Today, recommended diets should not consider only consumers’ health but also food system sustainability(Reference Fardet and Rock1). At first view, defining such a diet, covering all four securities at once (sanitary, health, nutritional and environmental), appears to be a tremendous task(Reference Gazan, Brouzes and Vieux2–Reference Ronzon, Paillard, Chemineau, Esnouf, Russel and Bricas4). In all cases, nutritional sciences should be included in a transdisciplinary approach to include all the four criteria. Therefore, adopting a more holistic perspective should be preferred over the present single-nutrient/food group reductionist approach(Reference Fardet and Rock5) for which contradictory results have been obtained at present. Indeed, some studies showed that high nutritional quality, healthy diets and/or diets adhering to food-based dietary guidelines are not necessarily associated with lower greenhouse gas emissions (GHGE)(Reference Vieux, Soler and Touazi6–Reference Hitaj, Rehkamp and Canning10), while other studies are more moderate(Reference Sjors, Hedenus and Sjolander11–Reference Musicus, Moran and Lawman16). Moreover, GHGE is not the only issue to consider to define sustainable diet; there are also land use, water footprint, biodiversity, socioeconomic aspects and animal well-being. FAO of the United Nations defines sustainable diets as those that protect human health, the environment (pollution, deforestation, GHGE, etc.), small famers, culinary traditions and socioeconomics (healthy foods accessible to everyone, social life, fair trade, etc.)(17,18) , to which animal biodiversity and welfare(Reference Scherer, Tomasik and Rueda19) can also be added. In addition, it is worth mentioning that the diets of each world region depend on different economic, pedo-climatic and agronomic conditions. Therefore, the design of a sustainable diet should be sufficiently generic and based on the specific local conditions of each country respecting culinary traditions first derived from local agricultural production.
Scientific evidence has shown that excess animal energy content, notably red and processed meats, is the main factor threatening both food system sustainability(Reference Caillavet, Fadhuile and Nichèle20–Reference Llonch, Haskell and Dewhurst23) and health(Reference Farvid, Stern and Norat24–Reference Händel, Cardoso and Rasmussen28), some studies investigating both at once(Reference Aston, Smith and Powles29–Reference Scarborough, Appleby and Mizdrak31). However, several recently published reports and papers demonstrated that a balanced consumption of meat is sustainable for both health and environment(Reference Bernardi, Capri and Pulina32–Reference Adesogan, Havelaar and McKune36) ‘as they play a critical role in improving nutrition, reducing poverty, improving gender equity, improving livelihoods, increasing food security, and improving health’(Reference Adesogan, Havelaar and McKune36). Therefore, the issue is not to suppress animal foods but to achieve a win–win–win approach to synergistically protect human health, the environment and animal welfare by significantly reducing animal food production and consumption(Reference Scherer, Behrens and Tukker37).
Similarly to animal energy content, those from ultra-processed foods (UPF) have been associated with an increased risk of several chronic diseases or conditions in over thirty epidemiological studies(Reference Monteiro and Cannon38,Reference Fardet and Rock39) . These results, together with previous evidence about the influence of the degree of food processing on health (e.g., whole-grain v. refined cereals, raw fruits v. sweetened fruit juices, or red v. processed meats), showed that the degree of processing matters for defining food health potential(Reference Fardet and Rock1,Reference Fardet40) more than their nutrient contents alone(Reference Monteiro41). It has been also suggested that they are not associated with food system sustainability, notably due to increased plastic pollution, deforestation, intensive monocultures, energy-demanding technological processes, GHGE and excess water use(42).
Besides focusing on animal and ultra-processed foods, it is well admitted now that ‘eating varied’ ensures an optimum supply of synergistic bioactive protective compounds against chronic diseases(Reference Jacobs and Tapsell43,Reference Jacobs and Steffen44) as well as increases and improves environmental biodiversity(Reference Fardet and Rock1) contributing to its sustainability. Indeed, in some world regions, monotonous diets are shown to result in micronutrient deficiencies (mainly vitamin A, proteins, Fe, iodine and Zn)(45) in developing/emerging countries, or in developed countries when the western diet includes excess ‘empty’ ultra-processed energy content(Reference Marron-Ponce, Flores and Cediel46–Reference Louzada, Martins and Canella52). Thus, dietary diversity in developing countries is shown to be positively associated with nutritional adequacy(Reference Rani, Arends and Brouwer53–Reference Nguyen, Huybregts and Sanghvi55). On the contrary, dietary diversity in developed countries has not been necessarily associated with positive health outcomes in contrast with diet quality(Reference de Oliveira Otto, Padhye and Bertoni56). The reason is that diversity in these countries is generally associated with consuming more diversified UPF, not fulfilling healthy eating criteria(Reference Monteiro and Cannon38). Therefore, the well-known ‘eat varied’ applies to diverse raw agricultural commodities and mildly processed nutrient-dense foods.
From the above considerations, a generic and sustainable complex diet should be holistic and composed of diversified and high-quality foods containing a certain animal:plant ratio and a lower degree of food processing. However, healthy diets advocated today are either regional, for example, the Mediterranean or Nordic diet, or national (e.g., food pyramids or plates). The basis of these diets is essentially nutritional, and many failed to consider the growing rate of UPF share worldwide(Reference Monteiro and Cannon38,57) , threatening traditional diets in return.
