Minimally processed foods are more satiating and less hyperglycemic than ultra-processed foods: A preliminary study with 98 ready-to-eat foods

Abstract

Beyond nutritional composition, food structure is increasingly recognized to play a role in food health potential, notably in satiety and glycemic responses. Food structure is also highly dependent on processing conditions. The hypothesis for this study is, based on a data set of 98 ready-to-eat foods, that the degree of food processing would correlate with the satiety index (SI) and glycemic response. Glycemic response was evaluated according to two indices: the glycemic index (GI) and a newly designed index, the glycemic glucose equivalent (GGE). The GGE indicates how a quantity of a certain food affects blood glucose levels by identifying the amount of food glucose that would have an effect equivalent to that of the food. Then, foods were clustered within three processing groups based on the international NOVA classification: (1) raw and minimally processed foods; (2) processed foods; and (3) ultra-processed foods. Ultra-processed foods are industrial formulations of substances extracted or derived from food and additives, typically with five or more and usually many (cheap) ingredients. The data were correlated by nonparametric Spearman's rank correlation coefficient on quantitative data. The main results show strong correlations between GGE, SI and the degree of food processing, while GI is not correlated with the degree of processing. Thus, the more food is processed, the higher the glycemic response and the lower its satiety potential. The study suggests that complex, natural, minimally and/or processed foods should be encouraged for consumption rather than highly unstructured and ultra-processed foods when choosing weakly hyperglycemic and satiating foods.

Full text

Food &
Function
PAPER
Cite this: Food Funct., 2016, 7, 2338
Received 26th January 2016,
Accepted 23rd April 2016
DOI: 10.1039/c6fo00107f
www.rsc.org/foodfunction
Minimally processed foods are more satiating and
less hyperglycemic than ultra-processed foods:
a preliminary study with 98 ready-to-eat foods
Anthony Fardet*
Beyond nutritional composition, food structure is increasingly recognized to play a role in food health
potential, notably in satiety and glycemic responses. Food structure is also highly dependent on proces-
sing conditions. The hypothesis for this study is, based on a data set of 98 ready-to-eat foods, that the
degree of food processing would correlate with the satiety index (SI) and glycemic response. Glycemic
response was evaluated according to two indices: the glycemic index (GI) and a newly designed index,
the glycemic glucose equivalent (GGE). The GGE indicates how a quantity of a certain food affects blood
glucose levels by identifying the amount of food glucose that would have an effect equivalent to that of
the food. Then, foods were clustered within three processing groups based on the international NOVA
classification: (1) raw and minimally processed foods; (2) processed foods; and (3) ultra-processed foods.
Ultra-processed foods are industrial formulations of substances extracted or derived from food and addi-
tives, typically with five or more and usually many (cheap) ingredients. The data were correlated by non-
parametric Spearman’s rank correlation coefficient on quantitative data. The main results show strong
correlations between GGE, SI and the degree of food processing, while GI is not correlated with the
degree of processing. Thus, the more food is processed, the higher the glycemic response and the lower
its satiety potential. The study suggests that complex, natural, minimally and/or processed foods should
be encouraged for consumption rather than highly unstructured and ultra-processed foods when choos-
ing weakly hyperglycemic and satiating foods.
Introduction
Meta-analyses clearly show that healthy and Mediterranean
diets can decrease the risk of developing type 2 diabetes by
15 to 23%.
1–5
Moreover, vegetarian and Mediterranean diets
significantly reduced fasting glucose by 0.36 mmol L
−1
and
3.89 mg dL
−1
, respectively, fasting insulin by 1.06 μUmL
−1
and the HOMA-IR index by up to 0.45, which contributes to
close control of carbohydrate metabolism.
6–8
What is particu-
larly significant is that these diets are predominantly based on
consuming raw and minimally processed and plant products
(fruits, vegetables, grains and seeds) in substantial quantities.
Moreover, concerning the food groups and on the basis of the
comparison of high versus low consumption, whole grains,
nuts, coffee, dairy products and legumes appear to be rather
protective with respect to type 2 diabetes, unlike sweetened
beverages and red and/or processed meat.
9,10
Conversely,
based on the results of meta-analyses, non-healthy diets (or
western-type diets) increase the risk of type 2 diabetes by 41 to
44%.
2,4
These diets are usually high in animal products and/or
ultra-processed foods that are high in energy and low in pro-
tective compounds. Beyond conventional food groups, these
results suggest that the degree of food processing also comes
into play when assessing the risk of type 2 diabetes.
11
Within the same food group, there are indeed foods with
diverse health values according to the technological processes
applied.
11
Thus, ready-to-eat breakfast cereals for adults, such
as muesli, and those for children, such as extruded cereals
enriched with sugars and fat, have very different nutritional
values.
12
Moreover, among 39 765 men and 157 463 women in
the Health Professionals Follow-up Study and the Nurses’
Health Study I and II, high intake of brown rice was associated
with an 11% lower relative risk of type 2 diabetes, whereas
high consumption of white rice was associated with a 17%
higher relative risk of type 2 diabetes.
13
In these two examples,
the degree of processing is likely to make the difference from a
nutritional point of view, not cereal as a botanical group.
