Defining the role of processing in food classification systems—the IUFoST formulation & processing approach

Background and objectives

The concept of UPFs has created controversies and, therefore, the use of food processing terminology in food classification should be comprehensively reviewed. IUFoST also recognized an international need for food classification underpinning dietary guidelines and policies as well as public health-related decisions that are based on scientific evidence and data to support objective decisions across the involved sectors and that consider global diversity in nutrition issues and food choices. Developing a research agenda for understanding the role of processed foods in the FBDG requires that food processing concepts be clearly defined and understood by scientists conducting such research across several disciplinary areas.

Thus, it was decided that the IUFoST task force on Processing for Nutrition, Diet and Health (P4NDH) will prominently address the question of how quantitative consideration of processing as an impact factor on specific quality characteristics or the overall quality profile of processed food products could be introduced.

Key definition 1: Classification of food properties

Food classification involves systematically arranging foods in groups or categories according to established criteria. The latter should be food quality criteria in the present case. Such are (i) safety, (ii) palatability, (iii) affordability (iv) nutrition value, (v) convenience and (vi) sustainability. The latter relates to the environment, health, economics and society.

As demonstrated in Fig. 1, the properties (i) to (vi) can be used to configure a property profile of a food product which is typically shifted by processing. In case optimization of one of these properties is intended some others may be compromised. A property profile for an exemplary base food product is shown in Fig. 1 (part A), which is indicated by a black line before processing treatment. After hypothetical treatments by two different processes (e.g. Process P1: high-pressure pasteurization (red line) and Process P2: thermal sterilization (blue line)) carried out to optimize either the nutrition value by more gentle treatment of heat-sensitive nutrients (P1) or the product shelf live reducing the microbial load by more intensive thermal decontamination treatment (P2), the respective changes in the property-profiles are shown in the hexagonal spiderweb arrangement in Fig. 1 (part B). This represents a typical balancing procedure of food product properties in an industrial environment if optimization of one or some of the properties shall be achieved. The processing toolbox is typically very flexible through the adjustment of processing type and parameters.

The NOVA classification system emphasizes the impact of ultra-processed foods with a focus on the nutritional quality of overall diets, and associated obesity and health/disease outcomes⁷. Palatability is mentioned in some places as a reason for overeating. Accordingly, the task force identified nutrition value (NV) and palatability as key target properties to be addressed in this context. For the first exemplary approach followed in this report, the nutrition value was the priority, due to its expected improved quantitative definability based on objective criteria. In contrast, considering palatability will require additional subjective criteria to meet a wider spectrum of consumer preferences.

Furthermore, the Nutrition Value (NV) constitutes the base for the derivation of dietary guidelines which, if applicable, should be taken into account for the implementation of processing criteria to be quantified. The NOVA system considers the addition of sugar, salt and saturated fats as indicators for food products classified as processed and ultra-processed. For the latter nutrients/components extracted from foods to imitate or enhance the sensory qualities of foods (“cosmetic ingredients”) come on top⁷.

The object of classification (OC) in the thematic context of the IUFoST task force work was the processed food product at the Point of Purchase (POP). In general classification sorts into classes of quantified Classification Parameters (CPs) which are functionally related to a selected PROPerty (PROP) of the targeted object. For a better understanding of classification definition and related designations, this is schematically demonstrated in Fig. 2.

To characterize a food product in its state at the POP, the parameters related to the PROP to be considered for classification (CPs) are determined preferably by quantitative measurement but may also be qualitatively described.

As the Point Of Classification (POC) in principle, each position along the food value chain can be selected for a food product in its local state. The definition of the Food Value Chain in the context of this report does not end on the consumer’s plate but includes the digestive and metabolic transformation steps of the consumed food as well as the resulting consumer’s status with respect to obesity risk and other health risk outcomes. The task force decided to refer to the food product at the point of purchase (POP). Home preparation may occur between the POP and the Point of Consumption (PCON), including processing steps. However, since both the conventional nutrient recommendation schemes used for dietary guidelines and the NOVA classification refer to processed food products at the POP, home processing being astonishingly exempt from the NOVA definition of ultra-processed food, is also not considered here in further detail but is understood to potentially further modify a food’s nutritional value depending on how the food is prepared which may not strongly differ from industrial processing steps but being in general less controlled.

An important boundary condition for proper classification is that the classification parameters (CPs) are determined at the POC (here POP). Other key specifications made by the task force for the introduction of classification parameters (CPs) for processing were:

1. (i)

   A second POC2 can be selected upstream of the food value chain from the POC1 (=POP), to describe a CP value change as the result of a processing sequence between these two points (ΔCP-values).

2. (ii)

   In such case (i) the definition of the CP has to be the same at each of the two POCs (CP1 @ POC1 = CP2 @ POC2).

3. (iii)

   If (ii) is not fulfilled, only correlations between classification results achieved at POC1 and POC2 are applicable.

4. (iv)

   An exception to (iii) is given in case a functional relationship between CP1 @ POC1 and CP2 @ POC2 has been identified.

5. (v)

   In case a second POC2 is selected downstream the food value chain from PCON, it has to be expected that the probability of point (iii) to occur is high as long as functional relationships between CPs at different POCs along the digestive and metabolic processing tracks are not quantitatively known.

Figure 3 demonstrates the ”POC cases” addressed in (i)–(v) before. It also includes indications of the different application frames for the suggested classifications by the IUFoST task force and the NOVA system.

As mentioned in (i) and demonstrated in Fig. 3, ΔCP-values are a measure of the treatment (processing) influence on the food product between two POCs. Accordingly, the ΔCP-values can serve as classifying parameters for the food processing impact experienced before the POP, hence describing the losses or gains of the nutrition value between the unprocessed or pre-processed state and the POP. In case the processing track before the POP consists of a series of n processing steps each of these steps could be analysed by respective ΔCP₁ to ΔCP_n-1 values. The Nova frame of classification as it is interpreted nowadays spans from the POP as a point of classification (POC) to the end of the extended food value chain where obesity and NCCD health risks are claimed to be considered. The NOVA classification at the POP considers four qualitatively described NOVA classes. These classification results are compared with results from observational studies on resulting obesity and health risk aspects. Such comparison is of a correlation nature since quantitative functional relationships between food product and nutrition value parameters along the digestive and metabolic tracks are so far only fragmentary available but not addressed by NOVA. Thus, it is suggested to designate such derived relationships between food product parameters at the POP and obesity and health risk data as “Food Impact Correlations” (Fig. 3).

