In recent years, vegetarianism has emerged as increasingly adopted dietary choice among people worldwide(1–3). This dietary pattern is based on excluding or substantially reducing the consumption of animal products, focusing instead on plant-based foods(4). In addition to its positive impact on reducing animal suffering and mitigating the effects of livestock on the environment(Reference González-García, Esteve-Llorens and Moreira5,Reference Springmann, Wiebe and Mason-D’Croz6) , the vegetarian diet has been linked to several health benefits, including a lower risk of chronic diseases such as CVD, type 2 diabetes and certain types of cancer, as well as promoting a healthy body weight and better control of cholesterol levels and blood pressure(Reference Chai, van der Voort and Grofelnik7–Reference Zhao, Zhan and Wang9).
Despite that, it is relevant to emphasise that not all plant-based diets may be equally beneficial to health(Reference Wickramasinghe, Breda and Berdzuli10–Reference Orlich, Sabaté and Mashchak14). This becomes even more significant in light of the transformations in food production and supply over the last decades, which have promoted an increase in the consumption of ready-to-eat or pre-made products conveniently labelled as vegetarian(Reference Curtain and Grafenauer15).
Ultra-processed foods are composed of substances derived from foods, including many that are exclusively used by the food industry, along with cosmetic additives that provide sensory attributes to these products. In addition, they lack whole foods in their composition and are high in sugar and fats and low in fibre and essential nutrients. These products are formulated to be highly attractive, convenient and widely available(Reference Monteiro, Cannon and Levy16). Several studies have shown an association between the consumption of these foods and lower overall nutritional diet quality(Reference Martini, Godos and Bonaccio17) and an increased risk of obesity and other chronic diseases, as well as mortality(Reference Louzada, Costa and Souza18,Reference Srour, Kordahi and Bonazzi19) .
Considering the increase in these foods, which are commonly sold in sophisticated packaging with claims related to nutrition, health and even environmental issues, alongside aggressive marketing, it is necessary to evaluate the quality of vegetarian food by considering the degree and purpose of industrial food processing. Therefore, this article aims to describe the consumption of foods according to the groups of the Nova classification among British vegetarians and to evaluate the association between the consumption of ultra-processed foods and the intake of nutrients recommended in international guidelines for this group.
Methodology
Sampling, data collection and food consumption
Data from the National Diet and Nutrition Survey (NDNS) were utilised, encompassing the years 2008–2019. Detailed descriptions of the survey’s methodology have been previously published(20,21) . Briefly, survey samples were randomly selected from the UK Postcode Address Archive, which contains a comprehensive list of all addresses in the four constituent countries (England, Wales, Scotland and Northern Ireland). From each randomly selected address, one child (aged 1·5–18 years) or one child along with an adult (aged 19 years or older) was selected. Data collection involved an interview with a researcher to obtain sociodemographic and food consumption information, as well as a visit from a nurse. Food consumption was assessed using a food diary completed by participants over four consecutive days. Participants who completed the diary for 3 or 4 d were included in the survey, resulting in a sample size of 15 643 individuals. In our study, we analysed data from individuals who identified as vegetarians, representing 2·3 % of the population (with 0·17 % identifying as vegan).
Food consumption was assessed using 4-d food diaries that included workdays and weekends, thus covering all days of the week. Individuals were instructed to record all food and beverages consumed on the day, inside and outside the home. It is important to note that children under 10 years of age had the help of their parents or caregivers for guidance when filling in the food record (and the ‘self-declaration’ of being vegetarian) with the help of the child. Portion sizes were estimated using homemade measurements or portions from labels. Once completed, the diaries were verified by interviewers with respondents, and missing details were added to enhance completeness. Diary days were randomly selected to ensure a balanced representation of all days of the week. Food consumption data from the complete registries were coded and edited using the DINO (Diet In, Nutrients Out) Program(Reference Fitt, Cole and Ziauddeen22), and nutrient intake was estimated using the NDNS (Nutrient Databank) food nutrition composition table(23).
