Content uploaded by Anthony Fardet
Author content
All content in this area was uploaded by Anthony Fardet on Sep 03, 2022
Content may be subject to copyright.
Food &
Function
PAPER
Cite this: DOI: 10.1039/c9fo02271f
Received 28th September 2019,
Accepted 28th January 2020
DOI: 10.1039/c9fo02271f
rsc.li/food-function
The holistico-reductionist Siga classification
according to the degree of food processing:
an evaluation of ultra-processed foods in French
supermarkets
Sylvie Davidou, * Aris Christodoulou, Anthony Fardet and Kelly Frank
The qualitative NOVA classification of foods according to their degree of processing is used worldwide by
researchers. NOVA defines ultra-processed foods (UPFs) by the presence of processed industrial ingredi-
ents and additives to modify the sensory properties (aroma, taste, colour and texture) of reconstituted
food, named ‘cosmetic’compounds, i.e., modifying food appearance. Some drastic processes directly
applied to food are also markers of ultra-processing. However, with the intent to develop an elaborate
tool for industries and retailers, the Siga classification was developed by combining the four holistic NOVA
groups with four more new reductionist subgroups considering the impact of processing on the food/
ingredient matrix; the contents of added salt, sugar and fat; the nature and number of markers of ultra-
processing (MUPs); and the levels of at-risk additives (groups are unprocessed, A0; minimally processed
foods, A1; culinary ingredients, A2; balanced foods, B1/C0.1; high salt, sugar and/or fat level foods, B2/
C0.2; processed/ultra-processed foods; and UPFs with more than one MUP, C1). The Siga algorithm was
used to characterize 24 932 packaged foods in French supermarkets (baby foods and alcohol excluded),
which were representative of the packaged food assortments. The main results showed that two-thirds of
the products were ultra-processed. Products with more than one MUP (C1) corresponded to the most
represented category, accounting for 54% of the products. Among foods with more than five ingredients,
75% were UPFs. Considering all products, the average number of ingredients, MUPs and at-risk additives
were 10.1, 2.6 and 0.5, respectively. Among food categories, some contained a high percentage of UPFs:
94, 95, 95, 81, 80, and 87% for salted meats, cooked dishes, flavoured yogurts/white cheeses, energy and
gourmet bars, breakfast cereals, and vegetarian dishes, respectively. Finally, the Siga algorithm presents a
useful tool for improving the health potential of packaged food and for decision-making on search
engine optimization (SEO) policy and assortment management in supermarkets.
Introduction
The health potential of foods is multiaxial, including “matrix”
and “composition”effects and in some cases, the presence of
processed ingredients and additives.
1
The food matrix results
from nutrient interactions, and can be characterized by
texture, chewiness, color, and other qualitative attributes.
1
While the “matrix”effect may impact the degree of chewing,
satiety, transit time and/or nutrient synergy, the “composition”
effect is the direct metabolic impact of food nutrients.
Most foods are processed in one way or another, and tech-
nological treatments as well as the addition of sugars, salt, fats
and additives impact both the food matrix and composition.
Today, the degree of food processing is increasingly
considered in epidemiological studies, especially since the
elaboration of the NOVA classification by Brazilian nutritional
epidemiologists.
2–4
NOVA ranks foods into four technological
groups: (1) un/minimally processed foods, (2) culinary ingredi-
ents, (3) processed foods, and (4) ultra-processed foods
(UPFs).
5
Beyond NOVA, the degree of processing rarely appears
in epidemiological studies and can be found mainly in the
form of binary comparisons such as red versus processed
meats,
6
fresh fruits versus fruit juices,
7
fruit juices versus swee-
tened beverages,
8
milk versus yogurts versus cheeses,
9
and
wholegrain versus refined cereals.
10
Except for the comparison
of milk versus yogurts versus cheeses, these previous studies
tend to show that more processed foods are less protective
against chronic diseases than minimally processed foods.
11–13
Based on the NOVA classification, new results have
notably underlined that populations consuming the most
Siga, 5 Avenue du Général De Gaulle, 94160 Saint-Mandé, France.
E-mail: sylvie@siga.care
This journal is © The Royal Society of Chemistry 2020 Food Funct.
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
View Journal
UPFs are the most at risk of chronic diseases and metabolic
deregulations, i.e., overweight/obesity,
14–16
adiposity,
17
meta-
bolic syndrome,
18,19
dyslipidaemia,
20
type 2 diabetes,
8
hyper-
tension,
21
total cancers and breast cancer,
22
cardiovascular
diseases,
23
irritable bowel syndrome and functional dyspep-
sia,
24
depressive symptoms,
25,26
and mortality.
27,28
Based on
NOVA, UPFs may be defined as food formulations character-
ized by the addition of cosmetic ingredients/additives primar-
ily for industrial use to mimic, mask, increase or restore
sensory properties (texture, taste, aroma and colour).
5
NOVA is now used worldwide, especially by academic
researchers, and is recognized as a relevant nutritional indi-
cator by the Food and Agriculture Organization of the United
Nations (FAO), the Pan American Health Organization (PAHO)
and the Brazilian dietary guidelines.
4,29–31
NOVA can be con-
sidered a qualitative and holistic classification in the same way
that each food group gathers heterogeneous foods of various
compositions, including nutrients, ingredients and/or addi-
tives. However, the holistic concept of UPF encompasses well-
identified common attributes such as the degradation of the
food matrix;
32
high levels of added sugars, fat and salt together
with low levels of protective micronutrients such as fibre, vita-
mins, minerals and antioxidants (i.e.,“empty”calories);
31–38
poor satiety and hyperglycaemia;
39–41
and the inclusion of arti-
ficial or xenobiotic compounds (e.g., aromas, newly formed
compounds and highly processed additives and ingredients).
32
To help consumers make the best food choice, several
scores have been developed, but all are based on a reductionist
approach considering foods as only sums of nutrients, e.g.,
SENS, Nutri-score, Nutrient Rich Food index, “traffic light”,
“health star rating”, modified reference intake indices and
other nutritional warnings.
42–44
These indices or scores
provide an overview of the overall nutritional balance of pack-
aged foods but fail to identify all unhealthy foods that can be
well rated nutritionally, although they are highly processed
and contain many additives (e.g., some light/vegan/gluten-free/
organic foods that are ultra-processed) or hidden sugars (e.g.,
dextrose, maltodextrins, invert sugar, etc.), and have lost their
“matrix”effect through fractionation (‘artificial’food matrices
resulting from the recombination of ingredients and additives),
or extrusion-cooking (e.g., extruded-cooked breakfast cereals for
children). In addition, we eat complex food matrices, not nutri-
ents, and two foods with the same nutritional composition, but
different matrices may have different physiological and meta-
bolic effects in the short term and health effects in the long
term,
1
e.g.,slowversus rapid sugars.
45
In a previous paper, the interconnectedness of holistic and
reductionist thinking was analysed, concluding that scientific
issues should be first considered holistically and then in a
reductionist way when necessary to dissect a particular mecha-
nism.
1
Applied to food classifications according to the degree
of processing, each holistic NOVA food group may be further
dissected in a more reductionist way, notably for different pur-
poses in destination to retailers and agro-food industries to
improve the quality of their assortments and products,
respectively.
46,47
Indeed, UPFs may exhibit from one to beyond
ten markers of ultra-processing (MUPs); processed foods may
contain different levels of added culinary ingredients, namely,
fat, sugar and salt, and minimally processed foods, and
depending on the processes applied, may exhibit different
degrees of food matrix unstructuration, e.g., through grinding,
pressing, and/or skimming. Such distinctions may be useful
for the practical applications of the NOVA classification.
