DR CHARLOTTE MILLS AND PROF GUNTER KUHNLE, DEPARTMENT OF FOOD AND NUTRITIONAL SCIENCES AT UNIVERSITY OF READING - WRITTEN EVIDENCE (FDO0033)

Written evidence submitted by Dr Charlotte Mills and Professor Gunter Kuhnle on behalf of the Department of Food and Nutritional, Sciences, University of Reading, with contributions from Professor Glenn Gibson, Professor of Food Microbiology; Professor Julie Lovegrove, Hugh Sinclair Chair in Human Nutrition; Justine Norris, Lecturer in Food Quality Management; Professor Jane Parker, Professor of Flavour Chemistry; Professor Vimal Karani, Professor in Nutrigenetics & Nutrigenomics; Kate Beall (research on UPF awareness among students)

Reason for submitting this evidence

The Department of Food and Nutritional Sciences at the University of Reading is at the nexus between food processing and nutrition research. It has a longstanding international reputation for research into Nutrition and Food Science and the investigation and transformation of Food Systems. In the latest Research Excellence Framework (REF), 95% of the research was acknowledged for its international standing across various disciplines, notably Agriculture, Food, and Veterinary Sciences. The Department specialises in investigating factors influencing dietary choices and the effects of nutrition on cardiovascular, metabolic, and neurodegenerative disorders. Moreover, the research extends to examining the broader impact of food production on the environment, the development of sustainable processing method and functional foods, initiatives to reduce food waste, and the utilisation of unavoidable food waste and processing by-products for beneficial purposes. The Department therefore has extensive expertise relevant to the enquiry by the House of Lords Select Committee on Food, Diet and Obesity.

Summary

Influence of nutrition on obesity Diets high in specific nutrients can exacerbate genetic predispositions to obesity, with variations observed across different ethnic groups. Interventions changing the school environment, in combination with nutritional and physical education programs, can mitigate childhood obesity trends.

Definition of UPF and HFSS foods UPF and HFSS adhere to distinct nutrition paradigms. HFSS primarily evaluates food based on quantifiable compositional criteria, while UPF employs a broader set of factors and often overlooks advancements in food science and technology.

Recognition of UPF and HFSS foods by consumers Consumers can only recognise both, UPF and HFSS food, with additional information on food labels.

Impact of UPF on health There is a paucity of reliable data from observational, clinical and in vitro studies to support the claims of an adverse effect of ultra-processed foods beyond their nutrient profile and possibly their texture.

Role of the food and drink industry The food and drink industry plays a multifaceted role in shaping food and diet trends and influencing policymaking processes. It influences consumer trends by developing strategic marketing plans and launching new products to align with evolving consumer preferences and regulatory requirements, such as those related to health and sustainability.

Question 4:              The influence of pre- and post-natal nutrition on the risk of subsequent obesity, and the specific influences on the diet of children and adolescents that contribute to the risk of becoming obese.

Genetic influences on the diet of children and adolescents that contribute to the risk of becoming obese:

1. Professor Vimal Karani's research focuses on the impact of genetic and dietary influences on the risk of obesity in children, adolescents and adults using large population-based studies in ethnically diverse populations (such as the UK, Europe, US, Australia, Africa, South Asia, Southeast Asia, Latin America, and West Asia) with a view towards implementing precision nutrition approaches to combat obesity. While diets high in saturated fatty acids have been shown to increase the genetic risk of obesity in Western populations, Professor Karani's research using data from GeNuIne (Gene-Nutrient Interactions) Collaboration has provided evidence for diets high in carbohydrates and animal proteins to increase the genetic risk of obesity in South Asian and Southeast Asian populations, respectively. Furthermore, his study in nearly 2 million individuals from the UK, US and European populations has provided genetic evidence for the association of milk consumption with higher BMI but lower serum cholesterol levels. If the interactions between genetic variations and nutritional requirements are better understood in various ethnic groups, dietary recommendations could be personalised according to genotype to ultimately promote health and reduce disease risk.

Specific influences on the diet of children and adolescents that contribute to the risk of becoming obese:

2. Our systematic review in Latin American and Caribbean populations showed that school environmental interventions, complementing nutritional and physical education can contribute to reduce incremental childhood obesity trends.

