Ultra-processed foods: A new holistic paradigm?

Journal Pre-proof Ultra-processed foods: a new holistic paradigm? Anthony Fardet, Edmond Rock

PII:

S0924-2244(19)30102-5

DOI:

https://doi.org/10.1016/j.tifs.2019.09.016

Reference:

TIFS 2636

To appear in:

Trends in Food Science & Technology

Received Date: 9 February 2019 Revised Date:

14 September 2019

Accepted Date: 18 September 2019

Please cite this article as: Fardet, A., Rock, E., Ultra-processed foods: a new holistic paradigm?, Trends in Food Science & Technology, https://doi.org/10.1016/j.tifs.2019.09.016. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2019 Published by Elsevier Ltd.

1

Ultra-processed foods: a new holistic paradigm?

2 3 4

Anthony Fardet1, Edmond Rock1

5 6

1

7

F-63000 Clermont-Ferrand, France

Université Clermont Auvergne, INRA, UNH, Unité de Nutrition Humaine, CRNH Auvergne,

8 9 10 11

*Corresponding author: Dr. Anthony Fardet, Université Clermont Auvergne, INRA, UNH,

12

Unité de Nutrition Humaine, CRNH Auvergne, F-63000 Clermont-Ferrand, France; tel. +33

13

(0)4 73 62 47 04, fax +33 (0)4 73 62 47 55, email: [email protected]

14 15 16

Short title: Ultra-processed foods: a holistic concept

17 18 19 20

Abbreviations: UPF, Ultra-processed food

21

1

22

Abstract

23

Background: The concept of ultra-processed foods (UPFs) is new, and it was proposed for the

24

first time in 2009 as group 4 of the NOVA classification to address the degree of food

25

processing. UPFs include not only “junk foods” but also foods marketed as healthy, such as

26

light, vegan, organic, or gluten-free products. UPFs are characterized by the presence of

27

highly-processed/purified “cosmetic” ingredients and/or additives to restore and/or exacerbate

28

organoleptic properties, i.e., taste, aroma, color and texture. Substantial industrial processing

29

techniques, e.g., puffing, extrusion cooking, and/or extreme fractioning/refining that greatly

30

breakdown the food matrix, may also be markers of ultra-processing. The UPF concept has

31

been consistently criticized for being an overly heterogeneous concept, and the NOVA

32

classification has been criticized for being qualitative only and too imprecise.

33

Scope and approach: This review is intended to discuss the UPF concept from a holistic

34

perspective and to analyze the scientific soundness of criticisms about UPFs and NOVA. The

35

UPF concept is first defined; then, its primary nutritional characteristics are described,

36

followed by their association with health based on human studies.

37

Key findings and conclusions: UPF criticisms differ between holistic and reductionist

38

perspectives. In a holistic concept, reductionist researchers view the proposed definition of

39

UPF as an imprecise, vague and heterogeneous technological group. However, from a holistic

40

perspective, the UPF concept has serious advantages, such as broad and common deleterious

41

health attributes (i.e., the loss of “matrix” effect, empty calories, poorly satiating,

42

hyperglycemic and containing artificial compounds foreign to the human body).

43 44

Keywords: Ultra-processed foods, food classification, holism, health potential, human studies

2

45 46

1. Introduction The act of ranking foods is not harmless, notably in relation to epidemiological studies

47

and the subsequent national dietary guidelines that are derived from them, and the potential

48

impact on public health area (Fardet et al., 2015). Generally, to date, foods have been ranked

49

according to their nutrient content, e.g., foods rich in protein such as fish, meat, dairy and

50

legumes and foods rich in sugars, salt and fat (as in many dietary pyramids worldwide), or

51

their botanical and animal origins, e.g., white and red meats, fruits, vegetables, cereals,

52

legumes, nuts, fish, dairy and eggs (Fardet et al., 2015), or other classifications (ANSES,

53

2017). In 2014, Brazil was the first country in the world to release dietary guidelines for

54

populations based on the degree of food processing (Ministry of Health of Brazil, 2014). They

55

distinguished un/minimally processed foods, culinary ingredients, processed foods and ultra-

56

processed foods (UPF) in what they call the “NOVA” (i.e., “new” in Portuguese)

57

classification (Monteiro et al., 2018).

58

Before NOVA, the degree of food processing was rarely taken into account during

59

binary comparisons such as whole versus refined grains (Aune et al., 2013), red versus

60

processed meat (Farvid et al., 2018), low fat versus whole dairy products (Benatar, Karishma,

61

& Stewart, 2013), or fruit versus fruit juices versus sweetened fruit juices (Imamura et al.,

62

2016). In epidemiological studies, the different relative risks of some chronic diseases were

63

calculated according to these first binary comparisons, with a generally higher risk when

64

foods were more processed, as shown in particular for fruit-based products (Fardet,

65

Richonnet, & Mazur, 2019). However, the degree of processing was not previously

66

considered within national dietary recommendations. Currently, it is clear that despite fifty

67

years of preventive nutrition and national dietary guidelines, the prevalence of chronic

68

diseases continues to increase worldwide, and several experts and researchers have recently

69

emphasized that unbalanced nutrition and its associated risk factors or deregulated

3

70

metabolisms are the leading cause of deaths and chronic noncommunicable diseases

71

worldwide (Development Initiatives, 2018; GBD 2015 Risk Factors Collaborators, 2016;

72

GBD 2013 Mortality and Causes of Death Collaborators, 2015). Notably, the WHO reported

73

that the proportion of deaths due to noncommunicable disease is projected to rise from 59% in

74

2002 to 69% in 2030 (Mathers & Loncar, 2006); and the GBD 2017 reported that in 2015 a

75

total of 107.7 million children and 603.7 million adults were obese, and that high BMI

76

accounted for 4.0 million deaths globally, nearly 40% of which occurred in persons who were

77

not obese (The GBD 2015 Obesity Collaborators, 2017). Where did we fail? Could our

78

nutrient-based approach be incorrect (Fardet & Rock, 2018)?

79

The concept of UPF is new, and it was first proposed in 2009 when, after observing

80

the substantial increase in obesity and type 2 diabetes prevalence within only a few years,

81

Brazilian epidemiologists suggested that the degree of food processing was overlooked in

82

favor of the nutrient contents of foods (Monteiro, 2009). In fact, we first consume complex

83

foods, not nutrients. Today, we have sufficient scientific data to recognize that the food health

84

potential combines both “matrix” and “composition” effects, and processing first impacts the

85

food matrix (Fardet & Rock, 2018). Since then, numerous papers using the NOVA

86

classification have been published in various contexts, i.e., epidemiological studies and

87

investigations of food consumption, food composition and nutrient density among others, as

88

previously reviewed (Monteiro et al., 2018). The accumulating evidence (see section 4. for

89

details) from these studies tends to show that a high consumption of UPFs is deleterious to

90

health, notably with increased risks of some chronic diseases and poor nutritional density

91

(Monteiro et al., 2018; 2019).

92

Since 2009, the concept of the UPF has been repeatedly criticized by the agro-food

93

industry as well as by some academic researchers (Gibney, Forde, Mullally, & Gibney, 2017;

94

Jones, 2018), but we cannot rule out the presence of potential conflict of interest as previously

4

95

reviewed (Mialon, Sêrodio, & Scagliusia, 2018). The primary criticisms are a lack of

96

scientific evidence for the validity of NOVA classification and UPF concept, the fact that it

97

does not address food nutrient compositions and that the UPF group is very heterogeneous in

98

terms of products and their composition (Mialon et al., 2018). Therefore, the objective of this

99

review is to present the state of the art about UPFs, notably regarding their exact definition,

100

their nutritional characteristics, and finally their link to health, while clarifying some issues or

101

criticisms within the context of a more holistic approach compared to the reductionist one.

102

Briefly, in contrast with reductionism, holism in nutrition is the way of seeing foods as whole

103

complex matrices, not only a sum of nutrients (Fardet & Rock, 2014). Therefore, a holistic

104

approach can be viewed as a integrated and global approach while reductionist approach is

105

more specific, and intend to study separately each part of the complex system. This can be

106

also applied to food classifications, being either global (few food groups) or very specific

107

(several food groups).

108 109

2. The concept of UPF

110

2.1 Definitions

111

The concept of UPF first appears in the scientific literature in 2009, in a paper by Monteiro et

112

al. (Monteiro, 2009). In 2014, the NOVA classification was officially released in a systematic

113

review about different food classifications around the world that addressed the degree of

114

processing (Moubarac, Parra, Cannon, & Monteiro, 2014). The definition of UPF given by

115

Monteiro et al. was as follows:

116 117

“Ultra-processed foods, such as soft drinks, sweet or savoury packaged snacks,

118

reconstituted meat products and pre-prepared frozen dishes, are not modified foods

5

119

but formulations made mostly or entirely from substances derived from foods and

120

additives, with little if any intact Group 1 food (i.e., unprocessed or minimally processed

121

foods). Ingredients of these formulations usually include those also used in processed

122

foods, such as sugars, oils, fats or salt. But UPFs also include other sources of energy and

123

nutrients not normally used in culinary preparations. Some of these are directly extracted

124

from foods, such as casein, lactose, whey and gluten. Many are derived from further

125

processing of food constituents, such as hydrogenated or interesterified oils, hydrolysed

126

proteins, soya protein isolate, maltodextrin, invert sugar and high-fructose corn syrup.

127

Additives in ultra-processed foods include some also used in processed foods, such as

128

preservatives, antioxidants and stabilizers. Classes of additives found only in UPFs

129

include those used to imitate or enhance the sensory qualities of foods or to disguise

130

unpalatable aspects of the final product. These additives include dyes and other colours,

131

colour stabilizers; flavours, flavour enhancers, non-sugar sweeteners; and processing aids

132

such as carbonating, firming, bulking and anti-bulking, de-foaming, anti-caking and

133

glazing agents, emulsifiers, sequestrants and humectants. A multitude of sequences of

134

processes is used to combine the usually many ingredients and to create the final product

135

(hence ‘ultra-processed’). The processes include several with no domestic equivalents,

136

such as hydrogenation and hydrolysation, extrusion and moulding, and pre-processing for

137

frying. The overall purpose of ultra-processing is to create branded, convenient (durable,

138

ready to consume), attractive (hyper-palatable) and highly profitable (low-cost

139

ingredients) food products designed to displace all other food groups. Ultra-processed

140

food products are usually packaged attractively and marketed intensively” (Monteiro et al.,

141

2018).