Our main objectives are therefore: (1) to take the NOVA
classification a step further through a holistico-reductionist
approach, taking into consideration the “matrix”effect (which
plays a substantial role in chewing, satiety, synergistic actions
of nutrients, transit time and nutrient bioavailability); the
quantity of salt, sugar and fat added in recipes; the degree of
processing of industrial/culinary food ingredients and addi-
tives; and the function, number, and potential health risk of
additives; (2) to rationalize the characterization of ultra-proces-
sing and the definition of UPFs; and (3) to propose to small
and large retailers and the agro-food industry a holistico-
reductionist score to highlight the health potential of foods in
line with the most recent scientific knowledge. Such a classifi-
cation is intended to help develop less processed foods but in
a gradual way. Finally, 24 932 packaged foods were evaluated
with the NOVA-based Siga classification.
Materials and methods
The Siga ‘philosophical’approach and basis
The Siga classification (“go forward”in Portuguese, meaning
“go further”,“improve the existent”) is based on the combi-
nation of holistic and reductionist approaches (Fig. 1). Holism
is defined globally by the school of thought that tends to
explain a phenomenon as being an indivisible whole: the
simple sum of its parts does not suffice to define it. As a
result, holist thinking is in opposition –or rather complemen-
tary –to the reductionist thought that tends to explain a
phenomenon by dividing it into parts. As shown in Fig. 1,
holism and reductionism coexist, but reductionist thinking or
studies should always be considered secondarily to holistic
thinking to be meaningful and should be included within a
holistic approach as a first step.
1
This idea applies to food
classification: the main holistic food gateway is the degree of
processing, followed by more specific and reductionist criteria
but not the other way around, e.g., beginning to study food’s
health potential by its parts –,i.e., nutrients, additives, and/or
other unknown xenobiotics and thereafter to move towards a
more holistic approach.
47
In the Siga classification, NOVA was considered as a holistic
classification and was further dissected into more reductionist
subgroups considering the matrix effects of the un/minimally
processed NOVA group, the degree of processing and un-struc-
turation of ingredients and additives, and added salt, sugar
and fat contents for the NOVA processed and ultra-processed
groups. Therefore, the Siga classification brings finer granular-
ity to NOVA groups. Siga has to be replicable, accurate, and
scientifically valid. Siga is also intended to be based on legal
Paper Food & Function
Food Func t. This journal is © The Royal Society of Chemistry 2020
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
information available for everyone, i.e., the list of ingredients
and the nutritional table on food packaging. In that way, the
main assumption is that “the degree of processing of the
ingredients that make up the whole complex food accounts for
its overall degree of processing”.
NOVA food groups
NOVA food groups have been recently defined as follows:
5
(1) “Minimally processed foods, that together with unpro-
cessed foods make up NOVA group 1, are unprocessed foods
altered by industrial processes such as the removal of inedible
or unwanted parts, drying, crushing, grinding, fractioning,
roasting, boiling, pasteurization, refrigeration, freezing,
placing in containers, vacuum packaging or nonalcoholic
fermentation”.
(2) “NOVA group 2 is composed of processed culinary ingre-
dients. These are substances obtained directly from group 1
foods or from nature, such as oils and fats, sugar and salt.
They are created by industrial processes such as pressing, cen-
trifuging, refining, extracting or mining, and their use is in the
preparation, seasoning and cooking of group 1 foods”.
(3) “NOVA group 3 is composed of processed foods. These
are industrial products made by adding salt, sugar or other
substances found in group 2 to group 1 foods; using preser-
vation methods such as canning and bottling; and, in the case
of breads and cheeses, using nonalcoholic fermentation. Food
processing here aims to increase the durability of group 1
foods and make them more enjoyable by modifying or enhan-
cing their sensory qualities”.
(4) “Ultra-processed foods (group 4) are formulations of
ingredients, mostly of exclusive industrial use, that result from
a series of industrial processes (hence ‘ultra-processed’)…
Ingredients that are characteristic of ultra-processed foods can
be divided into food substances of no or rare culinary use and
classes of additives whose function is to make the final
product palatable or often hyper-palatable (‘cosmetic
additives’)”.
The NOVA UPF definition has been shown to be useful for
academic studies, but it is not sufficiently specific for interfa-
cing with the industry or retailers. Therefore, there is a need
for a more specific approach to characterize ultra-processing,
based on the analysis of the ingredients and technological pro-
cesses leading to their production.
Developing an incremental characterization of UPFs
Among several food classifications according to the degree of
processing, NOVA is “the most specific, coherent, clear, com-
prehensive and workable”.
1
Within the context of the NOVA
food classification, a UPF definition was proposed. However,
the description of UPFs by NOVA is qualitative, notably lacking
an exhaustive list of the characteristic ingredients/additives of
UPFs.
5
To characterize UPFs, the Siga classification therefore is
intended to address this issue by defining a specific and
exhaustive list of MUPs.
In NOVA, several characteristic ingredients/additives of
UPFs are already mentioned: (1) food substances of no or rare
culinary use (…), ingredients originating from the fractionation
of whole food and those that have been subjected to hydro-
Fig. 1 The interconnectedness between holistic and reductionist approaches applied to the Siga classification (adapted from Fardet & Rock,
1
with
permission of ASN©).
Food & Function Paper
This journal is © The Royal Society of Chemistry 2020 Food Funct.
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
lysis, hydrogenation, or other chemical modifications, such as
hydrogenated or interesterified oils, hydrolysed proteins,
soy protein isolate, maltodextrin, invert sugar and high fruc-
tose corn syrup; and (2) cosmetic additives such as dyes and
other colours, colour stabilizers, flavours, flavour enhancers,
non-sugar sweeteners, and processing aids such as carbona-
tion; firming, bulking, anti-bulking, de-foaming, anti-
caking and glazing agents; emulsifiers; sequestrants; and
humectants.
2,48
Similar to foods, ingredients undergo different degrees of
processing, and the degree of transformation of an ingredient
accounts for its health potential. In Siga, a MUP must have
undergone highly deteriorating processes, such as chemical
extraction, purification, chemical synthesis, and hydrolysis,
leading to the un-structuration of the initial ingredient matrix
and the subsequent loss of protective bioactive compounds.
For example, concerning sugar-based ingredients, honey is
less processed and is healthier than table sugar,
49
and
glucose-fructose syrup is more processed than table sugar with
different potential metabolic effects,
50
notably leading to fatty
liver and liver triglyceride accumulation.
51,52
Similarly, table
sugars may exhibit different degrees of refinement from com-
plete/whole sugar to refined white sugar. The same is true for
other industrial sugary ingredients, such as glucose syrup,
invert sugar, dextrose, isolated fructose and maltodextrins,
which are all ultra-processed sugars. Concerning fats, hydro-
genated fats are more processed than refined oils, which are
more processed than virgin oils. For proteins, we can also
observe these different degrees of processing, e.g., protein
hydrolysates are more processed than protein isolates. The
idea behind this criterion is the loss of the ingredient matrix
effect. For example, at equal calorie and glucose unit compo-
sitions, isolated starch, modified starch, maltodextrin, glucose
syrup or dextrose do not necessarily have the same glycaemic
response
53
because the starch matrix has been unstructured
through several steps of hydrolysis.
Siga decrypts the degree of transformation for each ingredi-
ent listed on the ‘on pack’list based on regulatory texts, tech-
nical specifications and business practices. What matters is
not only the state of un-structuration of the ingredient but also
the result in terms of the depletion of protective bioactive com-
pounds in the final ingredient. Therefore, among ingredients,
Siga distinguishes non-MUPs (minimally processed ingredi-
ents) and MUPs.
