Nxac 150
Nxac 150
Nxac 150
Nutritional Epidemiology
1
MS-Nutrition, Marseille, France; 2 Université Clermont Auvergne, INRAE UNH, Clermont-Ferrand, France; 3 UMR INRAE-ACO
Pegase, St Gilles, France; and 4 MoISA, University of Montpellier, CIRAD, CIHEAM-IAMM, INRAE, Institut Agro, IRD, Montpellier,
France
ABSTRACT
Background: Shifting towards a more plant-based diet, as promoted in Western countries, will reduce the animal
protein contribution to total proteins. Such a reduction may not only impair protein adequacy, but also the adequacy in
other nutrients.
Objectives: We determined, for different adult subpopulations, the minimum total protein levels and the minimum
animal protein contributions to total proteins that are compatible with the fulfillment of all nonprotein nutrient-based
recommendations.
Methods: Mean nutritional contents and mean diet costs were estimated using a French, cross-sectional,
representative survey for 5 French subpopulations: 1) women < 50 y; 2) women 50–64 y; 3) women ≥ 65 y; 4) men < 65
y; and 5) men ≥ 65 y. For each subpopulation, linear programming optimization was used to assess the minimum protein
level (model set #1) and the minimum animal protein contribution to total proteins (model set #2) that are compatible
with the fulfillment of all nutrient-based recommendations (except proteins, for which levels were analyzed as outputs).
Total diet costs were not allowed to increase. Eating habits were considered in model set #2 only.
Results: The minimum amount of protein that was theoretically compatible with the fulfillment of nutrient-based
recommendations (model set #1) was below the minimum recommended protein intake for all subpopulations except
women < 50 y. In model set #2, for women and men ≥ 65 y, decreasing animal protein contributions to total proteins
below 55% and 60%, respectively, led to protein levels below recommended levels. For the other subpopulations
(women < 50 y, women 50–64 y, and men < 65 y), the lowest animal protein contributions to total proteins compatible
with a nutritionally adequate diet (including protein adequacy) were 55%, 50%, and 45%, respectively.
Conclusions: This study provides factual information about the animal protein contributions to total proteins compatible
with meeting all nutrient-based recommendations at no additional cost, and shows that they vary between 45% and
60% depending on the group of adults considered. J Nutr 2022;152:2514–2525.
Keywords: animal-to-plant protein ratio, nutritional adequacy, optimization, protein quality, affordability, diet cost,
France
C The Author(s) 2022. Published by Oxford University Press on behalf of the American Society for Nutrition. This is an Open Access article distributed under the
terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and
reproduction in any medium, provided the original work is properly cited.
Manuscript received April 6, 2022. Initial review completed June 13, 2022. Revision accepted July 1, 2022.
2514 First published online July 11, 2022; doi: https://doi.org/10.1093/jn/nxac150.
and less meat will inevitably reduce both total dietary proteins B6), minerals (magnesium, zinc), and water differ for men and
and the animal protein contributions to total proteins (4). women (Table 1). Then, for each sex, an age threshold of 65
Animal proteins have a more balanced indispensable amino acid y was used in order to take into account the differences in
profile and greater digestibility than plant proteins, although recommended levels of protein intake: 1.0 g of proteins/kg
plant proteins are commonly depicted as being of lower of body weight for adults aged 65 years old and older and
quality than animal proteins (5). Yet, in Western countries, 0.83 g of proteins/kg of body weight for younger adults (7).
where the quantities of protein consumed generally exceed The subpopulation of women was further divided according
minimum requirements, quality differences between proteins to iron needs, considering 50 y as the menopausal year (13).
from different food sources have negligible impacts on protein This led to 5 subpopulations, as follows: 1) women < 50
and amino acids adequacies (4, 6). In other words, for the years old; 2) women 50–64 years old; 3) women ≥ 65
majority of people in Western countries, who are typically years old; 4) men < 65 years old; and 5) men ≥ 65 years
omnivorous, the risk of inadequate protein intake is low, and old.
