The Consumption of Ultra-Processed Foods Is Associated with Abdominal Obesity in Individuals on Hemodialysis in Brazil

: The consumption of ultra-processed foods is associated with abdominal obesity, but this association has not yet been investigated in individuals on hemodialysis. This cross-sectional study, developed with 1014 individuals, aimed to analyze the prevalence of abdominal obesity and its association with the consumption of ultra-processed foods in individuals on hemodialysis in Brazil. The consumption of ultra-processed foods was measured using a questionnaire, whose list of foods was based on the Surveillance System of Risk and Protective Factors for Chronic Non-Communicable Diseases through Telephone Enquiry. Waist circumference was used to determine the presence of abdominal obesity. The difference in proportions between the independent variables and the outcome was verified using Pearson’s chi-square test. Hierarchical logistic regression was used to determine the association between the outcome and the independent variables. The prevalence of abdominal obesity was 77.22%. In the final model, higher levels of consumption of ultra-processed foods, being male and an adult, having between 8 and 11 years of schooling, and practicing physical activity below the recommended level increased the chances of having abdominal obesity. This research provides a basis for the creation and improvement of public policies that promote improvements in these health determinants.


Introduction
The group of ultra-processed foods (UPFs) consists of industrial formulations that are typically made with five or more ingredients, including substances and additives used in the manufacture of processed foods, such as sugar, oils, fats, and salt, in addition to antioxidants, stabilizers, and preservatives [1].An expansion of the UPFs available in the world's food supply, both in volume and variety, is occurring primarily in high-income countries, but this is also growing in middle-income countries [2].This provides evidence that a transition to a more processed diet is underway globally and accelerating [3].In Brazil, data from the latest Family Budget Survey show that the contribution of UPFs to the total calories purchased increased from 12.6% in 2002-2003 to 18.4% in 2017-2018.Furthermore, there was a decrease in the caloric participation of in natura foods [4].
In this context, statistics indicate that excess body weight affects more than two billion people, approximately 30% of the world's population [5].In Brazil, data from the Food and Nutrition Surveillance System reveal that the prevalence of obesity in adults increased from 15.6% in 2009 to 28.5% in 2019 [6].Excess fat deposition in the central region is called abdominal obesity and is associated with greater health risks, mainly cardiometabolic complications [7][8][9][10].For diagnosis, one of the simplest anthropometric measurements is waist circumference (WC) [11], recommended by the World Health Organization for screening the risk of metabolic and cardiovascular diseases [12], in addition to being a better predictor of these diseases than body mass index (BMI) [11,[13][14][15].Data from representative national samples in the United States [16], United Kingdom [17], Australia [18], and Korea [19], and a longitudinal study with civil servants in Brazil [20], found that higher levels of consumption of UPFs were positively associated with abdominal obesity.
Chronic kidney disease (CKD), as well as abdominal obesity, is a non-transmissible chronic disease and includes abnormalities in renal function or structure for a period greater than or equal to three months [21].It can be classified into five stages, and in the fifth stage the individual is considered an end-stage renal patient [21].At this stage, there is a need for renal replacement therapy, with hemodialysis being the main treatment modality [21,22].CKD affects 10% of adults worldwide and results in about 1.2 million deaths each year [23].It is estimated that in 2040 it will become the fifth-leading cause of death in the world [23].In Brazil, the total estimated number of individuals on dialysis during the year 2019 was 144,779, of which 92.6% were on hemodialysis [24].
However, the relationship between the consumption of UPFs and abdominal obesity in individuals with CKD undergoing hemodialysis has not yet been investigated in studies, and this is the first one to undertake this.Therefore, this study aims to analyze the prevalence of abdominal obesity and its association with the consumption of ultra-processed foods in individuals on hemodialysis in Brazil.

Design
This is an epidemiological, observational, and cross-sectional investigation, carried out with individuals undergoing hemodialysis in clinics in the metropolitan region of a state located in southeastern Brazil.

