Picture credit: Monticelllo/Getty

Who’s afraid of UPF? (Part 2)

Bad science is hard to swallow

Chris Van Tulleken begins his book by encouraging people to eat ultra-processed food (UPF) while they read it. This idea was inspired by Allen Carr’s Easyway, the stop-smoking book that promises to put you off smoking while you read it. Van Tulleken frequently compares UPF to cigarettes. He equates the food industry with the tobacco industry and portrays UPF as addictive. In part, this stems from his own experience. He recounts how he and his brother looked at the eleven diagnostic criteria for addiction in the Diagnostic and Statistical Manual of Mental Disorder, and both scored nine when thinking about their favourite (UPF) food. The threshold for a severe problem is six. 

This is one of several indications in Ultra-Processed People that the author does not have a particularly healthy relationship with food. He thinks about dinner while he’s having breakfast. He used to hassle his twin brother Xand about his eating so much that Xand dreaded seeing him. When he goes on a diet of 80 per cent ultra-processed food, he immediately eats five bowls of Coco Pops in a row. When he switches from Coke to Diet Coke, he is soon drinking six cans a day. In a slightly disturbing aside he recalls, “I would occasionally eat to the point of vomiting. This has never struck me as disordered so much as practical: it made a good night’s sleep easier if, having massively overeaten, I wasn’t bursting with food.”

To prove that UPF is addictive, he cites statistics from the USA:

The transition from trying UPF to being unable to stop using it is extremely high: 40 per cent of the US population live with obesity and we know that the majority of them will try to lose weight in a given year. Cessation rates are so low as to be non-existent. There is no other drug that, having tried it, 40 per cent of people will continue to use regularly despite negative health consequences (a definition of addiction).

This only makes sense if obesity is a direct proxy for UPF consumption, but that is only van Tulleken’s unevidenced assumption. What he is really saying is that people who are obese rarely lose weight. Giving up UPF, by contrast, seems to be surprisingly easy. Van Tulleken says that “a number of people involved in the process of making this book” have given up UPF, and he anticipates many readers of Ultra-Processed People doing the same (hence the Allen Carr reference). He himself “stopped eating UPF immediately and completely” once his pound shop Super Size Me experiment was over. His brother Xand “stopped eating UPF and hasn’t looked back”. There is no mention of any withdrawal effects.

This seems inconsistent with his claim that UPF is highly addictive and that people are trapped in an environment that doesn’t allow them to eat healthily. He spends a whole chapter explaining why willpower doesn’t work. Books like this usually do. If losing weight were as simple as eating less and moving more, there would be no need for diet gurus. There is, inevitably, also a chapter about why physical activity doesn’t work either. Van Tulleken acknowledges that: “There are large numbers of papers in respectable journals that contradict the hypothesis that UPF is a significant contributor [to weight gain]. These papers claim to show that inactivity is a primary driver of weight gain, and that increased calorie intake is less important.” He pointedly notes that there are “a few authors that come up repeatedly in these papers, including Steven Blair, Peter Katzmarzyk and James Hill”. His reason for naming these three individuals will be revealed at the end of the chapter. 

After outlining a number of studies showing that physical inactivity is linked to obesity and that people are, on average, less physically active than they used to be, he mentions a 2014 paper that had “enormous influence”. It used data from various sources, including DEFRA’s annual diet survey, to show that calorie consumption in the UK has gone down since the 1970s. I was the author of that paper. I don’t recall it having much influence at all, but his reason for focusing on my summary rather than on the underlying source material will also become clear by the chapter’s end.

Van Tulleken’s first reaction to the news that calorie consumption has gone down, not up, since the 1970s was that “it struck me as surprising the food industry could still be profitable with a 21 per cent reduction in calories”. It’s difficult to know where to begin with such a silly statement. Let us just say that it was no loss to economics when van Tulleken enlisted at medical school. 

His second reaction was to notice that the diet surveys suggested that people were consuming an average of around 2,000 calories a day which, for most people, would lead to weight loss. He points out that tests using doubly labelled water show that people are actually consuming around 2,500 calories a day. This can only mean that people under-report how much they eat. Moreover, the fatter people get, the more they under-report. Therefore, he writes, “Snowdon’s ‘Fat lie’ article misunderstood the data: calorie consumption isn’t in decline — it has been increasing for a long time.”

