹⁹19. Instituto Brasileiro de Geografia e Estatística. Censo demográfico 2000. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2003.. Assuming three persons per household and an average within household correlation in outcomes of 0.5, we would have had 80% power to detect differences as small as 0.2 standard deviations in the outcome when comparing two groups of 208 persons. However, we underestimated the number of eligible households; we ended up with a study sample of 4,048 households and only needed to sample one adult from each. Thus, power was even greater than expected due to the larger sample size and lack of clustering within household.

Funding for the study came from the National Institutes of Health (NIH) Fogarty International Center (grant n. D43 TW009315 and 5R03 TW008105), CNPq-409688/2006-1, FAPEMIG-CDS APQ 00677-08, and the National Health Fund (FNS)-25000.102984/2006-97/Brazil. The project was approved by the Ethics Research Committee of Federal University of Minas Gerais (protocol n. ETIC 253/06) and the Ethics Research Committee of the Department of Municipal Health (073.2008).

Data sources

Participants responded to a questionnaire made up of six modules: household information, sociodemographic data, health, habits and behaviors, anthropometric evaluation, and social determinants of health. Participants who did not report education (n = 4) or household income (n = 98), had no information on neighborhood income (n = 17), or who had missing BMI (n = 11) were excluded, leaving a total study sample of 3,918 participants. An additional 226 participants did not report physical activity and were excluded from the physical activity analyses.

The main outcomes of interest were BMI, obesity, low physical activity, and low fruit and vegetable intake. BMI was calculated as weight in kg/(height in m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657.). Interviewers used a standard scale (brand TANITA BF 542, Tanita Corp., USA) and stadiometer to measure height and weight. Interviewers were trained to correctly perform measurements following standard protocols. Obesity was defined as BMI ≥ 30kg/m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657. based on WHO recommendations.

Participants responded to questions regarding the frequency and duration of physical activity from the Portuguese version of the International Physical Activity Questionnaire (IPAQ), version 8. This instrument is recommended for national prevalence studies ²⁰20. Hallal PC, Victora CG. Reliability and validity of the International Physical Activity Questionnaire (IPAQ). Med Sci Sports Exerc 2004; 36:556.. The physical activity score utilized for analysis was calculated as the sum of minutes of moderate activity, plus twice the minutes of vigorous activity. Noncompliance with physical activity guidelines was defined as a score below 150 minutes/week as currently recommended by the U.S. Surgeon General’s Report, American College of Sports Medicine, and Centers for Disease Control and Prevention ²¹21. U.S. Department of Health and Human Services. Physical activity and health: a report from the Surgeon General. Atlanta: U.S. Department of Health and Human Services; 1996.,²²22. Hallal PC, Victora CG, Wells JC, Lima RC. Physical inactivity: prevalence and associated variables in Brazilian adults. Med Sci Sports Exerc 2003; 35:1894-900.,²³23. Pate RR, Pratt M, Blair SN, Haskell WL, Macera CA, Bouchard C, et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 1995; 273:402-7..

Diet was assessed using items adapted from other international and Brazilian surveys ²⁴24. Secretaria de Vigilância em Saúde, Ministério de Saúde. Vigitel Brasil 2011: Vigilância de fatores de risco e proteção para doenças crônicas por inquérito telefônico. Brasília: Ministério de Saúde; 2012. (Série G. Estatística e Informação em Saúde).. Consumption of less than three fruits and vegetables per day has been associated with increased risk of cardiovascular events ²⁵25. He FJ, Nowson CA, Lucas M, MacGregor GA. Increased consumption of fruit and vegetables is related to a reduced risk of coronary heart disease: meta-analysis of cohort studies. J Hum Hypertens 2007; 21:717-28., and we wished to explore the association of socioeconomic position with very low fruit and vegetable consumption. Thus, for our analyses, subjects who consumed fruits or vegetables less than three days per week were considered to have low levels of healthy food consumption. Fruit and vegetable intake was selected for its simplicity and because it has been shown to be a useful marker of dietary quality and related to social and economic factors in prior work ²⁶26. Morland K, Wing S, Diez Roux A. The contextual effect of the local food environment on residents’ diets: the atherosclerosis risk in communities study. Am J Public Health 2002; 92:1761-7..