The aim of the current study was to design a generic diet based on three dimensions at the same time, as identified previously(Reference Fardet and Rock1). Qualitative, rather than quantitative, recommendations allow the generic diet to be easily extrapolated and adapted to regional socioeconomic, climatic and agronomic specificities. Using a data mining approach including both original papers and foresight scenarios from several international institutions, we determined daily maximum animal and ultra-processed energy percentages for a healthy and sustainable diet while tending towards diversity among both animal- and plant-based foods. The definition of such a generic diet first intends to be disease-preventive, that is, increasing people’s healthy life-years while preserving food system sustainability.
Determination of animal:plant energetic ratio
A literature search was carried out through mining of scientific literature from two main sources:
1. Studies of the association between traditional a priori (i.e., defined prior to the study) and a posteriori protective diets and health: The diets were selected based on meta-analyses, reviews and/or prospective cohort studies (when meta-analyses not available) of their associations with health outcomes, including risks of all-cause mortality, overweight/obesity, type 2 diabetes, cancers, CVD, mental illnesses (e.g., depression, Alzheimer’s disease, Parkinson’s disease, cognitive decline) and/or other chronic conditions (e.g., osteoporosis, hypertension or glycaemic control). Our search was carried out with the ISI Web of Science up to December 2019. Diets with no proven health benefits, or not sufficiently studied, were not selected.
2. Scientific reports defining sustainable diets at horizon 2050 as protective of both human health and different environmental outcomes (i.e., GHGE, land and pesticide use, water footprint, food waste, etc.): A search was carried out for foresight collective expertise from the websites of international organisations, government institutions, research institutes and private foundations up to December 2019. Only those reports showing the animal energetic share and those taking into consideration environmental outcomes were selected.
Our objective was not to review scientific papers and reports but extract data about daily animal energy shares to reach both human health and environmental sustainability. Otherwise, it should be remembered that the health protectiveness of diets, as defined by the current study, is mainly based on their associations or correlated relationships, not causality. However, accumulating scientific, as generally used for defining national dietary recommendations, was considered sufficient to select such diets as protective of human health.
For each selected diet, the recommended minimum and maximum number of servings per day was collected and converted into daily energy percentages. The energy conversion, as given by either the French Ciqual(58) or American USDA(59) databases, was based on recommended serving sizes(60) of generic foods representative of the animal product category, that is, white meat, red meat, milk, cheese, yogurt, eggs, seafood and fish (combining fatty and lean fishes). Overall, by averaging all generic animal products, one serving of animal product corresponded to approximately 7·5 % of daily energy content (Table 1). However, serving sizes may vary according to countries, and not normalised at the European level. Serving sizes chosen by the current study correspond to those recommended by dietitians in France(60). Finally, a median serving of animal products was calculated based on the reported daily minimum serving of each diet.
* Based on LaNutrition.fr website (https://www.lanutrition.fr/bien-dans-son-assiette/bien-manger/les-recommandations-de-lanutrition.fr/une-portion-cest-combien).
† Based on French Ciqual database.
‡ Number of different foods.
§ Based on a daily basis of 2000 kcal for an adult, that is, (149 × 100)/2000.
Determination of non-ultra-processed: ultra-processed energetic ratio
In brief, UPF were derived from the NOVA classification(Reference Moubarac, Parra and Cannon61) and ‘are formulations of ingredients, mostly of exclusive industrial use, that result from a series of industrial processes (hence “ultra-processed”) … Processes include the fractioning of whole foods into substances, chemical modifications of these substances, assembly of unmodified and modified food substances, frequent use of cosmetic additives and sophisticated packaging’ (pp. 936 and 937)(Reference Monteiro, Cannon and Levy62). For example, UPF may include, among others, carbonated soft drinks, sweet or savoury packaged snacks, chocolate, candies (confectionery), ice cream, mass-produced packaged breads and buns, margarines and other spreads, biscuits, pastries, cakes and cake mixes, breakfast ‘cereals’, burgers, hot dogs and other reconstituted meat products, etc.(Reference Monteiro, Cannon and Levy62) To date, thirty-four epidemiological studies have reported about excess ultra-processed energy consumption and the risks of chronic diseases and/or metabolic dysregulations(Reference Monteiro and Cannon38,Reference Fardet and Rock39) . The most studied chronic disease was overweight/obesity, with three ecological studies, five cross-sectional studies and four longitudinal prospective studies(Reference Fardet and Rock39). Overweight/obesity was the most studied metabolic dysregulation, and is the first step to chronic diseases such as CVD and/or cancers(Reference Fardet and Boirie63). Therefore, obesity risk was the criterion used to select studies for the evaluation of two thresholds: (i) the median threshold of ultra-processed energetic intake at which the risk of overweight/obesity begins to significantly increase, and calculated from all selected studies; and (ii) the precautionary threshold, which corresponds to the minimum consumption of ultra-processed energy content at which the risk of overweight/obesity begins to significantly increase.