Key definition 2: Food processing

To focus on Food Product classification with respect to the Nutrition Value as targeted property and taking the possible impact of processing on such classification into account, the process-property relationship designated the main element of interplay to be addressed. As has been shown in the food engineering literature, a quantitative understanding of how processing creates or influences product properties must consider the food structure as a key element connecting process and properties.

This was established in the so-called Process-Structure-Properties scheme, also denoted as the S-Pro² scheme (Fig. 4). The core message of this scheme can be briefly expressed as: “Process creates structure and structure encodes properties”. Accordingly, when nutritionally relevant components within a formulation are treated in a process, that process is typically optimized to provide a processed food product structure that maintains or increases the nutritional value or at least minimizes losses in nutrient functionality with respect to digestibility, bio-accessibility, bioavailability and metabolic response.

Type and degree of processing decide the extent of structural changes and related impact on resulting properties to be tailored. In general, a Reverse Engineering approach is favoured to optimize selected food properties. This means that based on researched structure-property functions the structure which determines the optimal properties (e.g. nutrition value) is identified. This is followed by the selection of the process and adjustment of appropriate processing parameter settings according to process-structure functions derived from systematic investigations, in order to generate the identified structure representing the optimal quality characteristics intended.

In order to maintain the nutritional properties, present in food raw materials or ingredients, it may be important not to change the molecular structure of such nutrients when passing processing steps i.e. for their enrichment or separation. However, there are also cases where processing enhances digestibility, e.g. for proteins when partially denatured. Moreover, the generated or modified food matrix structure of processed foods in which the nutrients are integrated, determines their release kinetics in the digestive tract and the associated bio-accessibility and bioavailability.

The NOVA food classification system

The general reasoning for the NOVA classification system as given by its creator C. Monteiro in (ref. ¹⁰) is as follows: ”The significance of industrial processing, and in particular techniques and ingredients developed or created by modern food science and technology, on the nature of food and on the state of human health, is generally understated. This is evident in international and national policies and strategies designed to improve population nutrition and health, in dietary recommendations, and in public policies and actions guided by such recommendations. Until recently it has also been neglected in epidemiological and experimental studies concerning diet, nutrition and health”.

NOVA classifies all foods and food products into four groups as follows, with the key aspects adapted from Monteiro et al. (cited from refs. ^7,8):

GROUP 1: Unprocessed and minimally processed foods

Unprocessed (or natural) foods are the edible parts of plants (such as fruit, leaves, stems, seeds and roots) or from animals (such as muscle, offal, eggs, milk), and also fungi, algae and water, after separation from nature. Minimally processed foods are natural foods altered by methods that include the removal of inedible or unwanted parts and also processes that include drying, crushing, grinding, powdering, fractioning, filtering, roasting, boiling, non-alcoholic fermentation, pasteurization, chilling, freezing, placing in containers, and vacuum packaging. The distinction between unprocessed and minimally processed foods is not especially significant.

GROUP 2: Processed culinary ingredients

Processed culinary ingredients include oils, butter, lard, sugar and salt. These are substances derived from group 1 foods or else from nature by processes such as pressing, refining, grinding, milling, and drying. Some methods used to make processed culinary ingredients are originally ancient. But now they usually are industrial products, designed to make durable products suitable for use in home, restaurant and canteen kitchens to prepare, season and cook freshly prepared dishes and meals. In isolation, processed culinary ingredients are unbalanced, being depleted in some or most nutrients. Other than salt, they are also energy-dense, at 400 or 900 kilocalories per 100 g. This is around 3–6 times more than cooked grains and around 10–20 times more than cooked vegetables. They are used in combination with foods to make palatable, diverse, nourishing and enjoyable meals and dishes such as stews, soups and broths, salads, breads, preserves, drinks, and desserts. Thus, oils are used in the cooking of grains (cereals), vegetables and legumes (pulses), and meat, and are added to salads. Table sugar is used to prepare fruit- or milk-based desserts. It is misleading to assess their nutritional significance in isolation. They should always be assessed in combination with foods.

GROUP 3: Processed foods

These include canned or bottled vegetables or legumes (pulses) preserved in brine; whole fruit preserved in syrup; tinned fish preserved in oil; some types of processed animal foods such as ham, bacon, pastrami, and smoked fish; most freshly baked breads; and simple cheeses to which salt is added. They are made by adding salt, oil, sugar or other substances from group 2 to group 1 foods. Processes include various preservation or cooking methods, and with breads and cheeses, non-alcoholic fermentation. Processing here increases the durability of group 1 foods or modifies or enhances their sensory qualities. Most processed foods have two or three ingredients and are recognizable as modified versions of group 1 foods. They are generally produced to be consumed as part of meals or dishes and also may be consumed by themselves as snacks. Most are highly palatable.

Almost all processed foods are manufactured industrially. Processes include canning and bottling using oils, sugars or salt; and methods of preservation such as salting, salt-pickling, smoking, and curing. The ingredients infiltrate the foods and so the processes alter their nature. Processed food products usually retain the basic identity and most constituents of the original food. But when excessive oil, sugar or salt are added, they become nutritionally unbalanced. Except for canned vegetables, their energy density ranges from moderate (around 150–250 kilocalories per 100 g for most processed meats), to high (around 300–400 kilocalories per 100 g for most cheeses).

GROUP 4: Ultra-processed foods

Ultra-processed foods are formulations of ingredients, mostly of exclusive industrial use, typically created by a series of industrial techniques and processes (hence ‘ultra-processed’). Some common ultra-processed products are carbonated soft drinks; sweet, fatty or salty packaged snacks; candies (confectionery); mass-produced packaged bread and buns, cookies (biscuits), pastries, cakes and cake mixes; margarine and other spreads; sweetened breakfast ‘cereals’ and fruit yogurt and ‘energy’ drinks; pre-prepared meat, cheese, pasta and pizza dishes; poultry and fish ‘nuggets’ and ‘sticks’; sausages, burgers, hot dogs and other reconstituted meat products; powdered and packaged ‘instant’ soups, noodles and desserts; baby formula; and many other types of product.

Processes enabling the manufacture of ultra-processed foods involve several steps and different industries. It starts with the fractioning of whole foods into substances including sugars, oils and fats, proteins, starches and fibre. These substances are often obtained from a few high-yield plant foods (such as corn, wheat, soya, cane or beet) and from puréeing or grinding animal carcasses, usually from intensive livestock farming.

Some of these substances are then subjected to hydrolysis, hydrogenation, or other chemical modifications. Subsequent processes involve the assembly of unmodified and modified food substances with little if any whole food using industrial techniques such as extrusion, moulding and pre-frying. Colours, flavours, emulsifiers and other additives are frequently added to make the final product palatable or hyper-palatable. Sophisticated and attractive packaging is used, usually made of synthetic materials. Ingredients characteristic of ultra-processed foods are either food substances of no or rare culinary use or else classes of additives whose function is to make the final product sellable, palatable and often hyper-palatable.

