Ultra processed food and the NOVA system

The term Ultra-Processed Food (UPF) has moved mainstream into the consumer mindset, for better and worse. Expert SA food scientist and consultant, Nigel Sunley, has penned this article for SAAFoST’s Food Advisory Consumer Service, outlining the concept and its inherent scientific weaknesses.

SUMMARY

The term ‘ultra-processed’ food or ‘UPF’ has been used extensively in both academic publications and some more consumer-oriented literature in recent years. It has become a bit of a buzzword and is normally used in the context of criticising certain types of processed food as well as the food industry itself.

The concept uses the so-called NOVA system which attempts to classify food in terms of the level to which it has been processed rather than on the established classification principles using its nutritional composition.

While there have been many academic publications from the public health nutrition community claiming that consumption of UPF is a primary cause of certain diet-related medical conditions, an increasing volume of publications from the mainstream nutrition and food science communities have highlighted the scientific weakness of the concept and the inadequacy of the NOVA classification system.

Furthermore, at no stage has it been considered appropriate as the basis for regulatory measures and attempts to use it in some countries as part of dietary guidelines have also been controversial.

While it will undoubtedly remain a commonly used phrase in the context of the nutritional quality of food, it is unlikely for the foreseeable future to be viewed as a scientifically valid concept or a viable classification system.

BACKGROUND TO THE CONCEPT

The term ‘ultra-processed food’ was first proposed by the Brazilian public health nutritionist Carlos Monteiro in an article published in the scientific journal Public Health Nutrition in 2009. It postulated that the nutritional quality of food was impacted more by the extent to which it had been processed rather than in terms of its nutrient content. His original proposals classified foods into three categories, namely:

  • Group 1 – unprocessed or minimally processed foods
  • Group 2 – processed culinary or food industry ingredients.
  • Group 3 – ‘ultra-processed foods’

The Group 3 ‘ultra-processed’ products were initially defined as ‘Processing of a mix of Group 2 ingredients and Group 1 foodstuffs, in order to create durable, accessible, convenient and palatable ready-to-eat or ready-to-heat food products liable be consumed as snacks or to replace home-prepared dishes. Specific processes include baking, frying, deep frying, use of additives and cosmetics, addition of vitamins and minerals, salting, canning and sophisticated forms of packaging’.

The concept was immediately criticised due to its vagueness and the potential classification of a major proportion of food products as ‘ultra-processed’. It was, however, seized upon by a section of the public health community who regarded the food industry and industrially processed food as the principal cause of many diet-related medical conditions and particularly obesity.

Further definitions, including a further 4th group of ‘moderately processed food’ and the NOVA classification incorporating the various food groups were subsequently developed in an attempt to introduce supposedly greater scientific credibility to the concept, although the food science community and many sections of the nutrition science community were, and have remained, highly sceptical as to the quality of the science behind the concept.

The questions about the scientific acceptability of the NOVA concept did not, however, prevent the publication of a large number of peer-reviewed publications in established scientific journals which purported to find statistically significant links between consumption of ‘ultra-processed’ food as defined by the NOVA classification and a wide variety of adverse non-communicable disease (NCD) conditions such as obesity, diabetes, heart disease, certain cancers and even cognitive functions.

The prevailing theme throughout these papers was correlation with incidence of the various conditions, rather than any direct evidence that processing rather than nutritional composition was actually their cause.

Most of the publications also used them as a platform to criticise the food industry and called for restrictions and regulatory controls on the manufacture and sale of the so-called ‘ultra-processed foods’ without providing any suggestions as to how this might be achieved in practice.

Scientific papers with similar comments of this sort have continued to be published by certain sections of the nutrition community.

The proponents of the ‘ultra-processed’ concept have also made attempts to introduce it as part of formal public health policy in several countries, mainly in South America, where there has clearly been extensive lobbying of governments by the pro-NOVA lobby.

Unsurprisingly, Brazil, where the concept originated, was among the first countries to do this but even there its use is controversial and has been condemned as excessively vague, ambiguous and unworkable for use by consumers.

ISSUES RELATING TO VIABILITY OF THE ‘ULTRA-PROCESSED’ CONCEPT’

Most of the concerns relating to the ‘ultra-processed’ concept fall into one of three categories:

  • Poor definition and insufficiently scientifically rigorous.
  • Lack of clear causative evidence that processing rather than composition is the cause of diet-related medical conditions.
  • Impracticality of the concept as a means of improving public health.

DEFINITION ISSUES 

It can be seen from the definitions above that it is very difficult for scientists, let alone consumers with limited knowledge of food composition, to establish a simple and rigorous credible definition for ‘ultra-processed’ food or for the various food categories in the NOVA classification.