The 3V rule and nutritional needs
Based on the evaluated thresholds of animal and ultra-processed energetic percentages, we then tested the ability of a 3V-based diet to address nutritional needs by defining a weekly standard diet pattern based on common staple foods consumed in France, considered as an industrialised country (Table 2), from the French Ciqual(58) and American USDA(59) databases. This French generic diet also addressed food diversity, that is, at least two different food varieties were used among red meat, white meat, seafood, fish, eggs and dairy products and among cereals, nuts and seeds, legumes, fruits, vegetables and tubers. For example, for the cereal group, maize and wheat were used, not only wheat-based products; for fruits, more than one fruit type was used; for dairy products, cheese, yogurt and milk were used, and so on. From this theoretical and generic diet, the level of macronutrients, fibre and micronutrients (vitamins, minerals and trace elements) was calculated, and included at-risk nutrients when decreasing animal products or consuming too many UPF. These latter nutrients include essential amino acids (EAA, 13 g/d), SFA (12 % maximum of daily kilocalories), simple sugars from added sugars, fruit juices and honey as defined by the WHO (daily 10 % maximum energy content)(64), vitamin B12, vitamin D, vitamin A, Ca, iodine and salt, and linoleic acid (LA), α-linolenic acid (ALA), EPA, DHA, with a selected average conversion rates of ALA in EPA and DHA of 14 and 5 %, respectively(Reference Stark, Crawford and Reifen65).
DASH, Dietary Approach to Stop Hypertension.
* Number of minimum and maximum servings has been defined from data available in scientific papers and recommended food guide pyramids found on websites.
† Two servings of animal-based foods daily is around 15 % energy content based on a weekly 3 V-based diet (see Table 1).
The calculations were based on a standard 2000 kcal/d for a healthy adult, including 1·5 l/d of mineral water. The weekly supply of nutrients was compared with the dietary reference intake (DRI) in percentages and other recommendations. Finally, based on the French food guide pyramid, the diet was also designed to balance – among plant-based foods – (i) whole-grain cereals, legumes, nuts and seeds, that is, approximately 30–40 % daily energy content, (ii) fruits and vegetables, that is, approximately 20–30 % daily energy content and (iii) added fats and sugars, that is, approximately 15 % daily energy content.
Dimension 1: determination of a median optimal value of daily animal energy intake
The most studied diets were Okinawan, prudent, vegetarian, Mediterranean, Palaeolithic, Dietary Approach to Stop Hypertension (DASH), anti-inflammatory and Nordic diets (Table 2). Others such as Inuit, portfolio, flexitarian and ketogenic diets were very specific diets and/or not sufficiently studied. Scientific evidence towards their protective potentials was not enough relevant or consensual in order to include them in the current study. Notably, the ketogenic diet was tested in very specific pathological and clinical conditions, not as a preventive diet towards the abovementioned chronic conditions; the Inuit diet is confined to a very specific region; and there were no epidemiological studies on the portfolio diet.
GHGE, greenhouse gas emission; TYFA, Ten Years for Agroecology; FiBL, Swiss Research Institute of Organic Agriculture.
We found four a priori traditional diets (i.e., Okinawan, Palaeolithic, Mediterranean and Nordic/Baltic Sea diets) and four a posteriori generic diets (i.e., prudent, vegetarian, anti-inflammatory and DASH diets) as being potentially protective against the development of chronic diseases (Table 2):
1. The traditional Okinawan diet is a low-energy, nutrient-dense, antioxidant-rich diet pattern(Reference Willcox, Willcox and Todoriki66,Reference Willcox, Scapagnini and Willcox67) . The number of servings of animal products is 0–3 per day, that is, approximately 11·3 % energy content(Reference Willcox, Scapagnini and Willcox67).
2. The Palaeolithic diet is the richest in animal products (6–7 servings per day), allowing around equal proportions of animal/plant energy content daily(Reference Cordain68).
3. The Mediterranean diet(Reference Martinez-Lacoba, Pardo-Garcia and Amo-Saus69) provides an average of 2–4 servings daily corresponding to approximately 22·5 % energy content(Reference D’Alessandro, Lampignano and De Pergola70).
4. The Nordic/Baltic Sea diet is an umbrella term that encompasses any interpretation that combines food-based dietary guidelines with local Nordic foods(Reference Meltzer, Brantsæter and Trolle71,72) , corresponding to approximately 22·5 % animal energy content, that is, 2–3 servings of animal products per day(Reference Mithril, Dragsted and Meyer73).
Based on a posteriori scientific evidence, four generic diets have been proposed to prevent chronic diseases (Table 2):
1. The prudent diet – in which saturated/trans-fat intake is <10 % of total energy content and cholesterol is <300 mg/d and/or fibre intake is ≥25 g/d for women and ≥35 g/d for men(Reference Lichtenstein, Appel and Brands74) – is typically high in total fat (35–45 %) but low in SFA (7–8 % of energy), with an average of 1–4 servings of animal products per day, that is, approximately 18·8 % energy content(Reference Enas, Senthilkumar and Chennikkara75).
2. The vegetarian diet(Reference Parker and Vadiveloo76) includes approximately 1–3 servings per day of animal products, that is, an average of 15 % daily energy content.
3. DASH(Reference Rifai and Silver77) indicates 2–4 servings of animal products per day(Reference Rifai and Silver77), that is, an average of 22·5 % animal energy content.
4. The anti-inflammatory diet indicates fewer animal products with 0–1 daily serving, that is, approximately 9·4 % energy content.
Considering these eight traditional/a posteriori generic diets and the minimum number of recommended daily animal servings (Table 2), a median of 11·3 % animal energy content per day was reached, that is, 1·5 servings per day.