I didn’t misunderstand the data, though. Here’s what I said in the paper:

… it is well known that people tend to under-report the amount of food they consume due to a desire to deceive or — more commonly — a tendency to forget (over-reporting is also possible, though less common). The alternative method of keeping till receipts to check what food has been purchased is also problematic because some food is thrown away.

Researchers are well aware of these issues and have ways of testing the degree of under-reporting, notably with urine tests using ‘doubly labelled water’ which show how much energy a person has expended (and, therefore, how much energy a person of steady weight has consumed). Nevertheless, it is believed that Britons throw away about 10-20 per cent of the food they buy and under-report how much they eat by around 20 to 40 per cent (WRAP, 2013; Macdiarmid and Blundell, 1998).

When studying dietary trends over time the question is not whether people under-report but the extent to which under-reporting has changed over the years, if at all. Women and the obese are most likely to under-report and whilst the proportion of women in the population has remained stable, the proportion of obese people has clearly increased. It is therefore possible that more obesity has led to more under-reporting, but it is very unlikely that the population has become so forgetful and dishonest that the large, steady and virtually uninterrupted decline in calorie consumption reported in successive studies can be explained by misreporting alone.

In 2016, the Behavioural Insights Team produced a paper which attempted to correct for under-reporting of calorie intake using doubly labelled water. It estimated that calorie intake rose between 1990 and 2008 at a time when the survey data indicated a small decline but that the much greater decline between 1974 and 1990 was real. Overall, people were eating fewer calories in 2008 than they had in the 1970s.

It is possible that these estimates are also wrong and that calorie consumption has been steadily rising since 1974, but that would require taking the original graph and turning it upside down. When I said in the first part of this review that the general quality of dietary research was poor, this is the kind of thing I was talking about. If we’re not even sure how much food people have been eating, good luck trying to find out what type of food they have been eating.

Van Tulleken doesn’t mention the Behavioural Insights Team paper, nor does he provide any evidence that Britons are consuming more calories than they did in the 1970s when obesity is assumed to have been negligible. He simply asserts that: “The evidence is clear that we are eating more calories than ever and that trying to change our energy expenditure is not going to make a significant difference to weight.”

To support the second of these claims, he cites the Hadza hunter-gatherers of Tanzania. These people will be familiar to anyone who has read a book like this before, such as Henry Dimbleby’s Ravenous. A 2012 study found that their daily energy expenditure was not much different to that of your regular desk-bound Westerner, and yet none of them were obese. Van Tulleken admits that when the Hadza are not hunting or gathering, they rest a great deal but he incorporates this into his theory by inferring that “if we are active, our bodies compensate by using less energy on other things, so that our overall energy expenditure stays the same”. This is why he reckons that coal miners and athletes burn the same number of calories as the rest of us.

His claim about coal miners rests on a study of miners in the USA and Turkey which, he says, found that they burned “2,100 and 2,800 calories per day — the same as the rest of us.” Alas, he completely misread the study. That is how much they burned at work. In the course of a whole day, they burned an average of 3,658 calories. Furthermore, the number of calories burned depended on how physically active they were, with the most active miners burning 4,414 calories a day: 

Average energy expenditures according to activity levels were 3289.4 ± 356.64 kcal/d, 3910.0 ± 438.57 kcal/d, and 4413.8 ± 343.24 kcal/d (moderate, heavy, and above heavy activities; respectively).

This tells you all you need to know about the effect of physical activity on energy expenditure and it tells you a fair bit about the rigour of Chris van Tulleken’s research.

As for athletes, of course they burn more calories than desk workers. That’s why they don’t get fat despite eating thousands of calories a day and gorging on McDonalds in the Olympic village. Incidentally, soldiers also burn more energy than the average person, which is why their ration pack contains 4,000 calories while a civilian ration in World War II only contained 3,000 calories. You will notice that both of these exceed the 2,500 calories recommended for adults today (and WWII rations were supplemented by homegrown food, black market food and beer).

Van Tulleken’s assault on the laws of thermodynamics leaves him convinced that physical activity doesn’t prevent weight gain. This, combined with his conviction that willpower doesn’t work, leaves the public in a bit of a pickle. It seems there is nothing that can be done except ripping up the entire food supply and getting some philosopher kings to rebuild it in their own image.