Based on the NOVA classification ranking foods into
4 groups (see groups lines below) according to their extent of
processing, other studies have also shown that ultra-processed
food as a whole (Group 4), are nutritionally inferior to the com-
INRA, UMR 1019, UNH, CRNH Auvergne, F-63000 Clermont-Ferrand & Clermont
University, University of Auvergne, Human Nutrition Unit, BP 10448, F-63000
Clermont-Ferrand, France. E-mail: anthony.fardet@clermont.inra.fr;
Fax: +33 (0)4 73 62 47 55; Tel: +33 (0)4 73 62 47 04
2338 |Food Fun ct.,2016,7, 2338–2346 This journal is © The Royal Society of Chemistry 2016

bination of raw and minimally processed foods (Group 1),
culinary ingredients (Group 2) and processed foods (Group 3),
both for macro and micro nutrients.
14–18
Depending on the degree of processing, food can also regis-
ter differently on the glycemic index and have a different satiat-
ing effect,
19–21
as shown, for example, with raw, processed
(cooked, blended or refined) apples and carrots.
22–24
This
suggests that the more food is deconstructed, the higher its
glycemic response and the less satiating it is, but because data
on the subject remain scarce, this effect needs to be confirmed
for a larger number of foods.
As suggested above, any food has the potential to change
postprandial blood glucose after consumption. Until recently,
the glycemic index (GI) has been mainly used
25
and more
recently the glycemic load, which is related to the amount of
food consumed.
26,27
However, ten years ago a new concept was
developed and validated by Monro et al.: the glycemic glucose
equivalent (GGE).
28
It indicates how a quantity of a food
affects blood glucose levels by showing the equivalent effect of
ingested food glucose. Thus, if a serving of food contributes 15
GGE (g per 100 g), it has the same effect on the body as con-
suming 15 grams of glucose. Therefore, GGE behaves like a
compound of the food, and the relative glycemic impact is the
GGE consumption responsible for a glycemic response.
28
The
relative glycemic impact therefore differs from the GI because
it refers to the food and depends on the consumption of the
food (i.e., does not need to be restricted to equi-carbohydrate
comparison), whereas the GI refers to carbohydrates only and
is a unitless index that does not account for food intake. More-
over, because the consumption of the GGE is a function of the
consumption of food, it can be used quantitatively to give a
direct measurement of the glycemic impact of an amount of
food on the body rather than just carbohydrates.
28
By releasing
the constraint of equivalence in carbohydrates and with regard
to its reactivity to food intake, the GGE has significant advan-
tages over the GI and carbohydrate content in the manage-
ment of blood glucose. Thus, the content in the GGE should
allow individual objectives of meals to be clarified, realistically,
according to a glycemic effect once the individual GGE toler-
ance is established by measuring the blood glucose response
to a known consumption of GGE.
29
Otherwise, the satiety potential is often an overlooked
aspect of food. A feeling of prolonged satiety is beneficial
because it discourages snacking between meals, which is often
of refined foods, rich in energy and with a high GI. Being able
to choose foods with high satiety potential may therefore be
an advantage for preventing diabetes and obesity. However,
data on the satiety potential of foods are scarce.
19
Therefore, considering the possible role of processing in
the satiety potential and glycemic impact of foods, I hypoth-
esized, based on a much greater number of foods (n=98
ready-to-eat foods), that the degree of food processing would
correlate with satiety potential and glycemic response. Because
a higher prevalence of chronic diseases, e.g., dyslipidemia,
30
metabolic syndrome,
31
obesity,
32,33
and cardiovascular dis-
eases,
34
is associated with regular and high consumption of
ultra-processed foods, the foods in this study were classified,
not according to their botanical (i.e., fruits, vegetables, grains
and seeds) or animal (e.g., red and white meats, dairy pro-
ducts) affiliation, but according to the degree of their proces-
sing according to the NOVA classification.
35
The objective of
this study was to test the correlations among the degree of
food processing, satiety index and glycemic impact.
Materials and methods
Food selection
A table of 1224 foods and food ingredients consumed by dia-
betic individuals has been compiled by the French Paramedical
Society of Diabetes (SFD Paramédical, Paris, France). These
foods come from all conventional food groups, including bev-
erages, snacks, ready-to-eat meals, fats, seasonings, baked
goods, and confectionaries. Within them, 98 foods were selected
based on the data available in the literature for their glycemic
impact (GGE and GI) and/or satiety potential (Tables 1–3).
Degree of processing
For classification according to the degree and purpose of food
processing, we relied on the work of the Brazilian team of
Monteiro et al., who developed the NOVA classification of
foods based on the extent to which they are processed.
35,36
The international NOVA classification clusters foods into
4 groups according to the degree of processing from the least
to the most drastic transformation.
