Association between overweight and consumption of ultra-processed food and sugar-sweetened beverages among vegetarians Associação entre o excesso de peso e o consumo de alimentos ultraprocessados e bebidas açucaradas por diferentes grupos vegetarianos

1 Universidade Federal de Alagoas, Faculdade de Nutrição, Laboratório de Nutrição em Saúde Pública. Av. Lourival Melo Mota, s/n., Tabuleiro dos Martins, 57072-900, Maceió, AL, Brasil. Correspondência para/Correspondence to: JAC SILVEIRA. E-mail: <jonas.silveira@fanut.ufal.br>. 2 Nutricionista. Osasco, SP, Brasil. Association between overweight and consumption of ultra-processed food and sugar-sweetened beverages among vegetarians


A B S T R A C T Objective
To assess the consumption of ultra-processed food and sugar-sweetened beverages and to identify the association of this consumption with overweight among vegetarians.

Methods
A cross-sectional study with a convenience sampling method was conducted.Data were collected using an online questionnaire from 8/24/2015 to 10/8/2015.Subjects were male and female vegetarians aged >16 years.Using a food frequency questionnaire, we assessed the weekly consumption of ultra-processed food and sugar-sweetened beverages and described the frequency of daily consumption overall and according to type of vegetarianism.The association between overweight and excessive daily intake of ultra-processed food was analyzed by multiple logistic regression (OR [95CI%]).

I N T R O D U C T I O N
Although there are no population-level estimates -based on probabilistic sampling surveys -of the number of vegetarians in Brazil, interviews conducted by the Instituto Brasileiro de Opinião Pública e Estatística (IBOPE, Brazilian Institute of Public Opinion and Statistics) in the Brazilian metropolitan areas and in the interior of South and Southeast regions identified that 8% of the population claims to be vegetarian, corresponding to 15.2 million people.Of the sites surveyed, the São Paulo (SP) metropolitan region (Southeast region) -the most populated area in Brazil -had the largest number of vegetarians (n=792 thousand); however, Fortaleza (CE) (Northeast region) had a higher frequency of vegetarians (14%) [1].Therefore, as this group gains greater representation in the population, it is important to evaluate their health and nutritional status because they are characterized by a specific eating pattern related to the restriction of animal food intake [2].
Given these characteristics, research conducted with vegetarians has mostly focused on micronutrient deficiencies, especially regarding vitamin B 12 ; however, few studies have analyzed their food patterns related to the consumption of Sugar-Sweetened Beverages (SSB), Ultra-Processed Food (UPF) and ingredients such as simple sugars and sodium.
It is reasonable to consider that this lack of attention originates from the conceptual model related to the vegetarian diet [3], which evidence shows has positive effects on health outcomes, i.e., metabolic syndrome [4] and mortality [5].However, the assumption that solely being a vegetarian results in a healthy dietary pattern is problematic.
Adherence to this lifestyle is driven by many reasons other than health, including animal ethics, rights and welfare, environmental concerns and religion [2,6].However, in general, the characteristics of UPF, SSB and other highenergy, -fat, -carbohydrate and -sodium foods and ingredients [7] do not violate vegetarians' diet restrictions [8].
Consequently, considering the context of nutrition transition, which has indicated a significant increase in the consumption of UPF and SSB [9], without proper orientation, vegetarians are at risk of developing an eating pattern that is potentially harmful to their health.
Given this background, this study analyzed the UPF and SSB intake patterns among different types of vegetarianism to test whether the hypothesis regarding the "implicit" quality of vegetarians' diet is true; this analysis presents new information about vegetarians' eating habits, as to the authors' knowledge, there are no studies on this topic in Brazil.Finally, we investigated the association between excessive UPF intake and overweight, adjusting for potential confounders.