Non-MUPs are raw materials or ingredients that have not
been highly modified from the initial matrix, but they can be
altered by processes such as drying, crushing, grinding,
cooking, boiling, pasteurization, refrigeration, freezing, nonal-
coholic fermentation, roasting (i.e., exposing a food to direct
fire or to a suitable heat source, giving an aroma that recalls
the smell of slightly grilled, charred foods), infusion, etc.
These ingredients are thus not devoid of protective bioactive
compounds, even though their food matrix might have been
altered (e.g., juices, purées, powders). These ingredients have
some form of complexity in the nature of their constituent
interactions.
MUPs are deliberately added substances obtained by syn-
thesis or by a succession of physical, chemical, biological pro-
cesses leading to their purification and/or high deterioration
compared to the original material.
Genesis of the Siga classification
Including the “matrix”effect in minimally processed foods
(NOVA group 1). Scientific evidence converges to recommend a
generic and worldwide protective diet that obeys the 3 V rule:
Végétal (plant foods), Vrai (real foods), and Varié (varied
foods).
1,54
Indeed, epidemiological studies have shown that
protective diets worldwide for the most part obey the 3 V rule
and contain significant proportions of un/minimally processed
foods, e.g., the Mediterranean (high consumption of fruit,
both fresh and dried such as figs and raisins; nuts; vegetables;
whole-cereal products; fish; and seafood), Okinawa (high in
vegetables and legumes, low-GI grains, few fish and lean
meats, low in calories, and nutritionally dense, especially with
regard to phytonutrients in the form of antioxidants and flavo-
noids), Prudent (a higher intake of vegetables, fruits, legumes,
wholegrains, fish, and poultry, whereas the “western diet”is
characterized by a higher intake of red meat, processed meat,
refined grains, sweets, desserts), Baltic/Nordic (high consump-
tion of fresh fruit and vegetables, whole-cereal products, and
minimally processed animal products) or vegetarian (high in
plant-based foods) diets.
55–59
In addition, all national dietary
guidelines worldwide, based on collective and scientific exper-
tise, also recommend favouring raw and/or not overly pro-
cessed foods, with highly processed foods generally being at
the top of food pyramids.
However, raw foods, even if minimally processed, may have
their matrix altered by grinding, peeling, cooking, cutting,
refining and/or pressing, leading to different metabolic effects
as shown in apples,
60
almonds
61
and carrots.
62
The “matrix”effect can be simply defined as follows: “a
food with the same composition but different matrices, e.g.,
ground versus whole almond, does not have the same meta-
bolic effect or health effect in the long term”.
63
The food
matrix results from nutrient interactions and constitutes the
three-dimensional architecture of the food. A matrix can be
solid, semi-solid or liquid, with different impacts on nutrient
bioavailability and satiety potential.
64
In the NOVA group 1 of
un/minimally processed foods, mechanical, fermentation and
thermal treatments influence the matrix. For example, grind-
ing decreases the particle size and, hence, chewing time and
the secretion of satiety hormones; thermal treatment more or
less gelatinizes starch, such as in cooked pasta, and pre-fer-
mentation may radically change the food structure.
Based on scientific evidence related to food “matrix”
effects,
64
NOVA group 1 has been divided into unprocessed
and minimally processed foods (Siga groups A0 and A1–A2,
respectively; Fig. 2). Unprocessed foods are raw foods that have
not undergone any technological processes: they are raw milk,
fruits, vegetables, nuts, meat, and eggs. Minimally processed
foods encompass cooked, ground, fermented, and refined
Paper Food & Function
Food Func t. This journal is © The Royal Society of Chemistry 2020
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
grains, such as cooked grains, fruits, vegetables, meat, semi-
skimmed pasteurized milk, plain yogurt, 100% fruit juices, etc.
Processed culinary ingredients from the NOVA group 2,
when not too processed, have been included within the Siga
group “minimally processed ingredients”group (A2), e.g.,
butter, virgin vegetable oils, salt, table sugars, and honey;
however, refined and/or plant-based oils modified through
hydrogenation are considered ultra-processed ingredients and
therefore MUPs.
Impact of sugar, salt and fat contents in processed and
ultra-processed foods (NOVA groups 2–4). In NOVA groups 3
and 4 of processed foods and UPFs, the amount of added
culinary ingredients is not taken into account: it does not
amount to the same to add one or three servings of sugar to a
plain yogurt or one to three tablespoons of oil to a crudités
salad. In this case, the overly holistic and qualitative approach
does not take into account the nutritional quality, especially
the salt, sugar and fat contents, even if these must be con-
sidered secondarily and not primarily. The Food Standard
Agency (FSA) medium nutritional thresholds of 1.5 g salt per
100 g, 12.5 g sugars per 100 g and 17.5 g fat per 100 g for
foods and of 0.75 g salt per 100 g, 6.25 g sugars per 100 g and
8.75 g fat per 100 g for beverages
65
have been used to create
four Siga subgroups, i.e., nutritionally balanced (below FSA
thresholds) and high level salt/sugar and/or fat (above FSA
thresholds) processed foods and nutritionally balanced and
high level salt/sugar and/or fat UPFs (groups B1/B2 and C0.1/
C0.2, respectively; Fig. 2). Concerning the sugar threshold, it
was 22.5 g per 100 g in the most recent evaluation by the FSA;
however, we chose 12.5 g per 100 g, which appeared to us to be
more adequate and demanding, especially with the WHO rec-
ommendation of not exceeding 10% –ideally 5% –of daily cal-
ories from simple carbohydrates from added sugars, honey
and fruit juices.
The additives: health risk potential. The risk assessment of
additives is complementary to the evaluation of the degree of
transformation. Considering all the additive variants, there are
currently 375 authorized additives at the European level;
66
some additives have been noted to present a risk for health
when regularly consumed, e.g., phosphate additives
67
and
sodium nitrite.
68
Therefore, Siga developed a methodology to
evaluate additives with regard to health and based on
European Food Safety Authority (EFSA) reports, and additional
French ANSES opinions, and newly published studies since
the last EFSA reports. In the end, 240 additives were evaluated,
and the remaining additives had no EFSA evaluation available.
By default, they have been classified as “not at risk”.
Siga relies on the definition of the precautionary principle
of European food legislation (Regulation (EC) No. 178/2002,
Article 7), which states that: “In specific circumstances where, fol-
lowing an assessment of available information, the possibility of
harmful effects on health is identified but scientific uncertainty
persists, provisional risk management measures necessary to
ensure the high level of health protection chosen in the
Community may be adopted, pending further scientific infor-
mation for a more comprehensive risk assessment”.
69
In this
Fig. 2 From NOVA to Siga classifications according to the degree and extent of food processing.
Food & Function Paper
This journal is © The Royal Society of Chemistry 2020 Food Funct.
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
sense, decision-making is temporary and likely to evolve in
light of scientific advances and new published studies. The
evaluation methodology is based on two main criteria
(Fig. 3a):
- The general conclusion of the risk panel (no additive can
be assessed as being at risk if the panel formally concludes
that there is no safety concern);
- The adverse (certain or uncertain) effects observed in
humans and animals or in vitro (no additive can be assessed
as being at risk if there are no relevant adverse effects or no
observed effects) related to the precautionary principle.
69
The
consideration of uncertainties is an integral part of the meth-
odology. Thus, the assessment of the effect of the evaluated
additive takes two forms: “no safety concern”and “safety
concern”(Fig. 3b).