14,37,38
Briefly, as syn-
thesized from Monteiro et al.:
38
(1) raw and minimally pro-
cessed food: unprocessed foods are the parts of the animals
collected immediately after slaughter and the parts of plant
products after harvest or collection. Minimally processed foods
are unprocessed foods subject to a transformation, especially a
change in their physical properties that does not substantially
alter the nutritional properties or uses of the foods. These pro-
cesses are used to extend the storage time for unprocessed
foods and often to reduce the time and effort required for
their preparation. (2) Culinary ingredients: these are sub-
stances obtained directly from group 1 foods or from nature by
processes such as pressing, refining, grinding, milling, and
spray drying. The purpose of processing here is to make pro-
ducts used in home and restaurant kitchens to prepare, season
and cook group 1 foods and to make with them varied and
enjoyable hand-made dishes, soups and broths, breads, pre-
serves, salads, drinks, desserts and other culinary prep-
arations. Group 2 items are rarely consumed in the absence of
group 1 foods. (3) Processed foods: these are relatively simple
products made by adding sugar, oil, salt or other group 2 sub-
stances to group 1 foods. Most processed foods have two or
three ingredients. Processes include various preservation or
cooking methods, and, in the case of breads and cheese, non-
alcoholic fermentation. The main purpose of the manufacture
of processed foods is to increase the durability of Group 1
foods, or to modify or enhance their sensory qualities. (4)
Ultra-processed foods: these are industrial formulations of
Food & Function Paper
This journal is © The Royal Society of Chemistry 2016 Food Funct.,2016,7, 2338–2346 | 2339

substances extracted or derived from food and additives, typi-
cally with five or more and usually many (cheap) ingredients.
Such ingredients often include those also used in processed
foods, such as sugar, oils, fats, salt, anti-oxidants, stabilizers,
and preservatives. Ingredients only found in ultra-processed
products include substances not commonly used in culinary
preparations, and additives whose purpose is to imitate
sensory qualities of Group 1 foods or of culinary preparations
of these foods, or to disguise undesirable sensory qualities of
the final product. Group 1 foods are a small proportion of or
Table 1 Satiety index, available carbohydrates and glycemic potential of raw and minimally processed foods
Foods
Satiety index
(means ± SEM)
a
Serving
(g)
Available carbohydrates
b
(g per 100 g)
Relative glycemic
impact
Glycemic index
(means ± SEM)
d
GGE/
100 g
c
GGE/
serving
Red meat and pork 176 ± 50 Traces
Fish 225 ± 30 Traces
Egg 150 ± 31 0.5
Bulgur, cooked 150 18.6 8 12 48 ± 2
Pasta, cooked 119 ± 35 180 29.7 10 18 44 ± 3
Wholemeal pasta, cooked 188 ± 45 180 26.8 10 17.8 37 ± 5
Fava beans, cooked 80 6.1 7 5.6 79 ± 16
White beans, cooked 168 ± 42 150 13.6 6 9 29 ± 9
Kidney beans, cooked 150 14.4 4 6 28 ± 4
Yam, raw or cooked 150 27.9 10 15 37 ± 8
Lentils, cooked 133 ± 28 150 16.6 4 6 30 ± 4
Sweet corn on the cob, cooked 80 16.4 11 8.8 54 ± 4
Cassava 100 35.0 15 15 46
Split peas, cooked 150 14.0 7 10.5 32
Potato, boiled in water 323 ± 51 150 15.8 10 15 50 ± 9
Instant mashed potatoes, reconstituted 150 14.6 12 18 85 ± 3
White rice, cooked 138 ± 31 150 28.7 10 15 64 ± 7
Brown rice, cooked 132 ± 35 150 31.7 16 24 55 ± 5
Beetroot, raw or cooked 80 7.2 7 5.6 64 ± 16
Carrot, raw or cooked 80 6.6 3 2.4 47 ± 16
Turnip, raw or cooked 80 3.1 3 2.4 72
Parsnip, raw or cooked 80 17.0 12 9.6 97 ± 19
Green peas, cooked 80 8.3 2 1.6 48 ± 5
Pumpkin, cooked 80 1.9 3 2.4 75 ± 9
Milk (whole, semi-skimmed and skimmed) 250 5.0 1 2.5 27 ± 4
Plain yogurt 88 ± 23 200 4.6 2 3.3 36 ± 4
Muesli 100 ± 23 30 61.5 28 8.4 49 ± 9
Apricot 120 9.0 5 6 57
Pineapple 120 11.0 8 9.6 59 ± 8
Banana 118 ± 27 120 20.5 13 15.6 52 ± 4
Cherries 120 14.2 3 3.6 22
Strawberries 120 4.1 3 3.6 40
Kiwi fruit 120 9.4 5 6 53 ± 6
Melon 120 6.5 4 4.8 65 ± 9
Orange 202 ± 34 120 8.3 4 4.8 42 ± 3
Grapefruit 120 6.2 1 1.2 25
Watermelon 120 7.3 4 4.8 72 ± 13
Peach, nectarine 120 11.3 3 3.6 42 ± 14
Pear 120 10.8 5 6 33
Apple 197 ± 32 120 11.3 4 4.8 38 ± 2
Damson plum 120 9.6 5 6 39 ± 15
Black grape 162 ± 32 120 12.1 7 8.4 46 ± 3
Dried apricot 60 53.0 15 9 31 ± 1
Dried dates 60 62.5 70 42 103 ± 21
Dried figs 60 50.4 33 19.8 61 ± 6
Prunes 60 52.3 13 7.8 29 ± 4
Raisins 60 66.4 46 27.6 64 ± 11
Carrot juice 250 5.1 3 7.5 43 ± 3
Tomato juice 250 3.9 2 5 38 ± 4
Fresh orange juice 250 15.0 5 12.5 50 ± 4
SEM, standard error of the means.
a
Satiety index of white bread = 100%; data from Holt et al.
19
For information, the sample size (n-value for
means) for each food can be found in the table by Holt et al.
19 b
Data were collected primarily from the 2013 French Ciqual database (available
online at: https://pro.anses.fr/tableciqual/index.htm); the rest were taken directly from the product labels of specific brands.
c
Data from Monro.