Food substances of no or rare culinary use, employed in the manufacture of ultra-processed foods, include varieties of sugars (fructose, high-fructose corn syrup, ‘fruit juice concentrates’, inverted sugar, maltodextrin, dextrose, lactose), modified oils (hydrogenated or inter-esterified oils) and sources of protein (hydrolysed proteins, soya protein isolate, gluten, casein, whey protein, and ‘mechanically separated meat’). Classes of additives used only in the manufacture of ultra-processed foods, are flavours, flavour enhancers, colours, emulsifiers, emulsifying salts, artificial sweeteners, thickeners, and foaming, anti-foaming, bulking, carbonating, gelling and glazing agents. All of them, most notably flavours and colours, either disguise unpleasant sensory properties created by ingredients, processes or packaging used in the manufacture of ultra-processed foods, or give the final product intense sensory properties especially attractive to see, taste, smell and/or touch, or both.

Processes and ingredients used for the manufacture of ultra-processed foods are designed to create highly profitable products (low-cost ingredients, long shelf-life, powerfully branded). Their convenience (imperishable, ready-to-consume), hyper-palatability, and ownership by trans-national corporations using pervasive advertising and promotion, give ultra-processed foods enormous market advantages. They are therefore liable to displace all other NOVA food groups and to replace freshly made regular meals and dishes, with snacking anytime, anywhere.

Identification of ultra-processed foods (UPF)

Following instructions for UPF identification were also adapted from Monteiro et al.^7,8: “Food manufacturers do not have to state on food labels the processes used in their products, and even less the purposes of these processes. Sometimes this can make it difficult to identify ultra-processed foods with confidence. Generally, the practical way to identify if a product is ultra-processed is to check to see if its list of ingredients contains at least one item characteristic of the ultra-processed food group.

These are either food substances never or rarely used in kitchens or classes of additives whose function is to make the final product palatable or more appealing. Food substances not used in kitchens include hydrolysed proteins, soya protein isolate, gluten, casein, whey protein, ‘mechanically separated meat’, fructose, high-fructose corn syrup, ‘fruit juice concentrate’, invert sugar, maltodextrin, dextrose, lactose, soluble or insoluble fibre, hydrogenated or inter-esterified oil. The presence in the list of ingredients of one or more of these food substances identifies a product as ultra-processed.

Classes of additives exclusively used in ultra-processed foods include flavours, flavour enhancers, colours, emulsifiers, emulsifying salts, artificial sweeteners, thickeners, and anti-foaming, bulking, carbonating, foaming, gelling and glazing agents. Any example of these classes of additives, as shown on ingredients lists also identifies a product as ultra-processed” (end of cited descriptions from refs. ^7,8).

Such an identification approach is largely based on food additives and little to no actual processing technologies. In addition, the list of many “industrially produced food ingredients” in the above paragraph also includes proteins and fibre, which have nutritional value in addition to other functional properties. This should be recognized as such, rather than being negatively implied as the NOVA identification puts it.

Food classification systems considering processing

The NOVA and derived NOVA-type classification systems claiming to consider processing aspects for food product classifications at the point of purchase (POP) are exemplary listed in Table 1. As described within a comprehensive comparative overview by Ch. Sadler et al.⁹ these classification systems take into account qualitative conceptualization criteria, being: (a) the type of processing, (b) the intensity of processing, (c) the place of processing and (d) the purpose of processing. Figure 5 (adapted from ref. ⁹) schematically demonstrates these dimensions used in the conceptualization of processed foods.

C.R. Sadler et al.⁹ came to the conclusion that the classification systems including the processing aspects listed in Table 1, were mostly created to study the relationship between industrial products and health. No consensus was found on what factors determine the level of food processing. The classification systems embodied socio-cultural elements and subjective terms, including home cooking and naturalness. The authors conclude that: “so far processing in the considered food classification systems is a chaotic conception, not only concerned with technical processes. Most classification systems do not include quantitative measures but, instead, imply a correlation between processing and nutrition”. The authors derive that the concept of “whole food” and the role of the food matrix in relation to healthy diets needs further clarification and that the risk assessment/management of food additives also needs further debate.

NOVA ambiguities and uncertainties

As described in detail before the NOVA system sorts food products into four groups representing the “classification classes”, being: group I: fresh/minimally processed food; group II: processed ingredients; group III: processed food and group IV: ultra-processed food.

Both, the conceptualization dimensions and the NOVA group/class designations give the impression that from group I to group IV some degree of processing or processing intensity is increasing. However, the expectation that an assignable level of processing underlies the distinction of the NOVA classification groups is not met. In addition, and contrary to the expectation created, the main classification parameters (CPs) applied by the NOVA system to sort into the groups/classes I–IV marked with processing levels, follow the formulation-based addition of sugar, salt, saturated fats and ingredients of mostly exclusive industrial use. It is worth noting that most authoritative food-based dietary guidelines already include recommendations, including quantitative limitations, on foods, nutrients and food components to limit based on such formulation-based criteria.

Obviously, the ingredient composition for a food product is a matter of formulation and not of processing. Moreover, compared with proper formulation rules the NOVA system does not quantify group/class (I–IV) specific fraction ranges of the “critical” ingredients. Based on such formulation-determined but processing designated food product groups or classes I–IV, the NOVA system derives correlations between consumption of food products from these groups/classes (I–IV) with results from observational studies about obesity and health risks. This approach is not evidence-based and is not clear and incomplete.

For the IUFoST task force on Processing for Nutrition, Diet and Health the inclusion of processed food impact correlations with obesity and health risk outcomes was not intended. On the other hand, the stated aim of the task force was to create a more solid basis for the correlations intended by NOVA using a quantitative consideration of the processing influence on the nutritional quality of processed food products.

In this context, the following issues (1–10) with the NOVA/NOVA-type classification systems have been identified by the task force. The key aspects 1 and 2, to be avoided in food classification development and derived applications were selected for further detailed exploration.

1. 1.

   Confusion in the use of formulation and processing terms

2. 2.

   Lack of quantification of processing level

3. 3.

   Weak scientific approach

4. 4.

   Conflicting issues with nutritional advice

5. 5.

   Conflicting issues with fortified and personalized food

6. 6.

   Inadequacy with the needs of developing countries

7. 7.

   Differences in ingredients and food products

8. 8.

   Multifaceted complex non-linear relationships

9. 9.

   Ambitious terminology used (NOVA)

10. 10.