it is minimally influenced by the animal protein contribution
to the diet. Yet, caution is warranted for elderly people Dietary data and calculation of mean observed diets
because they have higher protein requirements than other adults Dietary data were derived from the Second French Individual
(estimated at 1 g/kg for adults over 65 y and 0.83 g/kg for and National Study on Food Consumption (INCA2), a 7-
younger adults) (7). Thus, for French adults aged 65 and over, day open-ended food-record representative survey conducted
a higher risk of frailty was observed for protein intakes lower between 2005 and 2007 by the French Agency for Food,
than 1 g/kg of body weight, independent of energy intake levels Environmental, and Occupational Health and Safety (14). For
(8). this study, all adults above the age of 18 y were studied,
Protein-source foods provide several nutrients other than leading to a total sample of 2624 individuals, distributed as
proteins and largely contribute to overall nutrient adequacy 922 women < 50 y, 418 women between 50 and 64 y, 197
(9, 10). Animal-based foods provide nutrients that are either women ≥ 65 y, 936 men < 65 y, and 151 men ≥ 65 y.
not found in plant-based sources (such as vitamins D and B12 The INCA2 survey was approved by the French National
and long-chain omega 3 fatty acids), found in small amounts Commission for Computed Data and Individual Freedom
(e.g., vitamin B6 and riboflavin), or found in less bioavailable (Commission Nationale de l’Informatique et des Libertés,
forms (e.g., iron and zinc), with the latter including unique CNIL).
sources of fiber, folate, vitamins E and C, and other antioxidants. For each of the 5 subpopulations, we estimated the average
In addition, diets with plenty of plant-based foods are not daily dietary intake, also called the observed diet below, based
necessarily more affordable than diets with more animal- on 212 frequently consumed food items (Supplemental Table 1),
based foods (11). More generally, food budget constraints following a previously described matching methodology (15).
are important determinants of food choices, and healthy Based on the French food composition table [CIQUAL 2013
food choices tend to be more expensive than unhealthy ones (16)], a sex-specific nutrient composition database containing
(12). energy, fiber, fats (including linoleic and alpha-linolenic acids;
The goal of this study was to determine the extents to EPA; DHA; and lauric, myristic, and palmitic acids, as well
which decreases in protein intakes, particularly from animal- as total SFA), proteins (including amino acids), carbohydrates,
based sources, could be made without impairing the nutritional 11 vitamins, and 11 minerals was derived for each of the 212
adequacy of the diet and at no additional cost. To reach food items (15). Phytate and amino acid contents in foods were
this goal, mathematical optimization models were developed derived from the International Network of Food Data Systems
to determine the theoretical minimum levels of total dietary databases (15). Average prices for the 212 food items were
proteins, as well as the minimum percentages of animal proteins calculated based on data from the 2006 Kantar Worldpanel (15,
in total proteins that are compatible with the fulfillment of all 17). After removal of alcoholic items from food consumption,
nutrient-based recommendations, without changing the total composition, and price databases, nutrient intakes and costs
energy content and at no additional cost, for 5 subpopulations associated with the consumption of 207 food items were
of French adults differing according to sex and age. estimated for each subpopulation, leading to 5 observed
diets.
Modeling
Material The models developed in this study used a mathematical
Populations of interest optimization technique based on the simplex algorithm. An
Five subpopulations of adults were defined according to nu- optimization model is comprised of variables, constraints, and
trient recommendation levels. First, segmentation was applied an objective function, and the algorithm finds the value that
based on sex because recommended levels of vitamins (A, E, each variable must have to comply with the constraints while
optimizing (i.e., minimizing or maximizing) the value given by
the objective function. In diet optimization models, the variables
MS-Nutrition and MoISA received financial support from the French National are the quantities of foods, and the algorithm finds the quantity
Interprofessional Association of Livestock and Meat (Interbev).
