Protein and Amino Acid Adequacy and Food Consumption by Processing Level in Vegans in Brazil

This cross-sectional survey study compares protein and essential amino acid intake levels against recommended levels among individuals in Brazil who adhere to a vegan diet and assesses whether adequate protein consumption is associated with greater intake of processed foods.


Introduction
Veganism is a lifestyle increasing in popularity worldwide that supports abstaining from the use of animal products.This leads to a dietary pattern that excludes meat, fish, poultry, dairy, eggs, and honey, among other animal-derived foods (henceforth, vegan diet). 1,2fraining from protein-rich animal foods has generated ongoing controversy as to whether individuals who adhere to a vegan diet can adequately meet protein requirements. 3Although studies show that these individuals can have borderline-adequate protein intake, albeit slightly reduced vs individuals consuming omnivorous diets, 4 it is unknown whether protein sources habitually consumed by vegans allow for adequate essential amino acid intake.
A range of plant-based meat and dairy substitutes have been formulated with the claim of being practical, protein-rich complements to a vegan diet.However, overconsumption of these products may disrupt the alimentary basis of vegan diets (unprocessed and minimally processed plant-based foods, such as fruits, vegetables, and whole grains) and lead to an increase in ultraprocessed food intake. 5,6The association of ultraprocessed food intake with overall health has been widely debated, [7][8][9] warranting the investigation of food consumption as a function of processing level in the vegan population.Therefore, the primary aim of this study was to describe protein and essential amino acid intake in a survey of Brazilian individuals who follow a vegan diet.The secondary aim was assessing food intake by processing level according to the Nova classification system and investigating potential factors associated with protein inadequacy.

Methods
This cross-sectional survey study was approved by the Centro Universitário Estácio de Sá de Santa Catarina ethical committee and conducted according to the Helsinki declaration.All participants signed a digital informed consent form.This manuscript adheres to the Strengthening the Reporting of Observational Studies in Epidemiology-Nutritional Epidemiology (STROBE-nut) reporting guideline and American Association for Public Opinion Research (AAPOR) reporting guideline for surveys. 10

Study Design and Participants
Data in this study derive from the Vegan Eating Habits and Nutritional Evaluation Survey (VEGAN-EATS), a cross-sectional survey conducted between September 2021 and January 2023.We previously published an analysis of disordered eating attitudes and food choice motives in this population stemming from this survey. 11This study expands on the topic by investigating the dietary profile of Brazilian vegan diets, with the topical issue of protein, essential amino acid intake, and food processing in this population.
Participants were recruited through advertisements on social media platforms and included males and females aged 18 years or older currently living in Brazil with internet access and the ability to read.Participants completed an online survey using the Google Forms platform (Google LLC).

Evaluation Tool
The survey included questions regarding participant general characteristics, self-reported anthropometric data, and vegan lifestyle.General characteristics included educational level (elementary school, high school, college, or postgraduate education), income (categorized according to the Brazilian Institute of Geography and Statistics in levels of A [ՆUS $4200], B [US $1350-$4200], C [US $550-$1350], and D or E [ՅUS $550]), 12 smoking (yes or no) and drinking status (none, 1-2 times/mo, 2-4 times/mo, 2-3 times/wk, or Ն 4 times/wk), and exercise habits (does not exercise, 1-2 h/wk, 3-4 h/wk, 5-6 h/wk, or Ն7 h/wk).Vegan lifestyle-related questions were the duration of adherence to a vegan diet (< 1 y, 1-2 y, 2-3 y, 3-4 years, and >5 y) and main motivation to shift toward a vegan diet (ethics, health reasons, environment, politics, life philosophy, medical restrictions, religion, or sports performance).Macronutrient and micronutrient, amino acid, and food intake by processing level were assessed by a self-administered 1-day food diary.This tool has been previously shown to perform adequately compared with the typical three 24-hour recalls, 13 and it was selected due to the impossibility of implementing repeated 24-hour recalls within our study.
Participants received an instructional video on how to fulfil the diary and fully report quantity and type of foods and beverages consumed within the previous 24 hours, including a precise level of detail on food preparation and ingredients, allowing best practices for Nova food classification. 14,15tal energy (kilocalories), macronutrient amounts (grams), and relative contributions of each Nova food processing category to the total energy intake were calculated.We also calculated the relative contribution stemming from proteins by food processing category.Examples of commonly reported foods and their classification are available in eTable 1 in Supplement 1.
Food diaries were quantified with Nutritionist Pro software version 7.3 (Axxya Systems) using the US Department of Agriculture database.When nutritional information was not available, we searched the literature or directly contacted food companies.Amino acid composition was available for most consumed proteins (94.2%).