Influence of pre- and post-natal nutrition on the risk of subsequent obesity:

3. Adverse effects of maternal vitamin D deficiency have been linked to adverse pregnancy outcomes. Our study in Indonesian pregnant mothers has demonstrated that the malnutrition status prior to pregnancy and inadequate or excessive gestational weight gain status during pregnancy as significant risk factors for developing adverse pregnancy outcomes. Our genetic study in these pregnant mothers showed that mothers of neonates with head circumference <35 cm had significantly lower vitamin D concentrations if they had a high genetic risk compared to those who carried a low genetic risk. In addition, pregnant mothers who had higher carbohydrate intake and who had a high genetic risk gave birth to babies with significantly lower birth lengths compared to babies born to mothers with low genetic risk. Birth size has been demonstrated to have an impact on cardio-metabolic health, both in childhood and adult life.

Question 5:              The definition of a) ultra-processed food (UPF) and b) foods high in fat, sugar and salt (HFSS) and their usefulness as terminologies for describing and assessing such products.

4. The identification of foods as “ultra-processed” (UPF) or “high in fat, sugar and salt” (HFSS) are both attempts to identify foods that have a potentially adverse effect on health and guide consumers in the food selection process.
5. Nutrient profiling systems, which define foods high in fat, salt and sugar (HFSS), and NOVA, which defines ultra-processed foods, are based on very different nutritional paradigms:

• nutrient profiling systems assume that the main impact on health is based on food composition, irrespective of processing,
• NOVA is based on the assumption that the main impact on health is based on processing, irrespective of food composition.

6. Nutrient profiling models use data on food composition (e.g. sugar, sodium, saturated fat, protein, fibre and energy density), and content of fruits, vegetables and nuts, to characterise foods. One of the most commonly used nutrient profiling system, Nutri-Score, is based on work by the Food Standards Agency[1] and uses an evidence-based approach based on UK dietary recommendations, Other nutrient profiling models are used, for example to provide OFCOM with a tool to regulate advertising.
7. Nutrient profiling tools use an objective measure to characterise foods, i.e. they rely on the actual food composition that can be determined by food analysis or estimated from food composition table. There is therefore no ambiguity in assigning nutrient profiles to individual foods.
8. Nutrient profiles are being used by many different countries, and thresholds are usually set by respective risk managers, for example in the UK by Public Health England or the Office for Health Improvement and Disparities (OHID).
9. NOVA, which has been introduced and developed by Carlos Monteiro, a Brazilian nutrition scientist, relies on processing methods and the purpose of processing to class foods into four categories. In contrast to nutrient profiles, NOVA does not take food composition into consideration and therefore does not make any a priori claims about impact on health.
10.      There is currently no authoritative definition of NOVA and multiple, often slightly differing, definitions have been published. For this evidence, the definition published 2019 by the Food and Agricultural Organisation (FAO) is used.[2]
11.      This definition describes ultra-processed foods (UPF, NOVA 4) as formulations of ingredients, mostly of exclusive industrial use, typically created by series of industrial techniques and processes. It includes a list of processes that are indicative of ultra-processing, but these are not exclusive to ultra-processed foods. Many foods listed in the FAO document as non-ultra-processed would meet some of these criteria:

12.      In a similar way, the FAO NOVA definition is contradictory regarding additives. It states that additives that prolong product duration, protect original properties or prevent proliferation of microorganisms can be part of all NOVA food groups and not just ultra-processed food. However, the same definition states that, for example, emulsifiers – used to prolong the shelf life of products such as breads – are an indicator of ultra-processed foods. There is thus a considerable amount of ambiguity regarding the role of emulsifiers and other additives.
13.      In addition to its ambiguity, there is a considerable amount of inconsistency in NOVA, for example:

• The FAO documents cites jam as an example for ultra-processed foods (as it is preserved with sugar), but fruits preserved in syrup are an example of non-UPF
• Bacon, ham and pastrami are NOVA 3 while sausages are NOVA 4
• Coffee, which undergoes many processing steps from bean to cup is NOVA 1, while spirits, which undergo considerably fewer processing steps, are NOVA 4
• Industrial breads made only from wheat flour, water, salt and yeast are processed foods, while those whose lists of ingredients also include emulsifiers or colours are ultra-processed. It is not clear whether for example traditional soda bread is therefore NOVA 3 or NOVA 4.