Siga definition of UPF. UPFs are characterized by the pres-
ence of at least one deliberately added substance obtained by
synthesis or by a succession of physical, chemical and/or bio-
logical processes leading to its purification and/or substantial
deterioration compared to the original material in the list of
ingredients. UPFs can also be created by the direct application
of a deterioration process (e.g., extrusion-cooking) to the food
Fig. 3 (a) Siga analysis criteria of EFSA reports; (b) the criteria for assigning the level of risk.
Paper Food & Function
Food Func t. This journal is © The Royal Society of Chemistry 2020
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
matrix. These substances are named MUPs and can be
indifferently an ingredient or an additive, most of which are
obtained by technological processes relating to cracking or
synthesis. Two levels of MUPs are distinguished:
- MUP1: obtained by chemical synthesis identical to natural
substances and/or by successive processes leading to purifi-
cation or obtained by successive processes leading to the high
deterioration of the ingredient matrix, such as isolated
protein, starches, natural flavouring, and yeast extract.
- MUP2: obtained by artificial chemical synthesis or by succes-
sive processes leading to the combined purification and substan-
tial deterioration of the matrix, as is the case for glucose syrup,
dextrose, hydrolysed proteins, carboxymethylcellulose, etc.
Purification processes are physical and/or chemical separ-
ation operations consisting of isolating a component from a
complex foodstuff(i.e., chemical solvent extraction, ultrafiltra-
tion, electrophoresis, etc.). High deterioration processes are
physical, chemical and/or biological operations that cause
strong modifications to the initial matrix (beyond thermal
and/or physical alteration processes leading to non-MUPs).
These processes may be hydrolysis processes (chemical or bio-
logical), chemical modifications such as hydrogenation, or
physical treatments such as mechanical separation of meat or
extrusion-cooking. These processes modify the initial matrix/
structure of the original material and result in more than a
simple alteration. Consequently, the substances obtained have
a potential deteriorated matrix effect together with a depletion
of associated protective micronutrients compared to the orig-
inal material. UPFs can be indifferently composed of ingredi-
ents or monoingredients. In the latter case, the food is a
UPF-MUP, such as refined oils.
Siga algorithm for food classification. Overall, the Siga
evaluation method works by downgrading substances (ingredi-
ents and additives) according to their degree of transformation
and the assessment of the associated risk. Each of these cri-
teria may result in a downgrading of the evaluated product.
The final class chosen corresponds to the most transformed
class with respect to all these criteria. Similar to NOVA, preser-
vatives and additives are not considered as cosmetic additives.
Therefore, even if a majority of them meets the definition of a
MUP, they are not identified in this way by the Siga algorithm.
However, Siga limits their number and considers their safety
assessment.
In addition, additives or ingredients that are minimally pro-
cessed are not included in the definition of UPFs, e.g., arabic
gum (solidified descending sap exudate) or virgin oils (cold
pressed). Based on fat/salt/sugar contents (if added, as identi-
fied from the ingredient list), and the nature and levels of
assessed risk of additives and/or ingredient degree of proces-
sing, the NOVA group 4 of UPFs was divided into three UPF
subgroups (Fig. 2: C0.1, C0.2, and C1). Groups C0.1 and C0.2
contain only one MUP1 (not assessed as at risk) but differ by
their salt, sugar or fat contents (if added, as identified from
the ingredient list). The presence of at least one MUP2 or an
additive with a safety concern indicates placement in the
C1 group.
The Siga algorithm is made of different homemade compu-
ter tools whose objective is to characterize each ingredient on
the packaging to score the food item according to a decision
tree –which corresponds to the Siga classification in the
8 groups presented above. A first computer tool allows to
browse the lists of ingredients, in order to isolate all the ingre-
dients from the list. More than 20 000 ingredients according to
their degree of processing and risk assessment (mainly for
additives) were analyzed and categorized as follows: (1) unpro-
cessed ingredients, (2) minimally processed ingredients, (3)
culinary ingredients (minimally processed added fat, sugars or
salt), (4) markers of ultra-processing (levels 1 and 2, i.e., MUP1
and MUP2, respectively), and (5) at-risk substances. Then, a
second computer tool detects ingredients not present in the
previous list of more than 20 000 ingredients. In this case, the
newly identified ingredient is assessed by a Siga expert accord-
ing to our criteria. The last computer tool scores the food
based on the detected ingredients, and the fat, sugar and/or
salt thresholds.
Application of the Siga classification
To generic food group (n= 9) pyramids. Generally, food pyra-
mids are all based on the same scheme including cereals,
legumes, fruits, vegetables, nuts, dairy products, meat, fish
and egg products but without mentioning the degree of pro-
cessing. The Siga algorithm was applied to each of these nine
food groups by selecting a specific food in its database and
categorizing it into the eight Siga technological groups
(A0–C1).
To the assortment of a supermarket. The Siga algorithm was
then applied to a sample of 24 932 packaged food products
(baby foods and alcohols excluded) representative of the food
offering in French supermarkets (n= 3296 brands of products
and n= 98 agro-food companies), i.e., with a barcode. The lists
of ingredients were collected online and directly on-site for
four French retailers. The percentages of products for each of
the eight Siga groups (A0–C1, Fig. 2) were calculated directly
from the Siga database (crude results, excluding fresh fruits
and vegetables, bread and cheeses, delicatessen, unlabelled
fishmongers packaged in advance). The results are presented
either for all products or for some food category. In this ana-
lysis, drastic processes directly applied to the food that were
considered MUPs (see definition above) were not considered
because they were not mentioned on the packaging as legal
information that Siga could check.
Results
Usual food groups based on the degree of processing
For each food group (cereals, legumes, fruits, vegetable, meat,
fish, dairy and egg products) of the usual pyramids, Siga cate-
gorized a specific food into the eight technological groups
(Fig. 4a–i). For example, concerning legumes, fresh peas are
unprocessed, traditional tofu is minimally processed, canned
peas with added salt are processed if the salt content is below
Food & Function Paper
This journal is © The Royal Society of Chemistry 2020 Food Funct.
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
Fig. 4 Application of the Siga algorithm to the nine usual food groups of pyramids. (a) Legumes, (b) vegetables, (c) cereals, (d) fruits, (e) nuts, (f )
fish, (g) eggs, (h) meats, and (i) dairy.
Paper Food & Function
Food Func t. This journal is © The Royal Society of Chemistry 2020
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
the FSA threshold, soya sauce with added salt above the FSA
threshold is a high salt, sugar and/or fat level processed food,
peas with added starch (MUP1) but nutritionally balanced is a
C0.1 UPF, hummus with a high level of added refined oil is a
high salt, sugar and/or fat level C0.2 UPF, pea-based meat sub-
stitutes with added pea proteins and aromas are C1 UPFs
(Fig. 4a), and so on for the following remaining eight food
groups (Fig. 4b–i).
Classification of packaged foods in the eight Siga groups
Among the 24 932 packaged foods, 67% were UPFs (from C0.1
to C1) and therefore contain at least one MUP (Fig. 5). Other
foods were either un-/minimally processed (17% in A0–A2) or
processed (16% in B1–B2). Among the UPFs, only 6% were in
C0.1 and 7% were in C0.2. Therefore, UPFs with more than
one MUP (C1) correspond to the most represented category,
with 54%.