The exposures of interest were years of education, household income, and neighborhood income. Education is a consistent predictor of population health, especially in developing contexts ²⁷27. Lynch J, Kaplan G. Socioeconomic position. In: Berkman L, Kawachi I, editors. Social epidemiology. New York: Oxford University Press; 2000. p. 13-35.. Household income reflects material conditions that impact health ²⁷27. Lynch J, Kaplan G. Socioeconomic position. In: Berkman L, Kawachi I, editors. Social epidemiology. New York: Oxford University Press; 2000. p. 13-35., and neighborhood income was utilized because contextual socioeconomic position measures have been shown to influence health after accounting for individual factors ⁷7. Diez Roux AV. Residential environments and cardiovascular risk. J Urban Health 2003; 80:569-89.. Education was classified into three categories: 0-8 years, 9-11 years, and 12+ years. Gross monthly household income was classified into three categories in terms of the minimum Brazilian wage (about US$ 290.00): < 2 minimum wages (< US$580), 2-5 minimum wages (US$580-1,450), and >5 minimum wages (> US$1,450). This categorization was based on the income distribution of participants in this study, previous studies, and the Brazilian Institute of Geography and Statistics (IBGE) 2010 census report ²⁸28. Camargos VP, César CC, Caiaffa WT, Xavier CC, Proietti FA. Imputação múltipla e análise de casos completos em modelos de regressão logística: uma avaliação prática do impacto das perdas em covariáveis. Cad Saúde Pública 2011; 27:2299-313.,²⁹29. Instituto Brasileiro de Geografia e Estatística. Censo demográfico 2010. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2012.. The neighborhood indicator examined was monthly income per capita obtained from the 2000 and 2010 Brazilian census ¹⁹19. Instituto Brasileiro de Geografia e Estatística. Censo demográfico 2000. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2003.,²⁹29. Instituto Brasileiro de Geografia e Estatística. Censo demográfico 2010. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatística; 2012.. We categorized this variable into tertiles. Census tracts were used as proxies for neighborhoods.

Data analysis

The distribution of BMI, obesity, physical activity, and dietary intake by income and education level was first summarized with descriptive and graphical analyses. Spearman correlations between income and education level were calculated. Multivariable linear and logistic regression models were run to examine the association between the outcomes of BMI, obesity, physical activity, and diet with education level, household income, and neighborhood income. We tested for effect modification by sex through stratified analyses and by including interaction terms in the regression models. The associations of BMI with education, household income, and neighborhood income differed significantly by sex (P for interaction terms introduced separately: < 0.0001 for sex and education, < 0.0001 for sex and income, and < 0.0001 for sex and neighborhood), thus, results are stratified by sex. The interaction terms between sex and education, household income, and neighborhood income for the physical activity and diet outcomes were not statistically significant; however, results are shown stratified by sex to facilitate comparison with BMI analyses.

All models were adjusted for age and ethnicity because we conceptualized these variables as possible confounders in the association between socioeconomic position and the outcomes of interest. In order to observe the associations of each socioeconomic position measure with the outcomes independent of the other two socioeconomic position measures, the final models were mutually adjusted for education, household income, and neighborhood income by adding these variables to the unadjusted model. Additionally, we ran intermediate models including only education and household income, but these results were very similar to the final models and were therefore not reported in the tables. Marginal models were used to account for clustering within census tracts ³⁰30. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986; 73:13-22.. We hypothesized that the socioeconomic position measures would be inversely associated with BMI, obesity, lack of compliance with physical activity guidelines, and low fruit and vegetable consumption.

Results

After excluding participants who were missing covariate data, the total sample size was 2,317 women and 1,601 men. The 130 excluded participants were similar in socioeconomic position and risk factor characteristics to those included in the analyses. Table 1 displays selected individual characteristics of the sample, stratified by sex. Men made up 40.8% of the sample and had an average age of 43.5 years, younger than women whose average age was 44.9 years. The ethnic composition of the sample was similar for men and women with the majority identifying as mixed ethnicity or white. Men reported more years of education than women with 18.4% of men reporting 12+ years of education compared to 16.4% of women. Women reported lower household income than men with 31.6% of women in the lowest income category compared to 19.5% of men. However, men and women had similar average neighborhood income. The average BMI of women was 26.5kg/m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657. compared to 25.5kg/m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657. in men, and 22.8% of women were classified as obese compared to 14.9% of men. Men were more likely to eat fruits or vegetables less than 3 days/week than women (40.5% vs. 31.2%, respectively), however, women were more likely to complete fewer than 150 minutes of physical activity per week than men (77.8% vs. 67.4%, respectively).