M E T H O D S
This was a cross-sectional study that applied a convenience sampling method because data were collected using an online questionnaire -developed on the Google Forms platform (<https://www.google.com/forms/about/>) -that was made available to the target population through social networks (Facebook, Twitter and WhatsApp groups) from 8/24/2015 to 10/8/2015.Eligibility criteria was defined as individuals of both sexes aged >16 years old who selfidentified as a vegetarian, independent of the type of vegetarianism (vegan, lacto vegetarian, ovo vegetarian and lacto-ovo vegetarian).
The outcome variables -UPF and SSB intake -were collected using a Food Frequency Questionnaire (FFQ) developed from the new Dietary Guidelines for the Brazilian Population [7] and the FFQ from both "Pesquisa de Orçamentos Familiares de 2008/2009" (POF, Consumers Expenditure Survey) [10] and "Pesquisa Nacional de Demografia e Saúde da Criança e da Mulher de 2006/2007" (PNDS, National Survey on Demography and Health of Women and Children) [11].
Daily consumption of UPF and SSB was calculated by dividing the sum of the weekly consumption of each food available in the FFQ by seven (the number of weekdays).The category "<1x/day" should be interpreted as individuals who presented sporadic consumption of UPF and SSB.
Sample size was calculated by the method of Fleiss et al. [13] with continuity correction performed at OpenEpi [14] to detect an Odds Ratio (OR), of at least 1.8, with a twotailed significance level (α) of 5%, 80% power (1-β) and a prevalence of overweight in the unexposed group (UPF <3x/day) of 25% [15].Setting these parameters resulted in a sample size of 458 persons.
The answers obtained from the online questionnaire were automatically organized into Microsoft Excel format.We then imported the database to Stata/IC 12.0 (StataCorp., College Station, Texas, United States), which was used to organize and analyze all variables.
Initially, extensive descriptive analyses were conducted to identify the best approach to analyzing food consumption variables.Thus, we chose to present data as relative frequencies of consumption of UPF and SSB according to the type of vegetarianism; comparisons between proportions were conducted by Pearson's Chi-squared test, considering an α=0.05.
We performed simple and multiple logistic regression analysis to test the association between overweight and excessive intake of UPF, adjusting for potential confounders.As UPF consumption includes SSB (except for 100% fruit juice), we selected the former as an independent variable in the multivariable analysis; furthermore, when both were introduced in the same model, SSB intake did not maintain its association with overweight (p=0.652), with only UPF intake remaining statistically significant.Thus, the association between SSB and overweight was presented only in the univariate analysis.
For the multiple logistic regression analysis, we considered eligible only the variables that presented p<0.20 in the univariate analysis, and the introduction of variables into the model was performed using the stepwise forward method.Following technical recommendations, the estimates were interpreted according to their 95% Confidence Interval (95%CI), which is related to the p-value [16].The estimates were presented as odds ratios with their respective 95%CI (OR [95%CI]).
This research was submitted and approved by an Ethics Committee in Human Research (CAAE 45685715.0.0000.5493).The Consent Form was shown on the home screen of the online questionnaire, and thus the participant could only access the next page after reading and expressing his/her agreement.Additionally, the answers were only saved after the participants had completed the whole questionnaire, allowing them to stop participating at any time.

R E S U L T S
In total, 503 individuals answered the questionnaire; their characteristics are presented in Table 1.The most frequent type of vegetarianism in our sample was ovo-lacto vegetarians (45.5%) and vegans (41.7%).The majority of the participants were women (83.7%), and the median time as a vegetarian was 3.5 years.Additionally, 82.3% reported that their main reason for adhering to a vegetarian diet was animal ethics, rights and welfare; only 1.6% indicated that they had become a vegetarian for health reasons.
The prevalence of overweight (obesity included) in our population was 23.5%; however, when stratified according to UPF consumption (≥3x/day), the prevalence increased to 38.3%.It is worth noting that 74.0% of the individuals reported that they had maintained or gained weight after following a vegetarian diet, and of those who reported weight maintenance or gain, the prevalence of overweight was 50.0% and 15.9%, respectively.Tables 2, 3 and 4 show the results of the FFQ (by food item and grouped as UFP and SSB), which indicated that 60% of the participants ate UPF and SSB at least once a day.Furthermore, the frequency of excessive daily intake of UPF (≥3x/day) and SSB (≥3x/day) was 16% and 20%, respectively.