142

6

143

In brief, UPFs are characterized by the modification of their matrix to render it highly

144

palatable through the use of industrial aromas, taste exhausters, and coloring and texturizing

145

agents. This definition of UPF also coincides with that given by Poti et al. for what they have

146

called “highly processed foods”, and adapted from NOVA:

147 148

“Multi-ingredient industrially formulated mixtures processed to the extent that they are no

149

longer recognizable as their original plant/animal source and not typically consumed as

150

additions” (Poti, Mendez, Ng, & Popkin, 2015).

151 152

Therefore, the definition of UPF is first based on the impact of processing on the food matrix

153

and its sensory properties, rather than the nutrient content.

154

UPFs are more than junk foods and may include foods marketed as “healthy” for

155

consumers, such as light, low-fat, low-sugar, low-salt, vegan, gluten/lactose-free and organic

156

products (Figure 1). It must also be emphasized that not all industrialized foods are ultra-

157

processed. Therefore, UPFs are only a part, but significant, of industrial foods.

158 159

2.2 How many technological groups can be created? Holism versus reductionism

160

Ultimately, a relevant question could be about the correct number of technological groups to

161

create. This number likely depends on our approach to technology, i.e., holistic or reductionist

162

(Fardet, 2018). If we think very holistically, a binary comparison is possible, including

163

unprocessed versus processed foods (Figure 2). This approach is probably unrealistic because

164

almost all foods are processed in one way or another. To eat unprocessed foods alone would

165

signify a return to the epoch of hunter-gatherers. Even in that case, cooking meat is still a

166

process. In contrast, if we employ a very reductionist perspective to unravel the technological

167

route of a product, a high number of technological groups can be created, including each

7

168

process parameter, i.e., the time, temperature, pressure, ingredients, etc. Again, this approach

169

also seems unrealistic because a food is generally the result of several technological

170

processes, each one acting differently on each food nutrient; therefore, this double complexity

171

is out of reach for researchers. In addition, each year, thousands of new UPFs are marketed

172

worldwide, rendering the study of their technological route an endless process. In addition,

173

much of the food processing information is not given on food packaging by the industry.

174

Therefore, if it is useful to deconstruct these food technological treatments to expand our

175

scientific knowledge and reinforce our expertise it is not useful to provide assistance to our

176

society and to make recommendations to the large public.

177

NOVA is a still very holistic classification, such that it distinguishes among four

178

qualitative technological groups, but it is far more realistic than the binary comparison of

179

“unprocessed versus processed foods”. It involves minimally processed foods, culinary

180

ingredients, processed foods combining minimally processed foods and culinary ingredients,

181

and UPFs including purified/highly processed ‘cosmetic’ ingredients/additives to exacerbate

182

organoleptic food properties. In any case, the NOVA system has been used extensively by

183

academic researchers since the 2010s, and it is acknowledged in reports, statements and

184

commentary from the FAO, WHO and PAHO, and in leading scientific journals (Monteiro et

185

al., 2018). On the ISI Web of Knowledge, more than 180 papers include the term ‘UPF’ in

186

title. However, because it is very holistic, UPF group 4 may appear too heterogeneous upon

187

first consideration; this assumption seems to be both “wrong” and “right”. It is “wrong”

188

because despite being apparently very heterogeneous, these foods have common attributes

189

(see section 3.), but it is “right” if we think in a more reductionist manner because we can find

190

all types of foods in the UPF classification, from confectionary, complex dishes, and sodas to

191

dairy products, meat-based foods and biscuits with very different compositions, and foods

192

with various numbers of ultra-processing markers, from one to more than fifteen. However,

8

193

heterogeneity also comes from different brands, not only from the nutritional composition and

194

markers of ultra-processing. Therefore, additional reductionist NOVA subgroups might be

195

developed to create a derived classification system for industrial foods. The new more or less

196

reductionist parameters could be the degree of ingredient processing (e.g., honey versus table

197

sugar versus glucose syrup, virgin versus refined versus hydrogenated oils, or protein extract

198

versus purified protein versus hydrolyzed protein), the “matrix” food effect, number, function,

199

and risk from additives, and the sugar/fat/salt contents (Fardet, 2018). This classification

200

would finally combine the four holistic NOVA groups with additional reductionist subgroups

201

based on a previously led discussion on the interconnection of holistic and reductionist

202

thinking in nutrition research (Figure 3) (Fardet & Rock, 2018).

203

It should emphasize that NOVA classification is clear, useful, understandable and

204

simple to apply. These potential additional groups are not intended to substitute for the

205

NOVA classification and its use as a health public tool, but might serve as a derived

206

classification for other purposes, notably for industrial purposes and niche contexts to develop

207

less processed foods, through a more gradual approach.

208

It is also possible, based on a thorough reductionist approach, to develop a very

209

detailed food classification based on the degree of processing that includes more than ten

210

groups. The food technological route can be very detailed, which is undoubtedly a relevant

211

study that would improve our knowledge about the impact of processing on food health

212

potential, but it would probably be inapplicable to the larger public in terms of assisting with

213

food choices.

214 215

3. Common UPF characteristics despite apparent heterogeneity

216

First, UPFs are very convenient (ready to eat products), aggressively marketed (highly

217

attractive and hyper palatable), and highly profitable (low cost ingredients, almost

9

218

imperishable). As suggested above, despite their apparent heterogeneity, UPFs also exhibit

219

other relevant common nutritional properties as follows:

220 221

3.1 Breakdown of food and ingredient matrices Recently, we developed a new definition of food health potential based on the most

222 223

recent scientific results (Fardet & Rock, 2018). It combines both the “matrix” and

224

“compositional” effect, with “matrix” being the food qualitative and holistic fraction, which

225

comes first, because we first eat matrices and not nutrients, and the “composition” is the

226

quantitative and reductionist fraction, which comes second. The health potential of UPFs

227

should be evaluated with respect to this definition because two foods with the same

228

composition but different matrices or structures do not have the same health effects (Fardet &

229

Rock, 2018). Therefore, the breakdown of foods and ingredients during ultra-processing is a

230

crucial parameter to consider. A good analogy is the structure of a pearl necklace; when the

231

thread connects the beads, the necklace is analogous to complex whole foods, but when the

232

thread is isolated from the beads, it is not the same as the entire pearl necklace, even if it has

233

the same composition. The idea behind this analogy is that the thread plays a role within food

234

health potential, i.e., the link between nutrients, but not so much about isolated nutrients as

235

such. For example, the starch and fully hydrolyzed starch that are processed into glucose

236

syrup have the same calorie and nutrient composition but not the same health effects because

237

the links between the glucose units have been broken down. The same reasoning can be

238

applied for intact versus finely ground almonds (Grundy, Lapsley, & Ellis, 2016), among

239

others.

240

UPFs are characterized by either food breakdown through dramatic processes or the

241

use of unstructured, purified, fractionated and/or highly processed ingredients, e.g., esterified

242

and hydrogenated oils, fructose-glucose syrup or isolated, purified and/or hydrolyzed

10

243

proteins. Even though they have equal carbohydrate and calorie compositions, it is not the

244

same to consume whole cane sugar, refined table sugar or fructose-glucose syrup.

245

The fractionation of food or ingredients is therefore the first common attribute and is

246

very characteristic of UPFs. These products are part of a reductionist vision of food when

247

considered as only the sum of nutrients, with no consideration of the “link” between nutrients

248

(Fardet & Rock, 2014, 2018). Indeed, if foods are only the sums of their nutrients, why not

249

fractionate them?

250 251 252

3.2 “Empty” calories The term “empty calories” designates foods with poor nutritional density in fiber and

253

micronutrients, and that are rich in energy derived from macronutrients, i.e., lipids, proteins

254

and carbohydrates (Nestle, 2000; Rugg-Gunn, Hackett, Jenkins, & Appleton, 1991; Walker &

255

Cannon, 1984). It is consistently stated that UPFs provide “empty” calories with a poor and

256

very low nutritional density, or at least lower than what is found in normally and/or minimally

257

processed foods (Cornwell et al., 2018; Crovetto, Uauy, Martins, Moubarac, & Monteiro,

258

2014; Fardet, Méjean, Labouré, Andreeva, & Féron, 2017; Julia et al., 2018; Louzada et al.,

259

2015a; Martínez Steele, Popkin, Swinburn, & Monteiro, 2017; Moubarac, Batal, Louzada,

260

Martinez Steele, & Monteiro, 2016; Rauber et al., 2018) unless they have been enriched with

261

added purified fiber, minerals, and vitamins, notably to correct highly refined and/or

262

fractionated products. However, added micronutrients are not the same as the micronutrients

263

that are naturally present in complex food matrices (Fardet, 2015a, 2015b; Talvas et al.,

264

2010). Notably, the natural synergy between micronutrients, such as antioxidants, is lost when

265

purified micronutrients are added (often at high doses), and they are detached from their

266

‘natural’ food matrix environment (Fardet, 2015a, 2015b), as it has been notably reviewed for

267

citrus fruits and vitamin C (Rock & Fardet, 2014), and eloquently shown with β-carotene

11

268

supplementation on increased cancer risk (Omen et al., 1996; The Alpha-Tocopherol, Beta

269

Carotene Cancer Prevention Study Group, 1994)

270

“Empty” calories, e.g., sweetened beverages or refined grain-based products, have

271

consistently been shown to be associated with an increased risk of chronic diseases (Joint

272

WHO/FAO Expert Consultation, 2003). Fiber and micronutrients are protective compounds

273

for preventing chronic diseases. They notably include antioxidants, anti-inflammatory and

274

anti-carcinogenic compounds, hypoglycemic and hypolipidemic compounds, and lipotropes.

275

Since chronic diseases are multifactorial and include several deregulated metabolisms at their

276

original root causes (Fardet & Boirie, 2013), it is not surprising that the regular consumption

277

of “empty” calories, and consequently micronutrient-poor UPFs, are associated with increased

278

risks of chronic diseases (see section 4.).

279

For example, we recently studied an elderly French population and found that UPFs

280

have an 81% lower nutritional density than minimally processed foods (the nutrient density

281

score decreased from 19.7 to 3.8, a score based on qualifying nutrients only) and the LIM

282

score (the mean percentage of the maximum recommended values for 3 disqualifying

283

nutrients) increased from 3.6 to 23.6 (Fardet, Méjean, Labouré, Andreeva, & Féron, 2017).

284

Similarly, Moubarac et al. studied 33,694 individuals from the 2004 Canadian Community

285

Health Survey aged 2 years and above, showing that consumption of UPFs predicts diet

286

quality, i.e., the higher the UPF consumption is, the lower the consumption of vitamins, fiber

287

and minerals (Moubarac, Batal, Louzada, Martinez Steele, & Monteiro, 2016). The same

288

results were obtained in 2,898 Brazilian subjects aged 10 years or older (Louzada, Martins, et

289

al., 2015a).