Table 2 shows the mean differences in BMI by education, household income, and neighborhood income. Introduction of interaction terms into the model indicated that associations of BMI with education, household income, and neighborhood income differed significantly by sex. Thus, results are stratified by sex. In the crude analysis for women, average BMI had an inverse association with education, a positive association with household income, and showed no clear trend with neighborhood income. After adjusting for age, ethnicity, education, and neighborhood, only the BMI and education association remained significant. After adjustment, the mean BMI of women with 0-8 years of education was 1.79 (95% confidence interval – 95%CI: 1.03-2.55) kg/m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657. higher than that of women with 12+ years of education. For men, average BMI had significant positive associations with income and neighborhood and showed no clear trend with education. After adjustment for all covariates, the mean BMI of low-income men was 1.26 (95%CI: 0.58-1.93) kg/m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657. lower than that of high-income men, and the mean BMI of men in the poorest neighborhoods was 0.76 (95% CI: 0.23-1.28) kg/m²2. Church TS, Thomas DM, Tudor-Locke C, Katzmarzyk PT, Earnest CP, Rodarte RQ, et al. Trends over 5 decades in U.S. occupation-related physical activity and their associations with obesity. PLoS One 2011; 6:e19657. lower than that of men in the wealthiest neighborhoods.

The social patterning of the odds of obesity showed similar trends to those shown with BMI in women only (Table 3). After adjusting for household and neighborhood income, the odds of obesity in women with 0-8 years of education were 2.03 (95% CI: 1.36-3.03) times the odds in women with 12+ years of education. In men, however, when BMI was dichotomized into obese and not obese, we no longer observed trends with any socioeconomic position measure after adjusting for all covariates. Low income did appear to have a slight protective effect for men, but this trend was non-significant. No significant associations were found between obesity and neighborhood income for either men or women.

Table 4 displays the odds ratios (OR) of less than 150 minutes/week of physical activity by education, household income, and neighborhood income, stratified by sex. Results are shown stratified by sex to facilitate comparison with BMI analyses. For women, less than 150 minutes/week of physical activity had a significant inverse association with income and education levels. After adjusting for covariates, the odds of less than 150 minutes/week of physical activity in women with 0-8 years of education were 2.12 (95%CI: 1.49-3.02) times the odds in women with 12+ years of education, and the odds of less than 150 minutes/week of physical activity in low-income women were 2.08 (95%CI: 1.47-2.96) times the odds in high-income women. A similar, albeit weaker, pattern was seen in men with regards to education. However, the income trend was not significant in men. Additionally, for men, less than 150 minutes/week of physical activity had a significant inverse association with neighborhood income. After adjusting for other covariates, the odds of less than 150 minutes/week of physical activity in men in the poorest neighborhoods were 2.04 (95%CI: 1.41-2.96) times the odds in men in the wealthiest neighborhoods. Less consistent neighborhood patterns were observed in women.

Table 5 displays the crude and covariate adjusted OR of consuming fruits or vegetables less than three days/week by education, household income, and neighborhood income, stratified by sex. For women, low consumption of fruits and vegetables had significant inverse associations with education and household income. After adjusting for covariates, the odds of consuming fruits or vegetables less than three days/week in women with 0-8 years of education were 1.61 (95%CI: 1.15-2.26) times the odds in women with 12+ years of education, and the odds of consuming fruits or vegetables less than three days/week in low-income women were 2.21 (95%CI: 1.65-2.96) times the odds in high-income women. Similar but weaker trends were found in men for both education and household income. However, the overall trend in fruit and vegetable consumption across categories of education in men was not significant (p-value = 0.0777). No association was found between fruit and vegetable consumption and neighborhood income in either men or women after adjusting for covariates.

