A study published in BMJ looks at the relationship between emulsifier consumption and cardiovascular risk.
Prof Kevin Whelan, Professor of Dietetics, Head of the Department of Nutritional Sciences, King’s College London, said:
“Emulsifiers are additives used by food industry to help oil and water to mix together and to stay mixed together. In doing so they make foods such as dairy desserts, ice creams and chocolate look more appealing and taste creamy. By allowing bread dough to mix better and give it a stronger structure they also prevent large air pockets in our loaves.
“Emulsifiers are widely used around the world, here in the UK there are more than 6000 foods with emulsifiers added to them. Emulsifiers are types of food additives usually, but not always, described on food labels ranging from E400 to E499.
“We now know from studies in animals and in the laboratory that some emulsifiers can change our microbiome, and cause inflammation in the gut, however there is very little research undertaken in humans.
“This recent study followed over 95,000 people in France and measured whether their background intake of emulsifiers was related to them developing cardiovascular disease.
“What is a first here, is that the researchers were able to measure emulsifier intake in such granular detail. Emulsifiers are in some breads and not others, some ice-creams but not all. But because the researchers not only knew what food people ate, but what exact brand of food it was, they were able to calculate how much total emulsifiers, and which type of emulsifier they were eating.
“Some emulsifiers were found to be eaten in higher amounts by people who then went on to develop cardiovascular disease, including cellulose (E460), carboxymethylcellulose (E466) and monoglycerides and diglycerides of fatty acids (E471, E472).
“This study included a large number of otherwise healthy people and they were followed up over a long period of time. However, with this type of research we cannot be certain whether emulsifiers caused the cardiovascular disease, or whether some other aspect of diet or lifestyle may be responsible.
“The evidence is not yet strong enough to advise everybody avoids eating emulsifiers. To identify a cause and effect we need to do clinical trials.”
Prof Kevin McConway, Emeritus Professor of Applied Statistics, Open University, said:
“This new study demands a lot of care in interpreting its results. The research paper begins (in its introduction) and ends (in its final sentence, also quoted at the end of the press release) in talking about ultra-processed foods. It seems clear that an important reason for doing the research was the concern that has been expressed about associations between high consumptions of ultra-processed foods and increased risk of several diseases, and in particular the hypothesis that these associations can be explained by the properties of additives in ultra-processed foods. Definitions of ultra-processed foods generally refer to them containing ingredients, often additives, that wouldn’t be seen in a home kitchen or in traditional cooking, and there has been concern in particular about emulsifiers.
“That final sentence in the paper and in the press release, “Meanwhile, several public health authorities recommend limiting the consumption of ultra-processed foods as a way of limiting exposure to non-essential controversial food additives,” isn’t really (in my view) building on the actual findings of this study. It’s true that public health authorities in some countries do make this recommendation, but this research would support it only in relation to quite a small minority of the additives they considered. So really they are only stating a fact about what some public health authorities have done, not about their own findings.
“This study, then, doesn’t look at ultra-processed foods in general or additives in general. It uses data from a large cohort study in France, the NutriNet-Santé cohort study, to investigate possible associations between the consumption of different amounts of emulsifiers in food and cardiovascular disease (CVD). And the researchers did find statistical evidence of associations between higher consumption levels of a small number of the emulsifiers that they studied, and CVD risk.
“One obvious snag with this is that, as the research paper and the press release make clear, the study can’t tell us that higher consumption of these emulsifiers actually causes the increased CVD risk. That’s because the study is observational. The problem is that there are many other differences between the groups of participants who consumed different amounts of emulsifiers, apart from the amounts of emulsifier they consumed. That’s made clear in Table 1 in the paper, which reports differences between the consumption groups in age, BMI, family history of CVD, education, smoking, physical activity, and other aspects of dietary intake. So an observed association between emulsifier consumption and CVD risk could be caused by the emulsifier consumption, or it could be caused by one or more of the other differences between the groups. Of course, the researchers realise this, and they made statistical adjustments to allow for differences in a wide range of other factors, including those in Table 1, and others. I think this adjustment was done appropriately – indeed I think the statistical analysis methods used in the research are generally good. But you can never be sure that everything relevant has been adjusted for – you can’t adjust for a factor on which you have no data, for example – so it remains possible that emulsifier consumption has no role in increasing CVD risk, or that it only has a very small role alongside other factors.