290 291

3.3 Poorly satiating

12

292

Ultra-processed foods were also less satiating than minimally processed foods (Figure

293

4) (Fardet, Méjean, Labouré, Andreeva, & Féron, 2017). Notably, before the UPF

294

classification was conceptualized, refined bakery products, snacks, and confectionary had a

295

significantly lower satiety index (< 100) in 11-13 healthy subjects than less processed,

296

protein- and carbohydrate-rich foods and fruits (> 165) (Holt, Miller, Petocz, & Farmakalidis,

297

1995). UPFs are richer in fat and sugars, and they are less satiating while being poorer in

298

protein and fiber, the more satiating nutrients (Blundell, 2010; Tremblay & Bellisle, 2015). In

299

fact, fiber and protein are very characteristic of raw foods, and protein and fiber notably

300

provide the structure of plant- and animal-based foods, respectively. In addition, solid foods

301

are more satiating than semisolid/viscous and liquid foods (Chambers, 2016; Fiszman &

302

Tarrega, 2017). The fractionated-recombined UPFs exhibit more liquid, semisolid and friable

303

structures, requiring less chewing time and subsequently lower satiety hormone stimulation

304

(Chambers, 2016). In his review, Chambers concluded, “As a final thought, the findings

305

described in this paper hint that the textural complexity of a healthy, varied diet (plenty of

306

fruit and vegetables and wholegrains, some dairy or dairy alternatives, and some beans,

307

pulses, fish, eggs, lean meat and other proteins) could mean that, in additional to all the other

308

health benefits, it is more satiating (and therefore more protective against overconsumption)

309

than diets based on processed foods and energy-containing beverages” (Chambers, 2016).

310

Lower satiety potentials in foods and diets may therefore lead to further increased

311

dietary intake between meals, notably of other UPFs, and satiety is a crucial parameter of

312

balanced dietary intakes (Chambers, 2016; Tremblay & Bellisle, 2015). The high contents of

313

added fat, sugars and salt in UPFs together with their hedonic attributes may lead consumers

314

to continue eating these foods even when satiated because the “pleasure to eat” overcomes the

315

feeling of satiety.

316

13

317 318

3.4 Hyperglycemia Based on Foster-Powell et al.’s glycemic index tables (Foster-Powell, Holt, & Brand-

319

Miller, 2002) and the data on one hundred ready-to-eat foods commonly consumed by the

320

French diabetic population, UPFs also tend to be more hyperglycemic than minimally

321

processed foods (Figure 5) (Fardet, 2016). The explanation behind this property is obvious. In

322

returning to the holistic definition of the food health potential combining “matrix” and

323

“composition” effects (Fardet & Rock, 2018), the breakdown of the food matrix through

324

dramatic processes (e.g., extrusion cooking and puffing) and ingredient recombination in

325

UPFs renders starch and sugars more easily accessible to digestive enzymes, i.e., α-amylases

326

and α-glucosidases, leading to a more rapid glucose release within the blood. Otherwise,

327

UPFs are often enriched with added simple sugars such as dextrose, glucose syrup, glucose-

328

fructose syrup, table sugar, etc., all of which are highly available for digestion.

329

Consequently, the combination of low satiety and high glycemic potentials in UPFs is

330

a potential characteristic that favors weight gain and type 2 diabetes through excess simple

331

sugar consumption and/or rapidly available sugars (see section 4).

332 333 334

3.5 Artificial ingredients/additives/compounds Another characteristic of UPFs, although not strictly exclusive to UPFs, is their

335

contents of artificial and other compounds that are new to the human organism. These are

336

primarily compounds that are newly formed under drastic processes, e.g., acrylamide,

337

artificial ingredients (e.g., purified/hydrolyzed proteins, refined sugars, hydrogenated fats,

338

etc.), artificial aroma (more than 2,800 are used in France) and some artificial additives that

339

humans have not previously encountered. Therefore, since the 1980s, the global human

340

population has been consuming massive amounts of new artificial compounds, primarily

341

“cosmetic” ones, and little is known about their effects on cell function and human health

14

342

over the long term, not to mention the “cocktail” effect and/or potential impact on the

343

digestive epithelium (Csaki, 2011; Lerner & Matthias, 2015) and gut microbiota (Roca-

344

Saavedra et al., 2018). Unfortunately, there are no long-term studies in humans on the

345

potential health impacts of these new artificial compounds, either alone or in combination.

346 347

4. Ultra-processed foods, health and chronic diseases

348

The association between UPF consumption and the risk of chronic diseases is under

349

investigation, although more than twenty studies have already been published since the

350

release of the NOVA classification at the beginning of the 2010s. Beyond NOVA, there were

351

studies about foods belonging to UPFs, but without them being named as such, e.g., sodas

352

(Wijarnpreecha, Thongprayoon, Edmonds, & Cheungpasitporn (2016), artificially sweetened

353

fruit juices (Imamura et al., 2016), etc. The epidemiological studies published on UPFs are of

354

heterogeneous quality, but the conclusions and accumulating scientific evidence all note the

355

potential detrimental effects of these foods (Table 1).

356 357

4.1 Ecological studies The scientific weight of ecological studies is low, but their results may provide some

358 359

indicative tendencies that can be used for more sound epidemiological studies. Concerning

360

UPFs, three studies have been published using NOVA classification, one on nineteen

361

European countries (Monteiro et al., 2017), one on eighty countries (Vandevijvere et al.,

362

2019) and the other on fourteen American countries (Pan American Health Organization,

363

2015).

364

In the European study, the average household availability of UPFs ranged from 10.2%

365

in Portugal and 13.4% in Italy to 46.2% in Germany and 50.4% in the UK (Monteiro et al.,

366

2017). A significant positive association was found between the national household

15

367

availability of ultra-processed foods from 1991 to 2008 according to country and the national

368

prevalence of obesity among adults. The adjusted R2 value was 0.63 with a P value of 0.02.

369

Among the different countries, Belgium and Portugal are exceptions since despite a high UPF

370

availability in Belgium (approximately 45% of available calories), the prevalence of obesity

371

was only ≈12%, and conversely, despite the low UPF availability in Portugal (approximately

372

10% of calories), the prevalence of obesity was quite high, at ≈15%.

373

In the American study, the annual retail sales per capita of UPFs and drink products

374

and the prevalence of obesity in 2013 (%) were correlated (R2 = 0.76, P < 0.001) for fourteen

375

countries (all Latin American countries except Argentina, plus Canada and the United States)

376

(Pan American Health Organization, 2015). After controlling for confounders (gross national

377

income, urbanization, and deregulation), the association remained significant

378

(R2 = 0.84; P < 0.001). The correlation was stronger for these countries than for European

379

ones, and no outlier country was found.

380

In the most recent study, increases in ultra-processed foods volume sales/capita were

381

significantly and positively associated with population level BMI trajectories (by

382

Vandevijvere et al., 2019).

383

More generally, in viewing the highest levels of retail sales for UPF and beverages

384

around the world in 2013 that corresponded to ten countries (i.e., USA, Canada, Germany,

385

Mexico, Belgium, Australia, Norway, UK, Japan, Switzerland, and the Netherlands, as given

386

by the Pan American Health Organization (Pan American Health Organization, 2015)), we

387

found that they are positively correlated with the obesity prevalence in 2015 (OECD, 2017)

388

(R2 = 0.48, P = 0.02; unpublished data).

389

Clearly, it must be noted that this is a correlation, not a causation, and that many other

390

factors are involved in the increased risks of obesity worldwide such as increased inactivity,

391

pollution, genetic factors, stress, and/or loneliness, since obesity is a multifactorial chronic

16

392

disease. However, because an unbalanced diet is the leading cause of deaths worldwide

393

(approximately 19% of all deaths) (GBD 2015 Risk Factors Collaborators, 2016), the effect of

394

UPFs on increased obesity prevalence should be seriously considered among all these factors.

395 396

4.2 Cross-sectional studies

397

Excess weight (Juul, Martinez-Steele, Parekh, Monteiro, & Chang, 2018), obesity (Canella et

398

al., 2014; Louzada et al., 2015b; Nardocci et al., 2019) and metabolic syndrome (Steele, Juul,

399

Neri, Rauber, & Monteiro, 2019; Lavigne-Robichaud et al., 2018; Tavares, Fonseca, Garcia

400

Rosa, & Yokoo, 2012; Nasreddine et al., 2017) were the most frequently studied of chronic

401

diseases in a cross-sectional design.

402

In the first study, which included 15,977 US adults aged 20-64 years, “consuming ≥

403

74.2 versus ≤ 36.5% of total energy from UPFs” was associated with 1.61 units of higher

404

body mass index (BMI) (95% CI: 1.11, 2.10), a 4.07 cm greater waist circumference (95% CI:

405

2.94, 5.19) and 48, 53 and 62% higher odds of having a BMI ≥ 25 kg/m2, BMI ≥ 30 kg/m2

406

and abdominal obesity, respectively (OR 1.48; 95% CI: 1.25, 1.76; OR 1.53; 95% CI: 1.29,

407

1.81; OR 1.62; and 95% CI: 1.39, 1.89, respectively; P for trend < 0.001 for all) (Juul et al.,

408

2018). The BMI and waist circumference increased linearly with each quintile of UPF

409

consumption, from 36.5 to ≥ 74.2%. Similar results were obtained in a sample of 55,970

410

Brazilian households from the 2008-2009 Household Budget Survey, indicating that the

411

household availability of ultra-processed products was positively associated with both the

412

average BMI and the prevalence of excess weight and obesity (Canella et al., 2014). All the

413

prevalence data increased linearly from quartile 1 (1.6%-21.8% energy from UPFs) to quartile

414

3 (28.5%-34.8%) and then reached a plateau at quartile 4 (34.8%-54.9%). In another cross-

415

sectional study on 30,243 individuals aged ≥ 10 years from the 2008-2009 Brazilian Dietary

416

Survey, the same research team found that mean differences in the BMI increased linearly

17

417

with each quintile of UPF consumption (% of total energy), that odds ratio for being obese

418

increased significantly from 14% energy from UPFs, and the odds ratio for bearing excess

419

weight increased significantly from 23% energy from UPFs (Louzada et al., 2015b). The last

420

study included 19,363 adults aged 18 years or more from the 2004 Canadian Community

421

Health Survey, and using multivariate logistic regression models, it primarily showed that

422

UPF consumption was positively associated with obesity (Nardocci et al., 2019). The main

423

results showed that a ten-percentage point increase in the relative energy intake from UPFs

424

increased the likelihood of obesity by 5% (OR = 1.05, 95% CI: 1.01, 1.08).