14.      NOVA does not consider many developments in food science and technology, such as processing aids or enzymes which are usually not declared on labels. This means that for example breads that do not contain any additives but have been made with flour improvers and enzymes – that do not have to be declared on a label – would be classed as NOVA 3.
15.      According to NOVA, the intention of maximising profit[3] is a characteristic of ultra-processed foods, which means self-same food could be NOVA 3 and NOVA 4, depending on intention of manufacturing.
16.      In summary, HFSS uses a clearly defined characteristic to characterise foods whereas UPF requires the interpretation of a complex, and sometimes contradictory, set of rules. The definition of UPF does not take the development of food science and technology of the last decades into consideration, which makes it impossible to classify many products such as certain breads, cereals or meat products.

Question 6:              How consumers can recognise UPF and HFSS foods, including the role of labelling, packaging and advertising.

17.      The ambiguity of the definition of UPF makes it difficult for the consumer to identify UPF unequivocally – this is exacerbated by the lack of information on processes and processing aids on food labels.
18.      There has been a tendency in public communication to refer to UPF as junk food, although this term is more commonly associated with foods high in fat, salt and sugar[4]. Indeed, communications about UPF often use pictures and images of HFSS foods instead of the full range of UPF. While there is an overlap between UPF and HFSS foods, this overlap is not complete and many more than one third of ultra-processed foods sold in the UK according to Open Food Facts have a positive nutrient profile (Figure 1). There is therefore a clear difference between the common perception of what ultra-processed foods are and the actual definition.
19.      Even experts are not always able to identify ultra-processed foods without difficulty. In a recent study by the developers of NOVA in Brazil, the researchers had to use a multi-stage process to assign different foods to the correct category.[5]
20.      A survey of 80 students at the University of Reading as part of a research project showed that while about 30% of students were familiar with the term ultra-processed foods, there were a lot of misconceptions about the actual definition and many students assumed that all non-organic or processed foods were ultra-processed.
21.      Due to the difficulty in interpreting the definition, there have been many simplified explanations in public-facing communications which fail to capture the actual definition and result in more confusion and concern by the public:

• everything wrapped in plastic – many fruits and vegetables are wrapped in plastic to increase shelf-life and maintain freshness. Cheese, bacon and ham (NOVA 3) are often wrapped in plastic;
• foods with more than 5 ingredients – many recipes include more than five ingredients, including a wide range of traditional foods;
• foods with a health claim – the Great Britain Nutrition and Health Claims (NHC) permits health claims for a range of foods that are not ultra-processed, such as oats or olive oil;
• foods that could not be made at home or that have not been used by past generations – many foods that are not-ultra-processed according to NOVA cannot be easily made at home, such as for example ham and bacon, or tinned foods. Many foods, including minimally processed foods, have only recently been introduced to the UK market and were therefore unknown in the past.

22.      The role of additives in the definition of ultra-processed foods is frequently discussed, and additives are often suggested as an indicator of UPF. However, this is not in agreement with NOVA which explicitly states that “additives that prolong product duration, protect original properties or prevent proliferation of microorganisms” can be found in all NOVA classes, including the group of minimally processed foods. Many of the additives described as characteristic for ultra-processed foods, such as emulsifiers, are added to prolong shelf-life and the definition is therefore self-contradictory.
23.      In summary, consumers can recognise both, UPF and HFSS food, only with additional information on food labels. Misleading information about UPF, such as simplified rules or conflation of UPF with HFSS foods, lead to confusion of consumers.

Question 7:               The cost and availability of a) UPF and b) HFSS foods and their impact on health outcomes.

