Overweight and obesity and associated factors in adults in a poor urban area of Northeastern Brazil

REV BRAS EPIDEMIOL 2020; 23: E200036 RESUMO: Introdução: As mudanças produzidas no processo saúde/doença, sobretudo no campo da nutrição, corroboram a substituição das carências nutricionais com emergência do excesso de peso (sobrepeso/ obesidade). Objetivo: Analisar a prevalência e fatores associados ao excesso de peso em adultos residentes em uma área urbana carente do Recife, Nordeste do Brasil. Métodos: Trata-se de um estudo transversal analítico, com uma amostra de 644 adultos de 20 a 59 anos. Analisaram possíveis associações do excesso de peso aos fatores demográficos, socioeconômicos, comportamentais e morbidades por meio de regressão de Poisson, considerando como estatisticamente significantes aquelas com valor de p < 0,05. Resultados: A prevalência do excesso de peso foi de 70,3%, sendo menor na faixa de 20–29 anos e maior na faixa etária de 30–39 anos, e estabilizando-se nas demais. No modelo de regressão multivariado, foi observado que a faixa etária, classe econômica, diabetes mellitus (DM) e hipertensão arterial sistêmica (HAS) mostraram-se diretamente associada ao excesso de peso, enquanto a variável consumo semanal de feijão mostrou-se inversamente associada. A alta prevalência do excesso de peso encontrado pressupõe que as comunidades carentes das quais os indivíduos fazem parte já se incluem no processo de transição nutricional que está em curso no país. Conclusão: Os resultados significativos de sobrepeso/obesidade detectados na área urbana carente estudada, impõe a necessidade de incluir esse problema como prioridade de saúde pública nessas comunidades.

Overweight and obesity and associated factors in adults in a poor urban area of Northeastern Brazil

INTRODUCTION
The great changes that occurred in the health/disease process from the second half of the twentieth century began to present a very peculiar configuration in the field of nutrition, typified by the overlapping of global and specific nutritional deficiencies due to the epidemic or pandemic emergence of overweight and obesity 1 .
Overweight and obesity are characterized by the accumulation of body fat, exceeding the acceptable standards of anthropometric normality in different degrees and belonging to the group of chronic non-communicable diseases (CNCDs) 2 .They act as important risk factors for the morbidity and mortality of adult populations, being associated with 63% of the global total of deaths caused by the CNCD.Of that amount, 78% of mortality occurs in middle and low income countries 3 .
The worldwide prevalence of overweight/obesity has shown a rapid and progressive increase in the last decades.Currently, 2.1 billion adults have this condition, which represents almost 30% of the world population.It should also be noted that, from 1980 to 2013, overweight increased by 27.5% among adults 4 .