425

Concerning the risk of metabolic syndrome (measured as prevalence ratio), it was

426

significantly increased by 2.5 in Brazilian adolescents who consumed high levels of UPF (≥

427

3rd quartile, i.e., more than 1 245 g of UPFs daily versus low consumers < 1 245 g daily)

428

(Tavares, Fonseca, Garcia Rosa, & Yokoo, 2012). More recently, Lavigne-Robichaud et al.

429

studied the Eeyouch adults (n = 811) from seven communities on Eastern James Bay for their

430

risk of metabolic syndrome (Lavigne-Robichaud et al., 2018). In a comparison of the highest

431

(83%) versus lowest (21%) quintiles for the contribution of UPFs to the total daily dietary

432

energy intake, the results showed that the adjusted odds ratio of metabolic syndrome was 1.90

433

(95% CI: 1.14, 3.17).

434

In the US adult population (6 385 participants from the cross-sectional National Health

435

and Nutrition Examination Survey 2009-2014, aged ≥ 20 years) a significant linear

436

association between the dietary contribution of UPF and the prevalence of metabolic

437

syndrome was found (a 10% increase in contribution was associated with a 4% prevalence

438

increase) (prevalence ratio = 1.04; 95% CI: 1.02, 1.07) (Steele, Juul, Neri, Rauber, &

439

Monteiro, 2019).

440 441

Finally, using data from a representative sample of 109,104 Brazilian adolescents enrolled in the National Survey of School Health (2012), it was reported that the UPF

18

442

consumption score was positively associated (P < 0.05 for linear trend between the categories

443

of UPF consumption) with the presence of asthma and wheezing in a dose-response manner

444

(Melo, Rezende, Machado, Gouveia, & Levy, 2018). The adjusted OR of asthma and

445

wheezing comparing highest (≥ 5 days/week) to lowest (0-2 days/week) quintile of ultra-

446

processed consumption scores (based on frequency of UPF consumption/week) were 1.27

447

(95% CI: 1.15, 1.41) and 1.42 (95% CI: 1.35, 1.50), respectively. According to the authors,

448

additives and excess weight are the primary possible mediators of the association between

449

UPFs and asthma.

450 451

4.3 Prospective studies

452

It is now important to check whether the results of the cross-sectional studies are supported,

453

or not, by longitudinal prospective cohort studies.

454

Regarding overweight and obesity, 8,451 middle-aged Spanish university graduates

455

(from the prospective Spanish cohort, the SUN (University of Navarra Follow-Up), who were

456

initially not overweight or obese, were followed for a median of 8.9 years (Mendonca et al.,

457

2016). After adjustments for potential confounders, participants in the highest quartile of UPF

458

consumption (6.1 servings/day) were at a higher risk of developing overweight or obesity

459

(adjusted HR: 1.26; 95% CI: 1.10, 1.45; P for trend = 0.001) than those in the lowest quartile

460

of consumption (1.5 servings/day). The HR began to increase significantly from 2.7

461

servings/day (adjusted HR: 1.15; 95% CI: 1.01, 1.32).

462

In addition to obesity, the relationship between UPF consumption and the risk of

463

hypertension in a prospective Spanish cohort, i.e., the Seguimiento Universidad de Navarra

464

project, was studied after a mean follow-up of 9.1 years in 14,790 Spanish adult university

465

graduates who were initially free of hypertension at the baseline (Mendonca et al., 2017).

466

Again, after adjusting for potential confounders, participants in the highest tertile of UPF

19

467

consumption (5.0 servings/d) had a higher risk of developing hypertension (adjusted HR,

468

1.21; 95% CI: 1.06, 1.37; and P for trend = 0.004) than those in the lowest tertile (2.1

469

servings/d).

470

Unlike obesity, the lipid profile has not been studied thoroughly. Only one study has

471

addressed this issue (Rauber, Campagnolo, Hoffman, & Vitolo, 2015). That study was limited

472

to 345 children of low socioeconomic status from São Leopoldo, Brazil. A linear regression

473

analysis was used to assess the relationship between processed products and UPF intake at 3-4

474

years on changes in lipid concentrations from preschool to school age at 7-8 years. The

475

primary result showed that UPF consumption at preschool age was a predictor of a higher

476

increase in total cholesterol (β = 0.430; P = 0.046) and LDL cholesterol (β =0.369; P = 0.047)

477

from preschool to school age; but the effect was at the limit of significance. Another study

478

about anthropometric and glucose profiles followed the same schema as the previous study,

479

with 307 children of low socioeconomic status from São Leopoldo, Brazil, who were

480

followed between the ages of 4 and 8 years (Costa et al., 2018). The adjusted linear regression

481

analyses showed that UPF consumption at preschool age was a predictor of an increase in the

482

delta waist circumference from preschool to school age (β = 0.07; 95% CI 0.01-0.14; P =

483

0.030), but not for glucose metabolism. Further work is clearly needed to complete these two

484

studies.

485

Concerning type 2 diabetes, there is no study using the NOVA classification, but only

486

studies on sweetened beverages, which are ranked as UPFs within the NOVA classification. It

487

is not within the scope of this paper to review all the studies about sweetened beverage

488

consumption and the risk of type 2 diabetes. A recent meta-analysis about this association

489

including eight prospective studies, 286,697 participants, and 29,264 cases of type 2 diabetes

490

was chosen (Wang, Yu, Fang, & Hu, 2015). The reported pooled risk was 1.30 (95% CI: 1.20,

20

491

1.39, no significant heterogeneity between studies) when comparing the highest versus lowest

492

intake of sugar-sweetened beverages.

493

Recently, five prospective studies from the French NutriNet-Santé cohort were

494

released. The first one was about cancer risk and UPF consumption and included 104,980

495

participants aged at least 18 years (Fiolet et al., 2018). The authors reported that a 10%

496

increase in the proportion of UPF in the diet was associated with a significant increase of in

497

the risks of overall cancer (+10%, 95%CI: 1.03, 1.17) and breast cancer (+15%, 95%CI: 1.03,

498

1.29); conversely, the +10% consumption of unprocessed/minimally processed foods was

499

associated with a 9% lower risk of overall cancers (P < 0.001) and a 58% lower risk of breast

500

cancer (P = 0.03). Interestingly, the results remained statistically significant after making

501

adjustments for several markers of the nutritional quality of the diet (notably the lipid,

502

sodium, and carbohydrate intakes), suggesting that other parameters may be included in this

503

effect. The authors proposed several hypotheses to explain these results, such as the generally

504

poorer nutritional quality of diets rich in UPFs, the use of a wide range of additives contained

505

in UPFs for which “the effect on health of the cumulative intake across all ingested foods and

506

potential cocktail/interaction effects remain largely unknown”, and the potential presence of

507

neoformed contaminants due to drastic heat treatments. In the second prospective study, the

508

association between UPF consumption and gastro-intestinal disorders (irritable bowel

509

syndrome, functional constipation, functional diarrhea, and functional dyspepsia) was

510

investigated in 33,343 participants from the web-based NutriNet-Santé cohort (Schnabel et

511

al., 2018). After adjustments for confounding factors, an increase in the UPF was associated

512

with a 25% higher risk of irritable bowel syndrome (OR Q4 versus Q1 = 1.25, 95%CI: 1.12,

513

1.39, P for trend < 0.0001) and a 25% higher risk of functional dyspepsia (OR Q4 versus Q1

514

= 1.25, 95%CI: 1.05, 1.47, P for trend = 0.004), and the risk significantly increased from

515

14.5-20.6% UPF in weight for both gastrointestinal disorders. No association was found for

21

516

functional constipation and functional diarrhea. The third study has investigated the

517

association between UPF consumption and mortality (Schnabel et al., 2019), finding among

518

44 551 participants that an increase in the proportion of ultra-processed foods (by weight)

519

consumed was associated with a higher risk of all-cause mortality (HR per 10% increment,

520

1.14; 95% CI: 1.04, 1.27; P = 0.008). In the fourth study about UPF consumption and incident

521

depressive symptoms the estimated hazard ratio for a 10% increase in UPF by weight was

522

1.21 (95% CI: 1.15, 1.27) (Adjibade et al., 2019). In the last and most recent study among 105

523

159 participants aged at least 18 years and followed upon on average 5.2 years, each

524

increment of 10% by weight of UPF consumption was positively and significantly associated

525

with a +12% increased risk of cardiovascular diseases (95% CI:1.05, 1.20), and conversely

526

with minimally-processed foods (-9% reduced risk for each 10% increment by weight) (Srour

527

et al., 2019). These results were confirmed in three recent other prospective studies

528

associating high UPF consumption with significant increased risks of mortality (Kim, Hu, &

529

Rebholz, 2019; Rico-Campà et al., 2019) and the incidence of depression (Gomez-Donoso et

530

al., 2019) in both The Third National Health and Nutrition Examination Survey and Spanish

531

SUN cohort. UPF consumption was also related to gestational weight gain and neonatal outcomes

532 533

in a sample of US pregnant women based on data from a longitudinal study performed in

534

20132014 at the Women's Health Center and Obstetrics & Gynecology Clinic in St. Louis

535

(Rohatgi et al., 2017). Results showed that percent of energy intake from UPF was associated

536

with, and may be a useful predictor of increased gestational weight gain and neonatal body

537

fat.

538

Beyond usual chronic diseases, UPF consumption is also strongly associated with

539

frailty risk in older adults (p for linear trend: <0.001 with Q4 versus Q1 OR = 3.67 [2.00-

540

6.73]) (Sandoval-Insausti et al., 2019).

22

Although it did not use the NOVA classification, a recent prospective study pointed

541 542

out the impact of processing on plant-based food health potential (Satija et al., 2017). In a

543

pooled multivariable analysis, higher adherence to a healthful plant-based diet (whole grains,

544

fruits/vegetables, nuts/legumes, oils, tea/coffee) was inversely associated with the coronary

545

heart disease risk (HR: 0.75; 95% CI: 0.68 to 0.83; P for trend = 0.001), whereas an

546

unhealthful plant-based diet (juices/sweetened beverages, refined grains, potatoes/fries,

547

sweets) was positively associated with coronary heart disease risk (HR: 1.32; 95% CI: 1.20 to

548

1.46; P for trend = 0.001).