Nutrition profile of UPFs

24.      The impact of individual ultra-processed foods on health outcomes is difficult to estimate due to a lack of reliable data from clinical intervention studies. However, based on their nutrient profile, approximately one third of ultra-processed foods recorded in the Open Food Data database have a Nutri Score rating of “A” or “B”, suggesting an overall beneficial nutrient profile. A comparison of salt, sugar and fat content between the four different NOVA classes show that many ultra-processed foods are in low category of the UK’s traffic light label (Figure 2).
25.      Individual ultra-processed foods, such as breads and cereals, can be an important source of fibre[6] and wholegrain, and fish fingers can provide a source of fish, especially for children.
26.      Processing – including processes that result in a classification as ultra-processed – are crucial for the reformulation of foods to improve nutrient profiles, for example by reducing salt, fat and sugar content in foods. One example is the reduction of salt in bread: according to data collected by Action on Salt, new processing methods, that often would fall within the criteria mentioned within the UPF-definition, have allowed a salt-reduction in breads of about 10%[7]. In contrast to these breads, with an average salt content below 1 g/100 g, many artisanal loaves – often recommended as being healthier than UPF breads – have considerably higher salt contents of up to 2 g/100 g. These reformulations were crucial to reduce the overall salt intake in the UK in the last two decades as part of the government’s salt reduction campaign.

Impact of UPF on health – data from observational studies

27.      The majority of research investigating associations between UPF intake and health is from observational studies, A recent umbrella review published in the British Medical Journal[8] found that the majority of data are either of very low or low quality, that is that the true effect might be markedly different from the estimated effect. The only associations between UPF intake and adverse health outcomes that were of moderate quality were those with obesity, type 2 diabetes and all cause mortality.
28.      A key limitation of these observational studies is the reliable estimate of intake. Many studies rely on food-frequency questionnaires which are often limited to a small number of foods and are therefore unable to differentiate between processed and ultra-processed foods. However, even more detailed methods do not collect sufficient information to distinguish between different degrees of processing and need to rely on estimates, such as assuming all bread consumed in the UK is ultra-processed while all bread consumed in France is processed.
29.      Studies investigating individual ultra-processed food groups often found that the majority of these groups are not associated with disease risk and only a small group, for example sugar-sweetened beverages, are associated with an increased risk. For example, Chen et al. (2023) found that among all ultra-processed foods, only artificially and sugar sweetened beverages, animal-based products and ready meals to be associated with an increased risk of diabetes, while ultra-processed cereals, snacks and other foods were associated with a reduced risk.
30.      There have been only limited studies investigating the health effect of individual aspects of ultra-processed foods on health, namely additives. However, like other observational studies, it is impossible to obtain reliable measures of exposure. NutriNet-Santé, the main observational study investigating associations between additive intake and adverse health outcomes[9], relies on a small number of samples to estimate overall intake of emulsifiers, and there is no information about the methods used – considering the difficulty measuring carbohydrate based-emulsifiers, it would be important to explain the methods used.
31.      The use of additives in foods is not standardised and depends on a range of factors, including properties of raw materials and processing environment. It is therefore impossible to estimate the actual intake of additives from food records alone and observational studies have to rely on average food content data to estimate intake of additives. We have shown that this introduces a considerable bias and makes estimated associations between intake and health outcomes unreliable[10]. While it is commonly assumed that measurement errors result in an attenuation of observed effect sizes, this is not always the case.
32.      There is a paucity of clinical studies to compare the impact of processed and ultra-processed foods on health. Most data currently available, in particular the study by Hall et al. (2019)[11], compared minimally processed (NOVA 1) with ultra-processed foods (NOVA 4). Without processed food as control intervention, it is impossible to infer any effect of ultra-processing on health.
33.      Current evidence suggests that ultra-processed foods are hyperpalatable and result in overconsumption, however, this can be explained by differences in the food texture[12]: homogenised food, for example as so-called smoothie, can often result in overconsumption.
34.      While studies investigating the effect of ultra-processed foods – and in particular additives – on the microbiome often show differences in the composition of the microbiome, it is not possible to infer a health effect from these data. Fluctuations in ostensibly health promoting genera like the bifidobacteria, lactobacilli and concomitant decreases in pathogens such as proteolytic bacteroides, clostridia may be useful but must be tallied with robust health biomarkers. These can include end products (organic acids, gases), immune modulators and inflammatory status (or symptoms in patients). Many studies have used unphysiological preparations of additives in water – which is not representative of their actual occurrence in foods and conducted studies in rodents. The latter is a meaningless model given gross differences in diet, anatomy, gut microbial composition, coprophagy and host-microbe interactions.
35.      In summary, there is a paucity of reliable data from observational, clinical and in vitro studies to support the claims of an adverse effect of ultra-processed foods beyond their nutrient profile and possibly their texture.