44
d
Glycemic index of glucose = 100; data for glycemic index from Foster-Powell et al.
20
For information, the sample size (n-value for means) for
each food can be found in the table by Foster-Powell et al.
20
When values were given without SEM, this means that they correspond to only one
measurement.
Paper Food & Function
2340 |Food Funct.,2016,7, 2338–2346 This journal is © The Royal Society of Chemistry 2016

are even absent from ultra-processed products. The main
purpose of industrial ultra-processing is to create products
that are ready to eat, to drink or to heat, liable to replace both
unprocessed or minimally processed foods that are naturally
ready to consume, such as fruits and nuts, milk and water,
and freshly prepared drinks, dishes, desserts and meals.
Common attributes of ultra-processed products are hyper-
palatability, sophisticated and attractive packaging, multi-
media and other aggressive marketing to children and adoles-
cents, health claims, high profitability, and branding and own-
ership by transnational corporations”.
38
According to NOVA, consumption of Group 4 ultra-pro-
cessed food predicts overall diet quality, obesity and other
chronic diseases, while Group 1, 2 and 3 taken together are the
basis of a healthy diet. Indeed, epidemiological studies have
provided evidence that foods within group 4 are primarily
responsible for the dramatic increase in the prevalence of
obesity, metabolic syndrome and dyslipidemia.
30–33,39–43
This study was based on ready-to-eat foods only. Therefore,
culinary ingredients from Group 2 were not considered. Then,
the 98 foods were ranked within Groups 1, 3 and 4 based on
NOVA descriptions. Finally, for the purpose of this study, food
groups were renamed as follow: Group 1, raw and minimally
processed foods (MPF, n= 49 foods; Table 1); Group 2, pro-
cessed foods (PF, n= 12 foods; Table 2); and Group 3, ultra-
processed foods (UPF, n= 37; Table 3).
The glycemic potential
Approximate GGE values may be obtained from:
28,29
GGE ¼ð%of available carbohydrates=100ÞGI ð1Þ
and:
Relative glycemic impact ¼food weight consumed
GGE per g ð2Þ
GGE values may be also subject to error imported from cur-
rently available carbohydrate values.
28
In this study, GGE was connected with the GI and the
degree of processing using tables for the GGE
44
and the GI.
20
GGE values not available in table by Monro were calculated
from formula (1) (see above).
The satiety potential
The only available data are the satiety index (SI) given for 38
foods grouped into six categories: fruits (average SI = 170),
bread products (average SI = 85), snacks and confectionery
(average SI = 100), starchy foods (average SI = 166), foods high
in protein (average SI = 166) and ready-to-eat breakfast cereals
(average SI = 134), with white bread used as a reference
(average SI = 100).
19
Briefly, SI score was calculated by dividing
the area under the satiety response curve for the test food by
the group mean satiety area under curve for white bread and
multiplying by 100.
In this study, the SI was connected with the glycemic
response (GGE and GI) and the degree of processing using a
table compiled by Holt et al.
19
Statistical analyses
The data for the GI, GGE, SI and degree of processing were cor-
related using the nonparametric Spearman’s rank correlation
coefficient (R
s
) for quantitative data (BiostaTGV, based on
R software, available online at: http://marne.u707.jussieu.fr/
biostatgv/?module=tests/spearman). This web tool was develo-
ped in 2000 by the Institute Pierre Louis of Epidemiology and
Public Health, which is affiliated with INSERM, and the Pierre
& Marie Curie University. A Pvalue <0.05 indicates a signifi-
cant correlation.
For calculations, the qualitative data for the degree of
processing were converted into quantitative data as follows:
MPF = 1; PF = 2; UPF = 3, with “3”being more processed than
Table 2 Satiety index, available carbohydrates and glycemic potential of processed foods
Foods
Satiety index
(means ± SEM)
a
Serving
(g)
Available carbohydrates
b
(g per 100 g)
Relative glycemic impact
Glycemic index
(means ± SEM)
d
GGE/100 g
c
GGE/serving
French fries 150 24.9 21 31.5 75
Fried potatoes, home-cooked 116 ± 35 150 30.0 21 31.5 75
Lebanese hummus 30 9.3 0 0 6 ± 4
Minestrone 250 4.8 3 7.5 39 ± 3
Cheese 146 ± 28 0–3.0
White bread 100 ± 0 30 52.3 37 11.1 95 ± 15
Wholemeal bread 157 ± 29 30 50.6 29 8.7 71 ± 2
Rye bread 30 49.8 25 7.5 58 ± 6
Pita bread 30 53.4 33 9.9 57
Pears in syrup 120 13.9 4 4.8 44
Peanuts, roasted, salted 50 9.7 2 1 14 ± 8
Cashew nuts, grilled, salted 50 21.8 5 2.5 22 ± 5
SEM, standard error of the means.
a
Satiety index of white bread = 100%; data given by Holt et al.
19
For information, the sample size (n-value for
means) for each food can be found in the table by Holt et al.
19 b
Data were collected primarily from the 2013 French Ciqual database (available
online at: https://pro.anses.fr/tableciqual/index.htm); the rest were taken directly from the product labels of specific brands.
c
Data from Monro.
44
d
Glycemic index of glucose = 100; data for glycemic index from Foster-Powell et al.
20
For information, the sample size (n-value for means) for
each food can be found in the table by Foster-Powell et al.