METHODS
This is a descriptive and analytical cross-sectional study, based on data from the research "Health, nutrition and assistance services in a slum population in Recife: a baseline study", developed by Instituto de Medicina Integral Professor Fernando Figueira (IMIP) in partnership with the Nutrition Department of Universidade Federal de Pernambuco (UFPE) and the Recife City Hall.Data collection was home based and took place between June and December 2014, in a poor urban area known as the Coelhos community, located in the Boa Vista neighborhood, in the municipality of Recife, capital of the State of Pernambuco.
To calculate the sample, a group of 3,816 adults aged 20 to 59 years was used as reference.As was estimated by the Primary Care Information System (SIAB) in Recife, there is a prevalence of 51.5% of overweight and obesity of adults in this age group in the Recife Metropolitan Region of Pernambuco 7 .An estimate error of 4% and a confidence interval of 95% were assumed, and 10% was added to compensate for possible losses, cases of non-response or questionnaires eventually invalidated by inconsistencies, thus obtaining an initial sample of 570 participants.The final sample included a total of 644 adults.The number of observations assessed here constitutes a subset of the representative sample of people over 20 years old, calculated to represent the universe of adults in the Coelhos community.
Sampling was probabilistic, and the adults were selected by simple random drawing, without substitution.Pregnant women, individuals with congenital or acquired physical limitations which made anthropometric measurements impossible to measure, and cases with visible edema or psychic disorders that hindered collaboration were excluded.
Weight was measured on a digital scale, Seca ® 876, with a capacity of up to 250 kg and a scale of 100 g, with individuals barefoot, wearing minimal clothing and without any objects in their pockets, hands or head.For height measurement, a portable millimeter stadiometer, brand Alturexata LTDA., with precision of up to (1 mm) was used.The volunteers were in an upright position, barefoot, with upper limbs dangling along the body.In order to ensure accuracy, measurements were taken in duplicates and, when the difference between them exceeded 0.5 cm for height and 100 g for weight, the two measurements were repeated and noted with closer values, and then the average of these measures was used.
The independent variables were presented categorically, classified as: • demographic: sex (male and female) and age group (20-29, 30-39, 40-49 and 50-59); • socioeconomic: economic class, assessed based on the criteria of the Brazilian Association of Research Companies (ABEP) 14 , defined based on a points system that considers the possession of goods and the education level of the head of the family (B1/B2, C1/C2 and D/E), schooling (illiterate/incomplete elementary school, complete elementary school/incomplete primary school, complete primary school/ incomplete secondary school and complete secondary school/complete higher education/incomplete higher education), race/ethnicity, which was self-reported (white, black, brown and others), occupation ((not working, unemployed, sporadic work, social benefits, employed/self-employed), housing -wall (brick/masonry and others), number of rooms (> 4 and ≤ 4), basic sanitation -garbage disposal (public collection and others), waste disposal and water supply (general network and others); • behavioral: weekly food consumption -beans, vegetables, fruits, soda or artificial juice, (1-2 times a week or never, 3-4 times and 5-7 times), meats with excess fat (no and yes) and physical activity (sufficiently active and insufficiently active); • diabetes and high blood pressure (no/yes).
Food consumption was assessed using the weekly consumption questionnaire used by the Ministry of Health 15 .To determine the level of physical activity, the International Physical Activity Questionnaire (IPAQ) was used, in its short version 16 .This instrument measures the frequency and duration of moderate, vigorous physical activities and walks performed in the last week for at least 10 continuous minutes, including standardized exercises, sports, occupational and recreational physical activities performed at home, in free time, as a means of transportation and at leisure.

REV BRAS EPIDEMIOL 2020; 23: E200036
The criteria established by IPAQ refer to four categories of the level of physical activity: very active, active, irregularly active and sedentary.For analysis purposes, a these variables were re-categorized into: • "sufficiently active" (very active + active), applied to people who reported practicing vigorous activity with a frequency greater than or equal to three times a week for 20 minutes or more, or who practiced moderate activity, or walking with a frequency greater than or equal to five times a week for at least 30 minutes, or any activity that added up was equivalent to a frequency greater than or equal to five times a week and greater than or equal to 150 minutes a week (walking + moderate activity + vigorous activity); • "insufficiently active" (irregularly active + sedentary), people who do not fit the aforementioned criteria.
Regarding morbidities, the diagnosis of diabetes was performed by biochemical examination, with individuals with blood glucose ≥ 126 mg/dL or with a report of use of a hypoglycemic agent being considered "cases" 17 .Blood pressure was measured according to standardized procedures 18 .Two measurements were taken at different times (interval of 15 minutes), and adults with systolic blood pressure ≥ 140 mmHg, and/or with diastolic blood pressure ≥ 90 mmHg, or with reports of use of anti-hypertensive agents.
Initially, descriptive analyzes were performed in order to characterize the frequency distribution of the variables under study.Subsequently, bivariate analyzes were performed using simple Poisson regression, to show possible associations with excess weight with the independent variables.In the adjusted model, the criterion for inclusion of the variables was the association in the crude analysis with excess weight, with a value of p < 0.20.The variables with p < 0.05, obtained through multivariate Poisson regression, with robust adjustment of the standard error, remained in the final model.The results were expressed by prevalence ratio (PR) and 95%CI.Statistical analyzes were performed using the SPSS software, version 13.0 (SPSS Inc., Chicago) and Stata, version 13.0 (StataCorp., College Station, United States).
This study was approved by the IMIP Research Ethics Committee, protocol No. 4017-14, in accordance with the requirements of CNS Resolution No. 466/12.All respondents were informed about voluntary participation in the research and signed the Informed Consent.