Discussion

In developed countries, chronic disease risk factors such as obesity, physical inactivity, and inadequate fruit and vegetable intake have long been associated with low socioeconomic position ⁴4. Kaplan GA, Keil JE. Socioeconomic factors and cardiovascular disease: a review of the literature. Circulation 1993; 88(4 Pt 1):1973-98.,⁵5. McLaren L. Socioeconomic status and obesity. Epidemiol Rev 2007; 29:29-48.,⁶6. Dinsa GD, Goryakin Y, Fumagalli E, Suhrcke M. Obesity and socioeconomic status in developing countries: a systematic review. Obes Rev 2012; 13:1067-79.. The social gradient in developing countries has been less consistent, and most studies in these contexts have investigated BMI and obesity exclusively ⁹9. Monteiro CA, Conde WL, Popkin BM. Income-specific trends in obesity in Brazil: 1975-2003. Am J Public Health 2007; 97:1808-12.,¹¹11. Sobal J, Stunkard AJ. Socioeconomic status and obesity: a review of the literature. Psychol Bull 1989; 105:260-75.,¹²12. Marins VM, Almeida RM, Pereira RA, Sichieri R. The association between socioeconomic indicators and cardiovascular disease risk factors in Rio de Janeiro, Brazil. J Biosoc Sci 2007; 39:221-9.,¹³13. Boissonnet C, Schargrodsky H, Pellegrini F, Macchia A, Marcet Champagne B, Wilson E, et al. Educational inequalities in obesity, abdominal obesity, and metabolic syndrome in seven Latin American cities: the CARMELA Study. Eur J Cardiovasc Prev Rehabil 2011; 18:550-6.,¹⁴14. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H. Social patterning of chronic disease risk factors in a Latin American city. J Urban Health 2008; 85:923-37.. Studies in Brazil over the past several decades suggest a shifting social gradient as the country progresses through the nutrition transition. In 1989, a positive association between income and obesity was found in both Brazilian men and women ¹⁶16. Monteiro CA, Benicio MH, Conde WL, Popkin BM. Shifting obesity trends in Brazil. Eur J Clin Nutr 2000; 54:342-6.. 14 years later, a nationally representative sample of Brazilian adults showed that high-income groups still had higher obesity prevalence than low-income groups, but low-income groups had a higher rate of increase ⁹9. Monteiro CA, Conde WL, Popkin BM. Income-specific trends in obesity in Brazil: 1975-2003. Am J Public Health 2007; 97:1808-12.. This shifting pattern is not unique to Brazil. A review of low-income countries found a higher prevalence of overweight in high-education groups in 35 of 39 countries; however, the rate of increase in overweight prevalence was higher in the low-education groups in 21 countries ³¹31. Jones-Smith JC, Gordon-Larsen P, Siddiqi A, Popkin BM. Is the burden of overweight shifting to the poor across the globe? Time trends among women in 39 low- and middle-income countries (1991-2008). Int J Obes (Lond) 2012; 36:1114-20.. Our study demonstrates a next step in the shift reported by previous studies in Brazil.

The social patterning of obesity greatly depends on the region of Brazil examined. In 1996 in the less developed northeastern region of Brazil, obesity in men was positively associated with income but not with education, and obesity in women was positively associated with income and inversely with education ³¹31. Jones-Smith JC, Gordon-Larsen P, Siddiqi A, Popkin BM. Is the burden of overweight shifting to the poor across the globe? Time trends among women in 39 low- and middle-income countries (1991-2008). Int J Obes (Lond) 2012; 36:1114-20.. In the more developed southeastern region, the same study found a positive association between obesity and income in men and a clear inverse relationship between obesity and education in women ³²32. Monteiro CA, Conde WL, Popkin BM. Independent effects of income and education on the risk of obesity in the Brazilian adult population. J Nutr 2001; 131:S881-6.. Our study was carried out in a developed Brazilian city and was consistent with the findings in the more developed southeastern region; the women with the lowest education in our sample had the highest average BMI and obesity prevalence.