Both food composition and the processes used in its manufacture at industrial and even domestic scale are extremely complex and the NOVA classification has been much criticised for its vagueness and lack of specificity. It also contains scientifically questionable components, among them:

  • Statements suggesting that a certain product is acceptable when made in a domestic kitchen, but the same product made to an identical recipe and process on an industrial scale is unacceptable. There is no scientific justification for this, and it has been criticised as being ideologically rather than scientifically based.
  • Any food product containing more than five ingredients is unacceptable. This is clearly illogical as many food products produced at home as well as industrially contain five or more ingredients and there is absolutely no scientific evidence that an increased number of ingredients would cause a food to be automatically detrimental to health. It would also suggest that food products fortified with vitamins and minerals for public health purposes (such as maize meal and bread flour in South Africa) would be unacceptable. The original Monteiro definition seen above also includes vitamins and minerals as unacceptable substances.
  • Any product containing a food additive or any substance that cannot be purchased ‘over the counter’ by consumers is unacceptable. This ignores the valuable contribution made by food additives and specialised food ingredients, all of which have undergone extensive safety evaluations before their introduction, and which ensure safety, palatability and in some instances improved nutritional quality in our food supply.  
  • The original definition of ‘ultra-processed food’ would also suggest that any products that are baked, fried or deep-fried and canned are unacceptable irrespective of their nature.

These sweeping and unsubstantiated generalisations have been extensively criticised by the food science community as well as many sections of the nutrition community as illogical, unscientific and seemingly intended to denigrate industrially processed food on ideological rather than rational grounds.

It has also been observed that no inputs were seemingly ever sought from the food science and technology professions when the concept was being put together and that its proponents appear to lack credible scientific knowledge of food and its manufacture.

CAUSATION VS CORRELATION

Nutrition science has managed to establish clear and well-substantiated links between both excessive and insufficient consumption of individual nutrients and the incidence of non-communicable diseases. For example, it is generally accepted that excessive consumption of sugars, sodium and saturated fat are undesirable while consumption of fibre, protein and some complex carbohydrates in suitable quantities is desirable for good health. There is ample credible and universally accepted scientific evidence for all these principles. Foods have consequently been historically assessed in terms of their nutritional composition and the ‘ultra-processed’ concept would therefore require a very different and more complex approach.

For the concept to have any credibility, it is therefore essential to establish clear evidence that the specific properties attributed to ‘ultra-processed’ foods are actually causing the medical conditions that are claimed by its supporters. This has not been the case so far, as the publications promoting the ‘ultra-processed’ concept have only highlighted correlation between its consumption (based on the questionable definitions above) and resulting incidence of the various medical conditions concerned.

They fail to take into consideration the fact that many of the foods supposedly defined as ‘ultra-processed’ may have unfavourable nutritional compositions and that it is these rather than processing which are likely to be the main cause of these conditions, not to mention other well established non-diet related contributory factors such as lack of exercise, sedentary lifestyle, smoking and alcohol consumption and genetic factors, all of which are well documented causes of NCDs.

In statistical terms, correlation is not generally seen as an acceptable basis for interpretation of data and acceptable conclusions can only be drawn where actual causality can be established.

However, it is still desirable to establish if the properties attributed to ‘ultra-processed’ foods (as opposed to nutritional composition) can directly cause NCDs. It is generally acknowledged throughout peer-reviewed scientific literature that evidence for this is currently lacking.

Due to the very wide range of food-related NCDs and the equally wide range of properties supposedly attributed to ‘ultra-processed’ food via the NOVA classification, it is a hugely complex task which is at present limited to a series of published review papers that summarise the work done in relation to the various possible mechanisms involved.

Perhaps the best summary to date is a paper published in April 2023 by nutrition scientists at Purdue University in the US. It focused on causes of obesity (which is a key driver behind many NCDs), is very comprehensive and contains no less than 366 references to work done in this area. Its main conclusion, expressed in the form of a ‘Statement of Significance’ was:

 ‘This review identified no mechanistic evidence directly linking ultra-processed food intake with increased body mass index and raises questions about the adoption of the NOVA system for dietary guidance at this time’.

Clearly this is not the end of the debate as, for example, factors other than obesity need to be considered (although the review concerned touched on some of these). However, the review systematically considers factors such as use of packaging, palatability and hunger stimulation, cost, use of low-energy sweeteners to replace sugar, use of food additives and changes in digestive processes that might result from the processing of foods.

In all cases no firm evidence could be found that the properties attributed to foods under the NOVA classification could conclusively be shown to be causes of obesity and, if anything, reiterated the desirability of sticking to established nutritional composition-related criteria as measures of the acceptability of different foods in relation to development of obesity.

There will no doubt be a great deal more research conducted in this area in future but the evidence to date suggests that it will require a huge amount of work to conclusively assess the specific properties of so-called ‘ultra-processed’ food in relation to NCD incidence.

In the meantime, simple common sense, as well as existing science, indicates that it is infinitely preferable for the foreseeable future to stick with existing systems of assessment that use easily measurable, unambiguous and relatively consumer-friendly nutrition compositional parameters to assess the acceptability of foods.

Nor would there appear to be sufficient evidence to invoke a precautionary principle as justification for regulatory or other initiatives in relation to ‘ultra-processed’ foods, as proposed by some public health activists. This leads us to the debate regarding the practicality of the concept as a whole, which will be discussed below.