549 550

4.4 Systematic reviews

551

A systematic review by Costa et al. confirms that the degree of food processing matters when

552

defining the food health potential (Costa, Del-Ponte, Assunção, & Santos, 2018). Twenty-six

553

studies that evaluated groups of UPFs (as defined by NOVA, such as snacks, fast foods, junk

554

foods and convenience foods) or specific UPFs (soft drinks/sweetened beverages, sweets,

555

chocolate and ready-to-eat cereals) were selected for their association with body fat during

556

childhood and adolescence in healthy subjects. Fifteen studies had a cohort design and the

557

others were primarily interventional studies, with only three cross-sectional studies. The

558

authors reported that most of the studies found positive associations between UPF

559

consumption and body fat, and all the studies that showed this association had a longitudinal

560

design.

561 562

4.5 Interventional studies

563

Up today there is only one interventional study led with UPF (Hall et al., 2019). Twenty

564

weight-stable (BMI = 27 ±1.5) inpatient adults received ultra-processed and unprocessed diets

565

for 14 days each, each diet being matched for presented calories, sugar, fat, fiber, and

23

566

macronutrients. Main results showed that ad libitum intake was ≈500 kcal/day (+20% on 14

567

days, coming mainly from fat and carbohydrates) more on the ultra-processed versus

568

unprocessed diet, and that body weight changes were highly correlated with diet differences

569

in energy intake (+1.1 kg on ultra-processed diet, and -0.9 kg on minimally-processed diet).

570

Probably the most important point to emphasize is that these effects were not due first to

571

compositions of the diets because both were matched for this. The effect has to be attributed

572

to other factors, notably the loss of the “matrix” effect in UPF, which are generally more

573

viscous, liquid, friable and soft matrices demanding less chewing, and therefore leading to a

574

lower feeling of satiety (Chambers, 2016), and also to the presence of cosmetic

575

ingredients/additives amplifying sensory properties of UPF, such as taste exhausters leading

576

to a higher caloric consumption, pleasure to eat outweighing satiety.

577

Otherwise, older interventional studies showed that the degree of food processing

578

matters with respect to the effects on nutrient bioavailability, satiety, other metabolisms, and

579

subsequently, food health potential (Berry et al., 2008; Burton & Lightowler, 2006;

580

Gaudichon et al., 1995; Granfeldt, Bjorck, & Hagander, 1991; Haber, Heaton, Murphy, &

581

Burroughs, 1977; Moorhead et al., 2006).

582

Hundreds of UPFs are marketed each year around the world, and therefore it is not

583

possible to test all the products in humans through interventional designs. The gold standard

584

intervention study would be to follow two groups of volunteers for several years, with one

585

group consuming minimally processed and processed foods as their dietary basis and the

586

other primarily consuming UPFs. However, it is very likely that this type of study will never

587

be undertaken, at least for ethical reasons. However, it would be relevant to select a large

588

group of subjects that already has a diet based on UPFs (i.e., probably more than 60% of daily

589

calories) and to intervene and decrease their consumption of UPFs and observe their

24

590

physiological parameters over the short term and their risk of chronic diseases over the longer

591

term.

592 593

4.6 Other studies

594

As suggested in the introduction, before the release of NOVA, the degree of processing has

595

primarily been considered in epidemiological studies through binary comparisons (e.g., whole

596

versus refined grains, red versus processed meat). However, these studies provided the first

597

interesting findings, notably, that whole-grain cereal-based products are protective against

598

chronic diseases and refined cereal-based products are either neutral or deleterious (Fardet &

599

Boirie, 2014). In addition, a recent systematic review of meta-analyses showed a

600

technological gradient for the health potential of fruit-based products, with whole complex

601

fruits (raw or dried) being more protective than 100% fruit juices or sweetened fruit juices

602

(Fardet, Richonnet, & Mazur, 2019). Regarding animal-based foods, the processed meat

603

intake was significantly positively related to overall mortality, while red meat intake was not

604

(van den Brandt, 2019). It is beyond the scope of this article to review all these studies, but

605

the abovementioned references have suggested a processing effect beyond the NOVA

606

classification.

607

In the end, it is worth mentioning two other studies that employ the NOVA

608

classification, but for other purposes related to health than the sole study of the associations

609

between UPF consumption and the risk of chronic diseases. In the first study, modeling

610

methodology was used to find that reducing the processed culinary ingredients and ultra-

611

processed foods in the Brazilian diet may reduce cardiovascular disease mortality by 5.5%

612

(the intakes of saturated fat, trans-fat, salt and added sugar from UPFs and processed culinary

613

ingredients were reduced by one quarter), 11.0% (the reduction of 50% of the same nutrients

614

in ultra-processed foods and processed culinary ingredients) and 29.0% (the reduction of the

25

615

same nutrients in UPFs by 75% and in processed culinary ingredients by 50%) (Moreira et al.,

616

2018).

617

The second study, again using the NOVA classification, explored the consumption of

618

UPFs and its association with food addiction (Yale Food Addiction Scale for Children) in

619

overweight children (Filgueiras et al., 2018). The primary results showed a significant

620

tendency towards the higher consumption of added sugar (refined sugar, honey, and corn

621

syrup) and UPF among children diagnosed with food addiction.

622 623

5. Conclusions and future trends

624

The concept of UPF is undoubtedly holistic, notably because processing impacts food as a

625

whole complex matrix, not only the nutrients (Fardet & Rock, 2018).

626

Why is it very important to be more holistic? Because the more holistic the food index

627

is, the broader and more global the impact. If we focus only on food nutrients (reductionist

628

thinking), we will have little impact on health, since chronic diseases are not being linked to

629

the excessive consumption of one nutrient. If infectious and deficiency-related diseases have

630

justified a reductionist approach because they are based on one identifiable causal factor, i.e.,

631

toxin intrusion and micronutrient deficiency, respectively, then the prevention of chronic

632

disease does not. Multifactoriality demands a more holistic approach. In another field, if the

633

same reasoning is used to address food system sustainability, from a holistic perspective, the

634

issue is not about animal versus plant proteins, but rather, animal versus plant-based foods,

635

with this latter comparison including environment protection and animal well-being rather

636

than focusing only on proteins. Therefore, the global challenge of food system sustainability

637

cannot be addressed with reductionist indicators or a nutrient-based approach.

638 639

In this way, the NOVA classification and UPF concept are quite holistic and, therefore, well adapted to the interface with society for nutritional guidelines and public

26

640

health policies, as has been done in Brazil (Monteiro et al., 2015; Ministry of Health of Brazil,

641

2014) and Uruguay (Ministry of Health of Uruguay, 2016). Indeed, it appears that a simpler

642

and more qualitative approach than a reductionist and quantitative one based on a nutrient-

643

based approach is more efficient in terms of facility in communication and implementation of

644

recommendations. Consequently, based on scientific evidence, it is clear that UPFs

645

consumption should be limited, similarly what is currently being done to prevent the use of

646

tobacco, alcohol or other drugs. Structural measures like mandatory front-package NOVA

647

classification should be implemented in the future.

648

Additionally, there is clearly a need for further studies using food classification based

649

on the degree of processing, notably longitudinal ones with a cohort design. As concluded by

650

Costa et al., “There is a need to use a standardized classification that considers the level of

651

food processing to promote comparability between studies” (Costa et al., 2018). However, all

652

the results summarized in this paper clearly show that the foods in epidemiological studies

653

should first be classified according to their degree of processing, not on a nutrient basis or on

654

their animal/plant origins, as previously thoroughly discussed and demonstrated (Monteiro,

655

Levy, Claro, de Castro, & Cannon, 2010; Moubarac, Parra, Cannon, & Monteiro, 2014;

656

Fardet et al., 2015). It is obvious that it is not the same to consume a whole intact piece of

657

boiled fish, canned fish with oil or fish nuggets containing more than 15 “cosmetic”

658

additives/ingredients. This conclusion is common sense. These examples could be repeated

659

infinitely with all the foods supplied by our environment. This is not to say that the food

660

composition or nutrient approach does not matter, but only for the second intention mentioned

661

above, not the first intention, all simply because we eat complex foods first, not nutrients. The

662

holistic matrix environment of nutrients is essential, meaning that it is not the same to

663

consume fructose from empty calorie sources such as sodas as it is to consume natural

27

664

fructose from complex fruits accompanied by satiety (due to chewing and the fiber matrix),

665

minerals, fiber, vitamins, antioxidants and the generally slower release of simple sugars. Finally, there seems to be a link between the high level of fractionation of our natural

666 667

foods and the explosion of chronic diseases in response to this excessive fractionation. The

668

destruction of the harmony within foods seems to have destroyed the harmony within the

669

human body, which reacts through a whole myriad of chronic diseases, intolerances,

670

hypersensitivity and other syndromes. In addition, if the price of UPFs can seem low for our

671

wallet, it is actually excessively expensive in terms of public health costs, and for the planet,

672

humankind, animals and the environment included.

673 674 675

Conflicts of interest

676

Anthony Fardet is member of the scientific committee of the Siga society since 2017.

677 678 679

Funding

680

There is no funding source for this work.

681 682 683

Acknowledgements

684

None.

685 686 687

References

28

688

Adjibade, M., Julia, C., Allès, B., Touvier, M., Lemogne, C., Srour, B., Hercberg, S., Galan,

689

P., Assmann, K. E., & Kesse-Guyot, E. (2019). Prospective association between ultra-

690

processed food consumption and incident depressive symptoms in the French NutriNet-

691

Santé cohort. BMC Medicine, 17(1), 78.

692

ANSES, French Agency for Food, Environmental and Occupational Health & Safety (2017).

693

Étude individuelle nationale des consommations alimentaires 3 (INCA 3). Anses Éditions.

694

Aune, D., Norat, T., Romundstad, P., & Vatten, L. J. (2013). Whole grain and refined grain

695

consumption and the risk of type 2 diabetes: a systematic review and dose-response meta-

696

analysis of cohort studies. European Journal of Epidemiology, 28(11), 845-858.

697

Benatar, J. R., Karishma, S., & Stewart, R. A. H. (2013). Effects of high and low fat dairy

698

food on cardio-metabolic risk factors: a meta-analysis of randomized studies. Plos One,

699

8(10), e76480-e76480.

700

Berry, S. E., Tydeman, E. A., Lewis, H. B., Phalora, R., Rosborough, J., Picout, D. R., &

701

Ellis, P. R. (2008). Manipulation of lipid bioaccessibility of almond seeds influences

702

postprandial lipemia in healthy human subjects. American Journal of Clinical Nutrition,

703

88(4), 922-929.