To identify participant groups, we used data from a questionnaire where participants self-declared as either vegetarians or vegans, collectively referred to as vegetarians. Initially, participants were asked, ‘Can I check whether you would describe yourself as vegetarian or vegan?’ with response options ‘vegetarian’, ‘vegan’ or ‘neither’. If participants identified as vegetarian, an additional question served as a data quality control: ‘Can I check if you eat meat, fish, poultry, or dishes containing these foods?’ They responded with ‘yes’ or ‘no’. For those identifying as vegan, another data quality control question followed: ‘Can I check if you consume any food of animal origin, such as meat, fish, poultry, milk, dairy products, eggs, or dishes containing these foods?’ Again, they responded with ‘yes’ or ‘no’. We compared responses to the initial question with these follow-up questions and found no discrepancies. All foods reported by participants who identified themselves as vegetarians were included in the analyses, regardless of whether they included meat in their diet.
Our outcomes were based on the values recommended by the WHO for nutrient intake for the prevention of chronic non-communicable diseases: proteins, carbohydrates, free sugars, total fats, saturated fats and dietary fibre(24–27) and the European Food Safety Authority (EFSA) for micronutrients and vitamins A, B6, B12, C and E, Na, potassium, Ca, Mg, phosphorus, Fe, Cu, Zn, iodine and Se(28). Protein content, carbohydrates, free sugars, total fats and saturated fats were expressed as a percentage of total energy intake, while fibre and micronutrients were expressed per 1000 kcal (g, mg or μg per 1000 kcal). We used the following cut-offs specified by the WHO for free sugars ≥ 10 % of total energy, saturated fats ≥ 10 % of total energy and dietary fibre < 12·5 g/1000 kcal. The recommended values for each micronutrient, based on EFSA guidelines and stratified by sex and age group, are presented in online Supplementary Table 1. Both the average requirements of the EFSA reference values and, in the absence of these, the adequate intakes were considered. Due to the scarcity of information about postmenopausal women, the recommended amount of Fe for this phase (6 mg/d) was adopted for individuals over 40 years of age. As for Zn, average requirements recommendations vary according to the amount of phytate present in the overall diet. Since these data are not included in the NDNS database, the intermediate Zn recommendations for moderate phytate levels (7·6 mg/d for women and 9·3 mg/d for men) were used.
Food classification according to industrial processing
All foods present in the food records were classified according to Nova, a food classification system based on the nature, extent and purpose of the industrial processing to which the food was subjected prior to its consumption(Reference Monteiro, Cannon and Levy16). The foods were classified exclusively into one of the four groups of Nova: fresh or minimally processed foods; processed culinary ingredients; processed foods; and ultra-processed foods; as well as their respective subgroups.
All foods in the NDNS database are coded as food numbers and grouped into subsidiary food groups (n 155). For the subsidiary food groups that include foods belonging to different Nova groups (n 52), the codes were classified individually. By doing so, it was possible to classify each ingredient of homemade preparations into its corresponding Nova group and subgroup. Further details regarding food categorisation methods can be found in previous publications(Reference Rauber, Louzada and Steele29,Reference Rauber, Louzada and Martinez Steele30) .
Socio-economic and demographic characteristics
The socio-economic and demographic variables of interest were sex, age (continuous), region (England, Wales, Scotland and Northern Ireland), ethnicity (White, mixed groups such as White and Black Caribbean, White and Black African, White and Asian, or other origins, Black, Asian, and other ethnicities) and occupational social class (management positions and higher specialists, lower management and specialist positions, intermediate occupations (clerk, sales and service) and self-employed, manual and routine service occupations, technicians and unemployed). A missing category was created for the ‘occupational social class’ variable due to 2·8 % of the sample having missing data for this variable.
Data analysis
For the analyses, we used the average of all available food record days for each person, with more than 96 % completing all four food diary days.
First, we estimated the distribution of the total energy consumed by the individuals according to the groups and subgroups of Nova (% of the total energy of self-declared vegetarians). Next, we examined how the energy share of each Nova food group and subgroup varied in the tertiles of the energy share provided by ultra-processed foods in all groups. We also estimated nutritional indicators of the overall diet and within all tertiles of the dietary share of ultra-processed foods. Linear regression models were used to test trends in tertiles of the dietary contribution of ultra-processed foods. Standardised regression coefficients (β) were estimated to allow for comparisons across variables with different units.
