Who’s afraid of UPF? (Part 1)
The ultra-processed food scare is a backwards step for science
If Jamie Oliver is the fun police, Chris van Tulleken is the Taliban. The selling point of books like Ultra-Processed People is the idea that everything you know is wrong. Van Tulleken, an infectious diseases doctor and television presenter, takes this to extremes. In this book, almost everybody is wrong, many of them are corrupt and almost no one is to be trusted. Only Dr. van Tulleken, a handful of researchers and anyone who pays £25 to read this book knows the real truth. The problem is not sugar. The problem is not carbs. Artificial sweeteners don’t work. Exercise doesn’t work. Willpower doesn’t work. Every scientist who has published research contradicting his theory is in the pay of the food industry or — how’s this for an ad hominem argument? — has cited studies by people who are. The British Nutrition Foundation, the Academy of Nutrition and Dietetics, the British Dietetic Association, the Centre for Social Justice, the Institute of Economic Affairs, Tortoise Media, Diabetes UK, Cancer Research UK and the British Heart Foundation are all tainted by food industry funding. Even Jamie Oliver – Saint Jamie, the Sage of Essex — is guilty by his association with Tesco and Deliveroo, and because he makes ultra-processed food (“albeit fairly marginal items”).
It is this ultra-processed food (UPF), argues van Tulleken, that is the real cause of obesity and diet-related diseases in the world today. Food is classified as UPF if it is wrapped in plastic and contains an ingredient you don’t have in your kitchen. This includes everything from mustard to Magnums but, counter-intuitively, doesn’t include sugar, salt or fat. Van Tulleken doesn’t quite put it like this but, in effect, anything you make at home is healthy while nearly anything you buy in a supermarket, aside from raw ingredients, is bad for you.
The evidence for this striking proposition can be briefly outlined, and van Tulleken deals with it swiftly in a single chapter. Firstly, there are a number of studies using observational epidemiology which find a correlation between diets which are high in UPF and various ailments, including not only obesity, heart disease and type 2 diabetes, but also dementia, depression, cancer and more. Secondly, there is a randomised controlled trial which gave a small group of volunteers a two-week diet of either ultra-processed food or minimally processed food. The nutritional profile of each diet was similar (the same levels of salt, sugar, etc.) and the volunteers were offered twice as much as they needed to maintain a healthy weight. The people on the ultra-processed diet ended up eating 500 calories more than the people on the minimally processed diet and put on nearly a kilogram of weight.
The randomised controlled trial was published in 2019 and already has over 1,200 academic citations. Van Tulleken considers it to be extraordinarily robust, but it only really stands out because the general standard of dietary research is so poor. The volunteers were not given ultra-processed versions of the same meals. They were given totally different meals, plus very different snacks, and they could eat as much as they wanted for free. What does it actually demonstrate? Arguably, all it shows is that if you give people unlimited quantities of tasty food, they will eat more of it than if you give them blander food. Van Tulleken assures us that “the two diets were equally delicious”, but this would seem to contradict his claims elsewhere that UPF is “hyper-palatable”, delicious and irresistible.
As for the epidemiological correlations, what is it that actually correlates? UPF is an incredibly broad category encompassing most foods that are known as HFSS (high in fat, sugar or salt) and many more besides. People who eat a lot of UPF tend to have lower incomes, which correlates with all sorts of health conditions. In the study van Tulleken cites to demonstrate that UPF causes cancer, the people who ate the most UPF had the highest smoking rate and were least likely to be physically active. Epidemiologists attempt to control for such factors, but with so much going on in the data, it is an heroic assumption to think that the effect of food processing can be teased out from the effects of fat, sugar, salt, obesity, smoking, stress, exercise and numerous socio-economic influences.
That is the key question and Ultra-Processed People never adequately answers it
What is the effect of processing anyway? That is the key question and Ultra-Processed People never adequately answers it. If, as van Tulleken argues, the issue is not about specific nutrients and it is not about food being high in calories per se, there must be something else going on. He asserts, rather contentiously, that a £10 pizza from his local pizzeria is “not associated with obesity or diet-related diseases” but a cheap UPF pizza is.
Both pizzas have roughly the same number of calories, fat, salt and sugar. But one is a traditional food not associated with obesity or diet-related, while the other isn’t.