Author disclosures: The authors report no conflicts of interest.
of each food in the modeled diet that is compatible with the
Interbev had no role in the design, implementation, analysis, or interpretation of simultaneous fulfillment of all constraints for the minimum (or
the data. maximum) value of the objective function.
Supplemental Figure 1 and Supplemental Tables 1–4 are available from the In the present study, 2 kinds of models were developed,
“Supplementary data” link in the online posting of the article and from the same
called model set #1 and model set #2. Model set #1 diets
link in the online table of contents at http://jn.nutrition.org.
Address correspondence to FV (e-mail: [email protected]). were developed to determine the minimum protein contents
Abbreviations used: INCA2, Second French Individual and National Study on compatible with the fulfillment of all nutrient recommendations
Food Consumption; NRV, nutrient reference values. (without imposing a minimum amount of total proteins),
Animal protein intake and nutritional adequacy 2515
TABLE 1 Constraints applied to model set #1 and model set #2, for each subpopulation1
2516
Women Men
Subpopulation <50 y 50–64 y ≥65 y <65 y ≥65 y
Nutritional constraints (applied to model set #1 and model set #2)
Vieux et al.
Energy, kcal/d = Observed (1731.6) = Observed (1711.5) = Observed (1675.4) = Observed (2207.1) = Observed (2082.8)
Carbohydrates, % energy [40; 55] [40; 55] [40; 55] [40; 55] [40; 55]
Fats, % energy [35; 40] [35; 40] [35; 40] [35; 40] [35; 40]
Linoleic acids, % energy ≥4 ≥4 ≥4 ≥4 ≥4
Alpha-linolenic acids, % energy ≥1 ≥1 ≥1 ≥1 ≥1
DHA + EPA, mg/d ≥250 ≥250 ≥250 ≥250 ≥250
SFA, % of energy ≤12 ≤12 ≤12 ≤12 ≤12
Lauric + myristic + palmitic acids, % energy ≤8 ≤8 ≤8 ≤8 ≤8
Total sugars, without lactose, mg/d ≤100 ≤100 ≤100 ≤100 ≤100
Sodium, mg/d ≤ Observed (2613.0) ≤ Observed (2624.7) ≤ Observed (2689.1) ≤ Observed (3459.9) ≤ Observed (3575.5)
Sodium-to-potassium ratio, molar ≤1 ≤1 ≤1 ≤1 ≤1
Water, ml/d 2000 2000 2000 2500 2500
Fibers, g/d ≥30 ≥30 ≥30 ≥30 ≥30
Vitamin A, μg/d [650; 3000] [650; 3000] [650; 3000] [750; 3000] [750; 3000]
Thiamin, mg/MJ ≥0.14 ≥0.10 ≥0.10 ≥0.14 ≥0.10
Vitamin B12, μg/d ≥4 ≥4 ≥4 ≥4 ≥4
Riboflavin, mg/MJ ≥0.17 ≥0.17 ≥0.17 ≥0.17 ≥0.17
Niacin, mg NE/MJ;mg/d [1.6; 900] [1.6; 900] [1.6; 900] [1.6; 900] [1.6; 900]
Pantothenic acid, mg/d ≥4.7 ≥4.7 ≥4.7 ≥5.8 ≥5.8
Vitamin B6, mg/d [1.5; 25] [1.6; 25] [1.6; 25] [1.8; 25] [1.7; 25]
Folates, μg/d ≥330 ≥330 ≥330 ≥330 ≥330
Vitamin C, mg/d ≥110 ≥110 ≥110 ≥110 ≥110
Vitamin D, μg/d [5; 50] [5; 50] [5; 50] [5; 50] [5; 50]
Vitamin E, mg/d [9.9; 300] [9.9; 300] [9.9; 300] [10.5; 300] [10.5; 300]
Calcium, mg/d [960; 2500] [950; 2500] [950; 2500] [960; 2500] [950; 2500]
Copper, mg/d [1.0; 5.0] [1.3; 5.0] [1.3; 5.0] [1.3; 5.0] [1.6; 5.0]
Iron, mg/d ≥162 ≥11 ≥11 ≥11 ≥11
Iodine, μg/d [150; 600] [150; 600] [150; 600] [150; 600] [150; 600]
Magnesium, mg/d ≥360 ≥360 ≥360 ≥420 ≥420
Phosphorus, mg/d ≥700 ≥550 ≥550 ≥700 ≥550
Selenium, μg/d [70; 300] [70; 300] [70; 300] [70; 300] [70; 300]