Population
This study was carried out in all hemodialysis units that serve chronic kidney disease patients in the metropolitan region of a state located in southeastern Brazil, issuing invitations to participate to all individuals.The region studied comprises 7 municipalities and 11 hemodialysis clinics.
The population of this study consisted of 1014 individuals (Figure 1).To participate in this research, the individuals were required to meet the following inclusion criteria: be 18 years old or older; have a confirmed diagnosis of CKD in the medical records; be on hemodialysis treatment for at least six months; and be ambulatory.We excluded the following groups from the survey: individuals under contact precaution; those with an acute condition that limited their participation in the study; those who did not reside in one of the municipalities in the studied region; those who were transferred to another clinic; and those with missing data on the outcome.
Out of the 1351 individuals undergoing hemodialysis in the region studied, after the exclusions, we obtained a final population of 1014 individuals (Figure 1).

Data Collection
Data collection took place from February to September 2019, after inviting all individuals undergoing hemodialysis, who also attended the units in the studied region, to participate.All assessments were performed by previously trained researchers.
Everyone was approached during their stay at the clinic, responding, during the hemodialysis session, to a semi-structured questionnaire on sociodemographic characteristics, lifestyle habits, and food consumption.Then, after the session, this was submitted for the collection of anthropometric data.

Sociodemographic and Clinical Characteristics
The variables of sex, age range, marital status, and race/color were dichotomized, respectively, into female and male, adult (18-59 years) and elderly (>60 years), with a partner and no partner, and white and non-white (self-declared race/color).
The categories used to define income were based on the family's minimum wage, defined as the lowest wage a company can pay an employee according to the number of hours worked.This has Brazilian validity by law and is reassessed annually based on the cost of living of the population.Its creation was based on the minimum amount that a person spends to ensure their survival.In the year this study was carried out, the minimum wage was R$998.00 or $235.37 USD.
The variables of education, work activity and smoking were separated into three categories, namely: <8 years, 8-11 years, and >11 years of study; with paid activity, no paid activity, and retired or on sick leave; and smoker, former smoker, and never smoked.

Data Collection
Data collection took place from February to September 2019, after inviting all individuals undergoing hemodialysis, who also attended the units in the studied region, to participate.All assessments were performed by previously trained researchers.
Everyone was approached during their stay at the clinic, responding, during the hemodialysis session, to a semi-structured questionnaire on sociodemographic characteristics, lifestyle habits, and food consumption.Then, after the session, this was submitted for the collection of anthropometric data.

Sociodemographic and Clinical Characteristics
The variables of sex, age range, marital status, and race/color were dichotomized, respectively, into female and male, adult (18-59 years) and elderly (>60 years), with a partner and no partner, and white and non-white (self-declared race/color).
The categories used to define income were based on the family's minimum wage, defined as the lowest wage a company can pay an employee according to the number of hours worked.This has Brazilian validity by law and is reassessed annually based on the cost of living of the population.Its creation was based on the minimum amount that a person spends to ensure their survival.In the year this study was carried out, the minimum wage was R$998.00 or $235.37 USD.
The variables of education, work activity and smoking were separated into three categories, namely: <8 years, 8-11 years, and >11 years of study; with paid activity, no paid activity, and retired or on sick leave; and smoker, former smoker, and never smoked.
We assessed the duration of CKD and the date of the initiation of hemodialysis from the medical records of the participants, which were categorized according to the number of years.We asked participants about their medical history, so that diabetes and hypertension were dichotomized into absence and presence.
Regarding the practice of physical activity, we questioned the individuals about the type performed and frequency.All individuals who accumulated at least 150 min of physical activity per week were considered physically active and those who did not fit these characteristics were considered physically inactive [39].