20
When values were given without SEM, this means that they correspond to only one
measurement.
Food & Function Paper
This journal is © The Royal Society of Chemistry 2016 Food Fun ct.,2016,7, 2338–2346 | 2341

“2”, and “2”more processed than “1”. Satiety index and glyce-
mic index were given as means ± SEM as indicated in the orig-
inal tables.
19,20
Results
Correlation between glycemic glucose equivalent and glycemic
index
A value for the GGE was determined for 83 foods in relation to
the GI (Tables 1–3). The GI and the GGE were significantly and
positively correlated (Fig. 1, R
s
= 0.56, P=4×10
−8
). However,
for low GGEs (below 15 g per 100 g), the range of GIs is highly
variable, i.e., between 6 and 100.
Relationship between processing group and glycemic impact
and satiety index
An SI was assigned to 33 foods (Tables 1–3). The SI is signifi-
cantly and inversely correlated with the degree of processing
(Fig. 2, R
s
=−0.60, P= 0.0002). Thus, the more the food is pro-
cessed, the less satiating it tends to be.
A GGE value was determined for 89 foods (Tables 1–3). The
GGE is significantly and positively correlated with the proces-
sing group (Fig. 3, R
s
= 0.45, P=8×10
−6
). Therefore, the more
a food is processed, the higher the GGE tends to be.
Table 3 Satiety index, available carbohydrates and glycemic potential of ultra-processed foods
Foods
Satiety index
(means ± SEM)
a
Serving
(g)
Available carbohydrates
b
(g per 100 g)
Relative glycemic
impact
Glycemic index
(means ± SEM)
d
GGE/
100 g
c
GGE/
serving
Pizza 100 27.7 8 8 51
Chicken McNuggets 100 17.0 7 21 46 ± 4
Pancake 80 28.0 19 15.2 67 ± 5
Fish’n dips 100 31.2 8 8 38 ± 6
Ravioli with tomato sauce 180 13.5 9 16.2 39 ± 1
Tomato soup 250 2.5 3 7.5 38 ± 9
Sweetened condensed milk 250 55.9 33 82.5 61 ± 6
Croissant (packaged) 47 ± 17 57 47.7 26 14.8 67
Kellogg’s all-bran fibre plus cereal 151 ± 30 30 48.0 17 5.1 42 ± 5
Kellogg’s coco pops cereal 30 85.0 67 20.1 77
Kellogg’s corn flakes 118 ± 19 30 78.3 69 20.7 81 ± 3
Kellogg’s special K cereal 116 ± 27 30 75.0 38 11.4 84 ± 12
Kellogg’s Frosties 30 87.0 49 14.7 55
Balisto bar (with fruits, honey,
milk and muesli)
30 56.0 35 10.5 61
Chocolate and cereal snack bar 30 65.7 36 10.8 50
Chocolate cookies 120 ± 24 61.3
Cookies 25 66.2 38 9.5 59 ± 2
Mini sponge cake 65 ± 17 63 60.7 28 17.6 46 ± 6
Shortbread 25 64.8 38 9.5 64 ± 8
Ice cream 96 ± 26 50 33.1 14 7 61 ± 7
Fruit or flavored yogurts 200 16.3 5 10 33 ± 7
Fruitcake 65 ± 17 55.7
Doughnuts 68 ± 20 42.0
Savoie sponge cake 63 68.3 26 16.4 46 ± 6
Chocolate muffin with bilberries 57 48.7 27 15.4 59
Fruit jelly or jam 30 60.0 33 9.9 51 ± 10
Dragees (chocolate and almond) 118 ± 26 52.0
M&M’s 30 60.1 17 5.1 33 ± 3
Mars bar 70 ± 25 60 79.2 41 24.6 65 ± 3
Sweetened popcorn 154 ± 40 20 62.0 45 9 72 ± 17
Snickers bar 60 60.2 23 13.8 55 ± 14
Chocolate milk 50 10.0 3 1.5 43 ± 3
Sweetened cocoa beverage 10.0 2 36
Sodas 10.0 7 63
Chips 91 ± 27 50 50.0 26 13 54 ± 3
Tortilla chips, salted 50 55.2 39 19.5 52
SEM, standard error of the means.
a
Satiety index of white bread = 100%; data from Holt et al.
19
For information, the sample size (n-value for
means) for each food can be found in the table by Holt et al.
19 b
Data were collected primarily from the 2013 French Ciqual database (available
online at: https://pro.anses.fr/tableciqual/index.htm); the rest were taken directly from the product labels of specific brands.
c
Data from Monro.
44
d
Glycemic index of glucose = 100; data for glycemic index from Foster-Powell et al.
20
For information, the sample size (n-value for means) for
each food can be found in the table by Foster-Powell et al.
20
When values were given without SEM, this means that they correspond to only one
measurement.
Paper Food & Function
2342 |Food Fun ct.,2016,7, 2338–2346 This journal is © The Royal Society of Chemistry 2016

Based on 89 foods, the GI, unlike the GGE, is not signifi-
cantly correlated with the processing group (Fig. 4, R
s
= 0.17,
P= 0.1166).
Relationship between glycemic glucose equivalent and satiety
index
Finally, based on 21 foods (Tables 1–3), the GGE is signifi-
cantly and inversely correlated with the SI (Fig. 5, R
s
=−0.58,
P= 0.0006). Therefore, foods with higher GGE content tend to
be less satiating.