Zinc, mg/d [7.5; 25] [7.5; 25] [7.5; 25] [9.4; 25] [9.4; 25]
≥0.0058 × phytates + 5.8 ≥0.0058 × phytates + 5.8 ≥0.0058 × phytates + 5.8 ≥0.0077 × phytates + 7.1 ≥0.0077 × phytates + 7.1
Nonnutritional constraints (applied to model set #1 and model set #2)
Diet cost, €/d ≤ Observed (5.16) ≤ Observed (5.77) ≤ Observed (5.52) ≤ Observed (6.33) ≤ Observed (6.30)
Fish, g/d <28.6 <28.6 <28.6 <28.6 <28.6
Quantity per food item,3 g/d ≤ 95th pctl ≤ 95th pctl ≤ 95th pctl ≤ 95th pctl ≤ 95th pctl
Fortified products and mineral waters,4 g/d ≤ Observed ≤ Observed ≤ Observed ≤ Observed ≤ Observed
(Continued)
minimum quantity of total proteins), without changing the total energy content, at no additional cost, and taking eating habits into account. Values in brackets are ranges e.g., in the case of fats, the recommended amount is between 35 and 40%
without changing the observed energy content and without
Fortified foods were defined as breakfast cereals (3 food items), multivitamin juice, soya-based drink, tomato soup, soya-based desserts (3 food items), cocoa powder, pineapple juice, a mixture of oils, instant drink, and low-fat margarine. Each
Model set #1 focused on determining the theoretical minimum level of total dietary proteins compatible with the fulfillment of all nutrient-based recommendations (without imposing a minimum quantity of total proteins), without changing the
Gradual decrease by 5%
exceeding the observed diet cost. Model set #2 diets were aimed
total energy content and at no additional cost. Model set #2 focused on determining the minimum percentage of animal proteins in total proteins compatible with the fulfillment of all nutrient-based recommendations (without imposing a
at determining the minimum percentages of animal proteins in
≥65 y
steps
total proteins compatible with the fulfillment of all nutrient-
based recommendations without changing the observed energy
content, without exceeding the diet cost, and taking eating
habits into account.
Men
Variables.
Variables were the same for the 2 kinds of models. All 207 food
items in the subpopulation-specific database were the variables
Gradual decrease by 5% for each model performed.
[5th pctl; 95th pctl]
[5th pctl; 95th pctl]
[5th pctl; 95th pctl]
<65 y
steps
Constraints.
Constraints applied to models are described in Table 1.
Nutritional constraints were applied to both model set #1
and model set #2. Each model included a subpopulation-
specific set of nutritional constraints imposing the fulfillment
of all nutrient-based recommendations (except proteins, for
which levels were analyzed as output). The energy contents
of modeled diets were kept equal to those of observed diets
Gradual decrease by 5%
Health and Safety reports (7, 13) available at the start of the
food item was constrained to remain below its observed amount. Mineral waters (8 food items) were constrained in the same way to favor tap water.
The 5th percentile and 95th percentile were estimated for all individuals, including nonconsumers of the food group.