704 705 706 707

Blundell, J. (2010). Making claims: functional foods for managing appetite and weight. Nature Reviews Endocrinology, 6(1), 53-56. Burton, P., & Lightowler, H. J. (2006). Influence of bread volume on glycaemic response and satiety. British Journal of Nutrition, 96(5), 877-882.

708

Canella, D. S., Levy, R. B., Martins, A. P. B., Claro, R. M., Moubarac, J. C., Baraldi, L. G.,

709

Cannon, G., & Monteiro, C. A. (2014). Ultra-Processed Food Products and Obesity in

710

Brazilian Households (2008-2009). Plos One, 9(3), e92752.

711

Canhada, S., Luft, V. C., Giatti, L., Duncan, B. B., Chor, D., Fonseca, M. J. M., Matos, & et

712

al., (2019). Ultra-processed Foods, Incident Overweight and Obesity, and Longitudinal

29

713

Changes in Weight and Waist Circumference? Public Health Nutrition. [Accepted in

714

Public Health Nutrition]

715

Chambers, L. (2016). Food texture and the satiety cascade. Nutrition Bulletin, 41(3), 277-282.

716

Cornwell, B., Villamor, E., Mora-Plazas, M., Marin, C., Monteiro, C. A., & Baylin, A.

717

(2018). Processed and ultra-processed foods are associated with lower-quality nutrient

718

profiles in children from Colombia. Public Health Nutrition, 21(1), 142-147.

719

Costa, C. S., Del-Ponte, B., Assunção, M. C. F., & Santos, I. S. (2018). Consumption of ultra-

720

processed foods and body fat during childhood and adolescence: a systematic review.

721

Public Health Nutrition, 21(1), 148-159.

722

Costa, C. S., Rauber, F., Leffa, P. S., Sangalli, C. S., Campagnolo, P. D. B., & Vitolo, M. R.

723

(2018). Ultra-processed food consumption and its effects on anthropometric and glucose

724

profile: a longitudinal study during childhood. Nutrition, Metabolism and Cardiovascular

725

Diseases, 29(2), 177-184.

726

Crovetto, M. M., Uauy, R., Martins, A. P., Moubarac, J. C., & Monteiro, C. (2014).

727

Household availability of ready-to-consume food and drink products in Chile: impact on

728

nutritional quality of the diet. Revista Médica de Chile, 142(7), 850-858.

729 730 731 732

Csaki, K. F. (2011). Synthetic surfactant food additives can cause intestinal barrier dysfunction. Medical Hypotheses, 76(5), 676-681. Development Initiatives (2018). 2018 Global nutrition report shining a light to spur action on nutrition. Development Initiatives Poverty Research Ltd.

733

FAO, Monteiro, C. A., Cannon, G., Lawrence, M., da Costa Louzada, M. L., and Machado, P.

734

P. (2019). Ultra-processed foods, diet quality, and health using the NOVA classification

735

system. Rome, Italy.

30

736

Fardet, A. (2015a). Complex foods versus functional foods, nutraceuticals and dietary

737

supplements: differential health impact (Part 1). Agro Food Industry Hi-Tech, 26(1), 20-

738

24.

739

Fardet, A. (2015b). Complex foods versus functional foods, nutraceuticals and dietary

740

supplements: differential health impact (Part 2). Agro Food Industry Hi-Tech, 26(3), 20-

741

22.

742 743 744

Fardet, A. (2018). Vers une classification des aliments selon leur degré de transformation : approches holistique et/ou réductionniste. Pratiques en Nutrition, 14(56), 32-36. Fardet, A., & Boirie, Y. (2013). Associations between diet-related diseases and impaired

745

physiological mechanisms: a holistic approach based on meta-analyses to identify targets

746

for preventive nutrition. Nutrition Reviews, 71(10), 643-656.

747

Fardet, A. (2016). Minimally processed foods are more satiating and less hyperglycemic than

748

ultra-processed foods: a preliminary study with 98 ready-to-eat foods. Food & Function,

749

7(5), 2338-2346.

750

Fardet, A., C. Méjean, H. Labouré, V. A. Andreeva, & Féron, G. (2017). The degree of

751

processing of foods which are most widely consumed by the French elderly population is

752

associated with satiety and glycemic potentials and nutrient profiles. Food & Function,

753

8(2), 651-658.

754

Fardet, A., Richonnet, C., & Mazur, A. (2019). Association between fruit or processed fruit

755

product consumption, chronic diseases and their risk factors: a systematic review of meta-

756

analyses. Nutrition Reviews, 77(6), 376-387.

757

Fardet, A., & Rock, E. (2014). Toward a new philosophy of preventive nutrition: from a

758

reductionist to a holistic paradigm to improve nutritional recommendations. Advances in

759

Nutrition, 5(4), 430-446.

31

760 761

Fardet, A., & Rock, E. (2018). Reductionist nutrition research has meaning only within the framework of holistic thinking. Advances in Nutrition, 9(6), 655–670.

762

Fardet, A., Rock, E., Bassama, J., Bohuon, P., Prabhasankar, P., Monteiro, C., Moubarac, J.-

763

C., & Achir, N. (2015). Current food classifications in epidemiological studies do not

764

enable solid nutritional recommendations to prevent diet-related chronic diseases: the

765

impact of food processing. Advances in Nutrition, 6(6), 629-638.

766

Fardet, A., & Boirie, Y. (2014). Associations between food and beverage groups and major

767

diet-related chronic diseases: an exhaustive review of pooled/meta-analyses and systematic

768

reviews. Nutrition Reviews, 72(12), 741-762.

769

Farvid, M. S., Stern, M. C., Norat, T., Sasazuki, S., Vineis, P., Weijenberg, M. P., Wolk, A.,

770

Wu, K., Stewart, B. W., & Cho, E. (2018). Consumption of red and processed meat and

771

breast cancer incidence: a systematic review and meta-analysis of prospective studies.

772

International Journal of Cancer, 143(11), 2787-2799.

773

Filgueiras, A. R., Pires de Almeida, V. B., Koch Nogueira, P. C., Alvares Domene, S. M.,

774

Eduardo da Silva, C., Sesso, R., & Sawaya, A. L. (2019). Exploring the consumption of

775

ultra-processed foods and its association with food addiction in overweight children.

776

Appetite, 135, 137-145.

777

Fiolet, T., Srour, B., Sellem, L., Kesse-Guyot, E., Allès, B., Méjean, C., Deschasaux, M.,

778

Fassier, P., Latino-Martel, P., Beslay, M., Hercberg, S., Lavalette, C., Monteiro, C. A.,

779

Julia, C., & Touvier, M. (2018). Consumption of ultra-processed foods and cancer risk:

780

results from NutriNet-Santé prospective cohort. BMJ, 360, k322.

781 782 783 784

Fiszman, S., & Tarrega, A. (2017). Expectations of food satiation and satiety reviewed with special focus on food properties. Food & Function, 8(8), 2686-2697. Foster-Powell, K., Holt, S. H., & Brand-Miller, J. C. (2002). International table of glycemic index and glycemic load values: 2002. American Journal of Clinical Nutrition, 76(1), 5-56.

32

785

Gaudichon, C., Mahe, S., Roos, N., Benamouzig, R., Luengo, C., Huneau, J. F., Sick, H.,

786

Bouley, C., Rautureau, J., & Tome, D. (1995). Exogenous and endogenous nitrogen flow-

787

rates and level of protein hydrolysis in the human jejunum after N-15 milk and N-15

788

yogurt ingestion. British Journal of Nutrition, 74(2), 251-260.

789

GBD 2015 Risk Factors Collaborators (2016). Global, regional, and national comparative risk

790

assessment of 79 behavioural, environmental and occupational, and metabolic risks or

791

clusters of risks, 1990-2015: a systematic analysis for the Global Burden of Disease Study

792

2015. Lancet, 388(10053), 1659-1724.

793

GBD 2013 Mortality and Causes of Death Collaborators. (2015). Global, regional, and

794

national age–sex specific all-cause and cause-specific mortality for 240 causes of death,

795

1990–2013: a systematic analysis for the Global Burden of Disease Study 2013. The

796

Lancet, 385(9963), 117-171.

797

Gibney, M. J., Forde, C. G., Mullally, D., & Gibney, E. R. (2017). Ultra-processed foods in

798

human health: a critical appraisal. The American Journal of Clinical Nutrition, 106(3),

799

717-724.

800

Gomez-Donoso, C., Sanchez-Villegas, A., Martinez-Gonzales, M. A., Gea, A., Mendoca, R.

801

D., Lahortiga-Ramos, F., & Bes-Rsatrollo, M. (2019). Ultra-processed food consumption

802

and the incidence of depression in a Mediterranean cohort: the SUN Project. European

803

Journal of Nutrition, DOI: 10.1007/s00394-019-01970-1.

804

Granfeldt, Y., Bjorck, I., & Hagander, B. (1991). On the importance of processing conditions,

805

product thickness and egg addition for the glycaemic and hormonal responses to pasta: a

806

comparison with bread made from 'pasta ingredients'. European Journal of Clinical

807

Nutrition, 45(10), 489-499.

33

808

Grundy, M. M., Lapsley, K., & Ellis, P. (2016). A review of the impact of processing on

809

nutrient bioaccessibility and digestion of almonds. International Journal of Food Science

810

& Technology, 51(1937-1946.

811

Haber, G. B., Heaton, K. W., Murphy, D., & Burroughs, L. F. (1977). Depletion and

812

disruption of dietary fibre. Effects on satiety, plasma-glucose, and serum-insulin. Lancet,

813

2(8040), 679-682.

814

Hall, K. D., Ayuketah, A., Brychta, R., Cai, H., Cassimatis, T., Chen, K. Y., Chung, S. T.,

815

Costa, E., Courville, A., Darcey, V., Fletcher, L. A., Forde, C. G., Gharib, A. M., Guo, J.,

816

Howard, R., Joseph, P. V., McGehee, S., Ouwerkerk, R., Raisinger, K., Rozga, I.,

817

Stagliano, M., Walter, M., Walter, P. J., Yang, S., & Zhou, M. (2019). Ultra-Processed

818

Diets Cause Excess Calorie Intake and Weight Gain: An Inpatient Randomized Controlled

819

Trial of Ad Libitum Food Intake. Cell Metabolism, 30, 1-11.

820 821 822

Holt, S. H., Miller, J. C., Petocz, P., & Farmakalidis, E. (1995). A satiety index of common foods. European Journal of Clinical Nutrition, 49(9), 675-690. Imamura, F., O'Connor, L., Ye, Z., Mursu, J., Hayashino, Y., Bhupathiraju, S. N., & Forouhi,

823

N. G. (2016). Consumption of sugar sweetened beverages, artificially sweetened

824

beverages, and fruit juice and incidence of type 2 diabetes: systematic review, meta-

825

analysis, and estimation of population attributable fraction. British Journal of Sports

826

Medicine, 50(8), 496-504.