RESULTS
In the final sample of the present study, there were a total of 644 adults, however, there were differences in sample values in some variables, due to the loss of responses for incomplete questionnaires and/or inconsistent data.
The nutritional status of the studied population is shown in Figure 1.The prevalence of overweight and obesity were similar, around 35%, exceeding the prevalence of eutrophy.Furthermore, it is detailed that 3.4% of the total obese people had severe obesity, which is practically twice the frequency of cases of weight deficit.The joint frequency of overweight/obesity, representing excess weight, was 70.3%.
The variables sex, age group, economic class, race/ethnicity, occupation, number of rooms, water supply (Table 1), weekly consumption of beans, consumption of meat with excess fat, high blood pressure and diabetes (Table 2) showed an significant association with excess weight (p < 0.20).
The PRs adjusted through Poisson's multivariate regression analysis showed that categories C1/C2 and B1/B2 presented an association with excess weight, in relation to the reference category (D/E), thus revealing that the higher the class, the the higher the prevalence of overweight.
Continue...The 30-39 years and 40-49 years age groups also showed an association with the outcome, in relation to the reference category (20-29 years), that is, as the age group increases, the prevalence of the problem was higher in relation to the reference category.The weekly frequency of bean consumption (≤ 2 times/week and never) showed an association with excess weight, in relation to the reference category (5-7 times/week), the lower the weekly frequency of bean consumption, the greater the prevalence of overweight.The referred morbidities diabetes and high blood pressure were also associated with the outcome, those who reported these morbidities had a higher frequency of excess when compared to those who did not (reference category).Such variables remained significantly associated with the outcome p < 0.05 (Table 3).

DISCUSSION
The high prevalence of excess weight found presupposes that poor or low-income communities are already included in the nutritional transition process.This result would be unusual in an urban environment of marked poverty, since a few years ago the forecast would be high frequencies of weight deficit 9 .
The most current study that can be taken as a reference was carried out in 2009, in a sample of 3,214 adults from deprived urban areas of Maceió (AL) 6 .This research revealed that 41.2% of adults were overweight/obese, while our study found a prevalence of 70.3%, well above the value found in Brazil in 2013 (56.9%) 5 .On an international level, practically identical prevalence rates to those of the Coelhos community (around 73%) were found in studies with poor populations in the United States 19 and Afghanistan 20 .
The much higher prevalence obtained in this assessment could result in three possible interpretations.As excess weight represents a rapidly progressive epidemic in Brazil, it is acceptable that a marked difference in five years may result in the very rapid pace of this problem's increase.The second version would accept that the situation of adults in the analyzed area may be very different from that found in Maceió, in a larger sample distributed  over several slum areas.A third conjecture would be that favela populations start to reproduce, and even exceed, a generalized pattern for the whole country, as part of the epidemiological homogenization expressed as the most updated scenario of the Brazilian population's nutritional status.This may represent the most consistent interpretation, although without the support of sequential, up-to-date and representative data on populations living in deprived urban areas.
In the Coelhos community, there was a statistically significant association between overweight/obesity and the age group, highlighting a higher prevalence among adults aged 30 to 39 years, and then stabilizing.In Brazil 4 , a study on the nutritional status of the beneficiaries of the Bolsa Família Program 21 and on low-income women in Rio de Janeiro 22 also found an association between the problem and the age group, showing a higher prevalence of overweight among adults aged 40 years or over and 50-59 years, respectively.This finding may be related to the decrease in the level of physical activity 23 , as well as basal metabolism and hormonal changes that happen with the aging process, which lead the body to store more fat 24 .Internationally, research carried out in the district of Kalutara, in Asia, found a higher prevalence of overweight in low-income adults, from the age of 40, with a reduction at the age of 50 25 .It is likely that the differences in life ecosystems specific to each low-income population may justify these mismatches in the results.
There was no association between overweight and education.According to VIGITEL 26 , the frequency of overweight tends to decrease with increasing schooling, however we did not observe this in this study, since the prevalence of overweight was similar in different age groups, regardless of the level of education.This is probably because the low socioeconomic condition and the social context in which the study population is inserted ends up favoring the acquisition of cheaper and caloric foods.
In the higher classes of the sample, the highest prevalence of overweight was identified (B1/B2).However, this represented only 5.8% of the sample.The size of the study sample may have influenced this result, since a higher prevalence of the problem was expected in the poorest classes.However, it should be considered that the population studied is a poor community and, therefore, a homogeneous population, so the expectations may have been found with a larger sample, such as the case of the studies conducted in Maceió 6 and in Ceará 27 , in which a higher prevalence of overweight was observed among adults with lower income.
A higher frequency of the problem was found among the black race/ethnicity and among adults without an occupation, although no significant association was observed.It is important to emphasize that being overweight has shifted its focus of risk to the most socially disadvantaged ethnicities, such as blacks and browns or other groups close to the condition of poverty, such as rural families and lower income strata 28,29 .
In the present study, approximately 70% of the population consumes beans five or more times a week.According to the Family Budget Survey, in 2008 and 2009, beans are still among the most consumed foods by the Brazilian population (≅ 70%) 30 .However, the low weekly consumption of beans (≤ 2 times) was shown to be associated with excess weight in this study, corroborating the study carried out in Belém (PA) on the consumption of beans in adults 31 , which may indicate a higher consumption of ultra-processed foods, in detriment to healthy foods such as beans.Furthermore, no associations were observed with other food consumption variables.
There was no association between the level of physical activity and the prevalence of overweight.The coexistence of 70.2% of sufficiently active individuals and approximately 70% of overweight/obesity are apparently conflicting results; similar findings were also observed in Pernambuco 7 .The high prevalence of active adults can be justified by the fact that the measurement of physical activity was performed by an instrument that considers activities performed in leisure, commuting, domestic and occupational activities, as well as by the socioeconomic condition of the population, which causes active commuting, occupational tasks and domestic activities to be the predominant type of physical activity, to the detriment of physical activity performed during leisure time, which is more common in developed countries 32,33 .
As expected, according to what other studies show, high blood pressure and diabetes were associated with overweight/obesity [34][35][36] .It is a trilogy of comorbidities (high blood pressure, type II diabetes and overweight/obesity) that are commonly observed together.
The cross-sectional design stands out as a limitation of the study, as it constitutes a limitation in the analysis of the association between the associated variables and the outcome, due to the impossibility of inferring a causal relationship, disregarding the before/after relationship which, due to formal logic, must condition the relationship.
The fact that the study was carried out in a poor community is considered to be a positive point, considering that few studies are still carried out in these communities, and the knowledge of the health situation of poor populations is important for planning interventions.