“Two quick points about the adjustments that were made. Importantly, there was an adjustment for the amount of ultra-processed food that the participants ate. That’s necessary, because it’s still plausible that some other ingredients in some ultra-processed foods might be a cause of increased CVD risk, and, as you’d expect, the data in Table 1 indicates that people who consume more emulsifiers also tend to consume more ultra-processed food generally. So this study is looking for associations between emulsifier consumption and CVD consumption that are independent of general consumption of ultra-processed foods. Also, there were adjustments for what the researchers call “markers of overall diet nutritional quality”. It’s not clear whether, by quality, they mean simply a property (good or bad) that something has, or whether they are saying that (for example) higher total fibre consumption means that a person’s diet is of higher quality in the sense that it’s better. The word ‘quality’ can mean either of these, in English (or (as ‘qualité’) in French). But this is irrelevant anyway – the point is that it makes sense to adjust for various aspects of diet that might be associated with emulsifier intake in some way, and might have strong links to CVD, whether or not they might mark what counts as a ‘good’ diet.
“It’s important to be clear that the researchers found statistically valid evidence of association between just five particular emulsifiers, and also two groups of emulsifiers, and CVD risk. Four of the specific emulsifiers where they found an association are within the two groups where they found an association, so to some extent the association in those groups will be reflecting the association for the specific emulsifiers within the groups. But the researchers examined in detail the evidence for 28 different emulsifiers, and eight groups of emulsifiers. (They also collected data on another 32 emulsifiers, and included consumption of those in their figures for total emulsifier consumption, but left out any more detailed analysis of the 32 because less than 5% of the participants had consumed each of them.) This means that they did not find statistically valid evidence of an association between 23 individual emulsifiers, that they did analyse, and CVD risk, or between six groups of emulsifiers and CVD risk. Now, not finding statistically valid evidence for an association does not mean for certain that there is no association – but these researchers did look in detail at over 95,000 participants and didn’t find evidence of association with CVD for a pretty big majority of the emulsifiers they looked at. On the whole, that seems to be evidence against the hypothesis that all, or most, emulsifiers are linked to bad health outcomes.
“The fact that the researchers looked at so many different emulsifiers does increase the chance that one or more of the associations that they did find could simply be the result of chance. In fact the researchers also report what happened when they made an adjustment to allow for the fact they examined many emulsifiers. What happens is that the evidence for an association with CVD becomes too weak to meet the usual threshold of evidence for two of the specific emulsifiers (E460 cellulose and E339 trisodium phosphate). Evidence does still remain for three of the specific emulsifiers, and the two overall groups into which those three fall, but really we’re down to solid evidence only on a few of the emulsifiers that were considered.
“Normally one might expect a research team to be very clear on the statistical methods they intend to use, including whether or not to make an adjustment for looking at many different emulsifiers. You’d expect that to be reported in the pre-registration of the study, that lays out the approach and methods before the study is carried out. The research paper does mention a registration number – but it’s for the entire, long-running, broad NutriNet-Santé cohort study, not for this particular use of data from it.
“So I certainly don’t take this as evidence that the existence of an emulsifier in the contents list of a food product means that eating a lot of it will increase your CVD risk – generally, there’s the issue that we can’t establish cause and effect anyway, but besides that, this study did not find statistically valid evidence for an association at all, for most emulsifiers that it included. I think the researchers realise this when, in the final paragraph of the paper, they call for further observational studies and also experimental studies.
“In that paragraph the researchers also mention that the associations they found are of “moderate magnitude”. To see what this actually means, consider the association that they found that is numerically strongest – that between total mono- and diglycerides of fatty acids and coronary heart disease. (Overall they found more evidence of associations with heart disease than with cerebrovascular disease (strokes etc.).) During the average length of follow-up in this study, 7 years, out of every 1,000 participants, about 11 developed coronary heart disease. Now suppose that everyone in the study had consumed an extra amount of mono- and diglycerides of fatty acids that amounts to one standard deviation of consumption. A standard deviation is a common statistical measure. To make sense of it here, note that the actual participants consumed about 200 milligrams of these additives per day, and one standard deviation is about 275 milligrams, so that’s a pretty major increase in consumption, particularly for those who weren’t consuming much to begin with. Also, assume for now that it is really the consumption of these emulsifiers that causes the increase in heart disease risk – even though it might actually not be the cause. Then the number of coronary heart disease cases in 1,000 participants during the follow-up period would go up from 11 to 12. Unfortunate for that one extra person out of 1,000, certainly, but ‘modest’ does seem an appropriate word here.