Quantities < 95th percentile estimated only for consumers of the food item.
steps
where the diet cost was kept below or equal to the observed diet
cost was applied to all models, in order to consider the realism
Under the assumption of high iron losses through menstrual blood.
for each subpopulation. The first model of the set did not
include any constraint on animal protein contributions to total
proteins. Then, starting from the percentage of animal proteins
in total proteins reached in the first model, that percentage was
1
2
3
4
In the observed diets, regardless of the subpopulation, the (despite no constraints applied to them), with animal protein
protein contributions to total energy were between 16% and contributions to total proteins as low as 40% (men < 65
17%. In contrast, in the model set #1 diets, the protein y), 45% (women 50–64 y), and 50% (women < 50 y). The
contributions to total energy dropped to between 12.5% (in cost constraint was binding in all the modeled diets (data
women < 50 y old) and 8.8% (in men < 65 y old). In not shown). All diets with total protein contents higher than
the observed diets, the animal protein contributions to total the recommended levels were qualitatively and quantitatively
proteins were around 68% to 72%, and they decreased to adequate in terms of their amino acid contents, so that full
percentages between 46.3% (in men < 65 years old) and 58.8% protein adequacy was ensured (Supplemental Table 4).
(in women ≥ 65 years old) in model set #1 diets. Supplemental Figure 1 shows the departures from the
Supplemental Table 3 shows nutrient contents, costs, and observed diets (i.e., the value, D, of the objective function; that
fish contents in observed diets and model set #1 diets for the is, the sum of the absolute differences between the quantity of
5 subpopulations. Some values were exactly equal to values im- each food in the observed diet and the corresponding modeled
posed by their respective (minimum or maximum) constraints. diet), as well as the numbers of remaining food items, induced
Such constraints are named “binding” or “active” constraints. by all model set #2 diets. Depending on the subpopulations,
Identifying them helps to point out which constraints are more departures between 1.0 kg/d and 1.8 kg/d were needed to obtain
difficult to fulfil than others. Binding constraints are also the the model set #2-NO diets. Then, a progressively decreasing
ones with the greatest influence on the kinds and amounts animal protein contribution to total proteins increased the
of food introduced in modeled diets (i.e., the selections or departure from the observed diet and decreased the number of
“food choices” made by the models). Constraints to energy foods in the modeled diet. The subpopulation’s specificities were
(equality), water (equality), SFA (maximum), fiber (minimum), observed as follows:
zinc (minimum), and vitamin D (minimum) were binding for
all 5 subpopulations, as were constraints to the diet cost 1) For women < 50 y, imposing a 50% animal protein
(maximum) and fish content (maximum). Minimum constraints contribution to total proteins induced a departure (D)
to alpha-linolenic acids and to calcium were binding for almost from the observed diet that was higher than the total mass
all subpopulations. Many other constraints were binding for of the observed diet.
some but not all subpopulations. Thus, the minimum constraint 2) For women 50–64 y, imposing a 45% animal protein
to iodine was binding in all diets modeled for women, but not contribution to total proteins induced a departure (D)
for men. Conversely, constraints to total sugars (maximum) and from the observed diet (2.8 kg/d) that was higher than
magnesium (minimum) were binding in all diets modeled for the total mass of the observed diet (2.5 kg/d).
men, but not for women. The constraint to iron was binding 3) For women ≥ 65 y, the modeled diet with a 50% animal
only in the subpopulation of women < 50 years old. protein contribution to total proteins did not reach the
Figure 1 shows the protein contents of observed diets and recommended protein level.
diets obtained with model set #2. In an initial step, and 4) For men < 65 y, the departure from the observed diet
compared to observed diets, fulfilling all the applied constraints was never higher than the total mass of the observed diet,
increased the total protein content for all subpopulations, and the protein contents in modeled diets were always
whereas no constraint was applied to proteins (model set #2- higher than the recommended level. However, the number
NO diets). Then, imposing a decrease in the animal protein of food items dramatically decreased (−17%) between
contribution to total proteins (other diets in model set #2) the modeled diets with 45% and 40% animal protein
induced a progressive decrease in the total protein content. contributions to total proteins.