Statistical Analysis
Due to missingness in body weight and height (available in 558 individuals), individuals with and without these data were compared and analyzed for potential associations between missingness and variables (eTable 2 in Supplement 1).There were only subtle differences between subgroups across all variables.Given that protein and amino acid intake adequacy requires assessing intake relative to body weight, observations with missing variables were excluded and complete cases were used for the main analysis.We also provided a complementary analysis in which missing body weight was imputed through multiple imputation 16 using the mice package 17 in R statistical software set to 5 iterations and the classification and regression trees method.A matrix of factors associated with body weight within the dataset was selected (age, sex, income class, exercise habits, motivation to shift to a vegan diet, total energy intake, and protein intake).The mean body weight from these 5 imputed datasets was used for descriptive statistics and model adjustment.
Descriptive data are presented as median and IQR for continuous variables and absolute and relative frequency (number and percentage) for categorical variables.We calculated 95% CIs surrounding the median using the bootstrap method through the boot package in R statistical software set to 10 000 iterations.
Protein and amino acid intake (milligrams per kilogram body weight per•day) were compared with Recommended Dietary Allowances from Dietary Reference Intakes. 18Nutrient adequacy ratios were calculated by dividing nutrient intake by its recommendation (with scores being truncated at 1) for each participant and then finding the mean of values across the entire sample for each nutrient. 19The mean adequacy ratio was calculated by taking the mean of all nutrient adequacy ratios. 19We obtained 95% CIs via bootstrapping (see previous description).
The median protein and amino acid contribution of individual food items was calculated, and the top 30 or 10 food items for protein and essential amino acids, respectively, were plotted.To further assess potential factors associated with inadequate protein intake, participants were classified as having inadequate (<0.8 g/kg) or adequate (Ն0.8 g/kg) protein intake, 18 and logistic regression models with Firth penalization 20 were used considering protein intake status as the outcome variable.Protein supplements or texturized soy protein consumption were used as binary variables (yes or no) in the main analysis.Quartiles were calculated for continuous variables (eg, unprocessed and minimally processed food, ultraprocessed food, and unprocessed and ultraprocessed protein intake) and used as categorical variables.Models were adjusted for age, sex, body weight, and energy intake relative to body weight.The α level was set at .05 for all analyses, and all P values assume a 2-sided test.All data cleaning, exploration, and visualization was performed using R statistical software version 4.2.2 (R Project for Statistical Computing) and RStudio (Posit Software, PBC) with dplyr and ggplot2 packages.

Results
Among 1014 participants who completed the survey, 43 individuals were excluded for not fully adhering to a vegan diet and 197 individuals were excluded due to insufficient report on foods and portions, resulting in 774 individuals (median [IQR] age, 29 [24-35]   1).Because 216 participants did not report body weight, relative protein and amino acid intakes were available for 558 individuals (complete case population; median [IQR] age, 29 [24-37] years; 459 female [82.3%]) (see eFigure 1 in Supplement 1 for study flowchart and eTable 2 in Supplement 1 for all demographic information).
Adjusted logistic regression models among the complete case population showed that consuming protein supplements (odds ratio [OR], 0.06 [95% CI, 0.02-0.14];P < .001)and textured soy protein (OR, 0.32 [95% CI, 0.17-0.59];P < .001)compared with not consuming those products was associated with reduced odds of displaying inadequate protein intake (Figure 3A).The second, third, and fourth quartiles of ultraprocessed food intake (eg, fourth vs first quartile of intake: OR, 0.16 [95% CI, 0.07-0.33];P < .001)and the third and fourth quartiles of ultraprocessed protein intake were associated with reduced odds of displaying inadequate protein intake, while the second, third, and fourth quartiles of unprocessed protein intake (eg, fourth vs first quartile of intake: OR, 12.42 [95% CI, 5.56-29.51];P < .001) was associated with increased odds of displaying inadequate protein intake compared with the respective first quartile (see Figure 3B for model coefficients and eTable 7 in Supplement 1 for quartiles).Results remained virtually unchanged in analyses performed with the imputed dataset (774 participants), not altering the original interpretation of our results (eTable 8 and eFigure 4 in Supplement 1).An exploratory analysis considering protein supplement and texturized soy protein intake as continuous rather than binary variables showed a negative association of protein supplement and texturized soy protein intake with the probability of displaying protein inadequacy (eFigure 5 in Supplement 1), which is in agreement with our main analysis.