Research requirements

36.      The disparity between the public perception of risk associated with ultra-processed food consumption and the limited availability of empirical data linking it directly to adverse health outcomes highlights the importance of high-quality research in this area.
37.      A key limitation of epidemiological research and the estimation of the prevalence of UPF intake is the lack of reliable methods to estimate the intake of ultra-processed foods; this limitation has also been highlighted in the SACN report on processed foods[13]. The development of better methods – including the development of objective measures using biomarkers – is crucial for future research and a better understanding of the impact of UPF on health.
38.      While there is a lot of focus on the difference between processed (NOVA 3) and ultra-processed (NOVA 4) foods, there is a paucity of data as most studies compare minimally processed (NOVA 1) with ultra-processed (NOVA 4) foods. Future research should focus on difference between processed (NOVA 3) and ultra-processed (NOVA 4) to provide a better understanding of the differences between those two categories.
39.      There are insufficient data on the effect of additives – in particular emulsifiers – on the human gut microbiome to understand the potential effect on health. There are two important research requirements to address this:

• Animal and in vitro models do not provide reliable data (see above) and human intervention studies are therefore crucial.
• A commonly accepted framework to interpret changes to the gut microbiome is needed to suitably interpret results from these studies.

40.      There is currently also a paucity of data regarding the health impact of different processing techniques. Such data would allow a better categorisation of foods and a better elucidation of underlying mechanisms.
41.      The development of a framework to integrate data from observational and intervention studies – similar to frameworks used in toxicology[14] –  is required in order to understand the potential impact of ultra-processed foods on health.

Question 8:              The role of the food and drink industry in driving food and diet trends and on the policymaking process.

42.      The role of the food and drink industry in driving food and diet trends and influencing policymaking process is complex. From the aspect of policymaking, industry representation may be visible using a number of approaches. For example, as key stakeholders, trade associations, members of scientific panels and advisory committee members as well as attendees at public hearings and providing opinion (industry and scientific) during the public consultation process. An example of respondents to public consultations would be the British Food and Drink Federation or the Institute of Grocery Distribution (IGD), industry recognised private standard schemes such as BRCGS, industry scientific experts and food manufacturers and businesses.
43.      The public consultation process allows the food and drink industry to voice concerns or opinions regarding the practicality (and timing) of implementation and any specific barriers foreseen which may affect the effectiveness of the proposal and impending change or creation of legislation. Clearly, the food and drink industry may lobby any proposal which would be detrimental (financial or impractical to implement) to their business sector (or individual businesses).  Industry can also advocate for change in policy and positively influence the need for legislative change as a result of scientific research, consumer trend data and technological advances. Moreover, it is the role of the government via regulators to make decisions in the interest of public health and provide the relevant guidelines and regulations.
44.      The Food and Drink Industry also plays a pivotal role in influencing consumer trends and creating new ones by developing strategic marketing plans and responding to changes in their sector. This may occur due to several factors including compliance with revised or new legislation such as HFSS. It may be achieved by modelling the business marketing strategy with consideration to future trends and influences (trends such as health and wellbeing, the economy, climate /sustainability, fair trade, organic, vegan, plant based, dietary requirements etc.) then developing and launching new and reformulated products to meet the consumer need / legislative requirements.
45.      To mitigate any negative business impact brought about by legislative changes, product strategy realignment, review of potential opportunities for new products and creation of new product sectors such as plant based and “children's cereals” with added fruit, fibre and vitamins, healthy HFSS compliant snacks is a key survival response but also helps to shape future trends and new market growth.

4 April 2024