Finally, we evaluated the prevalence of inadequate nutrient intake per tertile of the dietary share of ultra-processed foods. Prevalence ratios were estimated using Poisson regression. The models were adjusted for sex, age, ethnicity, occupational social class, region and year of research. We adopted a significance level of 5 % and for all analyses, the sample design of the research and its weighting factors were considered. Data were analysed using Stata version 16.0.
Results
Sociodemographic characteristics of the UK vegetarian population
A total of 359 participants who self-declared as vegetarians were eligible for the analyses, equivalent to 2·3 % of the general population in cycles 1–11 of the national survey. The majority of self-declared vegetarians were women (66·4 %), aged between 19 and 64 (73·6 %), of White ethnicity (70·5 %), held lower management or specialist positions (26·2 %) and lived in central England (55·7 %) (Table 1).
Table 1. Sociodemographic characteristics of the UK self-declared vegetarian population, NDNS 2008–2019 (percentages and 95 % confidence intervals)
† Other ethnicities: Included in this variable are ‘White and black Caribbean, white and black African, white and Asian or other origin’.
Distribution of the total energy consumed according to the groups of the Nova classification
The percentage of total energy consumed from fresh or minimally processed foods was 29·2 %, with the highest contributions coming from milk (4·6 %), fruits (4·4 %), grains (4·3 %) and roots and tubers (2·5 %). The percentage of total energy consumed from culinary ingredients was 5 %, with emphasis on animal fat (1·8 %), vegetable oil (1·5 %) and sugar (1·3 %). Processed foods contributed 9·4 % of the total energy consumed, with the highest consumption observed in cheeses (4 %), beers and wines (2·3 %), pickled vegetables (1·3 %) and processed breads (1 %). Ultra-processed foods contributed more than half of the total energy consumed, accounting for 56·3 %. The most consumed foods in this category were industrialised packaged breads (14 %), meatless ready-to-eat dishes (8 %), sweets and desserts (7·3 %), breakfast cereals (5 %), cookies (4 %) and ‘vegetarian’ meats (3·5 %) (Table 2).
Table 2. Distribution of average energy consumed according to Nova classification groups in the UK self-declared vegetarian population, NDNS 2008–2019 (Average and 95 % confidence intervals)
a Coffee, tea, mushroom, mixed dishes; b Fresh fruit smoothie and fruit smoothie; c vegetables preserved in brine and fruit in syrup; d condensed milk, salted chestnuts; e pizza, instant soups, meat-free ready-to-eat dishes; f pies and cakes, ice cream, popsicles, desserts, stuffed sweets; g refer to imitation processed ‘meats’ (e.g. ‘sausages’, ‘hams’ or vegetarian ‘hamburgers’). h soya-based beverages and other milk ‘substitute’ beverages; i distilled alcohol, artificial sweeteners; j burgers, sandwiches and ready-to-eat dishes that contain meats (such as bacon, sausage, beef, chicken or fish).
Distribution of groups and subgroups of the Nova classification according to tertiles of consumption of ultra-processed foods in the vegetarian population
The dietary share of ultra-processed foods ranged from 37·5 % of total energy content (first tertile) to 74 % (third tertile). Across the groups of fresh or minimally processed foods, culinary ingredients and processed foods, the dietary contribution of most subgroups decreased from the first to the last tertile of ultra-processed food consumption. Notably, reductions were observed in nuts and seeds, grains, and vegetables (fresh or minimally processed foods), vegetable oils (culinary ingredients), and beers and wines and processed breads (processed foods).