D I S C U S S I O N
This research aimed to evaluate the consumption patterns of UPF and SSB among different types of vegetarians through an online survey.Despite its limitations, this was the first Brazilian study with vegetarians to shift the focus from nutrient deficiencies to the specific intake of UPF and SSB.
It is interesting to note that studies that overcome the paradigm of nutritional deficiencies and manage to analyze the protective effect of vegetarian diets on health outcomes are based on the assumption that subjects "naturally" have a healthier eating pattern than omnivores.We believe that this assumption is due to a deductive reasoning process; namely, as omnivores who eat higher amounts of fruits, vegetables, whole grains and nuts (Mediterranean pattern) perform better on metabolic risk markers, this relationship must also be true for vegetarians, since these foods are not restricted and would be the most available food option.Therefore, regarding food habits, despite their other behaviors, the choice to practice vegetarianism -by itself -has been assumed to be an intrinsic protective factor to cardiovascular disease [4,5,8,15].
However, this universal statement cannot be considered true under all circumstances.In our study, we observed that being a vegetarian, regardless of the level of restriction, did not inherently protect individuals from overweight or obesity.Obviously, considering the crosssectional design, it is not possible to determine when the individual became overweight, and the decision to be a vegetarian could have been a therapeutic one; however, more than 80% of the study population reported "animal ethics, rights and welfare" and not "health" as their main reason for becoming a vegetarian.
On the other hand, when after examining the association of UPF with overweight as well as the prevalence of overweight among individuals who reported weight maintenance (50.0%) or weight gain (15.9%) after practicing vegetarianism, UPF consumption pattern could be a cause of overweight.Moreover, disregarding conjectures related to the causal pathway, the maintenance of this condition among vegetarians could have been mediated by non-healthy eating patterns.This assertion is supported by the fact that the multiple logistic regression model was adjusted by time as a vegetarian, and no inverse association with overweight was established.
Furthermore, we performed a multinomial regression analysis using "weight postvegetarianism" as the outcome variable (Table 1) and excessive UPF intake as the explanatory variable, adjusting the model for sex, age, time as a vegetarian and preparation of own meals at home.In this analysis, which used the index category of "maintained weight", UPF consumption was associated with weight gain post-vegetarianism (2.0 [1.06; 3.77 95%IC]), further strengthening the argument regarding its harmful effect on body weight among vegetarians.No variables were associated within the category "lost weight".
Considering the current global context regarding climate change, Brazilians' eating pattern [9] and the potential benefits related to vegetarianism [4,8], this type of diet may be considered an environmentally sustainable [17] and nutritionally adequate [3] food and nutrition recommendation.However, it is important for health professionals, especially dietitians (in Brazil, professionals who possess the private function of dietary prescription), to not quickly assume a relationship between vegetarianism and healthy diet.Instead, they should proceed with a thorough assessment of eating habits and behavior, given the complexity in managing the nutritional requirements of these individuals, both regarding micronutrients deficiencies and excessive consumption of UPF and SSB.
From this perspective, we also found that not cooking one's meals was associated with overweight, which corresponds with the findings of Gorgulho et al. [18] regarding the fact that adults who eat away from home present worse eating patterns.Their study showed that 24% of residents from the São Paulo (SP) metropolitan area ate at least one meal away from home/day, while in our sample, the prevalence was 15%.
National data from the 2002/2003 POF indicated that the overall prevalence of eating away from home was even greater (35.1%) [19].Another publication using the same dataset showed that such behavior was associated with overweight and obesity in men, but not women, independent of age and income; regarding the consumption of UPF and SSB away from home, the authors observed that in general, overweight men ate almost twice as much as overweight women [20].
Nevertheless, it is important to consider that eating away from home does not necessarily result in a worst eating pattern.As shown by Hoffman [21], using data from the 2008/2009 POF, eating away from home was negatively associated with overweight and obesity; however, when the logistic regression coefficient was exponentiated, the effect size was low, suggesting that the statistically significant association was due to the large sample size.
We also highlight in Hoffman's [21] paper that the average daily consumption of sugars, lipids, sodium and cholesterol was higher among individuals who ate away from home than in individuals who ate exclusively at home.Therefore, despite consuming a slightly higher amount of vegetables (+0.101g/100kcal; p<0.05), people who ate away from home tended to present a less healthy eating pattern, which corroborates our hypothesis regarding the relationship between UPF and SSB intake and overweight.Therefore, on this topic, we can state that although eating away from home is not synonymous with unhealthy meals or eating at fast-food restaurants, these environments may lead to consumption of unhealthy meals, especially among those who are not interested in improving their eating patterns [18,20,21].Another fact that cannot be ignored is that in our study, 61.6% of the vegetarians reported having issues eating away from home.
Finally, we found that the age group of those ≥35 years (up to 63 years) was associated with overweight among vegetarians, and this finding can be considered consistent with the arguments raised to date, as this is an economically active part of society.Furthermore, this result could be a reaction to what occurs within the Brazilian population, as the overweight prevalence increases as the age group strata become older, regardless of sex, as shown in the 2008/2009 POF [22].
In addition to the issues related to the cross-sectional design and the limitations of FFQs, we note the absence of variables describing the sample's economic, educational and lifestyle (i.e., level of physical activity/ sedentariness) characteristics, which could improve the understanding of the association between consumption of UPF and overweight among vegetarians as potential confounders.
Regarding the external validity of our results, it is important to consider that 86.6% of our sample was from the South and Southeast regions (the most developed regions in Brazil) and that the sampling process was designed for hypothesis testing and not for geographical representativeness.
One aspect of this research that is worth to highlight was raising relevant questions and concerns to the clinical practice and academic research about the quality of vegetarians' diets.Furthermore, because of the increasing number of vegetarians, future national surveys should scale for this group when developing questionnaires -i.e., POF, PNDS, Pesquisa Na-cional de Saúde (National Health Survey), Vigilância de Fatores de Risco e Proteção para Doenças Crônicas por Inquérito Telefônico (Surveillance of Protective and Risk Factors for Chronic Diseases by Telephone Survey), Pesquisa Nacional por Amostra de Domicílios (National Household Sample Survey).
As a methodological contribution, previous publications have reported problems identifying and selecting vegetarians for research; therefore, this study provides an efficient approach to overcoming this limitation.