In most modeled diets, as shown by the bars exceeding the 5) For men ≥ 65 y, the modeled diet with a 55% animal
horizontal black line, the recommended level of protein intake protein contribution to total proteins did not reach the
was achieved despite the absence of a constraint applied to recommended protein level (squares).
it. However, for older subpopulations, the modeled diets did Based on these results and the criteria described above to
not contain the recommended protein levels when the animal select 1 model set #2 diet per subpopulation, modeled diets
protein contributions to total proteins were ≤55% for men ≥ 65 with the following percentages of animal proteins in total
y and ≤50% for women ≥ 65 y. In other subpopulations, proteins were selected for further analysis (Figure 2): 55% for
the modeled diets fulfilled all nutritional constraints, and their women < 50 y, 50% for women 50–64 y, 55% for women ≥ 65
protein contents were higher than the recommended levels y, 45% for men < 65 y, and 60% for men ≥ 65 y.
Figure 3 shows the dietary shifts in food groups induced by diets, as did unrefined starchy foods, milk, and eggs (except
each selected model set #2 diet, and Table 3 displays the dietary the latter in men < 65 y). In comparison, intakes of
shifts induced by the selected models in more detail. “Water refined starchy foods, breakfast cereals (except in men < 65
and drinks” and “plant-based alternatives” were removed from y), yogurt (except in men ≥ 65 y), cheese, meat and
the graph for visualization purposes because they presented deli meat, plant-based dishes, biscuits and sugar, desserts,
quantities that were too high and too low, respectively, but animal fats (except in men ≥ 65 y), and spices and
were included in Table 3. In all selected modeled diets, fruits sauces (except in women ≥ 65 y) decreased. Fish intakes
and vegetables, dairy products, and starchy foods increased attained the maximum applied amount (200 g per week).
(except the latter in men ≥ 65 y), while other food groups Tea and coffee intakes remained constant. Other subgroups
decreased. from the water and drinks group had decreased intakes
All subgroups within the fruit and vegetable food group for women and remained constant for men. For other
had increased intakes between the observed and modeled food subgroups (animal-based dishes, cakes and tarts, and
2521
to their low average body weight), the strict minimum was
even higher than the recommended protein level. Numerous
nutritional constraints were binding in model set #1, showing
that foods that are a source of protein were needed for several
other nutrients. Achieving the recommended levels was difficult
for vitamin D, zinc, fiber, alpha-linolenic acid, and SFA in all
subpopulations and for iodine, calcium, magnesium, and total
sugars in some subpopulations. Regarding iron, the constraint
was binding only for young women. In that subpopulation, the
level of the iron constraint (i.e., 16 mg/d) imposed corresponded
to a hypothesis of high iron losses in menstrual blood. Note that
alternative models (data not shown) were conducted where only
11 mg/d of iron (recommended iron intake for young women
with normal iron losses) was required, and the minimized total
protein contents did not change (54 g).
Despite the absence of any constraint related to amino
acid contents in diets, indispensable amino acids were sys-
FIGURE 2 Animal and plant protein contributions to total proteins
(in percentages) in the model set #2 selected diets (i.e., modeled tematically adequate in all modeled diets fulfilling nutrient
diets with the lowest animal contributions to total proteins that recommendations, including the recommendation for total
are compatible with nutrient and protein adequacy, affordability, and proteins, whether their animal protein contribution to total
eating habits) for each subpopulation. The models imposed the proteins was decreased or not. This means that other nutrient
fulfillment of all nutrient-based recommendations except for the deficiencies would occur before amino acid deficiency would
recommendation for proteins, while minimizing the departure from become a problem. Accordingly, previous studies showed that
the observed diet. protein adequacy is slightly or not influenced by the amino
acid distributions in foods that are a source of protein (4), and
is more significantly related to protein quantity than quality,
vegetable fats), the directions of variation differed between except in diets containing high plant protein contributions to
subpopulations. total proteins (around 70%) from minimally diversified sources,
such as refined grains (20).