Several international agencies have proposed foresight scenarios to define sustainable diets to protect both human health and the environment at horizon 2050, mainly by reducing environmental impacts arising due to food systems (Table 3). Our search identified eight relevant scenarios that take into consideration such environmental impacts as climate change, land and water use and biodiversity, among others: Agrimonde-Terra(Reference Le Mouël, de Lattre-Gasquet and Mora78), EAT-Lancet(Reference Willett, Rockström and Loken79), WWF2050(80,81) , Nordic Sufficiency(Reference Karlsson, Röös and Sjunnestrand82), Swiss FeedNoFood2050(Reference Stolze, Schader and Müller83), Afterres 2050(84), IDDRI(Reference Poux and Aubert85) and Millennium Institute(86) (Table 3).
Concerning international foresight scenarios, Agrimonde-Terra, focusing on land use for a healthy diet, is based on 500 kcal of animal energy content among a total of 3000 kcal, that is, 16·7 % animal energy content(Reference Le Mouël, de Lattre-Gasquet and Mora78). EAT-Lancet designed a ‘universal’ healthy diet for a sustainable food system with approximately 13·6 % animal energy content(Reference Willett, Rockström and Loken79). The Millennium Institute, focusing on total food waste and losses and total harvested area, proposed four scenarios, among which scenario four takes the most environmental issues into consideration, with approximately 16·0 % daily animal energy content. At the European level, the IDDRI proposed the TYFA (Ten Years for Agroecology) scenario that corresponds to around 13·5 % daily animal energy content(Reference Poux and Aubert85).
Concerning regional scenarios, the French scenarios proposed approximately 24·3 % animal energy content for Afterres 2050(87) and approximately 25·7 % animal energy content for WWF France(80,81) . The Swiss Research Institute of Organic Agriculture (FiBL) proposed an optimised diet, called the FeedNoFood2050 scenario, that assumes an improved use of agricultural land by feeding only grass and by-products to livestock, which are not competing with direct human nutrition, that is, do not require arable land (neither in Switzerland nor abroad), thereby allowing good social, health, economic and environmental sustainability compared with their reference diet(Reference Stolze, Schader and Müller83). It corresponds to approximately 14·7 % daily animal energy content (Table 4). Finally, the Nordic Council of Ministers proposed a generic sufficiency scenario for all Nordic countries, with approximately 17·9 % daily animal energy content(Reference Karlsson, Röös and Sjunnestrand82).
UPF, ultra-processed food.
* As defined by NOVA classification.
† Servings were converted into kilocalories based on the 3 V-based diet (Table 2), that is, one serving is approximately 6·9 % kcal daily.
Taking into account traditional/a posteriori generic diets (median 11·3 %) and foresight diet scenarios (median 16·4 %) together, a median value of approximately 15·0 % animal energy content was reached. In comparison, actual animal energy consumption in France is approximately 37–45 %(88) and that recommended by French PNNS is approximately 25–30 %(89) (Table 2). Beyond France, considering thirty-seven countries, the average recommendation of daily animal energy intake is 21·9 v. 26·7 % for actual consumption (Table 2)(Reference Scherer, Behrens and Tukker37).
Dimension 2: determination of the maximum value of daily ultra-processed energy content
Six epidemiological studies – four cross-sectional(Reference Louzada, Baraldi and Steele90–Reference Juul, Martinez-Steele and Parekh93) and two longitudinal(Reference Canhada, Luft and Giatti94,Reference Mendonca, Pimenta and Gea95) – investigated the association between increased risk of overweight/obesity and excess ultra-processed energetic consumption (Table 4). In the first Brazilian study involving 55 970 households, obesity risk increased by 2·1 % beyond 17·3 % ultra-processed energy content(Reference Canella, Levy and Martins91). In the second Brazilian study involving 30 243 individuals aged ≥10 years, obesity risk increased by 29 % with the share of ultra-processed energy content increasing from 14 to 22 %(Reference Louzada, Baraldi and Steele90). In the third study conducted in the USA, an overweight risk of over 20 % was observed from 36·6 % UPF energy content(Reference Juul, Martinez-Steele and Parekh93). Finally, in a Canadian cross-sectional study involving 19 363 adults aged ≥18 years, obesity risk increased by 8 % from 36·1 % ultra-processed energy content, and by 3 % for each 10 % increase of ultra-processed energy content(Reference Nardocci, Leclerc and Louzada92). Concerning longitudinal studies, the first Brazilian cohort comprised 4525 civil servants aged 35–74 years at baseline, and obesity risk increased by 36 % from 23·9 % ultra-processed energy content, with a linear increase between 0 and 30 % ultra-processed energy content up to approximately 59 % risk of overweight/obesity(Reference Canhada, Luft and Giatti94). The second longitudinal study was carried out involving in 8451 middle-aged Spanish university graduates, where overweight/obesity risk significantly increased from 2·7 servings per day of UPF, that is, approximately 18·6 % of energy content(Reference Mendonca, Pimenta and Gea95).
From these studies, the median daily ultra-processed energetic percentage at which obesity risk begins to significantly increase is approximately 21·3 % (Table 5). Taking the lowest precautionary threshold, we obtained 14·0 % at which obesity risks significantly increased by over 29 %(Reference Louzada, Baraldi and Steele90). At 14 %, a cubic spline analysis in the longitudinal study by Canhada et al. (Reference Canhada, Luft and Giatti94) showed an increased overweight/obesity risk of approximately 29 % (Table 5).
* The diet comprises 14·1 and 15·0 % energy content from ultra-processed and animal foods, respectively. , animal foods; , ultra-processed foods; , both ultra-processed and animal-based foods.