For the ultra-processed food group, the dietary share of most subgroups increased from the first to the last tertile of the contribution of ultra-processed foods and was statistically significant for ready-to-eat dishes (meat-free), sweets and desserts, cookies, vegetarian meats, packet snacks, soft drinks, crisp, margarine, other sugary drinks, canned vegetables in sauce, ultra-processed cheeses and reconstituted meats. It is noteworthy that French fries (433·3 % increase), vegetarian meats (280 %), meatless ready-to-eat dishes (269·4 %), sweets and desserts (167·6 %) and packaged snacks (126·6 %) were the subgroups that showed the highest increases along the tertiles of the dietary participation of ultra-processed foods (Table 3).
Table 3. Distribution of the groups and subgroups of the Nova classification according to the tertiles of consumption of ultra-processed foods (UPF) in the self-declared vegetarian population of the UK, NDNS 2008–2019
a Coffee, tea, mushroom, mixed dishes; b fresh fruit juices and fruit smoothie; c vegetables preserved in brine and fruit in syrup; d ham, meat and fish, smoked, salted or canned; e pizza, instant soups, meat-free ready-to-eat dishes; f pies and cakes, ice cream, popsicles, desserts, filled sweets; g refer to imitation processed ‘meats’ (for example, ‘sausages’, ‘hams’ or vegetarian ‘hamburgers’), h soya-based beverages and other milk ‘substitute’ beverages; i distilled spirits, artificial sweeteners; j hamburgers, sandwiches, and ready-to-eat dishes that contain meats (such as bacon, sausage, beef, chicken or fish).
Nutritional indicators of the diet according to the contribution of ultra-processed foods
The average energy intake of the participants was 1758·5 kcal/d, being distributed in the total diet in 12·3 % protein, 53 % carbohydrate, 10·9 % free sugar, 32·2 % fat and 11·4 % saturated fat.
As the tertiles of the contribution of ultra-processed foods to total energy intake increased, the content of free sugars increased significantly (from 8·9 % in the first tertile to 13 % in the last tertile), while the dietary content of fibre (from 15·7 % to 13·4 %), potassium (from 1742·3 mg to 1443·5 mg), Mg (from 178·6 mg to 154·2 mg), Cu (from 0·8 mg to 0·7 g), vitamin A (from 712·4 μg to 443·3 μg) and vitamin C (from 68 mg to 47·7 mg) decreased.
In the analysis using the continuous variable, we observed that for each 10 % increase in the dietary contribution of ultra-processed foods, the free sugar content (β –0·25) increased, while the density of fibre (β –0·26), potassium (β –0·38), Cu (β –0·22), Mg (β –0·31), Zn (β –0·15), iodine (β –0·25), vitamin A (β –0·37) and vitamin C (β –0·22) decreased (Table 4).
Table 4. Nutritional indicators of the diet according to consumption of ultra-processed foods in the self-declared vegetarian population of the UK, NDNS 2008–2019
* Standardised regression coefficient: adjusted for sex, age, ethnicity, occupational social class, region and year of survey.
Prevalence of inadequate intake according to dietary contribution of ultra-processed foods
More than half of vegetarians did not meet the recommended values for saturated fat (65·3 %) and dietary fibre (42·9 %). Regarding micronutrients, 95 % of the participants had inadequate potassium intake and more than half had inadequate intake of Mg, Cu, Zn, iodine, vitamin B12 and E. As the contribution of ultra-processed foods to total energy intake increased (from the first to the last tertile of consumption), the prevalence of inadequate intake of free sugars increased significantly (from 32·9 % to 60·7 %, respectively), as well as the prevalence of inadequate fibre intake (from 26·1 % to 47·5 %).
In the analysis using the continuous variable, we observed that for every 10 % of the dietary contribution of ultra-processed foods, the prevalence of inadequate intake of free sugar (prevalence ratio = 1·12), dietary fibre (prevalence ratio = 1·12) and potassium (prevalence ratio = 1·01) increased (Table 5).
Table 5. Prevalence of inadequate intake according to consumption of ultra-processed (UPF) foods in the self-declared vegetarian population of the UK, NDNS 2008–2019
PR, prevalence ratio.
* PR: adjusted for sex, age, ethnicity, occupational social class, region and year of survey.
† The tertile analyses were not performed because 100 % of the participants exceeded the maximum recommended Na intake.