Foods that were sources of both animal and plant proteins
were needed to cover nutrient requirements, even when proteins
Discussion
were minimized. Shifting towards more plant-based diets in
It is known that foods that are sources of protein contain Western countries is often recommended by the scientific
several nutrients other than protein but, to the best of our community and public stakeholders for health and climate
knowledge, this study is the first that assessed to what purposes (1,21–23), but there is no consensus regarding the
extent total proteins and animal protein contributions to total adequate ratio of animal protein to plant protein in sustainable
proteins could be theoretically reduced without impairing diets. A 1:1 ratio (1 g of plant protein for 1 g of animal protein)
the fulfillment of all other nutrient-based recommendations, is often presented as a nutritional standard but, to the best of
excluding the use of nutritional supplements or fortified foods. our knowledge, it is not recommended by any official order. In
A mathematical optimization approach was used to model this study, models decreasing the animal protein contributions
population diets while simultaneously fulfilling a set of nutrient- to total proteins by 5% steps (model set #2) showed that animal
based constraints (not including the constraint on proteins), protein contributions of 60% (men ≥ 65 y), 55% (women < 50
without changing dietary energy and at no additional cost. The y and ≥ 65 y), 50% (women 50–64 y), and 45% (men < 50
results showed that in highly theoretical models where total y) would be fully compatible with adequate nutrients and
proteins were directly minimized without considering eating proteins, affordability, and eating habits, despite their relatively
habits, a strict minimum of at least 48 g/d of total proteins was low protein contents. Our results therefore imply that moving
needed to meet nutrient-based recommendations for nutrients towards diets with lower animal protein contributions or to
other than proteins. Models better at taking eating habits into fully vegan diets, like the ones included within the range
account showed that the animal protein contributions to total of diets recommended by the EAT–Lancet Commission (21),
proteins, which were approximately 70% in observed diets, would necessarily require food fortification and/or nutrient
could be reduced to contributions that were between 45% and supplementation to cover adult nutritional requirements. The
60%, depending on age and sex, while still being compatible minimum percentages of animal proteins found in the present
with complete nutritional adequacy and affordability. Lower study to be compatible with nutrient adequacy in the different
percentages would either be mathematically unattainable or subpopulations considered can be compared to those obtained
constraints would have to be relaxed or removed, therefore in 2 recent studies focused on designing sustainable diets: 1
impairing nutritional adequacy and/or realism in the resulting for French adults from the NutriNet-Santé cohort (24) and
modeled diets. 1 for older Dutch adults from the Longitudinal Aging Study
One important finding from this study is that in the absence Amsterdam cohort (25). In apparent contradiction to our
of any constraint imposing a given protein level, a strict results, for the French study, the mathematical optimization
minimum protein amount was needed to cover other nutrient- approach used to derive nutritionally adequate individual diets
based recommendations in adults (model set #1). This quantity with progressively lower environmental impacts, while also
was close to the minimum recommended protein intake level controlling for costs, resulted in diets with an animal protein
for each subpopulation. For the subpopulation of women < 50 contribution to total protein as low as 22% for the most so-
y (where the recommended protein level was relatively low due called “disruptive” scenario (24). Such a discrepancy can be
2522 Vieux et al.
FIGURE 3 Observed (x-axis) and modeled (y-axis) quantities from selected MOD2 diets of each food group for each subpopulation. For (A)
women < 50 y and (C) women ≥ 65 y, the MOD2 selected diets contained animal proteins as 55% of total proteins; for (B) women 50–64 y, the
MOD2 selected diet contained animal proteins as 50% of total proteins; for (D) men < 65 y, the MOD2 selected diet contained animal proteins
as 45% of total proteins; and for (E) men ≥ 65 y, the MOD2 selected diet contained animal proteins as 60% of total proteins. Abbreviation:
MOD2, model set #2.