Anthropometry
To assess height, we measured participants barefoot using a portable stadiometer, with a precision of one mm, obtained from the Sanny ® brand (São Bernardo do Campo, São Paulo, Brazil).The individuals were instructed to stand upright, arms extended along the body and gaze fixed on a point on the horizon [40].
In addition, we weighed participants on portable scales, with a capacity of 150 kg and accuracy of 0.1 kg, from the Tanita ® brand (Campinas, São Paulo).For measurement, they were invited to step onto the scale barefoot, in an upright position and with as little clothing as possible [40].
We used WC to determine the presence and absence of abdominal obesity.For the measurement, we used the Sanny ® brand (São Bernardo do Campo, São Paulo) inextensible measuring tape.The examinee was instructed to remain standing, with arms extended along the body and feet together.The tape was positioned at the midpoint between the lower margin of the last rib and the iliac crest [40].When it was not possible to measure the midpoint, we used the measurement two cm above the umbilicus [42].We considered the presence of abdominal obesity as ≥80 cm for women and ≥94 cm for men [41].

Consumption of Ultra-Processed Foods
To analyze the participants' consumption of UPFs, we applied a questionnaire, whose list of foods included was based on the Surveillance System of Risk and Protective Factors for Chronic Non-Communicable Diseases through Telephone Enquiry [43], conducted by the Brazilian Ministry of Health, which annually assesses the food in the capitals of the country.We used the NOVA food classification, proposed by Monteiro et al. [1], and in this study only the fourth group was considered.The questionnaire examined four food items by the frequency of the consumption of isolated foods and food groups, namely: soft drinks and artificial juices (powder or box); sausages (ham, mortadella, salami, sausage); biscuits (chips, filled); and sweets (ice cream, chocolates, cakes, biscuits, sweets).
To assess the reliability of the questions and the reproducibility of the food consumption questionnaire in this population, we conducted a pilot test in January 2019 [44].Details on the stipulated frequency categories and the conversion of reported consumption frequencies into daily rates are provided in Marques et al. [44].We evaluated the distribution of consumption of UPFs across quartiles.The first (Q1) and second (Q2) quartile corresponded to the lowest level of consumption, while the third (Q3) and fourth (Q4) quartile corresponded to the highest level of consumption of UPFs.

Ethical Aspects
This study was approved by Research Ethics Committee nº.4.023.221(CAAE 6852-8817.4.0000.5060)and all participants signed an informed consent form in accordance with the principles of the Declaration of Helsinki.

Statistical Analysis
To describe this study's variables, we used absolute and relative frequencies.To verify whether there was a difference in proportion between the independent variables and the outcome, we used Pearson's chi-square test (x 2 ) to qualitative variables.
We performed hierarchical logistic regression to associate the outcome with the independent variables, including the variables that had a value of p ≤ 0.1 in the bivariate analysis.In the first model, only sociodemographic variables (sex, age range, and education) were included; for the second model, the work activity variable was added; and in the third model, lifestyle, and health variables (CKD time, hemodialysis time, and physical activity) were included.Diabetes and BMI were not included in model three, respectively, because it was self-reported and showed general obesity only.In all models, the variable of UPF consumption was added.The final model was performed using the Forward LR method, in which the variables of work activity, CKD time, and hemodialysis time were removed by the mathematical model itself.For all of them, we respect the assumptions of the absence of multicollinearity and the absence of outliers.
We performed statistical analyses using the statistical software R (4.0.3) for Windows, with a significance level of α < 5%.