“The researchers are right to point out that this relatively small effect could affect a lot of people – even in this health-conscious cohort, 84% of participants consumed at least some of this particular group of emulsifiers, though they make up only 5% of their total consumption of emulsifiers. But given all the statistical and other uncertainty, and in particular the uncertainty about whether adjustments have been made for all relevant factors, we really do need some more research and better understanding of how these emulsifiers might operate to increase CVD risk, before we should be concerned about their safety, I’d say.
“The researchers also mention several limitations of their study, for example concerned with the accuracy of measurement of diet. It’s not clear what direction these limitations would, overall, bias the findings, though the researchers do argue that the findings might be stronger if these measurement issues were not present. I also wonder how far it matters that all the dietary measures were made towards the start of people’s participation in the study. This means they could not pick up any later changes in diet – and you might expect such changes in this cohort, given that it consists of people with an interest in diet and health. So doing more research that tries to overcome these limitations is important – but it certainly won’t be easy.
“One final point about definitions of ultra-processed foods. This study did take a wide definition of emulsifiers, and the researchers are not claiming that all food that contain any of these substances is ultra-processed. But to emphasise this, Figure 2 in the paper shows that the second most commonly consumed emulsifier consumed by the participants was sodium bicarbonate, E500, which made up 27% of the participants’ total emulsifier consumption. Sodium bicarbonate, or bicarbonate of soda or baking soda as it’s more commonly known in a cooking context, does indeed have an E number, but like some other substances with E numbers, it finds a place in home kitchens. So if I make some soda bread at home (as indeed I do sometimes), am I making an ultra-processed food? Baking soda, which is usually thought of as a raising agent rather than an emulsifier, has been in use in home kitchens for well over a century. And what about Victoria sponge cake? That is made with self-raising flour, or using baking powder, and both of those generally contain sodium bicarbonate. This research did not find any association between sodium bicarbonate consumption and CVD, nor between consumption of the largest group of emulsifiers consumed by the participants, modified starches (a third of total consumption), and CVD. Together sodium bicarbonate and modified starches amounted for well over half of the participants’ consumption of emulsifiers. The next largest set of emulsifiers by consumption is pectins (6% of consumption), which again are found in home kitchens, typically for making jams set, and where again no association with CVD risk was found. So what this certainly shows me is that the usual definitions of ultra-processed foods are far from clear.”
Tracy Parker, BHF Heart Health dietician, said:
“Although this research shows a link between higher intakes of some emulsifiers and a greater risk of cardiovascular disease (CVD), this type of study can only show an association and we need more research to properly understand the link between ultra processed foods (UPF) and heart disease.
“Cutting down on UPF foods like cakes and biscuits and cooking more from scratch are already things we know can help improve our diets and, in doing so, lower our CVD risk. It is also essential to create an environment that supports this, by implementing delayed policies that are already on the table to restrict the advertising and promotion of often highly processed foods which are high in fat, sugar and salt.”
Dr Gavin Stewart, Senior lecturer in Evidence Synthesis, Newcastle University, said:
“Although this study demonstrates a potential effect there is a need for cautious interpretation. The strength of evidence from a single observational study is inherently low because of uncertainty about confounders and causal relationships. Further studies and evidence synthesis are required to reduce this uncertainty.”
Dr Duane Mellor, Registered Dietitian and Senior Lecturer, Aston Medical School, Aston University, said:
“This is another report from the French group which suggests a less healthy dietary pattern is associated with an increased risk of heart disease and stroke. This study collected online dietary questionnaires from over 95,000 individuals and followed them up for 7 years.