    Processing misinterpreted as negative factor

Since the task force is of the opinion that the NOVA system has included processing as a significant influencing factor on the nutritional properties of processed food products in the classification of such products which has so far not been sufficiently considered, the following questions should be further examined and evaluated in order to propose and develop possible refinement and optimization solutions:

Question A: How can the main issues of the NOVA system recognized by the task force being (i) the confusion in using formulation and processing terms and (ii) the lack of quantification of formulation and processing levels, be tackled by the task force?

Question B: Which are the benefits or weaknesses of the NOVA classification system compared to existing systems of nutrient profiling and derived recommendations for dietary guidelines in the view of renowned experts in the field outside of the task force?

Question B is addressed in the next chapter. Question A concerns the main theme of the task assigned to the IUFoST Task Force, which is dealt with in very detail in the chapter after the following.

Dietary guidelines and NOVA

This chapter is meant to address question B posed before. As a part of the analysis of food classification systems, the task force also reviewed a number of recent publications that addressed the NOVA classification system and evaluated its claimed added value compared with so far successfully applied nutrient profiling of food products for dietary guideline recommendations^{11,12,13,14,15,16,17,18}. Some of the prominent experts in the human nutrition field are reflected by citations of their statements in the following.

OPINION 1: In a debate between Astrup, A. & Monteiro, C. A. published in Am. J. Clin. Nutr. 116, 1482–1488 (2022)¹¹,

A. Astrup comes to the conclusion that: ”The NOVA classification of ultra-processed foods (UPFs) rests on poorly defined food processes and the presence of food additives from a chemically heterogeneous group, easily leading to misclassification. UPFs are claimed to promote overconsumption of energy and obesity due to high palatability, but little evidence supports effects beyond those that can be accounted for by nutrient composition, energy density, and food matrices. Observational studies link dietary intake of UPFs with obesity, but none have demonstrated independent associations after controlling for likely confounders. A highly cited randomized controlled feeding study that compared a UPF diet with an unprocessed diet showed a rapidly weaning effect on energy intake that can be entirely explained by more conventional and quantifiable dietary factors, including energy density, intrinsic fibre, glycemic load, and added sugar. Clearly, many aspects of food processing can affect health outcomes, but conflating them into the notion of ultra-processing is unnecessary because the main determinants of chronic disease risk are already captured by existing nutrient profiling systems. In conclusion, the Nova classification adds little to existing nutrient profiling systems; characterizes several healthy, nutrient-dense foods as unhealthy; and is counterproductive to solve the major global food production challenges”.

C.A. Monteiro responded that: “The concept of ultra-processing, as contained within the NOVA food classification system, is well-founded, clear, supported by a wealth of investigations, and crucial to inform new dietary guidelines and other public policies and actions designed to reduce the production and consumption of ultra-processed foods”. For further detail on the description of the NOVA classification system in this report refer to the sub-chapters “The Nova Food classification” and “Identification of ultra-processed foods (UPF)” described above.

OPINION 2: by Drewnowski, A., Gupta, S. & Darmon, N. in Nutr. Today55, March/April (2020)¹² stated:

“The category of “ultra-processed” foods in the NOVA food classification scheme is ostensibly based on industrial processing. We compared NOVA category assignments with the pre-existing family of Nutrient Rich Food (NRF) indices, first developed in 2004. The NRF indices are composed of 2 sub-scores: the positive NR based on protein, fibre, vitamins and minerals, and the negative LIM sub-score based on saturated fat, added sugars, and sodium. The 378 foods that were components of the widely used Fred Hutchinson Cancer Center food frequency questionnaire were assigned to NOVA categories and scored using multiple NRF indices. Contrary to published claims, NOVA was largely based on the foods’ content of saturated fat, added sugars, and sodium. There were strong similarities between NOVA categories and NRF scores that were largely driven by the nutrients to limit. Nutrient density led to higher increased NRF scores but had less impact on NOVA categories. As a result, the NOVA scheme misclassified some nutrient-rich foods. We conclude that the NOVA classification scheme adds little to the pre-existing nutrient profiling models. The purported links between NOVA categories and health outcomes could have been obtained using pre-existing NRFn.3 nutrient density metrics.”

OPINION 3: by Hess, J. M. et al. in J. Nutr. 153, 2472–2481 (2023)¹³:

“The purpose of this proof-of-concept study was to determine the feasibility of building a menu that aligns with recommendations for a healthy dietary pattern from the 2020 DGA and includes 80% kcal from UPF as defined by NOVA. In the ultra-processed DGA menu that was created, 91% of kcal were from UPF, or NOVA category 4. The HEI-2015 score was 86 out of a possible 100 points. This sample menu did not achieve a perfect score due primarily to excess sodium and an insufficient amount of whole grains. This menu provided adequate amounts of all macro- and micronutrients except vitamin D, vitamin E, and choline. The authors conclude that: “Healthy dietary patterns can include most of their energy from UPF, still receive a high diet quality score, and contain adequate amounts of most macro- and micronutrients.

OPINION 4: by Braesco, V. et al. in Europ. J. Clin. Nutr.76, 1245–1253 (2022)¹⁴: “French food and nutrition specialists completed an online survey in which they assigned foods to NOVA groups. The survey comprised two lists: one with 120 marketed food products with ingredient information and one with 111 generic food items without ingredient information. We quantified assignment consistency among evaluators using Fleiss’ κ (range: 0–1, where 1 = 100% agreement). Hierarchical clustering on principal components identified clusters of foods with similar distributions of NOVA assignments. There were three clusters within the marketed foods: one contained 90 foods largely assigned to NOVA 4 (91% of assignments), while the two others displayed greater assignment heterogeneity. There were four clusters within the generic foods: three clusters contained foods mostly assigned to a single NOVA group (69–79% of assignments), and the fourth cluster comprised 28 foods whose assignments were more evenly distributed across the four NOVA groups.

Conclusions: Although assignments were more consistent for some foods than others, overall consistency among evaluators was low, even when ingredient information was available. These results suggest current NOVA criteria do not allow for robust and functional food assignments.”

From these human nutrition expert opinions cited before and a larger number of publications not cited here, the IUFoST task force could conclude that the NOVA classification in summary has also weaknesses concerning:

B1. Only little to no addition to existing nutrient profiling systems

B2. Not evidence-based statements on the harmfulness of approved processes/additives & their health risk impact

Additional information of relevance for the task force which could be gleaned from literature /e.g. ref. ¹¹ was, that there is strong opinion in the scientific community on:

B3. The extent of food processing significantly affects diet quality and health outcomes.

B4. Classification systems based on certain characteristics of food processing, including the degree to which the food matrix is affected and the use of certain additives, can distinguish categories of food with beneficial or adverse health outcomes in a useful manner.