PRACTICALITY OF THE CONCEPT IN RELATION TO PUBLIC HEALTH IMPROVEMENT

The desirability of improving public health and reducing the incidence of NCDs cannot be disputed. Nor can the importance of dietary practices in this regard – they are a significant, albeit not the only cause of NCDs, and improvement in eating practices by the general public would unquestionably be beneficial.

The challenge, however, is that behavioural considerations become critical, along with the harsh reality that most food is purchased and consumed on the primary basis of palatability and cost, with cost clearly the key factor in low-income groups and with very little consideration typically given to nutritional content, particularly by poorly educated consumers.

The question therefore arises of what is the best way to influence purchasing and consumption behaviours? A significant portion of the public health nutrition community, seemingly including those supporting the ‘ultra-processed’ concept, is of the opinion that the best way to achieve this is by means of formal government intervention using regulatory and taxation measures to effectively force behaviour change.

While a number of papers have been published suggesting that this would be a desirable approach, no conclusive evidence to this effect has yet been observed. Recent developments in Chile, where radical regulatory measures were introduced in 2016 including printing of warning messages on the labels of foods with high levels of sugar, sodium and saturated fat, coupled with extensive restrictions on the marketing of certain foods to children, have enabled the efficacy of these measures to be formally assessed.

It has been seen that some seven years after their implementation these initiatives have had at best a minimal effect on dietary habits among the population. It can be speculated that this is due to both the reluctance by consumers to reduce consumption of good tasting foods irrespective of their nutritional content, coupled with resentment towards what can be seen as government interference in free choice, particularly for food which is inherently an emotional topic.

The intervention in Chile and other similar initiatives, including one proposed in the form of regulations published for comment in South Africa in April 2023, are however, at least based on measurable parameters – in other words they use the established method of classification in terms of nutritional content.

It can be seen from the comments above on the definition of ‘ultra-processed’ food and the related parameters in the NOVA classification that any attempt to use these as formal parameters for imposition of, for example, warning labels or taxation on the products concerned would be impossible to implement in practice due to the vagueness of the parameters.

Even those parameters that could in principle be quantified would result in consumer confusion and scientific challenges. For example, any attempt to require warning labels on products containing food additives would immediately be challenged as unjustified both scientifically and legally due to the extensive evidence that exists in relation to the safety and indeed desirability of food additives.

Similarly, any attempt to impose additional taxation on baked, fried and canned products on the grounds that they are classified as ‘ultra-processed’ and therefore to be avoided would be rejected as not only heavily flawed scientifically but completely impossible to implement from a legal and administrative perspective.

There is also the potential consideration of using the ‘ultra-processed’ concept as part of dietary guidelines and as indicated above, this has been attempted in various countries. This brings into consideration the question of ease of understanding of the concept by the public.

The vagueness and lack of clear parameters in the concept have already been discussed above and a paper published by researchers from the University of Surrey and the European Food Information Council in 2022 is appropriately titled ‘‘’Even We Are Confused”: A Thematic Analysis of Professionals’ Perceptions of Processed Foods and Challenges for Communication’.

It points out that there is no consensus among informed food professionals of how processed food should be defined and divided into different processing level categories for the purpose of making food choices that will be beneficial to health. If such confusion exists among professionals, how can it reasonably be expected to be any better among consumers? The use of the ‘ultra-processed‘ concept for dietary guideline purposes is therefore, at best, questionable and is likely to only result in major consumer confusion.

The issue is further complicated by the publication in mass media and social media of misleading and sweeping, often factually incorrect statements on processed food, usually propagated by individuals with limited scientific knowledge and ideological agendas – these often originate from the more emotional activist members of the public health nutrition community.

In South Africa, the Daily Maverick, usually a source of excellent investigative journalism, is sadly one of the major culprits in this area and has published a series of poorly researched and one-sided articles on processed food written by public health activists, and without any attempt to solicit opinions from the food industry or the food science community. We suggest that consumers treat articles of this sort with caution.

CONCLUSIONS

The ‘ultra-processed’ concept is firmly entrenched in the vocabulary of the professional food community and also in that of some consumers. It is an extremely controversial concept with as yet very little scientific credibility but is typically thrown around by its supporters as a means of attacking the current system of commercial food supply which, although far from perfect, does a generally effective job of feeding the 8 billion plus global and rapidly urbanising population.

While there is certainly scope for some improvement in the nutritional quality of processed food, this is often easier said than done as fundamental questions of palatability and cost must be considered.

Use of the emotional and vague term ‘ultra-processed’ and the equally vague NOVA classification are of little value in this area. Until such time as clear evidence of causality rather than correlation between their classification parameters and the incidence of NCDs can be shown, the ‘ultra-processed’ concept will be confined to lengthy and complex academic debates coupled with emotional and often poorly researched and misleading statements in the media.

Consumers would be well advised rather to continue to use existing measurable parameters for both desirable and less desirable nutritional components when making food choices and to beware of populist statements that arbitrarily condemn processed foods.

Authored by Nigel Sunley, August 2023. Nigel can be contacted here