“The analysis suggested an association between a number of emulsifiers, a type of food additives, and risk of heart disease and stroke. However, the emulsifiers, which the paper names by both their E numbers and chemical names, represent a variety of compounds, some of which are naturally occurring in food (and may even have health benefits e.g. lecithin and cellulose), and others that are synthetic and have been suggested in animal studies to be associated with negative effects (e.g. polysorbate 80). The finding that cellulose (which is E460 and is a naturally occurring plant fibre and occurs in most foods), was associated with increased risk of heart disease, when fibre in previous studies was actually associated with health benefits (especially bowel health), suggests that the association reported in this study is unlikely to be directly related to the cellulose itself. It is perhaps more likely the overall composition of the foods it is contained in, along with the overall diet, that can be associated with an increased risk of heart disease. However, other emulsifiers including carboxymethylcellulose and some of the esters of mono-and diglycerides are not found naturally in foods and could plausibly have different health effects. This study cannot confirm any causal relationship, because it is an observational study. Also, for some of the emulsifiers there is limited plausibility of mechanisms that could be behind an association with cardiovascular disease. Finally, the size of the effects were small – only small effects on the chance or hazard ratio of higher intake of additives being associated with increased risk of heart disease or stroke. Overall, therefore, this study provides weak evidence of an effect of emulsifiers as a class of food additives on risk of heart disease or stroke.
“It is important to recognise that as this study estimated intakes of additives, whilst not accounting for naturally occurring levels of cellulose (which may have shown a very different effect), it is possible that what this study is actually showing is that a less healthy diet is linked to increased risk of heart disease and stroke. As such it does not change what is known about heart disease and stroke, as the foods which were high in these emulsifiers and were associated with increased risk were foods like pastries, cakes, biscuits and dairy desserts. Perhaps rather than focusing on which foods contain emulsifiers and theoretical links to an increased risk of heart disease, we should focus on helping people to consume dietary patterns that we know are healthy, which are based on vegetables, fruits, nuts, seeds and legumes along with small amounts of dairy foods, and unprocessed meats.”
Prof Gunter Kuhnle, Professor of Nutrition and Food Science, University of Reading, said:
“This is a very well conducted study that shows that people who consume foods that contain emulsifiers have a very slightly increased risk of heart disease. The observed increases are very small compared to other dietary risk factors of disease.
“It is not possible to attribute the findings to emulsifiers. The study relies on self-reported dietary data and estimates emulsifier intake from this information. Considering that emulsifiers are found in a wide range of foods, it is impossible to attribute observed effects to these compounds and not specific dietary patterns. Many foods with unfavourable nutrient profile contain emulsifiers and this might also explain some of the results.
“A key limitation of the study is that the group of emulsifiers is very diverse and likely to have very different effects on the human body. Some – such as cellulose and cellulose derivates – are not absorbed and act similar to dietary fibre. Others – such as fatty acids – are absorbed and metabolised like dietary fat. It is therefore unlikely that all emulsifiers have the same effect on health by increasing the risk of heart disease.
“Emulsifiers have been investigated by regulatory agencies such as EFSA for some time and found to be safe in the amounts used. It is important to note that many studies that found adverse effects of emulsifiers on health were conducted in animals with a very different digestive system using formulations that were different from the normal diet.
“The statement by the authors that several public health authorities recommend limiting exposure to “controversial food additives” is very misleading at least in the context of the UK and the EU: food additives are tightly regulated. The relevant agencies set acceptable daily intakes (ADI) that are based on a thorough risk assessment and take uncertainty factors into consideration. The ADI is the amount that can be consumed daily over a lifetime without presenting an appreciable risk to health. If the relevant public health authorities were aware of a risk from “controversial food additives”, it would result in a change of the ADI and not a vague recommendation to the general public.
“There are no recommendations against the consumption of ultra-processed foods in the UK or in the recently published Nordic dietary guidelines.”
‘Food additive emulsifiers and risk of cardiovascular disease in the NutriNet-Santé cohort: prospective cohort study’ by Laury Sellem et al. was published in the BMJ at 23:30 UK time on Wednesday 6 September 2023.
Dr Gavin Stewart: “No known COI.”
Prof Gunter Kuhnle: “Former member of EFSA panel that evaluated emulsifiers. Member of the COT (Committee on Toxicity).”
Dr Duane Mellor: “No conflicts of interest.”