While answers B1 and B2 should be addressed in the context of recommendations to be worked out by the task force, B3 and B4 additionally confirm the relevance of processing as a main influential factor to be considered in future food classification and the necessity of further research on the mentioned aspects.

Differentiation of formulation and processing in food classification by definition and quantification

Within this key chapter of the report, question A posed at the end of the previous chapter will be prominently addressed, which for making reading easier is repeated here being: “How can the main issues of the NOVA system recognized by the task force being (i) the confusion in using of formulation and processing terms and (ii) the lack of quantification of a processing degree, be tackled?”

Formulation (F) and Processing (P) were identified as independent but interacting factors in the food product manufacture, in order to optimize desirable food properties of relevance according to consumers Preference, Acceptance and Needs (PAN profiles¹⁹). Such property characteristics as mentioned before, are: nutrition value (task force focus), safety, palatability (sensory characteristics), sustainability, convenience and affordability to which F & P can contribute in a positively complementing or counteracting manner. Accordingly, it was stated that Formulation and Processing have to be treated as separate criteria if introduced into food classification systems and should consider quantitative parameters to express their specific “levels of execution”.

The following DEFINITIONS were decided by the task force:

Formulation: “Systematic selection of relative quantities of ingredients for a food product”

Processing: “Treatment of a food material to achieve a desired effect”

Enabling a quantitative approach to describe the influence of formulation and processing of a food product on a targeted food property, the latter has to be decided in the context of the question to be answered or the problem to be solved. If Food classification aims for supporting the development of dietary guidelines the nutrition value (NV) of a food product is confirmed as a key property. In the following a suggested 3-STEP PROCEDURE (3SP) is described for processed food products, treating formulation and processing separately but taking into account their mutual influence on the NV.

For quantification of the nutrition value (STEP 1), the so-far well-experienced Nutrition Rich Food Index (NRFx.y) was selected as the exemplary approach^1,20,21. The NRFx.y is a formulation-related value for nutrition profiling of food products considering the two groups of x recommended nutrients (NR) and y nutrients to be limited (LIM) according to Eq. (1), below¹. Typical numbers for x and y are 9 and 3 according to 9 recommended nutrients (NR): protein, dietary fibre, vitamin A, vitamin C, vitamin E, calcium, iron, potassium and magnesium and 3 nutrients to be limited (LIM): added sugar, sodium and saturated fatty acids.

with: (I) NR9_100 kcal = Σ_1–9 (m_{nutrient i}/m_DVi)/S_i · 100 denoting 9 nutrients recommended per product portion providing 100 kcal (418.4 kJ) of energy, m_DVi = mass of Daily Value (nutrient i) and S_i = calories per serving

(ii) LIM3_{100 kcal} = Σ_1–3 (m_{nutrient i}/m_MRVi)/S_i ·100 denoting (3) nutrients to be limited per product portion providing 100 kcal (418.4 kJ) of energy, m_MRVi = mass of maximum recommended value.

A logical connection can be seen between nutrient density (ND), energy density (ED) and NRF, since for the basis of a food portion with 100 kcal energy content, only products with high ND or high ED and high ND achieve a high NRF score. Those with (only) high ED or low ND are found in the low NRF score range. NRFx.y also has strengths in terms of its flexible adaptability to the preferred needs of certain consumer groups, as explained in more detail in the Supplementary Information to this report. Consequently, NRFx.y was chosen as the recommended quantitative base definition of nutritional value, the selected food property to be classified here.

Within STEP 2 of the 3SP the quantification of degrees of formulation and processing is treated. Since the NRFx.y describes the proportions of the nutritionally relevant components of a food formulation, it also represents a nutritional value-related “formulation level” that can be used as a formulation classification parameter (CP). In order to define an equally suitable, easy-to-determine processing parameter, it is suggested to refer to the chosen NRFx.y-based nutrition value definition and express the “degree of processing” by the difference in the NRFx.y value after and before a processing treatment, denoted as ΔNRFx.y. which is subsequently used to classify the influence of processing on the nutrition value (Eq. (2)).

The values x (number of nutrients recommended) and y (number of nutrients to be limited) were chosen as 9 and 3, thus configurating the NRF9.3 as has been described before.

The graphical representation of the IUFoST F&P Classification (abbreviated as IF&PC) approach is denoted as STEP 3 of the 3SP. Within this, a 2D-(F&P) Classification Matrix Diagram (CMD) is suggested for illustrating (i) the NRF9.3 as quantitative nutrition value-related “formulation level” classification parameter (CP1, x-axis) and (ii) ΔNRF9.3 as quantitative nutrition value-related “processing level” classification parameter (CP2, y-axis) generally demonstrated in Fig. 6 and complemented with data from processed food systems (²² and USDA-FNDDS²³) in Fig. 7. Figure 6 restricts to demonstrate shifts in positioning of a food system depending on principle changes in formulation or processing.

On the x-axis of Figs. 6 and 7 the NRF9.3 is displayed within the range of −100 to +200. Fresh berries and citrus fruits come close to NRF9.3 values of 200 while sugar-sweetened soda beverages can reach values below −50⁶. The processing degree ΔNRF9.3 can impact positively or negatively on the nutrition value within an estimated range of −100 to +100 based on initial experience. The coupled impact of formulation level (F) expressed by NRF9.3 and processing level (P) expressed by ΔNRF9.3 can be calculated by applying Eq. (3) which defines the so-denoted Formulation and Processing Food Index FPFI^N, the superscript N indicating the nutrition value as the considered food property being classified. Thus, FPFI^N values represent quantitative Nutrition Value classes, the magnitude of which is indicated in Figs. 6 and 7 by the Iso-FPFI^N lines (lines of constant FPFI^N).

A, B denoting scaling constants of the Classification Matrix Diagram (CMD) demonstrated in Figs. 6 and 7, where A = 25 (1 FPFI^N unit equivalency to 25 NRF9.3 units) and B = −2 (FPFI^N value in the origo). The subscript 1 represents the NRF9.3 before the considered processing step)

Tracking of formulation and processing in the classification matrix diagram (CMD)

Points located in the CMD are denoted as P_i (x_i/y_i) or P_i (NRF9.3_i/ΔNRF9.3_i), with NRF9.3 being the x-coordinate and ΔNRF9.3 the y-coordinate. A formulation for a food product to be processed before processing is described by a starting point P₁ (NRF9.3₁/0). This starting point P₁ is located on the ΔNRF9.3 (processing scale) zero line. If a pure formulation change is executed (e.g by adding sugar to glaze carrots; see example in Fig. 7), this causes a horizontal shift from P₁ (NRF9.3₁/0), to a P₂ (NRF9.3₂/0) in the diagram. With NRF9.3₂ < NRF9.3₁ due to the LIM-term increase in the NRF9.3 caused by the sugar addition (see Eq. (1)), the shift vector (arrow) is negatively directed. – Subsequent designations of indices for points in the CMD refer specifically to Fig. 7.