‡ The recommended values for each micronutrient, by sex and age group, are presented in online Supplementary Table 1.
Discussion
Ultra-processed foods contributed more than half of the energy consumed by the UK’s vegetarian population, with emphasis on packaged breads, meat-free ready-to-eat dishes and sweets and desserts. As the dietary share of ultra-processed foods increased, there was a decrease in the share of fresh and minimally processed foods, culinary ingredients, and processed foods, particularly in subgroups that are considered markers of healthy eating, such as grains and cereals, fruits, nuts, and seeds, and vegetables.
The higher consumption of ultra-processed foods was associated with poorer nutritional quality of the diet, with higher free sugar content and lower amounts of fibre, potassium, Mg, Cu, iodine, vitamin A and vitamin C. These results provide new insights the dietary contribution of ultra-processed foods and their impact on the overall diet quality of vegetarians, since few studies have evaluated the diet of this population considering the different levels of processing. The lack of statistically significant findings regarding Na levels across the tertiles of ultra-processed foods may be attributed to the high prevalence of elevated Na intake among self-declared vegetarians, as all participants have an excessive intake of this nutrient (see Table 5). Another salient observation pertains to specific micronutrients, notably Fe, for which we did not discern a pattern akin to those commonly observed in analogous studies. This discrepancy may arise from the consumption of ultra-processed foods fortified with said micronutrients, or from inherent disparities in nutritional composition among distinct subcategories of ultra-processed foods.
The significant dietary contribution of ultra-processed foods to the total energy intake of British vegetarians observed in this study aligns with findings in studies conducted in high-income countries. Similar trends have been reported in the general populations of the USA (57·5 %), the UK (56·8 %), Canada (47·7 %), Australia (42 %) and France (24·1 %)(Reference Rauber, Louzada and Steele29,Reference Martínez Steele, Popkin and Swinburn31–Reference Calixto Andrade, Julia and Deschamps34) . Regarding vegetarian populations, a French cohort study showed that the energy contribution of ultra-processed foods was higher for vegetarians (37 %) and vegans (39·5 %) than for meat eaters (33 %) and that this high intake was driven by the consumption of vegetarian meats and plant-based beverages(Reference Gehring, Touvier and Baudry13).
The negative impact of the greater dietary contribution of these foods on the nutritional quality of the diet observed in our study is consistent with studies conducted with representative samples of the general population from several countries(Reference Machado, Steele and Levy32,Reference Calixto Andrade, Julia and Deschamps34–Reference Rauber, Steele and Louzada36) . In the UK, analysis of general population consumption data showed that as the consumption of ultra-processed foods increased, the dietary content of carbohydrates, free sugars, total fats, saturated fats and Na increased, while the protein, fibre and potassium content decreased(Reference Rauber, Louzada and Steele29).
The possible reasons for the high consumption of ultra-processed foods among British vegetarians may be due to social and economic issues, given that most of this population is in management and specialist positions(Reference Baker, Machado and Santos37). Along with this, most of it is in the centre of England, where you can have more access to ultra-processed foods. The idea that vegetarian eating is inherently healthy due to its superior nutritional quality has been widespread due to the potential benefits associated with health(Reference Zhao, Zhan and Wang9,Reference Wickramasinghe, Breda and Berdzuli10,Reference Segovia-Siapco and Sabaté12,Reference Li, Zeng and Wang38–Reference Loeb, Fu and Bauer40) . However, the findings of our research offer a critical view of this assumption by revealing that the vegetarian population has experienced a high consumption of ultra-processed foods and an unfavourable nutritional profile, specifically, indicating an increase in sugar content. This can be explained by the high amount of sugars present in plant-based products. Excessive consumption of these foods by vegetarians raises concerns about potential negative impacts on health(Reference Orlich, Sabaté and Mashchak14).