Obesity was once a disease of the elite, but as a country’s GNP (Gross National Products) increases, the social gradient reverses. For women, the crossover takes place earlier than for men, when a country reaches a GNP per capita of approximately US$2,500 ³³33. Monteiro CA, Moura EC, Conde WL, Popkin BM. Socioeconomic status and obesity in adult populations of developing countries: a review. Bull World Health Organ 2004; 82:940-6.. Our findings of an inverse association of female BMI and obesity with education and a positive association between male BMI and income are consistent with other studies in Latin America that report similar gender differences in social patterning ¹⁴14. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H. Social patterning of chronic disease risk factors in a Latin American city. J Urban Health 2008; 85:923-37.,³⁴34. Fleischer NL, Diez Roux AV, Hubbard AE. Inequalities in body mass index and smoking behavior in 70 countries: evidence for a social transition in chronic disease risk. Am J Epidemiol 2012; 175: 167-76..Throughout Latin America, low socioeconomic position continues to protect against obesity in men while increasing the risk of obesity in women.

The opposite social patterning of BMI and obesity in men and women could be partly influenced by differing ideal body images for men and women. Women often find smaller body sizes more attractive while men are generally dissatisfied with small bodies ³⁵35. Feingold A, Mazzella R. Gender differences in body image are increasing. Psychol Sci 1998; 9:190-5., and well-educated women at any given BMI are more sensitive to this image ³⁶36. McLaren L, Kuh D. Women’s body dissatisfaction, social class, and social mobility. Soc Sci Med 2004; 58:1575-84.. The directionality difference in men and women could also be due to body image and selection factors, such as heavier individuals experiencing discrimination when pursuing higher paid jobs, playing a more important role in women than in men. The strong association of BMI and obesity with education in women could be due to increased exposure among well-educated women to media that promote thin body shapes; well-educated women may also have a stronger desire to take up less physical space in a patriarchal society ³⁶36. McLaren L, Kuh D. Women’s body dissatisfaction, social class, and social mobility. Soc Sci Med 2004; 58:1575-84.. The weaker associations observed for income than for education in women may be due to the fact that women in every education level made less money than men. Consequently, income may not accurately capture a woman’s social position.

In developed countries, prevalence of physical inactivity is higher in low socioeconomic position groups ²¹21. U.S. Department of Health and Human Services. Physical activity and health: a report from the Surgeon General. Atlanta: U.S. Department of Health and Human Services; 1996.,²³23. Pate RR, Pratt M, Blair SN, Haskell WL, Macera CA, Bouchard C, et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 1995; 273:402-7., and a similar pattern has begun to emerge in developing countries ³⁷37. Bauman A, Ma G, Cuevas F, Omar Z, Waganivalu T, Phongsavan P, et al. Cross-national comparisons of socioeconomic differences in the prevalence of leisure-time and occupational physical activity, and active commuting in six Asia-Pacific countries. J Epidemiol Community Health 2011; 65: 35-43.. Nevertheless, our study is the first in Brazil to find a social gradient of physical inactivity similar to that of developed countries. Studies in other regions of Brazil and Latin America found either a positive association ²²22. Hallal PC, Victora CG, Wells JC, Lima RC. Physical inactivity: prevalence and associated variables in Brazilian adults. Med Sci Sports Exerc 2003; 35:1894-900.,³⁸38. Alves JG, Figueiroa JN, Alves LV. Prevalence and predictors of physical inactivity in a slum in Brazil. J Urban Health 2011; 88:168-75. or no association ¹⁴14. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H. Social patterning of chronic disease risk factors in a Latin American city. J Urban Health 2008; 85:923-37.,²²22. Hallal PC, Victora CG, Wells JC, Lima RC. Physical inactivity: prevalence and associated variables in Brazilian adults. Med Sci Sports Exerc 2003; 35:1894-900. between socioeconomic position and physical inactivity. Our findings of strong income and education patterning of physical activity, especially in women, demonstrate the shifting patterns in physical activity practices in urban regions of Brazil where the elite may have resources permitting higher levels of activity, especially for leisure ³⁹39. Barros MV, Nahas MV. Comportamentos de risco, auto-avaliação do nível de saúde e percepção de estresse entre trabalhadores da indústria. Rev Saúde Pública 2001; 35:554-63..

Education and income had independent effects on physical activity in women and to an extent in men in Belo Horizonte. Those with less education may lack the health-knowledge and social capital to sustain healthy physical activity levels, and those with lower income tend to have less leisure time and opportunity for recreational activities ¹¹11. Sobal J, Stunkard AJ. Socioeconomic status and obesity: a review of the literature. Psychol Bull 1989; 105:260-75.,⁴⁰40. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 2000; 894:i-xii, 1-253.. Additionally, low-education and low-income populations have less financial and social resources to avoid the environments that promote physical inactivity ⁴¹41. World Health Organization. Diet, nutrition and the prevention of chronic diseases. Geneva: World Health Organization; 2003. (WHO Technical Report Series, 916)..