Considering all products, the average number of ingredi-
ents, MUPs and at-risk additives were 10.1, 2.6, and 0.5,
respectively (results not shown in tables or figures). Among
UPFs only, the average number of ingredients and MUPs were
13.2 and 3.8, respectively, and the average number of ingredi-
ents in non-UPF products was 3.7 (results not shown). The
most represented MUP1 (Table 1) were refined oils (≈30% of
the products), aromatic extracts and natural flavours (≈26%),
starches (≈12%), lecithins (≈10%), and isolated animal and
plant proteins (6%). The most represented MUP2 were hydro-
lysed sugars (≈23%), artificial/synthetic aromas (≈19%), and
modified starches (≈8%) (Table 1). The most represented at-
risk additive was sodium nitrite (E250, ≈5%) (Table 1).
Otherwise, among non-UPFs, ≈84% had from one to five
ingredients in their list (Fig. 6a), and ≈16% had more than
five ingredients; among all products (UPFs + non-UPFs), ≈67%
of products with ≥1 ingredient were UPFs (see also Fig. 5),
≈72% of products with ≥2 ingredients were UPFs, and ≈88%
of products with ≥5 ingredients were UPFs (Fig. 6b). In
addition, only ≈4% of products were UPFs when only one
ingredient was present, ≈13% for two ingredients, ≈25% for 3
ingredients, and ≈75% for 6 ingredients (Fig. 6c). Finally, the
higher the number of ingredients, the lower the number of
non-UPF products (Fig. 6d).
Classification of packaged foods in some food category
Among the 81 food categories initially defined in the Siga data-
base, some contained a high percentage of UPFs: 94, 95, 95,
81, 80, and 87% for salted meat, cooked dishes, flavoured
yogurts and white cheeses, energy and gourmet bars, breakfast
cereals, and vegetarian dishes, respectively (Fig. 7); of these,
98, 93, 82, 82, and 69%, respectively, contained more than one
MUP.
Discussion
The Siga technological groups, based on holistic NOVA food
groups, distinguish four new subgroups based on more
specific and reductionist criteria. The Siga definition of UPFs
is focused on the extent of processing and the loss of the
“matrix”effect, for either foods or ingredients, more than on
the function of added ingredients/additives. The Siga defi-
nition also includes at-risk additives and some drastic indus-
trial processes directly impacting the food matrix (i.e.,puffing,
pre-frying, extrusion-cooking). The holistic paradigm under-
lying the Siga classification is based on the assumption that
the “whole is more than the sum of the parts”
1,70
and therefore
that the food or ingredient matrix plays an important role in
good health, including a beneficial synergistic action of nutri-
ents, the presence of protective micronutrient co-factors, a
higher satiety potential and a more appropriate nutrient bio-
availability for the human organism.
39–41
In the Siga under-
lying paradigm, the degree of processing comes first and the
nutrient content second, not the contrary.
1
Indeed, we first
consume food matrices, not nutrients, and processing affects
not only the nutrient content but also, above all, the food
matrix.
Fig. 5 Distribution of 24 932 packaged foods marketed in France
according to the degree of processing with the Siga technological
groups. A0: unprocessed, A1: minimally processed foods, A2: minimally
processed culinary ingredients, B1: nutritionally balanced processed
foods, B2: high salt, sugar and/or fat level processed foods, C0.1: nutri-
tionally balanced UPFs level 0, C0.2: high salt, sugar and/or fat level
UPFs level 0, C1: UPFs.
Table 1 Top ten MUPs and additives with safety concerns encountered
in the 24 932 packaged foods
Ingredient/additive
Percentage
of products
MUP1 Refined oils 29.7
MUP1 Aromatic extracts and natural flavours 25.9
MUP2 Hydrolysed sugars 22.5
MUP2 Artificial/synthetic aroma 18.8
MUP1 Starches 12.4
MUP1 E322 (Lecithins) 10.3
MUP2 E14ii (modified starches) 8.1
MUP1 Isolated animal and vegetable proteins 6.0
R1 E250 (sodium nitrite) 5.1
MUP1 E415 (xanthan gum) 4.7
Food & Function Paper
This journal is © The Royal Society of Chemistry 2020 Food Funct.
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
In this study, we showed that approximately two-thirds of
packaged foods were UPFs, with 13% containing only one
MUP. Therefore, there is a very high level of UPFs among pack-
aged marketed foods. With Siga, UPFs were differentiated
according to the number of MUPs: considering that the level 0
of ultra-processing is acceptable (C0.1, C0.2), 54% of packaged
products are very highly processed (C1 group). A previous
study carried out in New Zealand supermarkets identified 83%
of packaged foods as UPFs according to NOVA, and these
foods were also the least healthy.
71
In another recent US study,
it was found that among 230 156 packaged foods and bev-
erages (data obtained through Label Insight’s Open Data data-
base), 71% were UPFs according to NOVA.
72
These percentages
are not so far from that found in this study, and the slight
differences probably result from both differences between
countries and discrepancies in the way NOVA and Siga charac-
terize UPFs and their MUPs.
Otherwise, Siga revealed that the UPF percentage among all
foods, including fresh and bulk foods, is approximately 50%
(results not shown). In a recent French study using NOVA
classification in which the authors evaluated 7883 households,
the percentage of UPF sales in French conventional stores was
approximately 39%.
73
Therefore, there seems to be good agree-
ment between the presence of UPF in supermarkets and pur-
chasing behaviours. In the same study, it was also evaluated
that among organic products, the percentages of UPF sales
were approximately 31% and 27% in conventional and organic
stores, respectively,
73
suggesting a lower percentage of UPFs
among organic foods. This result is not surprising because in
organic food, only 48 additives are authorized, which naturally
decreases the likelihood of being classified as a UPF, whose
many additives are MUPs. In addition, the specifications are
stricter in terms of process, e.g., no artificial aromas are
allowed in organic processing.
It was also shown that among the 24 932 packaged foods,
the average number of ingredients was 13.2 for UPF and 3.7
Fig. 6 (a) Cumulative percentages of non-UPF products according to the number of ingredients; (b) among all products (non-UPF + UPF), cumulat-
ive percentages of UPF products with N or more ingredients (i.e.,≥1, ≥2, ≥3, ≥4, etc.); (c) percentages of non-UPF and UPF products according to
the number of ingredients; and (d) number of non-UPF and UPF products according to the number of ingredients.
Fig. 7 Percentage of food according to the degree of processing cate-
gorised by the Siga classification for six food categories. A0: unpro-
cessed, A1: minimally processed foods, A2: minimally processed culinary
ingredients, B1: nutritionally balanced processed foods, B2: high salt,
sugar and/or fat level processed foods, C0.1: nutritionally balanced UPF
level 0, C0.2: high salt, sugar and/or fat level UPF level 0, C1: UPFs.
Paper Food & Function
Food Func t. This journal is © The Royal Society of Chemistry 2020
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
for non-UPF, suggesting that long lists of ingredients are very
indicative of ultra-processing: for example, among products
with 4 ingredients, 55% are UPFs, and this figure is 75% for
products with more than 5 ingredients (Fig. 6c). For all pro-
ducts, the average numbers of ingredients, MUPs and at-risk
additives were 10.0, 2.6, and 0.5, respectively. Since the average
number of MUPs among all UPFs was 3.8, this suggests that,
on average, food products with more than 6 ingredients are
very likely to be UPFs. The most frequently observed MUPs are
refined oils, aromatic extracts/natural flavours, hydrolysed
sugars (i.e., coming from starch hydrolysis), artificial/synthetic
aromas, and starches. Most of these MUPs are used for enhan-
cing taste (refined oils, flavours, aromas, and sugars) while
using a lower amount of real foods (sugars and starches may
replace, by weight, better ingredients that are more expensive),
allowing UPFs to be produced at very low costs.