Discussion
The aim of this study was to analyze the relationship among
the glycemic response, the satiety potential and the degree of
processing of 98 ready-to-eat foods. My hypothesis was that the
most processed foods are the least satiating and the higher the
glycemic response. I also wanted to investigate the relationship
between the GI and the GGE to investigate whether the GGE is
a better choice than the GI for evaluating the degree of food
processing. The GI is currently used primarily by dieticians to
advise their diabetic patients. In the present study, the values
of the GGE, GI and SI were determined from various tables,
where possible, for each food.
19,20,44
Then, the data were
correlated.
Fig. 2 Relationship between satiety index and processing groups (n=
33 foods).
Fig. 3 Relationship between GGE and processing groups (n=89
foods).
Fig. 4 Relationship between GI and processing groups (n= 89 foods).
Fig. 1 Relationship between GGE and GI (n= 83 foods).
Fig. 5 Relationship between GGE and satiety index (n= 21 foods).
Food & Function Paper
This journal is © The Royal Society of Chemistry 2016 Food Funct.,2016,7,2338–2346 | 2343

Although the number of foods is further reduced, the
results confirmed my hypothesis and showed strong corre-
lations between the GGE, the SI and the degree of food proces-
sing, whereas the GI is not correlated with the degree of
processing. Thus, the more the food is processed, the higher
the GGE and the lower its SI. This is therefore the first study
emphasizing significant correlations between the degree of pro-
cessing with satiety and glycemic potentials among 98 foods.
These results are in agreement with those of satiety studies
on certain foods or complex diets. For example, Haber et al.
showed in 10 healthy male and female volunteers that the
more an apple is unstructured (whole versus puréed versus
juice), the less satiating it is three hours after consumption.
24
Similarly, Flood-Obbagy and Rolls showed that whole apple
increased satiety more than applesauce or apple juice,
addition of naturally occurring levels of fiber to juice not
enhancing satiety;
45
and Gustafsson et al. reported a signifi-
cantly lower glycemic response and a significantly higher
satiety score with a raw carrot compared to carrots cooked in a
microwave for 10 healthy volunteers.
22
Besides, in 36 healthy
adult women, Moorhead et al. reported that, compared to a
meal containing the nutrients of the carrot (i.e., no matrix and
no fibers), meals including whole or mashed puréed carrots
resulted in significantly higher satiety over 3.5 hours after con-
sumption.
23
In addition to the structure effect, Haber et al.
24
and Moorhead et al.
23
specifically related satiety to the pres-
ence or absence of the fiber fraction.
Our findings point in the same direction as a study con-
ducted in 2009 in our laboratory in which we showed in
11 healthy volunteers that eating a less processed cereal break-
fast ( pre-fermented wheat flakes, not steamed and with less
sucrose) resulted in significantly higher satiety over 3 hours
compared to conventional and more processed wheat flakes.
46
Finally, it was shown in 14 adults with impaired glucose toler-
ance who consumed a breakfast (75 g available carbohydrate)
including either whole almonds, almond butter, defatted
almond flour, almond oil or no almonds that whole almonds
led to the most attenuated and delayed glycemic response,
continuing throughout the day, as well as to the strongest
feeling of satiety.
47
In the same way, on an equal-calorie basis (≈1600 kJ),
a recent study showed in 24 healthy adult subjects that a Paleo-
lithic diet (718 g), based on minimally processed foods that
therefore retained their structure more or less intact, was
approximately 4 times more satiating than the reference
control diet (248 g) based on white rice and containing 3 times
less fiber and 19 times less total polyphenols.
48
However, the
glycemic and insulinemic responses differed little between
these two diets.
Concerning bread, one of our basic staple foods, studies
show that the less processed breads that are denser and/or
contain more or less intact grains have a higher satiating
potential than typical white bread,
49,50
which demonstrates
the effect of the physical structure of that food on satiety.
Furthermore, our results show that, while the GGE is sig-
nificantly correlated with processing groups, the GI is not.
Although the number of foods analyzed is still limited, this
finding suggests that the GGE would be linked more closely to
food structure than the GI and therefore would better reflect
the impact on it of processing.
These results show a clear link between the degree of pro-
cessing, the satiating potential and the glycemic impact of
foods, which is in agreement with previous literature. There-
fore, in the absence of SI data, the GGE and processing group
(easier and more readily determined) constitute two valuable
indices for choosing foods, i.e., favoring processing groups
1 (MPF) and 2 (PF) with GGE less than 20 g per 100 g of food.
Obviously, these associations need to be confirmed for an even
greater number of foods.
The physical and structural characteristics of the food
matrix are therefore key players in health food potential
beyond nutritional composition.
51
In other words, for the
same nutritional composition, two different foods may give
very different glycemic and satiety responses, with particularly
important implications for diabetic individuals, which
suggests that we should encourage complex natural and mini-
mally processed foods over highly unstructured foods when
choosing foods with low glycemic response.
10
Our results also
showed that the satiating potential of a food should be a new
property considered in formulating or processing foods.