A processing step that results in a change of the NRF9.3 value from (NRF9.3)₂ before processing to (NRF9.3)₃ after processing and thus generates a corresponding ΔNRF9.3 (=(NRF9.3)₃−(NRF9.3)₂), causes a point shift from the starting point P₂ (NRF9.3₂/0) to the resulting P₃ (NRF9.3₂ + ΔNRF9.3/ΔNRF9.3₃). Thus, P₃ is placed along a straight line (SL) through P₂ inclined (for equidistant scaling units in x and y direction by +45°) relative to the NRF9.3 (x) axis (Fig. 6). If NRF9.3₂ > NRF9.3₃ then ΔNRF9.3 < 0, a negative impact of the processing step is resulting (see example: glazed carrots cooked; Fig. 7), and P₃ is lower left-shifted from P₂ along SL. In the case of NRF9.3₂ < NRF₃ the P₃ shift will be to the upper right along SL (see example, pea protein isolate processing from pea seeds; Fig. 7).

For each change in formulation or processing the FPFI^N can be calculated according to Eq. (3). Respective FPFI^N values are denoted for the exemplary processed product examples presented within Fig. 7.

The FPFI^N value shifts between starting- and endpoints correspond to the influence of coupled formulation and processing steps applied in the manufacture of a processed food product one can stepwise follow in the CMD. Which FPFI^N value (class) a finally processed food product reaches also depends on the starting point, given by the NRF9.3 of the initial formulation. Starting from base formulations of different NRFx.y (Nutrition Value) and coupling with different processing steps of positive or negative impact could lead to endpoints in the CMD of similar FPFI^N values and thus in the same F&P-coupled nutrition value class. A detailed interpretation of such results will also depend on the applied extensions to the NRF9.3 as suggested in the Supplementary Material method 1.2.

The FPFI^N values scale from −10 to +10. FPFI^N_i and can be calculated according to Eq. 3 for each point P_i in the CMD. Iso-FPFI^N lines have been inserted into the CMD for orientation. These Iso-lines are inclined by −45° to the NRF9.3-axis (for equidistant scaling of both CMD axes) thus being oriented rectangular to the straight SL-lines along which the “processing impact vectors” with a length of (sqrt{2,})∆NRF9.3 are displayed. The factor (sqrt{2}) results from the fact that the processing vector endpoint (e.g. P₃ in Fig. 7) is received from equal shifts of the starting point (e.g. P₂) by ∆NRF9.3 = (NRF9.3)₃ – (NRF9.3)₂) into the x and y directions. For each processing change resulting in a ∆NRF value and related point shift in the diagram from P_i to P_i+1 the FPFI^N in the endpoint P_i+1 is received from Eq. (3) (for further extended details see Supplementary Method 1.2).

Key benefits in applying the IF&PC scheme

The CMD has been created in order to support the practical application of the two separated and quantified levels of formulation (F) and processing (P) impacts on the nutrition value (NV). As expected, such quantifying two-parameter classification system initially places increased demands on handling compared to only qualitatively descriptive or partially quantifying single-parameter systems.

As the exemplary (apple, carrot and pea) demonstration of formulation changes and processing modifications in Fig. 7 indicates, practical handling for the two-parameter IF&PC approach can be well assisted by the CMD. Moreover, the quantitative F&P coupling parameter FPFI^N (Food Processing and Formulation Index) as depicted in the CMD indicates a simplified base for possible future consumer communication and product labelling approaches considering quantitative F and P aspects appropriately. For assistance in refining the NOVA classification systems, it may be advantageous to apply the NRF9.3 (or modifications of it introduced as NRF*x.y.z in Supplementary Information to this report) as quantitative formulation characteristics and ΔNRF9.3 (or respective modifications ΔNRF*x.y.z) separately, or in the coupled FPFI^N format, for the NOVA-specific correlations with obesity and related health risk outcomes. Main handling rules and resulting benefits in applying the CMD are summarized in Table 2.

Extension of the IF&PC application frame

There is a range of desirable extensions of the IUFoST Formulation & Processing Classification (IF&PC) approach. These concern aspects of (a) considering other or further nutrients, (b) weighing nutrients included in the NRFx.y differently, (c) including additional nutrients or nutrient groups which are influenced by processing treatments leading to their reduction or increase and (d) introduce additional food matrix function criteria of nutritional relevance which may also be influenced by processing. In Table 3 a detailed overview of these extension aspects (a–d) is given, including contextual measures taken, being under exploration or of future research interest. A more detailed overview of IF&PC-based extensions on application scenarios for different consumer groups and additional categories of processing treatments is given in Supplementary Information to this report.

The exploration of relevant formulation and processing step combinations is expected to complement part of future project work for refinement and validation of the introduced IUFoST Formulation and Processing Classification (IF&PC) scheme with its integrated 2D-(F&P) Classification Matrix Diagram (CMD) tool. Main future requirements will be (i) the acquisition of nutrient data for ingredients and food component mixtures before and after different processing as well as (ii) the identification of processes or processing steps to be specifically considered in the classification. Related challenges and opportunities are expected in this context for modified definitions of the NRF to extend the application range of the IF&PC approach (see Supplementary Method 1.2 to this report).

Recommendations for NOVA refinement

This chapter is meant to extract part of the toolbox elements that have been developed by the IUFoST task force for differentiation and quantification of formulation and processing impacts on the nutrition value, in order to suggest the application of such elements for refinement of the NOVA classification system. In this context further exemplary literature supporting the NOVA classification (e.g. refs. ^{24,25,26,27,28,29}) and such that takes a critical position (e.g. refs. ^{30,31,32,33,34,35}) is considered.

The proposed steps to refine the NOVA classification system aim to eliminate ambiguities and inaccuracies that were identified by the task force and further analysed, with regard to their NOVA-based causes. Table 4 provides an overview of the results of these analyses and the proposed measures derived from these, including conclusions derived from the nutrition experts’ opinions considered in the chapter denoted as “Dietary Guidelines and NOVA” before^11,12,13,14.

In Table 4 (column 3) refinement measures for the NOVA classification system are listed to tackle the identified ambiguities and inaccuracies in the “NOVA Ambiguities and uncertainties” chapter of this report repeated in abbreviated form (as M1 to M7).