Food categories, such as salted meats, cooked dishes, fla-
voured yogurts and white cheeses, energy and gourmet bars,
breakfast cereals, and vegetarian dishes, contain very high
levels of UPFs (more than 80%), of which 98, 93, 82, 88, 82,
and 69% contain more than one MUP, respectively. In this
study, we did not give the results for the 81 food categories,
but other food groups also contain high levels of UPFs, e.g.,
fresh creams, vinegars, sandwiches, sodas, ice creams, pizzas,
and confectionaries. These food groups are also described as
UPFs in the NOVA classification, but the Siga classification
does not rank 100% of the products in each category as a UPF,
resulting in some further classification and differentiating
UPFs from the C0.1 and C0.2 groups from those of the
C1 group.
Concerning their health potential, UPFs generally have a
lower nutrient density profile in terms of protective bioactive
compounds than un-/minimally processed foods, i.e.,fibre,vita-
mins, minerals, trace elements and antioxidants.
38,40,41,74–76
These bioactive food compounds are the first line of prevention
against the development of chronic diseases that are multifac-
torial and involve several metabolic deregulations as triggering
factors.
77
Therefore, it is not surprising that regular and exces-
sive consumption of UPFs is associated with increased risks of
chronic diseases.
4,30
In addition, UPFs tend to be more hyper-
glycaemic and less satiating, which can lead to overweight,
obesity, hepatic steatosis and type 2 diabetes when regularly
consumed.
32
When such metabolic dysregulations are triggered,
they can then lead to more serious diseases such as some
cancers, steatohepatitis and cardiovascular diseases.
77
Otherwise, as recently shown in the only interventional study of
UPFs, regularly consuming UPFs led to a higher caloric intake,
probably due to the low satiating potential of these foods.
78
In
this study, a +20% increase in caloric intake was measured over
only two weeks, and this increase was attributed mainly to fat
and carbohydrates.
78
Since the diet composition was the same
at the beginning of the study, this increase had to be attributed
to other factors, notably the food matrix and the presence of
“cosmetic”ingredients/additives to amplify sensory properties,
finally leading to the overconsumption of these foods. Pleasure
to eat is stronger than satiation, probably due to some form of
‘food addiction’as shown in overweight children.
79,80
Indeed,
UPFs tend to exhibit textures that demand less chewing (i.e.,
soft, friable, viscous, semi-solid, liquid) and therefore less satia-
tion.
81
Indeed, among 139 solid/semi-solid foods, it was shown
that UPFs have a significantly “lower energy at break”(i.e.,the
energy needed until the food texture breaks).
39
In the end, the
hyper-palatability (both in texture, composition and use of cos-
metic ingredients/additives) may well be the first root cause of
their excessive consumption and subsequent excess calorie, salt,
sugar, fat and/or xenobiotic consumption, leading to chronic
diseases in the long term.
Therefore, the Siga classification may present a useful tool
for improving the health potential of packaged foods, e.g.,
through reformulation and/or less drastic processes. For
example, the replacement of glucose syrup by table sugar in a
product ranked C0 may lead an UPF to be switched to a
different category of processed foods, i.e., B1 or B2 according
to the sugar content. Then, a food ranked as a high salt, sugar
and/or fat level processed food (B2) may be switched to a
balanced processed food by decreasing added fat, salt and/or
sugar. Due to the low health properties of UPFs
5,30
and based
on the precautionary principle, it is therefore recommended
not to exceed 15% of daily calories from these foods
1
and
among them, to choose UPFs from Siga categories C0.1–C0.2
(acceptable/occasional UPFs with one MUP1), which in this
study, constituted approximately 12–13% of the packaged
foods. In addition, the Siga score may also be a decision-
making tool for search engine optimization (SEO) policy and
assortment management in supermarkets.
Concerning the consumer, the Siga score fits into a more
holistic scheme based on what we have defined as the 3 V Rule
for, in French, Végétal (Plant foods, maximum 15% daily
animal calories), Vrai (Real foods, maximum 15% daily UPF
calories), and Varié (Varied, if possible organic, local and sea-
sonal), meaning for the consumer that UPF may be consumed
without risk, if in moderation.
1
Then, in a supermarket, the
idea is to encourage consumers selecting non-UPF foods when-
ever possible. However, buying an UPF within the context of
the above-mentioned 3 V Rule is not problematic for health.
In conclusion, it is important to note that Siga is not
intended to substitute the NOVA classification but rather is
intended to be a pragmatic tool and a rational, objective and
actionable score for retailers and the agro-food industry to
improve the quality of both diets and processed foods in a
more gradual or step-by-step manner. NOVA is necessary and
sufficient for research and consumers. Siga only brings new
scientific elements for UPF definition and UPF sub-classifi-
cation based on the most recent data, notably to allow for the
identification of less processed ingredients as substitutes of
MUPs in UPFs. Otherwise, the Siga algorithm is scalable
depending on new published scientific evidence and risk evalu-
ations of additives. In the end, due to the high level of UPFs
among labelled-packaged products, different types of measures
should be put in place at different levels, including preventive
food education at school from 3 to 15 year-olds, consumer
awareness, and even the taxation of certain UPFs which are very
Food & Function Paper
This journal is © The Royal Society of Chemistry 2020 Food Funct.
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
harmful for health in order for real healthy foods to be less
expensive than UPFs –which overall is not the case.
82
Abbreviations
EFSA European Food Safety Authority
FAO Food and Agriculture Organization
FSA Food Standard Agency
MUP Marker of ultra-processing
PAHO Pan American Health Organization
SEO Search engine optimization
UPF Ultra-processed food
Author contributions
The present study was developed by all authors, who formu-
lated the research questions and designed and carried out the
study. Sylvie Davidou and Anthony Fardet took the lead for
writing the manuscript. Aris Christodoulou and Kelly Frank
collected and analysed the food data. Aris Christodoulou, Kelly
Frank and Sylvie Davidou contributed to the health evaluation
of additives and other food ingredients and participated in the
development of the Siga algorithm. All authors reviewed and
approved the final manuscript.
Financial support
This study received funds from Siga Society only.
Conflicts of interest
Anthony Fardet has been the president of the Siga scientific
committee since May 2017.
Acknowledgements
Pamela Ebner from Siga Society is greatly acknowledged for
her valuable help in analysing food data.
References
1 A. Fardet and E. Rock, Adv. Nutr., 2018, 9, 655–670.
2 J.-C. Moubarac, D. C. Parra, G. Cannon and C. A. Monteiro,
Curr. Obes. Rep., 2014, 3, 256–272.
3 C. A. Monteiro, G. Cannon, J. C. Moubarac, A. P. Martins,
C. A. Martins, J. Garzillo, D. S. Canella, L. G. Baraldi,
M. Barciotte, M. L. Louzada, R. B. Levy, R. M. Claro and
P. C. Jaime, Public Health Nutr., 2015, 18, 2311–2322.
4 C. A. Monteiro, G. Cannon, J.-C. Moubarac, R. B. Levy,
M. L. C. Louzada and P. C. Jaime, Public Health Nutr., 2018,
21,5–17.
5 C. A. Monteiro, G. Cannon, R. B. Levy, J.-C. Moubarac,
M. L. C. Louzada, F. Rauber, N. Khandpur, G. Cediel,
D. Neri, E. Martinez-Steele, L. G. Baraldi and P. C. Jaime,
Public Health Nutr., 2019, 22, 936–941.
6 M. S. Farvid, M. C. Stern, T. Norat, S. Sasazuki, P. Vineis,
M. P. Weijenberg, A. Wolk, K. Wu, B. W. Stewart and
E. Cho, Int. J. Cancer, 2018, 143, 2787–2799.
7 D. Aune, E. Giovannucci, P. Boffetta, L. T. Fadnes,
N. Keum, T. Norat, D. C. Greenwood, E. Riboli, L. J. Vatten
and S. Tonstad, Int. J. Epidemiol., 2017, 46, 1029–1056.