Fifteen per cent maximum animal and ultra-processed energy content v. nutritional needs
Based on generic staple foods and including both maximum 15 % animal and ultra-processed energy content and a variety of food groups (≥42 different animal and plant-based foods), a standard French generic diet was designed (Table 5). This diet comprises 36·1, 51·6 and 12·4 % lipid, carbohydrate and protein energy content, respectively. Whole grains and nuts, fruits and vegetables and added plant oils and sugars account for 33·7, 20·2 and 14·5 % of total energy content, respectively.
With reference to a daily energy requirement of 2000 kcal for a healthy adult, this standard diet supplies 2054 daily energy content, including 14·5 and 15·4 % energy content provided by ultra-processed and/or animal foods, respectively, that is, some animal-based foods were also ultra-processed (Table 5). A further analysis of this diet proceeded to address the levels of nutrients according to dietary recommended intake (Table 6). Sparing a few, most of the nutrients were found to be close to DRI. Added sugars were slightly below the maximum 10 % recommended by the WHO, that is, 9·3 %. SFA contributed to 9·1 % compared with the maximum recommended intake of 12 % daily energy content. Essential amino acids were largely addressed, that is, >13 g/d. If EPA and DHA were below recommendations (45 and 94 %, respectively), the contribution from ALA converting into EPA (approximately 14 % of ALA) and DHA (approximately 5 % of ALA) allowed reaching the DRI with 163 and 136 %, respectively. In addition, salt consumption and the LA:ALA ratio were below the maximum recommended level of 5 g/d and < 5, respectively. Without the addition of dried seaweed (nori), iodine DRI was not reached, at only 88 %. If veal liver contributed greatly to the vitamin A supply (106·8 % of DRI without veal liver, instead of 306·8 %), then β-carotene remains a relevant supply of provitamin A. Finally, vitamin D was only 19·1 % of the DRI, without considering vitamin D synthesis through exposure to sun.
LA, linoleic acid; ALA, α-linolenic acid; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid.
* See Table 5: 15 % ultra-processed foods represent 2·25 % of ultra-processed animal energy content among the 15 % (i.e., two servings per week) and 12·75 % of plant energy content among the 85 % (i.e., twelve servings per week); the diet comprises 12·1, 35·2 and 52·7 % of protein, lipid and carbohydrate energy content, respectively; 20·2, 33·7 and 14·5 % of fruits/vegetables, whole grains/legumes/nuts and added fat/sugar energy content, respectively, based on 1·5 l/d of mineral water.
† For a healthy adult.
‡ For an adult weighing 70 kg.
§ Daily intakes of macronutrients are only indicative because recommendations are 10–20 % for proteins, 40–55 % for carbohydrates and 35–40 % for lipids.
∥ Simple sugars from added sugars, fruit juices and honey.
¶ Conversion rates of ALA to EPA and DHA are on average 14 % (8–20) and 5 % (0·5–9), respectively(Reference Stark, Crawford and Reifen65).
** Including β-carotene converted to vitamin A (where 12 mg β-carotene = 1 mg vitamin A equivalent).
Our results showed that it is possible to design a complex diet compatible with both protection of human health and food system sustainability. Based on the ‘3V rule’ – for Végétal (plant), Vrai (real) and Varié (varied, if possible organic, local and seasonal), in French – these three dimensions are interconnected in the proposed diet. Therefore, a 3V-based diet may range between 15 and 30 % of both animal and/or ultra-processed energy content. For example, if all animal products are ultra-processed, the remaining 85 % energy content are all non-ultra-processed, plant-based products; and if all animal energy content are non-ultra-processed, then the remaining energy content are 15 % ultra-processed, plant-based energy content and 70 % are non-ultra-processed, plant-based energy content.
Otherwise, although tending towards a certain form of genericity and converging towards similar conclusions of EAT-Lancet (see key message no 4)(Reference Willett, Rockström and Loken79), the current study had different purposes. While EAT-Lancet notably focused on the percentage shares of different food groups for reaching ‘universal’ sustainability, the current study aimed to extract the potential generic and fundamental dimensions that would help in defining a healthy and sustainable diet, and that can be adapted and also used as levers by different public health policies. Thus, none of the diets considered by the current study (to propose percentage shares of animal and ultra-processed energy content) have taken into consideration the three dimensions of the 3V rule. In addition, the degree of food processing, based on the UPF concept, was not directly considered in EAT-Lancet. Finally, the 3V rule appears more to be an ‘umbrella concept’ for a downstream (rather than upstream) appropriation by the society at large through dieticians, nutritionists, restaurateurs, consumers and encompassing the abovementioned studies and reports.