For the suggested refinement measures presented in column 3 of Table 4 the base for measures M1, M2, M4 and M5 is described by the IF&PC scheme as explored before in this report. Measures M3 and M6 of future perspective concern suggested 2nd (and/or 3rd) properties besides the nutrition value to be considered. Since the NOVA classification system reasons strongly about the palatability of a food product and related overeating by the consumer, the “Satiation Kinetics Potential (SKP)” is recommended as such 2nd target property. A related parameter relevant for consideration in the modified NRF*x.y.z could be the calorie-specific food matrix volume. This is strongly dependent on (i) water content and water binding characteristics of the food product which in turn is expected to significantly influence the filling degree and residence time in the gastric volume upon eating as well as (ii) on the rheology of the gastric content. NRF*x.y.z could consider a related volume-parameter term and processing can largely impact on such, e.g by drying or concentration (volume and water content reduction) or by water incorporation supported by gelling/thickening (volume and water content/water binding increase).

This is an attempt of suitable parameter selection but is still far from comprehensively covering the satiation/overeating/obesity relationships in all their complexity. However, this approach can be seen as a first prospective step towards the extended application of the IF&PC scheme for another target property (Satiation Kinetics Potential) of processed foods, which, in addition to the nutrition value (NV) influences obesity and related health risks. This approach could therefore be understood as a future complementary step in a refinement procedure for the NOVA classification system using the IF&PC scheme as an efficient tool.

Hence, the IF&PC scheme would enable refining the NOVA classification base at the point of purchase (POP) based on which supporters of an accordingly refined NOVA system could then build an intended food impact correlation with obesity and related health risk data. NRF*x.y.z for the formulation level and ΔNRF*x.y.z for the processing level or FPFI^N as recoupled formulation and processing level would thus be applied as quantitative classification base parameters. Such synergistic IF&PC-NOVA connection denoted here as suggested IUFoST- refined NOVA (IUVA) approach is schematically demonstrated in Fig. 8.

Summarizing this chapter, the IUFoST task force is of the opinion that the application of the IF&PC scheme will have the potential to bring the following improvements to the NOVA system, which will, in turn, assist in eliminating major ambiguities and uncertainties that exist up to now:

1. (1)

   Clear separation of formulation and processing influences on the nutritional value of processed foods which can be optimized separately.

2. (2)

   Quantification of levels of formulation and processing impacts on the nutrition value.

3. (3)

   From (1) and (2) provision of a quantitative food product classification base to apply either separate F & P levels or quantitatively recoupled F & P levels expressed by the nutrition value-related (superscript N) Formulation and Processing Food Index FPFI^N as classification parameters.

4. (4)

   Use of the classification base according to (3) for improved correlative comparison with data on obesity and associated health risks (Food Impact Correlation).

5. (5)

   Differentiating conclusions on F & P or FPFI^N influences with regard to determined food impact correlation results according to (4).

6. (6)

   Indication of the possibility to extend considerations for processed foods concerning other product properties in addition to the nutritional value (e.g.: satiation kinetics and palatability in the NOVA context, but as well sustainability, safety, convenience and affordability; see Fig. 1 and Table 5) and their handling according to the IF&PC scheme.

7. (7)

   Further development of the classification base for processed foods at the point of Purchase (POP) in accordance with (6) for appropriate quantitative use in further refined correlation analyses according to (4)

8. (8)

   Iterative refinement of the classification and correlation procedures according to steps (1)-(7), applying appropriately extended classification parameters for levels of formulation (NRF*x.y.z) and processing (ΔNRF*x.y.z) which account for quantitative digestive and metabolic response characteristics, thereby refining food impact correlations with health promotion/health risk data to ultimately derive evidence-based causal functional relationships.

Transferability of the IF&PC Approach

As introduced and discussed throughout this report, the core of the IUFoST Formulation and Processing Classification approach (IF&PC) is the separation and quantification of formulation and processing impacts on the exemplarily selected Nutrition Value as the most relevant food property parameter underlaid to the NOVA classification system. The extended Nutrition Rich Food Index NRF*x.y.z and its difference between food product states before and after a processing treatment (∆NRF*x.y.z) were successfully pre-validated as suitable Classification Parameters (CP) for quantitative formulation and processing impacts. This sub-chapter intends to demonstrate the prospective generalizability of this approach with respect to other food product characteristics than the nutrition value. As addressed in the “Key definition 1” chapter further main food product properties of consumer relevance are: (a) safety, (b) sustainability, (c) palatability, (d) convenience and (e) affordability. To apply the IF&PC scheme representatively, the property-related classification parameters (CPs) have to be selected first. Table 5 suggests CPs for the named food product properties.

Digestibility (f) and (g) metabolic response aspects could also be approached e.g. in case of motivation for a further refinement and causalization steps to be addressed to NOVA’s correlation approach between nutritional food characetristics and related obestity and health risk outcomes.

For all of the main food properties named before (a)–(g) one can explore differentiated property-specific formulation (F) and processing (P) classification parameters, derive a two-parameter (F&P) classification and consolidate to a one-parameter (FPFI^x) classification.

Such a holistic approach creates a quantifiable basis for the realistic development of food products, which usually also has to take into account other product properties in addition to the nutritional value.

For both scenarios, the NOVA System refinement and the more holistic consideration of several essential food product properties, there are demanding future R&D and industrial application requirements which relate to the systematic acquisition of data (classification parameters CP, like e.g NRFx.y.z and ΔNRFx.y applied for the nutrition value) for food product formulations before and after execution of processing steps.

Such data will support research on the IF&PC scheme application in a wider food product framework to validate and refine the scheme. Related investigations should include all suggested food properties of major relevance.

In order to successfully manage accessing and orchestrating the required interdisciplinary expertise the IUFoST task force highly recommends to the communities of food scientists, food engineers, nutrition scientists and related professionals as they are, e.g., represented by IUFoST and IUNS to join forces and encourage concerted international R&D actions.

From our perspective as the IUFoST Task Force on Processing for Nutrition, Diet and Health, we hope to have efficiently nucleated an ongoing development process that provides good prospects for successful collaboration between food scientists, food engineers, and nutrition scientists in jointly developing future classifications of processed foods and, building on this, establishing evidence-based functional relationships between the properties of processed foods and their impact on important consequences for consumers and the environment.

Concluded R&D requirements in food science/engineering and nutrition

1. (a)

   Systematic studies should consider interactions between food components/ingredients and processes and quantify NRF and ∆NRF values systematically following the IF&PC scheme approach. Such should also be carried out for serially arranged processing steps. Food databases should be complemented by such data sets for processed foods/food components considering different processes and processing intensities without mixing such up with variations in formulation.

2. (b)

   The IF&PC scheme should be systematically checked and validated based on various formulation and processing scenarios and modifications/improvements be implemented.