8 F. Imamura, L. O’Connor, Z. Ye, J. Mursu, Y. Hayashino,
S. N. Bhupathiraju and N. G. Forouhi, Br. J. Sports Med.,
2016, 50, 496–504.
9 J. Guo, A. Astrup, J. A. Lovegrove, L. Gijsbers, D. I. Givens
and S. S. Soedamah-Muthu, Eur. J. Epidemiol., 2017, 32,
269–287.
10 D. Aune, T. Norat, P. Romundstad and L. J. Vatten,
Eur. J. Epidemiol., 2013, 28, 845–858.
11 A. Fardet, E. Rock, J. Bassama, P. Bohuon, P. Prabhasankar,
C. Monteiro, J.-C. Moubarac and N. Achir, Adv. Nutr., 2015,
6, 629–638.
12 A. Fardet, C. Richonnet and A. Mazur, Nutr. Rev., 2019, 77,
376–387.
13 A. Fardet and Y. Boirie, Nutr. Rev., 2014, 72, 741–762.
14 D. S. Canella, R. B. Levy, A. P. B. Martins, R. M. Claro,
J. C. Moubarac, L. G. Baraldi, G. Cannon and
C. A. Monteiro, PLoS One, 2014, 9, e92752.
15 M. L. Louzada, L. G. Baraldi, E. M. Steele, A. P. Martins,
D. S. Canella, J. C. Moubarac, R. B. Levy, G. Cannon,
A. Afshin, F. Imamura, D. Mozaffarian and C. A. Monteiro,
Prev. Med., 2015, 81,9–15.
16 F. Juul, E. Martinez-Steele, N. Parekh, C. A. Monteiro and
V. W. Chang, Br. J. Nutr., 2018, 120,90–100.
17 C. S. Costa, B. Del-Ponte, M. C. F. Assunção and
I. S. Santos, Public Health Nutr., 2018, 21, 148–159.
18 L. F. Tavares, S. C. Fonseca, M. L. Garcia Rosa and
E. M. Yokoo, Public Health Nutr., 2012, 15,82–87.
19 M. Lavigne-Robichaud, J.-C. Moubarac, S. Lantagne-Lopez,
L. Johnson-Down, M. Batal, E. A. Laouan Sidi and
M. Lucas, Public Health Nutr., 2018, 21, 172–180.
20 F. Rauber, P. D. B. Campagnolo, D. J. Hoffman and
M. R. Vitolo, Nutr. Metab. Cardiovasc. Dis., 2015, 25, 116–
122.
21 R. D. Mendonca, A. C. S. Lopes, A. M. Pimenta, A. Gea,
M. A. Martinez-Gonzalez and M. Bes-Rastrollo,
Am. J. Hypertens., 2017, 30, 358–366.
22 T. Fiolet, B. Srour, L. Sellem, E. Kesse-Guyot, B. Allès,
C. Méjean, M. Deschasaux, P. Fassier, P. Latino-Martel,
M. Beslay, S. Hercberg, C. Lavalette, C. A. Monteiro, C. Julia
and M. Touvier, Br. Med. J., 2018, 360, k322.
23 B. Srour, L. K. Fezeu, E. Kesse-Guyot, B. Allès, C. Méjean,
R. M. Andrianasolo, E. Chazelas, M. Deschasaux,
S. Hercberg, P. Galan, C. A. Monteiro, C. Julia and
M. Touvier, Br. Med. J., 2019, 365, l1451.
24 L. Schnabel, C. Buscail, J.-M. Sabate, M. Bouchoucha,
E. Kesse-Guyot, B. Allès, M. Touvier, C. A. Monteiro,
Paper Food & Function
Food Func t. This journal is © The Royal Society of Chemistry 2020
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
S. Hercberg, R. Benamouzig and C. Julia,
Am. J. Gastroenterol., 2018, 113, 1217–1228.
25 C. Gomez-Donoso, A. Sanchez-Villegas, M. A. Martinez-
Gonzalez, A. Gea, R. D. Mendonca, F. Lahortiga-Ramos and
M. Bes-Rastrollo, Eur. J. Nutr., 2019, DOI: 10.1007/s00394-
019-01970-1.
26 M. Adjibade, C. Julia, B. Allès, M. Touvier, C. Lemogne,
B. Srour, S. Hercberg, P. Galan, K. E. Assmann and
E. Kesse-Guyot, BMC Med., 2019, 17, 78.
27 L. Schnabel, E. Kesse-Guyot, B. Allès, M. Touvier, B. Srour,
S. Hercberg, C. Buscail and C. Julia, J. Am. Med. Assoc.,
2019, 179, 490–498.
28 A. Rico-Campà, M. A. Martínez-González, I. Alvarez-Alvarez,
R. d. D. Mendonça, C. de la Fuente-Arrillaga, C. Gómez-
Donoso and M. Bes-Rastrollo, Br. Med. J., 2019, 365, l1949.
29 Ministry of Health of Brazil, Primary Health Care
Department, Dietary guidelines for the Brazilian population,
Ministry of Health of Brazil, Secretariat of Health Care,
Primary Health Care Department, São Paulo, 2014.
30 FAO, C. A. Monteiro, G. Cannon, M. Lawrence,
M. L. d. C. Louzada and P. P. Machado, Ultra-processed
foods, diet quality, and health using the NOVA classification
system, Rome, Italy, 2019.
31 PAHO and WHO, Ultra-processed food and drink products in
Latin America: Sales, sources, nutrient profiles, and policy
implications, 2019.
32 A. Fardet and E. Rock, Trends Food Sci. Technol., 2019, 32,
174–184.
33 F. Rauber, M. L. da Costa Louzada, E. Steele, C. Millett,
C. A. Monteiro and R. B. Levy, Nutrients, 2018, 10, 587.
34 M. L. d. C. Louzada, C. Z. Ricardo, E. M. Steele, R. B. Levy,
G. Cannon and C. A. Monteiro, Public Health Nutr., 2018,
21,94–102.
35 G. Cediel, M. Reyes, M. L. da Costa Louzada, E. Martinez
Steele, C. A. Monteiro, C. Corvalán and R. Uauy, Public
Health Nutr., 2018, 21, 125–133.
36 E. Martínez Steele, L. G. Baraldi, M. L. d. C. Louzada,
J.-C. Moubarac, D. Mozaffarian and C. A. Monteiro, BMJ
Open, 2016, 6, e009892.
37 M. L. Louzada, A. P. Martins, D. S. Canella, L. G. Baraldi,
R. B. Levy, R. M. Claro, J. C. Moubarac, G. Cannon and
C. A. Monteiro, Rev. Saude Publica, 2015, 49,1–8.
38 M. L. Louzada, A. P. Martins, D. S. Canella, L. G. Baraldi,
R. B. Levy, R. M. Claro, J. C. Moubarac, G. Cannon and
C. A. Monteiro, Rev. Saude Publica, 2015, 49, 38.
39 A. Fardet, S. Lakhssassi and A. Briffaz, Food Funct., 2018, 9,
561–572.
40 A. Fardet, C. Méjean, H. Labouré, V. A. Andreeva and
G. Féron, Food Funct., 2017, 8, 651–658.
41 A. Fardet, Food Funct., 2016, 7, 2338–2346.
42 G. Ares, F. Varela, L. Machin, L. Antúnez, A. Giménez,
M. R. Curutchet and J. Aschemann-Witzel, Food Qual.
Prefer., 2018, 68, 215–225.