C O N C L U S I O N
Although vegetarianism is frequently associated with healthier diets and, consequently, with the prevention of poor health outcomes, this study identified 1) a high frequency of excessive daily consumption of UPF and SSB and 2) that this eating pattern was associated with overweight and obesity.Therefore, we stress the need to progress beyond the paradigm of traditional nutritional deficiencies (iron and cyanocobalamin) in nutritional counseling or surveillance of vegetarian individuals and to offer at least the same attention to the consumption of UFP and SSB.Additionally, we found that male sex, age group ≥35 years and not cooking one's own meals were factors associated with overweight among vegetarians.This research adds to the body of evidence indicating the harmful effects related to excessive UPF and SSB consumption on human health and endorses the need for regulatory proposals through public health policies that effectively promote reduction of this type of intake.

A C K N O W L E D G E M E N T S
We would like to thank the entire vegetarian community, who not only answered the questionnaire but also disseminated it among their peers.

Table 3 .Table 4 .
Consumption of ultra-processed food and sugar-sweetened beverage items according to types of vegetarianism (N=503) -ultra-processed food and sugar-sweetened beverage consumption among Brazilian vegetarians survey (2015).Consumption of ultra-processed food and sugar-sweetened beverage according to types of vegetarianism (N=503) -ultra-processed food and sugar-sweetened beverage consumption among Brazilian vegetarians survey (2015).ItemsFrequency of daily consumption (%) *Vegan x/day (n=210)Ovo-lacto vegetarian x/day (n=229)Ovo vegetarian x/day (n=22)Lacto vegetarian x/day (n=42) The sum of the weekly consumption of each food available in the Food Frequency Questionnaire divided by seven.

C
O N T R I B UT O R S JAC SILVEIRA study conception and design, statistical analysis and interpretation, drafting of the manuscript and final approval of the submitted version.SS MENESES, PT QUINTANA, VS SANTOS study conception, data collection, drafting of the manuscript and final approval of the submitted version.

Table 1 .
Characteristics of the study participants -ultra- Note: * Based on body mass index.SD: Standard Deviation; IQR: Interquartile Range; BMI: Body Mass

Table 2 .
Consumption of ultra-processed food and sugar-sweetened beverage among vegetarians (N=503) -ultra-processed food and sugar-sweetened beverage consumption among Brazilian vegetarians survey (2015).
Note: * Percentage calculated excluding the group of vegans (n=293); ** The sum of the weekly consumption of each food available in the Food Frequency Questionaries' divided by seven.

Table 5 .
Factors associated with overweight among vegetarians -ultra-processed food and sugar-sweetened beverage consumption among Brazilian vegetarians survey (2015).