Discussion
This cross-sectional survey study analyzed self-reported data from 1-day food diaries on protein and amino acid intake and food processing level among individuals who adhered to a vegan diet.The main findings were that the mean adequacy ratio for protein and essential amino acids was high, participants had a high level of consumption of unprocessed and minimally processed food and a low level of consumption of processed and ultraprocessed food, and consumption of ultraprocessed protein sources was associated with decreased odds of showing inadequate protein intake, while the opposite was true for unprocessed and minimally processed protein sources.
Although previous evidence from smaller samples indicated that vegan diets may provide adequate protein, albeit lower than that of omnivorous diets, 4 it was still unclear whether they delivered enough essential amino acids.In our sample, a high mean adequacy for protein and essential amino acids was observed.To the best of our knowledge, the largest previous study to address amino acid intake in vegans was a cross-sectional analysis of the European Prospective Investigation into Cancer and Nutrition (EPIC)-Oxford cohort (98 individuals). 22The authors found lower amino acid intake among individuals who adhered to a vegan diet compared with meat-eaters; however, whether this difference was associated with relevant deficiencies was unknown. 22Based on a much larger sample and using a higher-resolution instrument, our study offers novel evidence that the most vegans can meet essential amino acid recommendations.
However, a considerable segment of participants did not meet recommendations for particular essential amino acids.For instance, lysine and the combination of methionine and cystine showed the lowest nutrient adequacy in our sample.Indeed, these amino acids are generally lower in plantvs animal-based proteins. 23,24Importantly, all amino acids are needed in adequate amounts to support de novo tissue protein synthesis, 25 reinforcing the need for properly planned, well-balanced combinations of different plant-based protein sources to attain an optimal essential amino acid profile.
7][28][29][30][31] Although vegan diets are traditionally based on fresh and in natura foods, there has been a tendency of increased ultraprocessed food consumption in vegans due to the proliferation of plant-based meat and dairy substitutes, possibly compromising the quality of vegan diets. 5A previous study in a French cohort 6 found ultraprocessed food consumption to be 39.5% of total energy intake among individuals who adhered to a vegan diet vs 33% in meat-eaters.In our study, unprocessed and minimally processed foods represented the highest caloric contribution in vegan diets, with a smaller presence of ultraprocessed foods.One may argue that compared with other countries, the plant-based meat and dairy substitute industry in Brazil may be less active, providing fewer or less attractive options.
Considering the global trend in the protein market toward an increasing popularity of plant-based protein and the growing demand for high-quality, natural, and sustainable protein sources, 32 an increase in consumption of ultraprocessed foods among Brazilian individuals who adhere to a vegan diet may be expected in the future, warranting follow-up surveys to monitor possible health outcomes.Recently, the discussion on whether all subgroups of ultraprocessed foods are equally harmful compared with one another has gained traction.Indeed, a 2023 study 33 found that distinct subgroups of ultraprocessed foods were differently associated with type 2 diabetes risk; subgroups, such as fruit-and dairy-based ultraprocessed desserts, had a risk reduction.Another study 34 found that although ultraprocessed food consumption was associated with multimorbidity related to cancer and cardiometabolic diseases, this association was not seen in the subgroup of plant-based ultraprocessed foods, suggesting that Nova may not capture possible nuance in quality among ultraprocessed items.In our study, textured soy protein was an important contributor to ultraprocessed food consumption.This is a challenging food item to classify given that there is significant variability in formulations considering the presence of food additives and cosmetics.This prompted us to use a more conservative approach of classifying this food as ultraprocessed, which is in accordance with previous research. 6In a sensitivity analysis considering textured soy protein as unprocessed and minimally processed, the energy contribution from ultraprocessed food decreased greatly.The reduction in relative contribution of ultraprocessed food to protein intake was even more pronounced, clearly suggesting that reclassification of individual foods may be associated with impactful changes in the contribution of ultraprocessed foods to vegan diets.
We found that protein supplement, textured soy protein, ultraprocessed food, and ultraprocessed protein intake were associated with reduced probabilities of protein inadequacy.
Conversely, unprocessed and minimally processed protein intake was associated with an increased probability of protein inadequacy.This may be partially explained by the lower energy and protein density in plant-vs animal-derived foods, 35 suggesting that it may be challenging for vegans fully avoiding ultraprocessed foods to reach higher levels of protein intake without substantially increasing food (and perhaps calorie) intake.While these findings do not imply that ultraprocessed foods are essential for individuals who adhere to a vegan diet to meet protein recommendations, it reveals a significant reliance on these foods to attend protein requirements.One may suggest that certain ultraprocessed foods, such as textured soy protein, may be recommended for this population.7][38] This holds true for protein supplements, an evidencebased strategy to support muscle health 39 also associated with protein adequacy in this study.At least regarding individuals who adhere to a vegan diet, unrestricted advice to avoid ultraprocessed foods may have unintended consequences, such as protein intake inadequacies, that warrant further investigation.This also suggests that vegans may benefit from public policies aimed to facilitate access to more natural and healthy foods and amplify nutritional support and education for adequacy of overall food intake.Simultaneously, our data reinforce the urgent need for the development of affordable, healthier, better-quality, cleaner-label, and protein-rich plant-based food options by the industry.