He doesn’t provide a reference for this and I doubt evidence exists for such a specific claim. More likely, he has simply inferred that a UPF pizza is more fattening because the epidemiological evidence suggests that UPF as a category is more fattening. The words “associated with” are doing a lot of heavy lifting. Is it that people eat more pizza when it’s cheap, or is it that the kind of people who eat £1 pizzas tend to be less healthy in general, or is it that there is something inherently dangerous about industrial food processing?
Van Tulleken clearly favours the last of these explanations, but what inherent dangers do ultra-processed products hold? Is it the sugar? Sugar consumption tends to rise in line with UPF consumption and UPF accounts for around 90 per cent of sugar intake. But van Tulleken dedicates a whole chapter debunking the idea that the rise in obesity can be blamed on sugar (or carbs). Worrying about sugar and carbs is so last decade.
Is it because UPF tends to be soft and therefore easier to over-eat? Van Tulleken thinks that is probably a factor — he describes UPF as being “pre-chewed” — but when it is suggested that manufacturers could change the texture of food to slow down consumption, he goes off the idea, saying “I didn’t like this. It felt a bit off.” Why? Because it involved more processing “rather than a shift to whole food”.
Is it that UPFs simply taste great and so people eat more of them?
Is it that UPFs simply taste great and so people eat more of them? This seems the obvious answer and while Van Tulleken doesn’t address this question directly, he clearly thinks so too. But where is the mileage in complaining about tasty food? No government is going to try to regulate that. Moreover, if it’s all about irresistible flavours, it is no longer about ultra-processing. He would have to admit that people can get fat from grandma’s cakes as well as Mr Kipling’s.
The one unique feature of ultra-processed food — its defining feature, in fact — is the presence of additives, those substances you don’t have in your kitchen. Van Tulleken mentions that the cheap UPF pizza “contains preservatives, stabilisers and antioxidants”. So is this where the danger lies?
It is possible, even likely, that there are additives in processed food that are doing us no good. There are historical precedents for ingredients turning out to be less wholesome than we thought, such as trans fats. This is why we have food regulators. If scientists think there is something dodgy about an additive, they should speak up. Van Tulleken raises concerns about two emulsifiers — polysorbate 80 and carboxymethylcellulose — which he says adversely affect the guts of mice in experiments. Both are deemed safe by the U.S. Food and Drug Administration (FDA) but this means nothing to van Tulleken who claims that “there is no functional regulation of food additives in the USA” and that the food industry essentially regulates itself. He then discusses one flavouring (isoeugenol) which he thinks should not be in food because a study found “clear evidence” that it caused liver cancer in mice. This, he argues, further demonstrates the failures of the FDA.
Even if this were true — and I suspect he is exaggerating the extent to which the US food industry self-regulates — the FDA is not the only food regulator in the world and it is not the only scientific body looking at food additives. The most relevant organisations for British readers are the European Food Safety Authority (EFSA) and the Food Standards Agency. Van Tulleken only mentions the former once (when he misnames it as the European Food Standards Agency) and barely mentions the latter at all except when he says that dimethylpolysiloxane was “first evaluated by the Food Standards Agency in 1969”, an impressive feat for an organisation that wasn’t created until 2000. He does not mention such bodies as the International Agency for Research on Cancer (IARC), the Committee on Toxicity, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) or any national regulators in other countries.
The EFSA, in particular, looks very carefully at food additives and in recent years has conducted further reviews into such substances as bisphenol A and nitrosamines, both of which it considers to be safe at the levels found in food. As new science emerges, safe levels are occasionally reduced to err on the side of caution. This is how regulators are supposed to work. Van Tulleken briefly acknowledges that food regulation is “somewhat better” in Europe but mumbles something about how chronic effects of additives on the microbiome are hard to test for. And so, as usual, we should simply assume the worst.
The fact of the matter is that polysorbate 80 has not just been approved by the FDA but by the EFSA, JECFA and Food Standards Australia New Zealand, among others. Carboxymethylcellulose is recognised as safe by the EFSA and the European Chemicals Agency. Both emulsifiers are on the Food Standards Agency’s list of approved additives. Five years after the study of isoeugenol in mice was published, an EFSA panel said the flavouring was safe to use in all animal feed (it had already concluded that it “does not have adverse effects on animal health, human health or the environment”.)