Discussion
The results of our study demonstrate that the prevalence of abdominal obesity in individuals undergoing hemodialysis is high.In addition, a higher level of consumption of UPFs, being male and an adult, having less education, and practicing physical activity below the recommended level increased the chances of having abdominal obesity in these individuals.
In our study, the prevalence of abdominal obesity was 77.22%.Studies have found prevalences of abdominal obesity in the North American [16], British [17], and Australian [18] adult populations of 53%, 40.7%, and 40.2%, respectively.In individuals with CKD, the prevalence values of abdominal obesity were 30.7% in stages four and five [45], and 60% in stages three to five in pre-dialysis clinics [46].In those treated with hemodialysis, the values were 39% in Norway [47] and 44.77% in Brazil [48].Thus, we noticed that the prevalence of abdominal obesity in our study is high and is higher than that of others found in the literature, including those with a population like that in our study [49].
Those with a higher level of consumption of UPFs were 1.72 times more likely to have abdominal obesity.Other studies have already shown that a higher level of consumption of UPFs increases the chances of having abdominal obesity.For example, when analyzing nationally representative samples of North Americans [16] and Australians [18], those with higher consumption of UPFs were 1.62 and 1.32 times more likely to have abdominal obesity, respectively.In Brazil, a longitudinal study noted that public servants with higher levels of consumption of UPFs had a 1.33 times greater risk of WC gain [20].Several mechanisms can explain the relationship between the consumption of UPFs and abdominal obesity.The formulation and ingredients of these products make them convenient, accessible, hyperpalatable, and attractive, leading to ingestion in larger quantities [1,50,51].As for their nutritional profile, they are denser in energy [3,[52][53][54] and have higher levels of fat, mainly saturated and trans fats, simple sugars, and sodium, in addition to lower levels of protein, fiber, vitamins, and minerals [55][56][57][58][59][60].As a result, they are nutritionally unbalanced [1,50,51], contributing to the increased prevalence of obesity in the population [52].
In our study, men were 2.21 times more likely to have abdominal obesity compared to women, corroborating the work by Jang and Oh [82], which found an increase of approximately 52% in the prevalence of abdominal obesity in Koreans after 20 years, but with less change observed in women.Men and women differ in terms of the distribution of adipose tissue, given that in men, adipose tissue preferentially accumulates in the visceral deposit, while the accumulation of fat occurs mainly in the subcutaneous deposit in women [83,84].The magnitude of this difference is amplified from the end of puberty to early adulthood, as males develop the typical android body shape, while females develop the gynoid form [83].There is a hypothesis that for survival, through hunting and escape, prehistoric man would need a more easily mobilized source of energy, hence the preference for storing fat in the visceral deposit [7,83].Furthermore, these differences in fat distribution between genders can be attributed to the influence of sex hormones [85] since studies have shown a positive association between testosterone and android adiposity, as well as between estrogen and gynoid fat deposition [86].
We also found that adults were 2 times more likely to have abdominal obesity than the elderly.This finding is like that discovered in studies in men that found a higher risk [87,88] and chance [89] of abdominal obesity in adults when compared to the elderly.For example, in a population-based cohort from Tehran, those aged 30 to 39 years had a 1.05 times greater risk of developing abdominal obesity, while those aged 50 to 69 years had a negative association with abdominal obesity, and so being in this age range was considered a protective factor [87].Furthermore, in Norway, during eight years of followup, those aged between 35 and 69 years were the ones who developed an increase in WC, while those aged 70 to 80 years showed a reduction [88].And finally, in China, the annual OR for abdominal obesity decreased significantly with age [89].A possible justification for the results found in ours and other studies is that adipose tissue increases in middle age and decreases at the end of life [90], and during the aging process fat is redistributed from the subcutaneous tissue to abdominal deposits, as well as to skeletal muscle, the heart, liver, pancreas, or blood vessels [91,92].
In our study, lower levels of education increased the chances of having abdominal obesity, corroborating representative data from Korea [45] and Portugal [93].Accordingly, higher education was considered a protective factor for abdominal obesity in Spanish [94] and Iranian adults [95].Several factors must be considered to explain the effects of education on abdominal obesity.People with low levels of education are less concerned with the consequences of abdominal obesity, mainly because they do not receive formal education, awareness about cardiovascular risk factors [96], and are exposed to the selection of unhealthy diets [97].In contrast, educated people are more aware of the consequences of the disease [98,99] and tend to have a healthier diet, characterized by higher consumption of fruits and vegetables, together with lower consumption of fats, having preventive effects in terms of abdominal obesity [95].We observe, therefore, that health professionals need to use health education-based strategies to make the information passed on more understandable, since individuals with less education have less perception of the consequences of abdominal obesity.
Finally, those who practiced physical activity below the recommended level were 1.95 times more likely to have abdominal obesity.Similar results were found in: Burkina Faso [100], Ivory Coast [96], the United Kingdom [17], Ethiopia [97], and Korea [101].Individuals with low and moderate levels of physical activity were 1.51 and 1.19 times more likely to have abdominal obesity, respectively [100].In Ivory Coast [96] and the United Kingdom [17], the prevalence of abdominal obesity was higher among those with less physical activity.Lastly, a sedentary lifestyle increased the chances of having abdominal obesity in Ethiopia [97] and Korea [101].Physical inactivity plays a central role as an independent and strong risk factor for the accumulation of visceral fat [102], with repercussions for general and abdominal obesity [103,104].The literature also shows that stress is associated with a lack of physical activity, which tends to increase cortisol secretion and the risk of abdominal obesity [100].In this sense, the practice of physical exercise is important in helping to prevent abdominal obesity [97], since the contraction of skeletal muscles releases myokines, which function similarly to a hormone, exerting specific endocrine effects on visceral fat, while other myokines act locally within muscle, through paracrine mechanisms, exerting effects on the signaling pathways involved in fat oxidation [102].Physical activity is, therefore, an important determinant of health for the intervention of health professionals working in hemodialysis clinics.
The strengths of our study include originality, considering that this is to our knowledge the first study to analyze the relationship between the consumption of UPFs and abdominal obesity in individuals undergoing hemodialysis.Also, this study's population consisted of patients from all hemodialysis clinics in the analyzed region.Another strength was the use of the NOVA food classification system, which has been recognized as a valid tool in nutrition and public health research and policy by the Pan American Health Organization [105].As limitations, we see the possible reverse causality due to the cross-sectional design of the research.In addition, the instrument used to assess the consumption of UPFs, which did not collect the ingested portions and was subject to memory bias, may not have allowed for an accurate estimate of consumption.We cannot disregard the anthropometric-based method of assessing abdominal obesity and the presence of confounding variables.However, waist circumference, used to measure the outcome, is a measure recommended by the World Health Organization for screening the risk of metabolic and cardiovascular diseases [12].In addition, we adjusted for potentially confounding variables when analyzing the association between exposure and outcome.