Pragmatically, since GGE values are significantly correlated
with both degree of processing (i.e., technological groups of
the NOVA classification) and SI values, and since SI values are
long and difficult to measure in humans, the use of GGE as a
food component on labelling (in g per 100 g) might be a first
rough and indirect reflection of the food satiety potential and
degree of processing. Otherwise, based on values from nutri-
ents that have been shown experimentally to have the greatest
impact on satiety, a Fullness Factor™has been developed to
calculate the satiating effect of a food, including protein, fat,
fiber and energy content (see at: http://nutritiondata.self.com/
topics/fullness-factor). Therefore, beyond the only nutrient
composition, indicating GGE (g per 100 g), Fullness Factor™
and NOVA group on food labelling or packaging would be an
important first step to help large public better choosing heal-
thier foods in a simple way.
The main limitation of this study was the number of foods
tested. However, all the main food groups were represented
(i.e., fruits, vegetables, legumes, cereals, nuts, dairy, meats and
snacks), and the chosen foods are quite representative of a
typical western diet. In addition, the three technological
groups all contained an adequate number of foods. Another
limitation was the small number of SI values (given for only 38
foods),
19
which restricts the generalizability of the correlations
with SI and other data. For example, the relationship between
the GGE and SI, even if significantly correlated, is based on
only 21 foods, and this combination therefore needs to be vali-
dated with a wider variety of foods. Finally, the NOVA classifi-
cation is based on degree of processing and end use (e.g.,
ingredient) rather than on structure per se, which might con-
tribute to the large amount of scatter in the data. If the foods
were to be classified using a more fine-grained system specifi-
Paper Food & Function
2344 |Food Funct.,2016,7, 2338–2346 This journal is © The Royal Society of Chemistry 2016

cally related to food structure, rather than to degree of proces-
sing –that is generally associated with loss of food structure –
the correlations would probably have been stronger. Unfortu-
nately, to the best of my knowledge, such a database relative to
physical parameters of food structure –notably as a function
of processing –does not yet exist worldwide. Such quantified
food structure parameters might be then correlated with the
degree of processing, e.g., hardness, softness, porosity, frag-
mentability and/or starch cristallinity, among others. In the
end, this may also provide more useful data for the use in
dietary management of glycaemic impact and satiety.
In conclusion, the main result of this study demonstrates
the important role played by the structure of a food on its
health characteristics, which is particularly useful in helping
diabetic individuals to choose protective foods rather than
basing a choice simply on GI. However, the implications can
also extend towards prevention of other chronic diseases for
all consumers. Otherwise, because degree of processing may
not always be apparent to consumers making food choices,
this paper again shows why a qualitative classification based
on processing, such as NOVA classification, is needed for
healthy food choices to be made.
Conflict of interest
None.
Abbreviations
GGE Glycemic glucose equivalent
GI Glycemic index
HOMA-IR Homeostasis model assessment of insulin
resistance
MPF Raw and minimally processed foods
PF Processed foods
SI Satiety index
UPF Ultra-processed foods
Acknowledgements
I acknowledge the Société Francophone du Diabète Paramédi-
cal (SFD Paramédical, Paris, France) and the French Associ-
ation of Nutritionist Dieticians (AFDN, Paris, France) for
supplying the list of foods consumed by diabetic individuals.
References
1 K. Esposito, C. M. Kastorini, D. B. Panagiotakos and
D. Giugliano, Metab. Syndr. Relat. Disord., 2010, 8, 471–476.
2 A. Alhazmi, E. Stojanovski, M. McEvoy and M. L. Garg,
J. Hum. Nutr. Diet., 2014, 27, 251–260.
3 E. Koloverou, K. Esposito, D. Giugliano and
D. Panagiotakos, Metabolism, 2014, 63, 903–911.
4 C. T. McEvoy, C. R. Cardwell, J. V. Woodside, I. S. Young,
S. J. Hunter and M. C. McKinley, J. Acad. Nutr. Diet., 2014,
114, 1759–1775.
5 L. Schwingshackl, B. Missbach, J. Koenig and
G. Hoffmann, Public Health Nutr., 2015, 18, 1292–1299.
6 Y. Yokoyama, N. D. Barnard, S. M. Levin and M. Watanabe,
Cardiovasc. Diagn. Ther., 2014, 4, 373–382.
7 C. M. Kastorini, H. J. Milionis, K. Esposito, D. Giugliano,
J. A. Goudevenos and D. B. Panagiotakos, J. Am. Coll.
Cardiol., 2011, 57, 1299–1313.
8 A. J. Nordmann, K. Suter-Zimmermann, H. C. Bucher,
I. Shai, K. R. Tuttle, R. Estruch and M. Briel, Am. J. Med.,
2011, 124, 841–851.
9 A. Fardet and Y. Boirie, Nutr. Rev., 2014, 72, 741–762.
10 A. Fardet, Méd. Mal. Métab., 2014, 8, 608–611.
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 D. Lioger, A. Fardet and C. Remesy, Cah. Nutr. Diet., 2007,
42, 309–319.
13 Q. Sun, D. Spiegelman, R. M. van Dam, M. D. Holmes,
V. S. Malik, W. C. Willett and F. B. Hu, Arch. Intern. Med.,
2010, 170, 961–969.
14 J. C. Moubarac, A. P. B. Martins, R. M. Claro, R. B. Levy,
G. Cannon and C. A. Monteiro, Public Health Nutr., 2013,
16, 2240–2248.
15 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.
16 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.
17 C. M. Luiten, I. H. Steenhuis, H. Eyles, C. Ni Mhurchu and
W. E. Waterlander, Public Health Nutr., 2016, 19, 530–538.
18 R. M. Bielemann, J. V. Santos Motta, G. C. Minten, B. L. Horta
and D. P. Gigante, Rev. Saude Publica, 2015, 49,28–28.