Strengths and Limitations
Strengths of this study include the large sample size and use of food diaries to quantify food intake. 40e study also has several limitations, including the cross-sectional design, self-reporting of information, and use of a 1-day food diary instead of repeated measurements of food intake.Our study features a convenience sample, predominantly composed of females, with eutrophic BMI and a high educational level.2][43] Nonetheless, our conclusions cannot be extrapolated to more population samples with a lower socioeconomical status or with different dietary patterns.Further studies are warranted to answer such questions.

Figure 1 .
Figure 1.Nutrient Intake and Adequacy Ratios and Proportion Meeting Recommendations

Figure 2 .
Figure 2. Caloric and Protein Intake by Nova Food Processing Category

Table 1 .
Study Population Characteristics

Table 1 .
Study Population Characteristics (continued) 12This value is among the 558 participants who provided body weight and height data.bIncome was categorized according to the Brazilian Institute of Geography and Statistics in classes A (ՆUS $4200), B (US $1350-$4200), C (US $550-$1350), and D or E (ՅUS $550).12

Table 2 .
Caloric Contribution and Protein Intake by Nova Food Processing Category and Amino Acid Intake a Results are presented as medians with 95% CIs, Association between binary variables and inadequate protein intakeThe association between quartiles of continuous variables and inadequate protein intake is shown.Results are presented as odds ratio coefficients and 95% CIs for having inadequate protein intake given the variable in comparison with the reference level.NA indicates not applicable.Intake of nonessential amino acids eTable 6. Caloric contribution and protein intake according to Nova food processing category and amino acid intake considering textured soy protein as unprocessed and minimally processed food eTable 7. Coefficients from adjusted logistic regression models estimating protein inadequacy (complete case analysis, n=558) eTable 8. Coefficients from adjusted logistic regression models estimating protein inadequacy (imputed dataset, n=774) eFigure 1. Study flowchart eFigure 2. Caloric and protein intake according to Nova food processing categories considering textured soy protein as unprocessed and minimally processed eFigure 3. Main food sources of protein and essential amino acids eFigure 4. Protein and essential amino acids intake, nutrient adequacy ratios and proportion of individuals meeting recommended intakes in the imputed (n=774) dataset eFigure 5. Exploratory logistic regression models estimating the probability of protein inadequacy according to protein supplement intake or texturized soy protein intake as continuous variables eTable 5.