827 828 829 830

Joint WHO/FAO Expert Consultation (2003). Diet, nutrition and the prevention of chronic diseases. In WHO Technical Report Series 916. Jones, J. M. (2018). Food processing: criteria for dietary guidance and public health? Proceedings of the Nutrition Society, 78(1), 4-18.

34

831

Julia, C., Martinez, L., Alles, B., Touvier, M., Hercberg, S., Mejean, C., & Kesse-Guyot, E.

832

(2018). Contribution of ultra-processed foods in the diet of adults from the French

833

NutriNet-Sante study. Public Health Nutrition, 21(1), 27-37.

834

Juul, F., Martinez-Steele, E., Parekh, N., Monteiro, C. A., & Chang, V. W. (2018). Ultra-

835

processed food consumption and excess weight among US adults. British Journal of

836

Nutrition, 120(1), 90-100.

837

Kim, H., Hu, E. A., & Rebholz, C. M. (2019). Ultra-processed food intake and mortality in

838

the USA: results from the Third National Health and Nutrition Examination Survey

839

(NHANES III, 1988-1994). Public Health Nutrition, DOI: 10.1017/S1368980018003890.

840

Lavigne-Robichaud, M., Moubarac, J.-C., Lantagne-Lopez, S., Johnson-Down, L., Batal, M.,

841

Laouan Sidi, E. A., & Lucas, M. (2018). Diet quality indices in relation to metabolic

842

syndrome in an Indigenous Cree (Eeyouch) population in northern Québec, Canada. Public

843

Health Nutrition, 21(1), 172-180.

844

Lerner, A., & Matthias, T. (2015). Changes in intestinal tight junction permeability associated

845

with industrial food additives explain the rising incidence of autoimmune disease.

846

Autoimmun Reviews, 14(6), 479-489.

847

Louzada, M. L., Martins, A. P., Canella, D. S., Baraldi, L. G., Levy, R. B., Claro, R. M.,

848

Moubarac, J. C., Cannon, G., & Monteiro, C. A. (2015a). Impact of ultra-processed foods

849

on micronutrient content in the Brazilian diet. Revista de Saude Publica, 49, 45.

850

Louzada, M. L., Baraldi, L. G., Steele, E. M., Martins, A. P., Canella, D. S., Moubarac, J. C.,

851

Levy, R. B., Cannon, G., Afshin, A., Imamura, F., Mozaffarian, D., & Monteiro, C. A.

852

(2015b). Consumption of ultra-processed foods and obesity in Brazilian adolescents and

853

adults. Preventive Medicine, 81, 9-15.

35

854

Martínez Steele, E., Popkin, B. M., Swinburn, B., & Monteiro, C. A. (2017). The share of

855

ultra-processed foods and the overall nutritional quality of diets in the US: evidence from a

856

nationally representative cross-sectional study. Population Health Metrics, 15(1), 6.

857

Mathers, C. D., & Loncar, D. (2006). Projections of global mortality and burden of disease

858 859

from 2002 to 2030. PLoS Medicine, 3(11), e442. Melo, B., Rezende, L., Machado, P., Gouveia, N., & Levy, R. (2018). Associations of ultra-

860

processed food and drink products with asthma and wheezing among Brazilian

861

adolescents. Pediatric Allergy and Immunology, 29(5), 504-511.

862

Mendonca, R. D., Lopes, A. C. S., Pimenta, A. M., Gea, A., Martinez-Gonzalez, M. A., &

863

Bes-Rastrollo, M. (2017). Ultra-Processed Food Consumption and the Incidence of

864

Hypertension in a Mediterranean Cohort: The Seguimiento Universidad de Navarra

865

Project. American Journal of Hypertension, 30(4), 358-366.

866

Mendonca, R. D., Pimenta, A. M., Gea, A., de la Fuente-Arrillaga, C., Martinez-Gonzalez, M.

867

A., Lopes, A. C. S., & Bes-Rastrollo, M. (2016). Ultraprocessed food consumption and

868

risk of overweight and obesity: the University of Navarra Follow-Up (SUN) cohort study.

869

American Journal of Clinical Nutrition, 104(5), 1433-1440.

870 871 872

Mialon, M., Sêrodio, P., & Scagliusia, F. B. (2018). Criticism against the NOVA classification: who are the protagonists? World Nutrition, 9(3), DOI: 10.26596. Ministry of Health of Brazil, Secretariat of Health Care, Primary Health Care Department

873

(2014). Dietary guidelines for the Brazilian population. In E. A. M. Editora Senac (Ed.), 80

874

pages. São Paulo.

875 876 877 878

Ministry of Public Health of Uruguay (2016). Dietary guidelines for the Uruguayan population: for a healthy, shared and enjoyable diet. 52 pages. Montevideo. Monteiro, C. A. (2009). Nutrition and health. The issue is not food, nor nutrients, so much as processing. Public Health Nutrition, 12(5), 729-731.

36

879

Monteiro, C. A., Levy, R. B., Claro, R. M., de Castro, I. R. R., & Cannon, G. (2010). A new

880

classification of foods based on the extent and purpose of their processing. Cadernos de

881

Saúde Pública, 26(11), 2039-2049.

882

Monteiro, C. A., Cannon, G., Moubarac, J. C., Martins, A. P., Martins, C. A., Garzillo, J.,

883

Canella, D. S., Baraldi, L. G., Barciotte, M., Louzada, M. L., Levy, R. B., Claro, R. M., &

884

Jaime, P. C. (2015). Dietary guidelines to nourish humanity and the planet in the twenty-

885

first century. A blueprint from Brazil. Public Health Nutrition, 18(13), 2311-2322.

886

Monteiro, C. A., Cannon, G., Moubarac, J.-C., Bertazzi Levy, R., Louzada, M. L. C., &

887

Jaime, P. C. (2018). The UN Decade of Nutrition, the NOVA food classification and the

888

trouble with ultra-processing. Public Health Nutrition, 21(1), 5-17.

889

Monteiro, C. A., Moubarac, J.-C., Levy, R. B., Canella, D. S., Louzada, M. L. d. C., &

890

Cannon, G. (2017). Household availability of ultra-processed foods and obesity in nineteen

891

European countries. Public Health Nutrition, 21(1), 18-26.

892

Moorhead, S. A., Welch, R. W., Barbara, M., Livingstone, E., McCourt, M., Burns, A. A., &

893

Dunne, A. (2006). The effects of the fibre content and physical structure of carrots on

894

satiety and subsequent intakes when eaten as part of a mixed meal. British Journal of

895

Nutrition, 96(3), 587-595.

896

Moreira, P. V. L., Hyseni, L., Moubarac, J.-C., Martins, A. P. B., Baraldi, L. G., Capewell, S.,

897

O’Flaherty, M., & Guzman-Castillo, M. (2018). Effects of reducing processed culinary

898

ingredients and ultra-processed foods in the Brazilian diet: a cardiovascular modelling

899

study. Public Health Nutrition, 21(1), 181-188.

900

Moubarac, J.-C., Batal, M., Louzada, M. L., Martinez Steele, E., & Monteiro, C. A. (2016).

901

Consumption of ultra-processed foods predicts diet quality in Canada. Appetite, 108, 512-

902

520.

37

903

Moubarac, J.-C., Parra, D. C., Cannon, G., & Monteiro, C. A. (2014). Food Classification

904

Systems Based on Food Processing: Significance and Implications for Policies and

905

Actions: A Systematic Literature Review and Assessment. Current Obesity Reports, 3(2),

906

256-272.

907

Nardocci, M., Leclerc, B.-S., Louzada, M.-L., Monteiro, C. A., Batal, M., & Moubarac, J.-C.

908

(2019). Consumption of ultra-processed foods and obesity in Canada. Canadian Journal of

909

Public Health, 110(1), 15-16.

910

Nasreddine, L., Tamim, H., Itani, L., Nasrallah, M. P., Isma’eel, H., Nakhoul, N. F., Abou-

911

Rizk, J., & Naja, F. (2017). A minimally processed dietary pattern is associated with lower

912

odds of metabolic syndrome among Lebanese adults. Public Health Nutrition, 21(1), 160-

913

171.

914 915

Nestle, M. (2000). Soft drink "pouring rights": marketing empty calories to children. Public Health Reports, 115(4), 308-319.

916

OECD (2017). Obesity Update 2017.

917

Omenn, G. S., Goodman, G. E., Thornquist, M. D., Balmes, J., Cullen, M. R., Glass, A.,

918

Keogh, J. P., Meyskens, F. L., Valanis, B., Williams, J. H., Barnhart, S., Cherniack, M. G.,

919

Brodkin, C. A., & Hammar, S. (1996). Risk factors for lung cancer and for intervention

920

effects in CARET, the beta-carotene and retinol efficacy trial. Journal of the National

921

Cancer Institute, 88(21), 1550-1559.

922

Pan American Health Organization, & WHO (2015). Ultra-processed food and drink products

923

in Latin America: Trends, impact on obesity, policy implications. Washington, D.C. 76

924

pages.

925

Poti, J. M, Mendez, M. A., Ng, S. W., and Popkin, B. M. (2015). Is the degree of food

926

processing and convenience linked with the nutritional quality of foods purchased by US

927

households? The American Journal of Clinical Nutrition, 101(6), 1251-1262.

38

928

Rauber, F., Campagnolo, P. D. B., Hoffman, D. J., & Vitolo, M. R. (2015). Consumption of

929

ultra-processed food products and its effects on children's lipid profiles: A longitudinal

930

study. Nutrition Metabolism and Cardiovascular Diseases, 25(1), 116-122.

931

Rauber, F., da Costa Louzada, M. L., Steele, E., Millett, C., Monteiro, C. A., & Levy, R. B.

932

(2018). Ultra-processed food consumption and chronic non-communicable diseases-

933

Related dietary nutrient profile in the UK (2008-2014). Nutrients, 10(5), 587.

934

Rico-Campà, A., Martínez-González, M. A., Alvarez-Alvarez, I., Mendonça, R. D., de la

935

Fuente-Arrillaga, C., Gómez-Donoso, C., & Bes-Rastrollo, M. (2019). Association

936

between consumption of ultra-processed foods and all cause mortality: SUN prospective

937

cohort study. BMJ, 365, l1949.