The 3V rule and nutritional needs
Our calculations were made within the framework of a French-style diet characteristic of developed countries. However, similar simulations could be done within other contexts where different types of animal- and plant-based products are consumed. Fifteen percentage of animal energy content may appear small at first view, but it is sufficient to supply all the essential macro- and micronutrients, provided the third rule (varied) is correctly applied (Tables 5 and 6). In addition, a reduction in ultra-processed energy content (‘empty’ energy content) will bring in new, minimally processed nutrient-dense foods rich in bioactive compounds that are protective against chronic diseases (i.e., more fibre, vitamins, minerals and antioxidants)(Reference Marron-Ponce, Flores and Cediel46–Reference Luiten, Steenhuis and Eyles51,Reference Fardet, Lakhssassi and Briffaz96,Reference Fardet, Méjean and Labouré97) . Notably, iodine and vitamin A DRI are easier to address when including a weekly serving of seaweeds/seafood (e.g., dried seaweeds) and offal/giblets (e.g., animal liver), respectively. Concerning EPA and DHA, a sufficient supply in ALA, notably through adequate use of plant oils, allows the DRI to be reached. Concerning vitamin D, the issue is different because many diets worldwide do not address its DRI, and many people in most countries are more or less vitamin D-deficient(Reference Palacios and Gonzalez98). Therefore, vitamin D deficiency is not due to the proposed 3V rule-based diet. Approximately one billion people worldwide have low levels of vitamin D in their blood(Reference Sahota99). Notably, according to a 2011 study, 41·6 % of adults in the USA are deficient, although their omnivorous diet is still rather rich in animal energy content(Reference Forrest and Stuhldreher100). Such deficiency may be attributable to a high consumption of ‘empty’ energy content, that is, a nutrient-poor diet(Reference Slining and Popkin101), and low sun exposure. Normally, adequate exposure to sun (UV B radiation) should allow addressing the vitamin D DRI(Reference Palacios and Gonzalez98), with differences in the needed exposure time between summer (10–20 min) and winter (2 h) periods. Beyond the mere quality of the diet, the question of vitamin D supplementation deserves to be addressed. Another solution for people not getting enough exposure to sun would be to consume approximately 35 g of cod liver oil weekly to achieve 100 % of DRI(58).
Reducing animal energy content to 15 % daily
The first ‘Végétal’ rule is probably the most problematic and controversial, for which it is very difficult to reach consensus. Globally, a consensus over achieving the balance between animal- and plant-based food energy content to improve the healthiness of a diet is still lacking. A theoretical value of 15 % animal energy content was proposed based on literature and determining a median value from combined minimal values proposed within these supposedly healthy diets. First, the median animal energy intake of eight recognised protective diets is 13·1 %, but only 11·3 % when considering the minimum number of daily servings. The eight selected scenarios for reaching diet sustainability at horizon 2050 converge to a median value of 16·4 % daily animal energy content. Combining both observational and foresight diets, a level of approximately 15 % daily animal energy content appeared to be optimum, particularly when respecting the two other rules, that is, ‘Vrai’ and ‘Varié’. Therefore, including environmental outcomes does not substantially modify the animal energy share of a diet to be protective of human health.
In developed countries where animal foods take a high level of energy share, plant-based foods are reported to be a relevant alternative(Reference Sjors, Hedenus and Sjolander11,Reference Seconda, Baudry and Alles12,Reference Macdiarmid, Kyle and Horgan15,Reference Scarborough, Appleby and Mizdrak31,Reference Scherer, Behrens and Tukker37) , even if some geographical regions, such as the Arctic, are more hostile for plant growth. In France where animal energy content are reaching approximately 37–45 % of energy share(88), at least 20 % can be replaced by under-consumed niche foods such as whole-grain cereals(Reference Bellisle, Hébel and Colin102), nuts and seeds(88) and legumes(88), all currently consumed under 14 g/d. In addition, this could contribute to sustainable crop production, especially legumes(Reference Foyer, Lam and Nguyen103). A diet based on a maximum of 15 % animal energy content is close to the recently emphasised healthy flexitarian diet(Reference Derbyshire104), meaning that animal foods would become an exception, which is not commonly the case today in western countries. In addition, as recently pointed out, ‘an increase in the adoption of plant-based diets presents an opportunity for the world to re-evaluate how meat can be sustainably produced, with greater emphasis on animal welfare, nutritional value, product safety, better utilisation, and distribution channels’(Reference Hicks, Knowles and Farouk105). Notably, Shepon et al. (Reference Shepon, Eshel and Noor106) demonstrated that by replacing all animal-based foods in the US diet with plant-based alternatives will add enough food to feed 350 million more people, while pointing out the importance of dietary shifts to improve food availability and security. This dietary shift may well correspond to the well-known flexitarian diet (for which, however, the dimension ‘real’ (‘Vrai’) is lacking in its accepted definition, notably through the level of UPF, and for which the dimension ‘varied’ is incomplete through the absence of ‘local’ and ‘seasonality’ criteria). Applying the 3V rule is also in agreement with the fact that a shift towards 85 % energy content being provided by plant-based foods should be very effective in reducing GHGE(Reference Springmann, Clark and Mason-D’Croz107). Other researchers suggested that moving to a flexitarian diet will lead to reduced GHGE (−54 %), land use (−8 %) and water footprint (−11 %)(Reference Batlle-Bayer, Aldaco and Bala108). The impact of dietary change was reported to be the highest on GHGE by partially replacing meat with plant-based foods, with effects being reduced through a replacement with mixed foods or dairy(Reference Hallström, Carlsson-Kanyama and Börjesson109).
However, the impact of dietary changes on sustainability is also driven by the national income levels(Reference Behrens, Kiefte-de Jong and Bosker110). In high-income countries, the nationally recommended diets are associated with reductions in GHGE, eutrophication and land use from 13·0 to 24·8 %, 9·8 to 21·3 %, and 5·7 to 17·6 %, respectively, while in upper-middle-income nations, such a recommended diet is only associated with slight decreases in impacts of 0·8–12·2 %, 7·7–19·4 % and 7·2–18·6 %(Reference Behrens, Kiefte-de Jong and Bosker110). The authors concluded that ‘reduced environmental impact in high-income countries is driven by reductions in energy content (≈54 % of effect) and a change in composition (≈46 %)’ while in low- and middle-income nations, the increased environmental impacts are associated with increased intake in animal products’(Reference Behrens, Kiefte-de Jong and Bosker110). Therefore, the adoption of a 3 V-based diet would mean a drastic reduction in the use of animal products in developed countries, stabilisation in emerging countries and an increase in developing countries where the daily animal energy share is under 10 % most of the time.