CONCLUSION
The 70% prevalence of overweight/obesity in a poor area of Recife is well above the results identified in other urban populations with similar characteristics and even much higher in representative samples from Brazil.The high prevalence of the problem, standing at around 30% above the frequencies found in studies published in the country after 2003 implies that the population analyzed is included in the rapid process of nutritional transition that the country has experienced in the last 40 years.
From the analytical point of view, of the 22 groups of variables investigated, age group, economic class, weekly bean consumption, diabetes and high blood pressure, were the variables that made up the final model, adjusted to define the risks associated with overweight.
Conclusively, the high prevalence of overweight/obesity detected in the studied underprivileged urban area imposes the need to include this problem as a public health priority in these communities.Furthermore, it imposes the recommendation to expand similar studies and also a qualitative approach to other similar communities that are spread across the national territory, in order to get a glimpse of a still somewhat unknown situation and to signal priority interventions to them.

Table 1 .
Prevalence and crude prevalence ratio (PR) of overweight in adults (20-59 years) in an poor urban area, according to demographic and socioeconomic variables.Recife, 2014.
95%CI: 95% confidence interval; MW: minimum wage; a sporadic/odd jobs/street work; b never worked and housewives; c has worked before, but had been unemployed for 30 days or more; d retired, pensioner, provisional benefits; *differences in sample values in some variables are due to the loss of observations: schooling (n = 633), race/ethnicity (n = 641), occupation (n = 642), water supply (n = 643).

Table 2 .
Prevalence and crude prevalence ratio (PR) of overweight in adults (20-59 years) in a poor urban area, according to behavioral variables and morbidities.Recife, 2014.

Table 3 .
Adjusted analysis of excess weight in adults (20-59 years) in a poor urban area.Recife, 2014.