938

Roca-Saavedra, P., Mendez-Vilabrille, V., Miranda, J. M., Nebot, C., Cardelle-Cobas, A.,

939

Franco, C. M., & Cepeda, A. (2018). Food additives, contaminants and other minor

940

components: effects on human gut microbiota-a review. Journal of Physiology and

941

Biochemistry, 74(1), 69-83.

942 943 944

Rock, E., & Fardet, A. (2014). Les antioxydants des agrumes : action en solitaire ou matricielle ? Phytothérapie, 12, 66-75. Rohatgi, K. W., Tinius, R. A., Cade, W. T., Steele, E. M., Cahill, A. G., & Parra, D. C.

945

(2017). Relationships between consumption of ultra-processed foods, gestational weight

946

gain and neonatal outcomes in a sample of US pregnant women. PeerJ, 5, e4091.

947

Rugg-Gunn, A. J., Hackett, A. F., Jenkins, G. N., & Appleton, D. R. (1991). Empty calories –

948

Nutrient intake in relation to sugar intake in English adolescents. Journal of Human

949

Nutrition and Dietetics, 4(2), 101-111.

950

Sandoval-Insausti, H., Blanco-Rojo, R., Graciani, A., Lopez-Garcia, E., Moreno-Franco, B.,

951

Laclaustra, M., Donat-Vargas, C., Ordovas, J. M., Rodriguez-Artalejo, F., & Guallar-

952

Castillon, P. (2019). Ultra-processed Food Consumption and Incident Frailty: A

39

953

prospective Cohort Study of Older Adults. The journals of gerontology. Series A,

954

Biological sciences and medical sciences. DOI: 10.1093/gerona/glz140

955

Satija, A., Bhupathiraju, S. N., Spiegelman, D., Chiuve, S. E., Manson, J. E., Willett, W.,

956

Rexrode, K. M., Rimm, E. B., & Hu, F. B. (2017). Healthful and unhealthful plant-based

957

diets and the risk of coronary heart disease in U.S. Adults. Journal of the American

958

College of Cardiology, 70(4), 411-422.

959

Schnabel, L., Buscail, C., Sabate, J.-M., Bouchoucha, M., Kesse-Guyot, E., Allès, B.,

960

Touvier, M., Monteiro, C. A., Hercberg, S., Benamouzig, R., & Julia, C. (2018).

961

Association between ultra-processed food consumption and functional gastrointestinal

962

disorders: Results from the French NutriNet-Santé Cohort. The American journal of

963

gastroenterology, 113(8), 1217-1228

964

Schnabel, L., Kesse-Guyot, E., Allès, B., Touvier, M., Srour, B., Hercberg, S., Buscail, C., &

965

Julia, C. (2019). Association Between Ultraprocessed Food Consumption and Risk of

966

Mortality Among Middle-aged Adults in France. JAMA, 179(4), 490-498.

967

Steele, E. M., Juul, F., Neri, D., Rauber, F., & Monteiro, C. A. (2019). Dietary share of ultra-

968

processed foods and metabolic syndrome in the US adult population. Preventive

969

Medicine. 125, 40-48

970

Srour, B., Fezeu, L. K., Kesse-Guyot, E., Allès, B., Méjean, C., Andrianasolo, R. M.,

971

Chazelas, E., Deschasaux, M., Hercberg, S., Galan, P., Monteiro, C. A., Julia, C., &

972

Touvier, M. (2019). Ultra-processed food intake and risk of cardiovascular disease:

973

prospective cohort study (NutriNet-Santé). BMJ, 365, l1451.

974

Steele, E. M., Juul, F., Neri, D., Rauber, F., & Monteiro, C. A. (2019). Dietary share of ultra-

975

processed foods and metabolic syndrome in the US adult population. Preventive Medicine,

976

9(125), 40-48

40

977

Talvas, J., Caris-Veyrat, C., Guy, L., Rambeau, M., Lyan, B., Minet-Quinard, R., Lobaccaro,

978

J. M. A., Vasson, M. P., George, S., Mazur, A., & Rock, E. (2010). Differential effects of

979

lycopene consumed in tomato paste and lycopene in the form of a purified extract on target

980

genes of cancer prostatic cells. American Journal of Clinical Nutrition, 91(6), 1716-1724.

981

Tavares, L. F., Fonseca, S. C., Garcia Rosa, M. L., & Yokoo, E. M. (2012). Relationship

982

between ultra-processed foods and metabolic syndrome in adolescents from a Brazilian

983

Family Doctor Program. Public Health Nutrition, 15(01), 82-87.

984

The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group (1994). The effect of

985

vitamin E and beta carotene on the incidence of lung cancer and other cancers in male

986

smokers. New England Journal of Medicine, 330(15), 1029-1035.

987 988 989 990 991

The GBD 2015 Obesity Collaborators (2017). Health Effects of Overweight and Obesity in 195 Countries over 25 Years. New England Journal of Medicine, 377(1), 13-27. Tremblay, A., & Bellisle, F. (2015). Nutrients, satiety, and control of energy intake. Applied Physiology Nutrition and Metabolism, 40(10), 971-979. van den Brandt, P. A. (2019). Red meat, processed meat, and other dietary protein sources

992

and risk of overall and cause-specific mortality in The Netherlands Cohort Study.

993

European Journal of Epidemiology, DOI: 10.1007.

994

Vandevijvere, S., Jaacks, L. M., Monteiro, C. A., Moubarac, J.-C., Girling-Butcher, M., Lee,

995

A. C., Pan, A., Bentham, J., & Swinburn, B. (2019). Global trends in ultraprocessed food

996

and drink product sales and their association with adult body mass index trajectories.

997

Obesity Reviews, DOI: 10.1111/obr.12860

998 999

Walker, C., & Cannon, G. (1986). The food scandal: What's wrong with the British diet and how to put it right. London: Arrow Books Publisher. Pages 210-228.

41

1000

Wang, M., Yu, M., Fang, L., & Hu, R. Y. (2015). Association between sugar-sweetened

1001

beverages and type 2 diabetes: A meta-analysis. Journal of Diabetes Investigation, 6(3),

1002

360-366.

1003

Wijarnpreecha, K., Thongprayoon, C., Edmonds, P. J., & Cheungpasitporn, W. (2016).

1004

Associations of sugar- and artificially sweetened soda with nonalcoholic fatty liver disease:

1005

a systematic review and meta-analysis. QJM, 109(7), 461-466.

42

1006

Fig. 1. Ultra-processed foods among all foods: what is their place?

1007 1008

Fig. 2. Food classifications: holism versus reductionism

1009 1010

Fig. 3. Holism versus reductionism and their interconnection: a theoretical framework applied

1011

to food classifications according to degree of processing (Adapted from Fardet & Rock,

1012

2018). Holistic and reductionist approaches should coexist, but reductionist studies should be

1013

always considered within the framework of holistic thinking to feed back the holistic and

1014

scientific issue. This can be applied to food classification, subdividing holistic food groups in

1015

more reductionist sub-groups.

1016 1017

Fig. 4. The transitivity relationship between the degree of processing, satiety and glycemic

1018

potentials: the higher the degree of food unstructuration and addition of fat and/or simple

1019

sugars the higher the glycaemic index and the lower the satiety potential.

43

1020

Table 1 Summary of the associations between high UPF1 consumption and the risk of chronic diseases and mortality Type of study Chronic diseases

Ecological

References

Cross-

References

Prospective

References

(Canella et al., 2014; Juul

+ (4)

(Costa et al., 2018;

sectional Excess weight, WC and obesity

+ (3)2

(Monteiro et al., 2017;

+ (5)

Pan American Health

et al., 2018; Louzada et

Rohatgi et al., 2017;

Organization, 2015)(

al., 2015; Nardocci et al.,

(Mendonca et al.,

Vandevijvere et al.,

2019)(Julia et al., 2018)

2016)(Canhada et al.,

2019) Metabolic syndrome

2019) + (4)

(Lavigne-Robichaud et al., 2018; Nasreddine et al., 2017; Steele, Juul, Neri, Rauber, & Monteiro, 2019; Tavares et al., 2012)

Hypertension

+ (1)

(Mendonca et al., 2017)

Hyperlipidemia (TC and LDL)

+ (1)

(Rauber et al., 2015)

⊗ (1)

(Nasreddine et al., 2017)

⊗ (1)

(Costa et al., 2018)

Hyperglycemia

44

Cardiovascular diseases

+ (1)

(Srour et al., 2019)

Cancers: -

Overall

-

Breast

-

Prostate and colorectal3

+ (1) + (1)

(Fiolet et al., 2018)

⊗ (1)

Gastro-intestinal disorders: -

-

Irritable bowel syndrome &

+ (1)

functional dyspepsia

⊗ (1)

(Schnabel et al., 2018)

Functional constipation & diarrhea

Asthma and wheezing Depressive symptoms

+ (1)

(Melo et al., 2018) + (2)

(Adjibade et al., 2019; Gomez-Donoso et al., 2019)

Frailty

+ (1)

(Sandoval-Insausti et al., 2019)

Mortality

+ (3)

(Kim, Hu, & Rebholz, 2019; Moreira et al., 2015; Rico-Campà et al., 2019)

45

1021

1

1022

colorectal cancer (HR = 1.23 [1.08 to 1.40]), but the P for the trend is not significant. Abbreviations: LDL, Low-Density Lipoproteins, TC, Total Cholesterol, UPF, Ultra-Processed Food, and WC, Waist Circumference. +,

1023

indicates a positive association, i.e., increased risk; ⊗ indicates an absence of significant association

As defined by NOVA food classification in four technological groups; 2Number of studies; 3For colorectal cancers, the comparison “high versus low” UPF consumption is positively associated with a significantly higher risk of

46

Junk foods

UPFs

Non UPFs

Non-industrial foods

Unprocessed foods

Industrial foods

Processed foods

Fake “healthy” foods, most of industrial dishes and infant formula, biscuits…

Reductionist study

Holism Reductionism

Holistic and scientific issue

Reductionist study

Reductionist study

Application to technological classifications

+

Holistic NOVA (n = 4 groups)

Reductionist subgroups Including “composition/matrix” effects, ingredient processing, additives and others

Reductionist study

+

↑ Degree of processing → degrading food matrix structure + added simple sugars and fat

+ ↓

↑ Foods

+

-

Glycemic impact

Satiety potential

+

↑

-

Highlights

Ultra-processed foods (UPF) is a new nutritional concept released in 2009. UPF is a holistic concept including food matrix and nutrient composition degradation. UPFs are empty calories, poorly satiating, and hyperglycaemic. Excess UPF consumption is associatied with increased risks of many chronic diseases.