The lower content of fibre, potassium, Mg, Cu and vitamins A and C associated with the higher consumption of ultra-processed foods may result from the substitution of fresh and minimally processed foods with ultra-processed alternatives. A study conducted on the French NutriNet-Santé cohort demonstrated that most vegetarians have a preference for unhealthy products over healthier options, as assessed through a plant-based diet index(Reference Gehring, Touvier and Baudry13). This may explain, in part, the unfavourable nutritional profile of the diet observed in our analysis.
Finally, considering the energy contribution of some meat-containing dishes, we can observe that self-reported vegetarian diets and vegetarian diet status based on dietary data may not be consistent. This discrepancy may arise because vegetarians have their own individual definitions of vegetarianism, a term that can be interpreted broadly.
In recent years, there has been a rise in plant-based meat and dairy substitutes, with various companies capitalising on the growing consumer preference for vegetarian options. These products are prominently displayed in supermarkets and often marketed alongside traditional meat and dairy products. While vegetarian or plant-based diets are generally associated with health benefits, not all plant-based diets offer the same advantages. Our study focuses on how ultra-processed foods affect the energy intake of self-declared vegetarians in the UK and the nutritional quality of their diets. This research suggests the need for policies that encourage informed food choices among vegetarian consumers. We used data from the NDNS, which employs a high-quality dietary assessment method providing a detailed analysis of the foods consumed. This method considers variations in consumption between different days of the week and seasons. Additionally, we utilised the Nova food classification system, which is widely recognised as a valid tool for nutrition research, public health and policy.
Some potential limitations also need to be considered. The data that were used are self-reported and therefore susceptible to social desirability bias; in essence, it constitutes a systematic research bias wherein participants tend to provide responses that align more closely with socially desirable norms rather than accurately reflecting their genuine opinions or behaviours. Regarding the methods, although food records are already recognised as one of the most comprehensive methodologies for assessing food consumption, one limitation is the possibility of abstaining from reporting some foods, especially ultra-processed foods. This may be due to the wide dissemination of the findings that the consumption of these foods causes in health. However, the data that NDNS provides are validated and accurate, where reviews are made by trained professionals to minimise the possibility of incorrect records(21). Even though the data in the NDNS database do not consider the degree of food processing, the standardisation methods minimise possible errors and biases. Lastly, we acknowledge that the cut-off values in UK guidelines differ slightly from those of the WHO and EFSA, which may lead to minor variations in prevalence rates of nutrient inadequacy.
Conclusion
In this cross-sectional study, it was possible to analyse food consumption in detail, considering the different levels of industrial processing among British vegetarians, and to evaluate the impact of the dietary contribution of ultra-processed foods on the intake of nutrients recommended in the international guidelines for the prevention of chronic non-communicable diseases.
Our findings highlight that the impact of ultra-processed foods on the diet of the vegetarian population in the UK constitutes more than half of the energy consumed. Therefore, a higher consumption of ultra-processed foods is associated with a poorer nutritional quality of the diet.
We reinforce the need for public policies that include fiscal and regulatory measures on these foods in order to reduce their consumption and thus increase the consumption of healthy preparations from minimally processed foods.
Acknowledgements
The author B. M. A. T. was supported by the Coordenacao de Aperfeiçoamento de Pessoal de Nível Superior (CAPES). CAPES had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
B. M. A. T.: Conceptualisation, Formal analysis, and Writing – original draft. M. M.: Writing – review and editing. M. L. C. L.: Writing – review and editing. R. B. L.: Conceptualisation and Writing – review and editing. F. R.: Conceptualisation, Data curation, Methodology; Writing – review and editing, and Supervision. All authors have read and agreed to the final manuscript version.
The authors have no conflict of interest to declare.
Data and files used in this study were acquired under licence from the UK Data Archive found at https://ukdataservice.ac.uk/. NDNS was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the UK National Health Service (NHS) Health Research Authority Research Ethics Committee (REC) Approval (Years 1–5, Oxfordshire REC A, REF 07/H0604/113; Years 6–10 and 11, East of England-Cambridgeshire South REC, REF 13/EE/0016). Written and/or verbal informed consent was obtained from all participants.
Supplementary material
For supplementary material/s referred to in this article, please visit https://doi.org/10.1017/S0007114524001909