19 S. H. Holt, J. C. Miller, P. Petocz and E. Farmakalidis,
Eur. J. Clin. Nutr., 1995, 49, 675–690.
20 K. Foster-Powell, S. H. Holt and J. C. Brand-Miller,
Am. J. Clin. Nutr., 2002, 76,5–56.
21 K. Foster-Powell and J. B. Miller, Am. J. Clin. Nutr., 1995, 62,
871S–890S.
22 K. Gustafsson, N. G. Asp, B. Hagander and T. Schweizer,
Int. J. Food Sci. Nutr., 1995, 46,3–12.
23 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.
24 G. B. Haber, K. W. Heaton, D. Murphy and L. F. Burroughs,
Lancet, 1977, 2, 679–682.
25 D. J. A. Jenkins, T. M. S. Wolever, R. H. Taylor, H. Barker,
H. Fielden, J. M. Baldwin, A. C. Bowling, H. C. Newman,
A. L. Jenkins and D. V. Goff,Am. J. Clin. Nutr., 1981, 34,
362–366.
26 J. Salmeron, J. E. Manson, M. J. Stampfer, G. A. Colditz,
A. L. Wing and W. C. Willett, JAMA, J. Am. Med. Assoc.,
1997, 277, 472–477.
Food & Function Paper
This journal is © The Royal Society of Chemistry 2016 Food Funct.,2016,7,2338–2346 | 2345

27 J. Salmeron, A. Ascherio, E. B. Rimm, G. A. Colditz,
D. Spiegelman, D. J. Jenkins, M. J. Stampfer, A. L. Wing
and W. C. Willett, Diabetes Care, 1997, 20, 545–550.
28 J. A. Monro and M. Shaw, Am. J. Clin. Nutr., 2008, 87, 237S–
243S.
29 J. A. Monro, Asia Pac. J. Clin. Nutr., 2002, 11, 217–225.
30 F. Rauber, P. D. B. Campagnolo, D. J. Hoffman and
M. R. Vitolo, Nutr., Metab. Cardiovasc. Dis., 2015, 25, 116–
122.
31 L. F. Tavares, S. C. Fonseca, M. L. Garcia Rosa and
E. M. Yokoo, Public Health Nutr., 2012, 15,82–87.
32 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.
33 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.
34 P. V. L. Moreira, L. G. Baraldi, J.-C. Moubarac,
C. A. Monteiro, A. Newton, S. Capewell and M. O’Flaherty,
PLoS One, 2015, 10, e0118353.
35 J.-C. Moubarac, D. C. Parra, G. Cannon and C. A. Monteiro,
Curr. Obes. Rep., 2014, 3, 256–272.
36 C. A. Monteiro, R. B. Levy, R. M. Claro, I. R. R. de
Castro and G. Cannon, Cad. Saúde Pública, 2010, 26,
2039–2049.
37 Ministry of Health of Brazil, Primary Health Care Depart-
ment, Dietary guigelines for the Brazilian population, Sec-
retariat of Health Care, São Paulo, 2014.
38 C. A. Monteiro, G. Cannon, R. B. Levy, J.-C. Moubarac,
P. Jaime, A. P. Martins, D. Canella, M. Louzada and
D. Parra, World Nutr., 2016, 7,28–38.
39 C. A. Monteiro, G. Cannon and J.-C. Moubarac, World
Nutr., 2014, 5, 140–168.
40 C. A. Monteiro, J.-C. Moubarac, G. Cannon, S. W. Ng and
B. Popkin, Obes. Rev., 2013, 14,21–28.
41 C. A. Monteiro, R. B. Levy, R. M. Claro, I. R. R. de Castro
and G. Cannon, Public Health Nutr., 2011, 14,5–13.
42 C. A. Monteiro, World Nutr., 2010, 1, 237–269.
43 C. A. Monteiro, Public Health Nutr., 2009, 12, 729–731.
44 J. Monro, Glycaemic Glucose Equivalents in New Zealand
Foods, New Zealand Institute for Plant & Food Research,
2004.
45 J. E. Flood-Obbagy and B. J. Rolls, Appetite, 2009, 52, 416–
422.
46 D. Lioger, A. Fardet, P. Foassert, M.-J. Davicco, J. Mardon,
B. Gaillard-Martinie and C. Remesy, J. Am. Coll. Nutr., 2009,
28,30–36.
47 A.Mori,R.ConsidineandR.Mattes,Nutr. Metab., 2011, 8,6.
48 H. F. J. Bligh, I. F. Godsland, G. Frost, K. J. Hunter,
P. Murray, K. MacAulay, D. Hyliands, D. C. S. Talbot,
J. Casey, T. P. J. Mulder and M. J. Berry, Br. J. Nutr., 2015,
113, 574–584.
49 J. Keogh, F. Atkinson, B. Eisenhauer, A. Inamdar and
J. Brand-Miller, Appetite, 2011, 57, 707–710.
50 P. Burton and H. J. Lightowler, Br. J. Nutr., 2006, 96, 877–
882.
51 A. Fardet, Food Funct., 2015, 6, 363–382.
Paper Food & Function
2346 |Food Funct.,2016,7,2338–2346 This journal is © The Royal Society of Chemistry 2016