Conclusions
The results of our study demonstrate that the prevalence of abdominal obesity in individuals undergoing hemodialysis is high.In addition, they point out that a higher level of consumption of UPFs, being male and adult, having less education, and practicing physical activity below the recommended level increased the chances of having abdominal obesity in these individuals.Our work highlights health determinants regarding which health professionals who work in this population should pay attention and intervene.In addition, it provides a basis for the creation and improvement of public policies that promote changes in these determinants.We suggest that future research investigate whether the associations remain in individuals undergoing hemodialysis in other regions of the world.In addition, researchers should perform longitudinal studies to establish causality between the variables.

Table 1 .
Descriptive characterization and bivariate analysis between the independent variables and the outcome in individuals undergoing hemodialysis.

Table 2 .
Hierarchical logistic regression between the associated variables in the bivariate analysis and the outcome of individuals undergoing hemodialysis.
a Model 1: Adjusted for UPF consumption + sociodemographic variables (sex, age range, and education).b Model 2: Adjusted for model variables 1 + work activity.c Model 3: Adjusted for model variables 2 + lifestyle and health variables (CKD time, hemodialysis time, and physical activity).Diabetes and BMI were not included in model 3, respectively, because it was self-reported and showed general obesity only.d Final model: Performed using the Forward LR method, in which the variables work activity, CKD time, and hemodialysis time were removed by the mathematical model itself.e p-value: Hierarchical logistic regression for the association of the outcome with the independent variables, including the variables that presented a value of p ≤ 0.1 in the bivariate analysis.f OR: odds ratio.g CI: 95% confidence interval.h UPFs: ultra-processed foods.i Q1: Quartile 1. Q2: Quartile 2. Q3: Quartile 3. Q4: Quartile 4. j CKD: chronic kidney disease.