These additives could be killing thousands of people a year for all I know, but at some point the layman has to defer to the experts and van Tulleken has given me no reason to trust him over the regulators, especially since his summary of the isoeugenol study is misleading. It only found “clear evidence of carcinogenic activity” among male mice. It found “equivocal evidence of carcinogenic activity” among female mice and male rats, and “no evidence of carcinogenic activity” among female rats. Besides, as he says a few pages later, people are not rodents and various studies have shown that “animal testing translates poorly to humans”. In case the reader hasn’t notice that Van Tulleken has just undermined his own argument, he goes on:
I acknowledge that I’m happy to use mouse and rat data to support my own point, but there is a difference: I am trying to reduce risk to life, while the makers of food colouring are trying to sell food colouring.
This admission of confirmation bias would be endearing if it wasn’t so pompous. The only reason he mentions the problems with rodent studies is that he doesn’t think it’s fair to kill thousands of rats to test whether food additives are safe, and even if no evidence of harm emerged from tests on animals, that wouldn’t make the additives safe in his eyes. He doesn’t think the government should be involved in such experiments in the first place: “The burden of proof should be on the companies that make and use additives to demonstrate long-term safety.” Perhaps so, but since van Tulleken rejects “industry science” out of hand, this would create a Catch-22 in which safety could never be proven.
Even if it were proved that a certain emulsifier was more harmful to health than previously thought, it would not justify demonising a vast range of foods on the basis that they include one of thousands of approved additives. Van Tulleken says that Alpen is UPF because it contains whey powder. Very well, call it UPF if you want, but what is wrong with whey powder? If the answer is that you don’t have it in your kitchen then this is not science, but rather some circular parlour game. At one point, van Tulleken says that the problem with additives is “not that they are themselves harmful, it’s that the additives are a proxy for UPF.” So it’s not the additives either. Everything turns out to be a proxy for something else. It’s like Russian dolls. It is the triumph of observational epidemiology over every other kind of scientific knowledge.
People who buy petrol are more likely to drink-drive and people who buy matches are more likely to get lung cancer. These associations are real and a clue lies within them, but you need to dig deeper to really understand what’s going on. Similarly, I suspect that UPF is ultimately just a proxy for a bad diet, but that doesn’t mean that all UPF is inherently bad. The idea that Alpen, of all things, is a hazardous food product is preposterous and it illustrates how silly the UPF definition is, and yet it is van Tulleken who brings it up. Although he complains that “foods around the margin of UPF are used by the UPF industry to rubbish the whole concept”, it wasn’t the UPF industry who put a photo of wholemeal bread on the cover of his book. His definition of UPF actually gets vaguer as the book goes on until he declares that his “personal rule of thumb is: if I’m struggling with whether to call a food UPF, then it probably is UPF.” By this stage, a classification system that started with the loosest connection to nutritional science has degenerated into vibes-based guesswork.
Every now and then, someone says the quiet part out loud. Da Costa Louzada, a Brazilian scientist who has been instrumental in driving the UPF scare, tells him: “Some products are not technically UPF, but they use the same plastics, the same marketing and development processes and they’re made by the same companies as UPF … They are not home-made foods” (my emphasis). Van Tulleken’s wife considers mineral water to be UPF because a mineral water company “markets it aggressively for no other reason than financial gain.”
This is surely the crux of it. The real objection to so-called ultra-processed food is that it is mass produced by corporations for profit using ingredients that feel unnatural to some people. Of dimethylpolysiloxane (E900), an anti-foaming agent that has been studied for more than half a century and which the EFSA confirmed was safe again in 2020, van Tulleken writes:
Dimethylpolysiloxane may very well be safe. Or it may be subtly harmful over a long period through some mechanism yet to be discovered. Either way, it occurs nowhere in nature. Whatever it does or doesn’t do to the body, we’ve never encountered it previously, and evolution has had no time to accommodate it.
This is not dissimilar to the kind of rhetoric used by “sceptics” about mRNA vaccines. It is not a scientific argument and the whole UPF craze is a backwards step for science. At the practical level, it bluntens the tools of observational epidemiology by constructing a huge category of food to be studied as if it were a single risk factor. More broadly, it creates mistrust of science by treating “chemicals” as inherently suspicious and, as we shall see in parts two of this review, by portraying thousands of scientists as corrupt.
Enjoying The Critic online? It's even better in print
Try five issues of Britain’s newest magazine for £10Subscribe