43 M. Egnell, P. Ducrot, M. Touvier, B. Allès, S. Hercberg,
E. Kesse-Guyot and C. Julia, PLoS One, 2018, 13, e0202095.
44 A. Drewnowski, Am. J. Clin. Nutr., 2010, 91, 1095S–1101S.
45 A. Fardet, Food Funct., 2015, 6, 363–382.
46 A. Fardet, A. Christodoulou, K. Frank and S. Davidou, IAA,
2019, 59,22–25.
47 A. Fardet, Prat. Nutr., 2018, 56,32–36.
48 C. Monteiro, G. Cannon, R. Levy, et al.,World Nutr., 2016,
7,28–38.
49 S. Rana, M. Mishra, D. Yadav, S. K. Subramani, C. Katare
and G. Prasad, Prog. Nutr., 2018, 20,5–14.
50 K. L. Stanhope and P. J. Havel, Am. J. Clin. Nutr., 2008, 88,
1733S–1737S.
51 L. Sanchez-Lozada, W. Mu, C. Roncal, Y. Sautin,
M. Abdelmalek, S. Reungjui, M. Le, T. Nakagawa, H. Lan,
X. Yu and R. Johnson, Eur. J. Nutr., 2010, 49,1–9.
52 X. Ouyang, P. Cirillo, Y. Sautin, S. McCall, J. L. Bruchette,
A. M. Diehl, R. J. Johnson and M. F. Abdelmalek,
J. Hepatol., 2008, 48, 993–999.
53 C. Pelkman, C. Lawton, F. Croden, D. Camidge,
E. T. Finocchairo and L. Dye, FASEB J., 2013, 27, lb310.
54 H. Cena and P. C. Calder, Nutrients, 2020, 12, 334.
55 J. Renzella, N. Townsend, J. Jewell, J. Breda, N. Roberts,
M. Rayner and K. Wickramasinghe, What national and sub-
national interventions and policies based on Mediterranean
and Nordic diets are recommended or implemented in the
WHO European Region, and is there evidence of effectiveness
in reducing noncommunicable diseases?, WHO, 2018.
56 R. Martinez-Lacoba, I. Pardo-Garcia, E. Amo-Saus and
F. Escribano-Sotos, Eur. J. Public Health, 2018, cky113–
cky113.
57 D. F. Romagnolo and O. I. Selmin, Nutr. Today, 2017, 52,
208–222.
58 D. C. Willcox, B. J. Willcox, H. Todoriki and M. Suzuki,
J. Am. Coll. Nutr., 2009, 28, 500S–516S.
59 L. G. Snetselaar and R. M. Lauer, The prudent diet: preven-
tive nutrition, in Nutrition in pediatrics: basic science and
clinical applications, ed. W. A. Walker, J. B. Watkins and C.
Duggan, 2003.
60 G. B. Haber, K. W. Heaton, D. Murphy and L. F. Burroughs,
Lancet, 1977, 2, 679–682.
61 M. M. L. Grundy, F. Carriere, A. R. Mackie, D. A. Gray,
P. J. Butterworth and P. R. Ellis, Food Funct., 2016, 7,69–78.
62 S. A. Moorhead, R. W. Welch, M. Barbara, E. Livingstone,
M. McCourt, A. A. Burns and A. Dunne, Br. J. Nutr., 2006,
96, 587–595.
63 A. Fardet, Prat. Nutr., 2017, 13,37–40.
64 A. Fardet, I. Souchon and D. Dupont, Structure des aliments
et effets nutritionnels, Quae edn, 2013.
65 Uk Department of Health and Food Standard Agency,
Guide to creating a front of pack (FoP) nutrition label for pre-
packed products sold through retail outlets, Uk Department
Of Health, Food Standard Agency devolved administrations
in Scotland, Northern Ireland and Wales in collaboration
with the British Retail Consortium, 2013.
66 EUR-Lex, Commission Regulation (EU) No 231/2012,
European Commission, 2012.
67 E. Ritz, K. Hahn, M. Ketteler, M. K. Kuhlmann and
J. Mann, Dtsch. Arztebl. Int., 2012, 109,49–U29.
Food & Function Paper
This journal is © The Royal Society of Chemistry 2020 Food Funct.
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
68 H. Jiang, Y. Tang, H. K. Garg, D. K. Parthasarathy,
A. C. Torregrossa, N. G. Hord and N. S. Bryan, Nitric Oxide,
2012, 26, 267–273.
69 P. Cox (European Parliament) and J. Piqué i Camps
(Council of the European Union), in Regulation (EC)
No. 178/2002 of the European Parliament and of the Council
of 28 January 2002 laying down the general principles and
requirements of food law, establishing the European Food
Safety Authority and laying down procedures in matters
of food safety, ed. E. Parliament, Journal officiel des
Communautés européennes, Brussels, 2002, vol. 178/2002,
pp. 1–24.
70 A. Fardet and E. Rock, Adv. Nutr., 2014, 5, 430–
446.
71 C. M. Luiten, I. H. Steenhuis, H. Eyles, C. Ni Mhurchu and
W. E. Waterlander, Public Health Nutr., 2016, 19, 530–
538.
72 A. S. Baldridge, M. D. Huffman, F. Taylor, D. Xavier,
B. Bright, L. V. Van Horn, B. Neal and E. Dunford,
Nutrients, 2019, 11, 1704.
73 M. Desquilbet, E. Maigné and S. Monier-Dilhan, Ecol.
Econom., 2018, 150, 194–203.
74 C.Julia,L.Martinez,B.Alles,M.Touvier,S.Hercberg,C.Mejean
and E. Kesse-Guyot, Public Health Nutr., 2018, 21,27–37.
75 D. C. Parra, M. L. d. Costa-Louzada, J.-C. Moubarac,
R. Bertazzi-Levy, N. Khandpur, G. Cediel and
C. A. Monteiro, Salud Pública Méx., 2019, 61,1–8.
76 J.-C. Moubarac, M. Batal, M. L. Louzada, E. Martinez Steele
and C. A. Monteiro, Appetite, 2016, 108, 512–520.
77 A. Fardet and Y. Boirie, Nutr. Rev., 2013, 71, 643–656.
78 K. D. Hall, A. Ayuketah, R. Brychta, H. Cai, T. Cassimatis,
K. Y. Chen, S. T. Chung, E. Costa, A. Courville, V. Darcey,
L. A. Fletcher, C. G. Forde, A. M. Gharib, J. Guo, R. Howard,
P. V. Joseph, S. McGehee, R. Ouwerkerk, K. Raisinger,
I. Rozga, M. Stagliano, M. Walter, P. J. Walter, S. Yang and
M. Zhou, Cell Metab., 2019, 30,67–77.
79 A. R. Filgueiras, V. B. Pires de Almeida, P. C. Koch
Nogueira, S. M. Alvares Domene, C. Eduardo da Silva,
R. Sesso and A. L. Sawaya, Appetite, 2018, 135, 137–145.
80 E. M. Schulte, N. M. Avena and A. N. Gearhardt, PLoS One,
2015, 10, e0117959.
81 L. Chambers, Nutr. Bull., 2016, 41, 277–282.
82 S. Gupta, T. Hawk, A. Aggarwal and A. Drewnowski, Front.
Nutr., 2019, 6, 70.
Paper Food & Function
Food Func t. This journal is © The Royal Society of Chemistry 2020
Published on 30 January 2020. Downloaded by RSC Internal on 2/21/2020 4:23:03 PM.
View Article Online