In developed countries, fruit and vegetable consumption is associated with higher levels of education ⁴²42. Dehghan M, Akhtar-Danesh N, Merchant AT. Factors associated with fruit and vegetable consumption among adults. J Hum Nutr Diet 2011; 24: 128-34., and similar patterns are beginning to appear in developing nations ⁴³43. Hosseinpoor AR, Bergen N, Kunst A, Harper S, Guthold R, Rekve D, et al. Socioeconomic inequalities in risk factors for non communicable diseases in low-income and middle-income countries: results from the World Health Survey. BMC Public Health 2012; 12:912.,⁴⁴44. Colón-López V, Banerjee G, Gertz AM, Ortiz AP, Calo W, Finney-Rutten LJ, et al. Behavioral correlates of fruit and vegetable intake in Puerto Rico: results from the Health Information National Trends Survey. P R Health Sci J 2013; 32:194-9.,⁴⁵45. Doku D, Koivusilta L, Raisamo S, Rimpelä A. Socio-economic differences in adolescents’ breakfast eating, fruit and vegetable consumption and physical activity in Ghana. Public Health Nutr 2013; 16:864-72.. Our results of low fruit and vegetable consumption in low-income and low-education groups of men and women were consistent with other findings in Brazil ⁴⁶46. Mondini L, Moraes SA de, Freitas IC, Gimeno SG. Fruit and vegetable intake by adults in Ribeirão Preto, Southeastern Brazil. Rev Saúde Pública 2010; 44:686-94.,⁴⁷47. Jaime PC, Monteiro CA. Fruit and vegetable intake by Brazilian adults, 2003. Cad Saúde Pública 2005; 21:S19-24.. In recent years, sugary and processed foods have become more available in Brazil ⁴⁸48. Levy-Costa RB, Sichieri R, Pontes NS, Monteiro CA. Household food availability in Brazil: distribution and trends (1974-2003). Rev Saúde Pública 2005; 39:530-40., and healthy foods, such as fruits and vegetables, are more costly than low-quality calorie dense foods ⁴⁹49. Drewnowski A, Darmon N. The economics of obesity: dietary energy density and energy cost. Am J Clin Nutr 2005; 8 Suppl 1:S265-73.. High-income populations have greater purchasing power for more expensive healthy foods, while low-income populations are more constrained in their choices ⁴¹41. World Health Organization. Diet, nutrition and the prevention of chronic diseases. Geneva: World Health Organization; 2003. (WHO Technical Report Series, 916).. The additional independent inverse effect of education on fruit and vegetable consumption indicates that education captures aspects of socioeconomic position that affect access to fruits and vegetables that income alone does not.

In men we found an inverse association between physical inactivity and poor diet and socioeconomic position, a seemingly contradictory positive association between BMI and socioeconomic position, and no association between obesity and socioeconomic position. Although the behavioral risk factors for obesity have begun to change in both men and women in Brazil, it is possible that the effects on social patterning of BMI have not yet become apparent in men. Additionally, the significant positive association in men of income with BMI but not with obesity is plausible because BMI as a continuous measure is more sensitive to differences that the dichotomized obesity variable could not detect. While our study suggests that men with higher incomes may have higher BMI, our findings also indicate that higher income men have the resources to consume healthy diets and participate in physical activity to prevent obesity. If the pattern of physical inactivity and poor diet in men of low socioeconomic position persists, we may observe an eventual shift in the BMI and obesity trends in men in Brazil.