3. (c)

   Complementary studies should look at interactions between additives and processes and also study matrix effects impacting nutrient release and digestion kinetics. A comprehensive literature (including work from the Codex Alimantarius) already exists on several ingredients and additives, and for most of these the totality of evidence does not support harmful health effects. Such additives should not be part of a novel classification system. By contrast, the evidence used to develop current authoritative guidelines supports recommendations on nutrients and food components of public health concern, which varies by country, but can include the adequacy of certain nutrients (e.g. fibre or calcium) as well as nutrients and food components to limit (e.g. added sugars, trans fatty acids, sodium) and these criteria should be included.

4. (d)

   Harmful processes and additives should be identified. A revised definition of “Formulation- or Process-Induced Health Risk Foods” (FI-HRF; PI-HRF) needs to start with the identification of the harmfulness of processing treatments and/or single, specific ingredients using reliable health biomarkers, and by conducting dose–response relations to identify thresholds.

5. (e)

   An international balanced panel of experts from food science, food process engineering, human nutrition, and medicine should be gathered to draft the future development of the IF&PC scheme and its possible application in NOVA refinement actions. Subsequently, there should be a hearing phase to receive suggestions for improvements. The panel should also produce a research plan for studies that need to have their usefulness confirmed before being launched. Such a research plan should involve animal studies, mechanistic human studies, observational studies, and randomized controlled trials. Clear criteria for positive study outcomes should be defined, as well as the totality of evidence required to endorse the further launch of a refined classification system.

6. (f)

   Close R&D collaborations between food science, food engineering, biological chemistry, clinical nutrition, public health and toxicology researchers and those involved in regulatory and quality sciences should be encouraged and setup addressing the before-mentioned actions. Such collaborations should also include industrial partners (food producers, food processing developers, and equipment manufacturers).

Nutrition value

For Nutrition Value quantification the Nutrition Rich Food Index (NRFx.y) was applied. The Nutrient Rich Food (NRF) Index is a nutrient profiling method that scores foods based on their nutrient content per 100 kcal (418.4 kJ). It considers both beneficial nutrients (protein, fibre, vitamins A, C, E, B12, calcium, iron, zinc, potassium, and others) and nutrients to limit (saturated fat, added sugar, and sodium). The most common version, NRF9.3, includes 9 beneficial nutrients and 3 nutrients to limit. The formula calculates the percentage of daily recommended values for each nutrient, capping beneficial nutrients at 100% to avoid overvaluation of fortified foods. The final score is determined by subtracting the sum of limiting nutrient percentages from the sum of beneficial nutrient percentages. Higher scores indicate more nutritious foods. This index is used in nutrition profiling and policy-making to evaluate and compare the nutritional quality of different foods and diets.

The NRF Index was initially developed by A. Drewnowski and colleagues. Key literature sources include: Drewnowski A. (2010) “The Nutrient Rich Foods Index helps to identify healthy, affordable foods” published in The American Journal of Clinical Nutrition, introducing the NRF9.3 algorithm and validating it against the Healthy Eating Index¹ and Drewnowski A. (2009); “Defining nutrient density: development and validation of the nutrient rich foods index” in the Journal of American Colledge of Nutrition²⁰. The NRF9.3 calculations typically use nutrient composition data from national food composition databases, such as (i) USDA’s National Nutrient Database for Standard Reference, (ii) National food composition tables specific to different countries and (iii) Commercial nutrition databases. Foods are generally analysed in their retail/purchased form, though some studies specify different preparation states (raw, cooked, prepared). Hence the NRF describes the nutritional value of a food formulation. The state of the food should be clearly defined since cooking and preparation can affect nutrient content significantly. The reference amounts for daily values typically come from national dietary guidelines like the US FDA’s Daily Values or similar national standards. The description of the influence of processing methods on the nutritional value of processed food products using the NRFx.y is initially limited to changes in composition due to the enrichment or loss of qualifying nutrients (NR term; see Formula 1) and disqualifying nutrients (LIM term; see Formula 1). However, the NRFx.y definition allows for extensions to include additional terms that take into account other formulation and processing influences that contribute to the nutritional value (see chapter “Extension of the IF & PC application frame”. As an example, in further detail, the Supplementary Information addresses the group of anti-nutrients using exemplarily phytic acid, which can be present in higher concentrations in plant-based food raw materials. Phytic acid can form complexes with nutritionally important minerals (e.g., Fe, Zn) and with proteins, leading to a reduction in the nutritional value due to partial to complete loss of bioavailability of such nutrients. Processing (thermal, mechanical, and enzymatic) can inactivate anti-nutrients and thus contribute to a significant improvement in the nutritional value of correspondingly processed foods³⁶. Consequently, if an anti-nutrient term (ANz) is introduced into the extended NRF* definition with a negative sign, the representation of a modified NRF*x.y.z allows the formulation-based influence of anti-nutrients (phytic acid/phytate in the example) and the degree of influence through inactivating processes to be captured by ∆NRF*x.y.z. Accordingly, the NRF can be expanded to include additional terms that allow for further aspects of the nutritional value (e.g., protein digestibility, see Table 5) to be considered.

Processing impact on Nutrition value

As developed within the work for this report in the chapter “Differentiation of Formulation and Processing in Food Classification by definition and quantification”, the processing impact on the nutritional value is determined by calculating ∆NRFx.y, which is the difference between the NRFx.y values after and before processing. Thus ∆NRFx.y is a quantitative measure of how processing affects a food’s nutritional value defined and quantified based on the NRFx.y by comparing its nutrient profile before and after the processing treatment. The resulting value serves as a classification parameter for the degree of processing’s impact on the nutritional quality. Similar to how energy and mass balances measure the quantitative change across process boundaries (input vs. output), the ∆NRFx.y thus results from a “nutritional value balance”.

Formulation and processing food index FPFI^N

The coupled impact of formulation level (F) expressed by NRF9.3 and processing level (P) expressed by ∆NRFx.y can be quantitatively expressed by the “Food Processing and Formulation Index” FPFI^N introduced in the work reported here, the superscript N indicating the nutrition value as the considered food property being classified. Thus, FPFI^N values represent quantitative Nutrition Value classes, which are demonstrated in a Classification Matrix Diagram (CMD) introduced here with Iso-FPFI^N lines (lines of constant FPFI^N). The Classification Matrix Diagram (CMD, see figs. 6 and 7) is suggested for the graphical representation of (i) the NRFx.y as quantitative nutrition value-related “formulation level” classification parameter (CP1) dependency of (ii) ΔNRF9.3, the quantitative nutrition value-related “processing level” classification parameter (CP2).