An important but neglected issue when considering food system sustainability is animal welfare. Scherer et al.(Reference Scherer, Tomasik and Rueda19,Reference Scherer, Behrens and Tukker37) were among the first to quantify the impacts of animal consumption on animal welfare through the development of an animal protection index. They found that animal welfare loss, associated with the current omnivorous average diet, is mostly driven by poultry and egg consumption (for animal life-years suffered and loss of morally adjusted animal lives) and by seafood consumption (for loss of animal lives)(Reference Scherer, Tomasik and Rueda19). The 3V rule, leading to the consumption of fewer animal products, but of higher quality animals raised in extensive conditions, is necessarily associated with improved animal wellbeing.
Limiting ultra-processed energy content to 15 % daily
The worldwide consumption of ultra-processed energy content is high, especially in western and Anglo-Saxon countries, often >40–50 % of daily energy content, and is growing rapidly in emerging countries(57,Reference Monteiro, Moubarac and Levy111) . In France, it accounts for at least 36 % of daily energy content(Reference Julia, Martinez and Alles112), even more when including the poorest who consume most of these products that are low in protective micronutrient density(Reference Maillot, Darmon and Vieux113). Based on the NOVA classification(Reference Moubarac, Parra and Cannon61), a decreased intake of energy content would imply more home cooking of bulk plant-based foods and returning to mildly/minimally processed foods such as pasta, plain yogurts, artisanal cheeses, breads and delicatessen, frozen or canned plant/animal-based products, dried fruits, nuts and seeds, whole fruits, muesli, butter, and so forth. As for animal energy content, a decreased intake of UPF for improving healthiness is recommended by international institutions based on evidence from prospective epidemiological studies(17,Reference Monteiro and Cannon38) . According to literature, a precautionary level of 15 % of daily energy content is proposed.
From epidemiological studies on obesity risk (Table 4), this 15 % threshold is only a proposition of a precautionary maximum threshold at which overweight/obesity risk was found to increase by approximately 29 % in two studies(Reference Louzada, Baraldi and Steele90,Reference Canhada, Luft and Giatti94) , awaiting other studies to be reported. Fifteen per cent of ultra-processed energy content correspond to approximately 1–2 servings per day; these foods might be generally more energy-dense than minimally processed foods due to added fats and sugars and lower water activity(Reference Fardet, Lakhssassi and Briffaz96). Disparity among the selected studies was observed, wherein overweight/obesity risks began to significantly increase from the higher daily UPF energy shares in USA and Canada (≥36 %) than in Brazil and Spain (≥14 %). However, in the Canadian(Reference Nardocci, Leclerc and Louzada92) and American(Reference Juul, Martinez-Steele and Parekh93) studies, the chosen reference quintiles were rather high for the energetic percentage of UPF (36·5 and 20·1 %), suggesting that significant effects may have been found with a lower reference quintile, as in Spain and Brazil where references were all <18 % daily energy content.
Besides obesity risk, other studies linked UPF with mortality risk(Reference Moreira, Baraldi and Moubarac114–Reference Rico-Campà, Martínez-González and Alvarez-Alvarez117). In a longitudinal study involving a restricted cubic spline analysis of the association between UPF and all-cause mortality, it was found that for each additional serving of UPF, all-cause mortality increased by 18 % (adjusted hazard ratio 1·18, 95 % CI 1·05, 1·33)(Reference Rico-Campà, Martínez-González and Alvarez-Alvarez117). In a recent study employing restricted cubic spline analyses, a dose–response analysis showed that replacing only 10 % UPF energy content (isoenergetic substitution) with either processed or un-/minimally processed foods might significantly decrease the risk of mortality by approximately 20 and 56 %, respectively(Reference Blanco-Rojo, Sandoval-Insausti and López-Garcia115). For CVD risk, a similar dose–response analysis showed that an increase of 15 % by weight of UPF in the diet increased the risk by approximately 26 % (P-value for non-linearity = 0·39)(Reference Srour, Fezeu and Kesse-Guyot118). In a Brazilian modelling study, the authors reduced the intake of saturated fat, trans-fat, salt and added sugar in ultra-processed foods by 75 %, substituting this reduction with a 75 % increase in unprocessed or minimally processed foods(Reference Moreira, Hyseni and Moubarac119). The results showed that such a scenario might lead to a reduction in CVD risk by 29 %.
The interconnectedness of the 3V rule
In the current study, three main dimensions were identified to holistically reach food system sustainability: Végétal (plant), Vrai (real) and Varié (varied), preferably organic, local and/or seasonal(Reference Fardet and Rock1). These three rules are clearly interconnected (Fig. 1), meaning that no one dimension can be excluded without impairing both human health and food system sustainability:
1. First, except for niche populations (e.g., the Inuits), too many animal energy content would threaten both human health(120,Reference Boutron-Ruault, Mesrine, Pierre and Mariotti121) and the environment(Reference Tilman and Clark122), also impacting animal welfare.