Area-level income was not a consistent predictor of health outcomes in this study. Neighborhood income was not associated with any of the three outcomes in women; however, there was evidence of a positive association with BMI and an inverse association with physical activity in men. Studies of neighborhood effects on health are almost exclusively limited to developed countries where residential environments with higher socioeconomic position are associated with better health, after adjusting for individual socioeconomic position ⁷7. Diez Roux AV. Residential environments and cardiovascular risk. J Urban Health 2003; 80:569-89.. Less deprived neighborhoods may have better access to health promoting resources such as recreational facilities and infrastructure, transportation, and healthy food stores. Few studies have examined this association in developing countries. In Argentina, lower area-level education was associated with higher BMI after adjusting for individual income and education in men and women, even though the association with individual-level socioeconomic position was not significant in men ¹⁴14. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H. Social patterning of chronic disease risk factors in a Latin American city. J Urban Health 2008; 85:923-37.. This same study found no area-level association with physical activity or diet in Argentina ¹⁴14. Fleischer NL, Diez Roux AV, Alazraqui M, Spinelli H. Social patterning of chronic disease risk factors in a Latin American city. J Urban Health 2008; 85:923-37.. The neighborhood socioeconomic position indicator we used may not appropriately capture the environmental variation relevant to the risk factors we studied. It is also possible that there is less environmental heterogeneity in our sample than in other contexts or studies.

The findings of our study have important social implications for the city of Belo Horizonte. Socioeconomic disparities are highly prevalent and persistent in Brazil, especially in urban areas ⁵⁰50. World Bank. Inequality and economic development in Brazil. Washington DC: World Bank; 2004.. The social patterning of cardiovascular disease risk factors and behaviors in Brazil is evolving to mirror the cardiovascular health profile of developed countries with the poor being at the greatest disadvantage. These shifts may result in a double burden of high chronic and infectious disease levels in these populations ⁵¹51. World Health Organization. Global health risks: mortality and burden of disease attributable to selected major risks. Geneva: World Health Organization; 2009., serving only to widen the already large gap between the poor and wealthy in Brazil and decrease the social mobility of those with the lowest socioeconomic position ⁵⁰50. World Bank. Inequality and economic development in Brazil. Washington DC: World Bank; 2004.. Without policies and interventions to address the cycle of poverty and poor health, this ever-increasing inequality could have substantial future economic and social costs for Brazil.

Our study had a number of strengths including a large sample size that was representative of an urban center in Brazil, the use of a validated survey to measure both leisure and occupational physical activity, the measurement of height and weight by a trained interviewer, and the inclusion of both individual and contextual measures of socioeconomic position. Our study also had limitations. The cross-sectional design precludes us from studying the possible causal interrelations between education, household income, and neighborhood income. These three variables are undeniably linked, and cross-sectional data prohibits the investigation of their sequencing. Furthermore, education, household income, and neighborhood income are limited measures of socioeconomic position, and future analysis could include additional variables such as occupation, wealth, and more specific and targetable neighborhood measures. In addition, we restricted our analysis to the Barreiro and West districts of Belo Horizonte, which may affect the external validity of our findings. We also used census tracts, which are variable in size, as a proxy for neighborhoods. However, this has been done in previous studies, and we were still able to detect associations at the neighborhood level. Additionally, the scope of the survey allowed us to gather only limited data on diet. Furthermore, due to missing data, we excluded 130 participants from our final sample; this exclusion has the potential to introduce bias if the examined associations in the excluded participants differ from the associations in the included participants. It is also worth noting that the survey was carried out in 2009, and the magnitude of the associated variables, or even the definition of variables themselves, may have undergone changes from 2009 to the present.

Our study identified the social patterning of various cardiovascular disease risk factors and behaviors in a large Brazilian city. BMI and obesity were inversely associated with education in women, and BMI was positively associated with income in men. In addition, physical inactivity and poor diet were associated with lower incomes and less education independently in both genders. In contrast, neighborhood income was not consistently related to health. Longitudinal analysis of this patterning is necessary to understand the causes of non-communicable diseases in the Latin American context. Understanding the drivers of these patterns will allow for the development of appropriate policy and interventions that address these health disparities.

Acknowledgments

The authors would like to thank the staff at the Belo Horizonte Observatory for Urban Health and the University of Michigan’s Center for Social Epidemiology and Population Health as well as the data collectors and respondents in the participating districts of Belo Horizonte. This work was supported by the National Institutes of Health (NIH) Fogarty International Center (grant n. D43 TW009315 and 5R03 TW008105), CNPq-409688/2006-1, FAPEMIG-CDS APQ 00677-08, and the National Health Fund (FNS)-25000.102984/2006-97/Brazil.

References

Publication Dates

History