2. Second, replacing real foods with too much of UPF has been consistently shown to increase the risk of several chronic diseases(Reference Monteiro and Cannon38,Reference Fardet and Rock39) . Therefore, as suggested for CVD risk in the US adult population, it is not only sufficient to consume more plant-based foods; they must be minimally/normally processed(Reference Satija, Bhupathiraju and Spiegelman123). Actually, for each point increase in highly processed plant food serving, it has been shown that CHD risk similarly increased as for each point increase in animal food serving(Reference Satija, Bhupathiraju and Spiegelman123).
3. Finally, replacing the third rule ‘Varied’ by ‘Monotonous’ inevitably leads to micronutrient deficiencies, as observed in some developing countries, for example, sub-Saharan African and South Asian countries, where staple foods are generally not very varied, such as white glutinous rice in Asia or cereal variety in Africa, which may constitute up to >50 % of daily energy content(124). This rule also includes ‘local’ and ‘seasonal’ criteria to be conscious that eco-friendly, healthy foods produced in one country and transported into and consumed by another country may negatively impact environmental factors.
Limitations of the current study
Calculations used in the current study were based on French standard serving sizes. Therefore, they are not representative of what can be found worldwide, depending on countries and their traditional habits. On average, a French standard animal serving size is 95 g, corresponding to around 7·5 % of daily energy requirement for an adult (Table 1). The converging figure of 15 % daily animal energy content is therefore only indicative, and allows for some variability around it.
Otherwise, the simulated 3V-based diet is based on French culinary uses, questioning its extrapolation to other culinary traditions. However, the results clearly emphasise the importance of consuming less of UPF with ‘empty energy’ and more of nutrient-dense whole-grain breads, virgin plant oils and various vegetables, legumes and fruits to adequately address all nutritional needs. Food diversity (i.e., varied) is certainly the key point of this 3V rule-based diet.
The median 15 % daily UPF energy threshold is based on six epidemiological studies only, with four cross-sectional studies being of lower quality compared with longitudinal ones. Therefore, 15 % is only indicative. However, a cubic spline analysis in Canhada et al. (Reference Canhada, Luft and Giatti94) showed a linear relationship between UPF energy consumption and overweight/obesity risk within a range 0·5–30 % energy content, and the risk begins to increase below 15 %.
Conclusions and perspectives
The ‘3V rule’ diet is a generic diet with three fundamental dimensions: animal:plant food ratio, degree of processing and food variety. In most of the healthy diets analysed here, the second rule – ‘Vrai’ (‘real’) – was neglected, probably because the degree of food processing only received attention recently. However, such a diet would probably not be applicable to niche populations represented by more traditional ethnicities such as Masai, Inuit, Aborigines and other hunter-gatherers whose diets (generally close to a keto-Palaeolithic diet) had long beed adapted to their environment(Reference Fumagalli, Moltke and Grarup125,Reference Hockett and Haws126) . It can be noticed that these diets fit with at least ‘Vrai’ and – to a lesser extent – ‘Varié’ dimensions of the 3V rule, since their environment seems to be less favourable to developing food crops, while providing wild animals. Nevertheless, if these particular diets may have been conceived when the worldwide population was under one billion, they may no longer be sustainable today when applying such diets to nearly eight billion people by 2030, and probably 9–10 billion by 2050, many of whom may be living in very large cities. Additionally, the 3V diet is also compatible with a livestock polyculture system, implying animal and plant biodiversity and a circular bio-economy approach to improve the agricultural system.
As discussed previously(Reference Fardet and Rock1), the 3V rule is sufficiently generic to be regionalised according to specific climatic, environmental, traditional and culinary uses worldwide and, finally, be part of a holistic territorial food plan, involving several stakeholders from society, science and politics. Therefore, a generic 3V-based diet is not in opposition to actual traditional worldwide protective diets. This generic diet does not intend to substitute regional diversity due to different agronomic, climatic, cultural, socioeconomic and religious beliefs, but, rather, to serve as the basic generic rules for improving their sustainability at horizon 2050 while considering regional specificities and growing population worldwide. Notably, the animal:plant energetic ratio and the food diversity may be adapted to different regions worldwide.
Otherwise, this qualitative diet appears to be easy to adopt without an extensive knowledge of nutrition. This diet does not refer to any nutrient while indirectly naturally offering all nutritional needs, allowing the majority of the population to appropriate a healthy diet. Difficulties will probably arise in replacing animal foods with more whole-grain cereals, legumes, nuts and seeds, while limiting highly palatable UPF. A vigilant point of this diet is to consume sufficient virgin plant oils and varied foods.
An interesting interventional study could consist to progressively moving a significant number of individuals consuming an omnivorous western diet (diet rich in animal and ultra-processed energy content) to a 3V rule-based diet and to measure short-term physiological/metabolic parameters with a follow-up of chronic disease prevalence over several years. Another perspective would be to use the lever of the 3V-based diet to develop healthy and sustainable regional diets and to adapt each rule to traditional, climatic, agronomic and socioeconomic specificities through modelling of three-dimensional abacuses.
Acknowledgements: None. Financial support: The current research received no specific grant from any funding agency, commercial or not-for-profit sectors. Conflict of interest: A.F. is a member of the Siga Society since 2017, which is a start-up dedicated to the development of a food scoring tool based on the degree of processing. Authorship: A.F. led the study and defined the 3V rule based on scientific literature. A.F. wrote the manuscript. E.R. contributed to the interpretation of results and to the final writing of the manuscript. Ethics of human subject participation: None.