How did anyone get the crazy notion that burning calories can prevent a condition caused by excess calories? Or, as van Tulleken puts it, “why is the scientific literature so confused about an issue that seems fairly easy to resolve?” Here comes the big reveal.

In the case of my little paper, it is because I am “paid a salary by the Institute of Economic Affairs” which has “received funding from sugar giant Tate & Lyle”, a company that “has an interest in promoting the narrative that inactivity, rather than food, is the problem”. He gives two references for this, both of which relate to panel discussions held by the IEA and sponsored by Tate & Lyle at Conservative Party conferences in 2016 and 2018. The topic discussed at both events was post-Brexit trade agreements. This was of great interest to Tate & Lyle because they use sugar cane sourced from Central and South America, which the EU put heavy tariffs on to protect sugar beet producers in Europe. This is why Tate & Lyle were one of the few companies to publicly support Brexit, and it is why they sponsored those events (they sponsored the entire conference in 2017). Assuming the IEA made a small profit from these events, a fraction of the money will have indirectly ended up in my pocket, thereby influencing a paper I wrote a few years earlier. That’s how it works, right?

Van Tulleken closes the chapter by returning to the scientists Steven Blair, Peter Katzmarzyk and James Hill, who van Tulleken reveals had undisclosed conflicts of interests due to funding from Coca-Cola. This was a bit of a scandal when it was made public in 2015, and it shouldn’t have happened, but conflicts of interest do not disprove research findings. The scientific literature on obesity and energy expenditure consists of a lot more than studies produced by these three academics, although you wouldn’t get that impression from reading Ultra-Processed People. Van Tulleken says that “whatever work has been funded by them [Coca-Cola] should be disregarded”, but that is not how science works. At any rate, it is not how it should work. These academics were doing original scientific research published in peer-reviewed journals. If it could be shown that there were serious methodological flaws or misconduct that invalidated the findings, the source of funding would be more interesting, but van Tulleken doesn’t even attempt to critique the papers themselves. He simply finds some academics with a conflict of interest and dismisses everything they say on that basis alone. 

This is a running theme of the book. Van Tulleken is willing to acknowledge that there are scientists who disagree with his theories so long as he can imply that they have been bought off. He mentions a thoughtful critique of the UPF concept, but mainly so he can point out that two of its four authors have worked with Nestlé. He admits that “not every single paper critical of NOVA [the classification system that spawned UPF] has identifiable conflicts of interest”, but he leaves the reader in little doubt that those which do should be ignored.

He does this so often that it becomes as boring as it is fatuous

He does this so often that it becomes as boring as it is fatuous. After admitting that some studies show that artificial sweeteners help reduce weight, he quickly adds that “many of the studies’ authors have declared relationships with food companies”. When a study finds that people eat UPF faster than other food, you might expect van Tulleken to be happy, but he is displeased by the authors’ suggestion that the food industry could address this by adding texture to their products. He implies that this is because they have received expenses for speaking at meetings sponsored by food companies.

When an industry-linked study finds “no relationship between low-calorie sweeteners and body weight”, he does not ask why Big Soda failed to rig the findings on this occasion. Instead, he has his unprocessed cake and eats it by saying that if even a “statistical analysis undertaken by industry can’t find a significant benefit to low-calorie sweeteners, that should be cause for alarm”. (In fact, the study was a meta-analysis which found no evidence of benefit in 9 cohort studies but found consistent benefits in 15 randomised controlled trials which, the authors correctly note, “provide the highest quality of evidence”.)  

There is no taboo about scientists working with the food industry

It is as if van Tulleken doesn’t need to read the studies at all. He just needs to know whether any of the authors have ever been in the same room as someone from Unilever. There is no taboo about scientists working with the food industry, nor should there be. If you are a food scientist, it is the obvious place to work. Not everybody wants to live off government grants. 

The problem with industry-funded science is not that it is unreliable — it is often of higher quality than non-industry research — but that it tends to focus on issues that are most helpful to industry. There is also a suspicion that research that doesn’t come up with helpful findings is suppressed, which is why the All Trials project is so important. It wouldn’t be surprising if Coca-Cola put more money into researching the benefits of physical activity than the harms of tooth decay, but that doesn’t make what it says about physical activity wrong.

For van Tulleken, a study doesn’t even need to be funded by the industry for it to be dismissed as industry science. All it takes is one of the authors sitting on a committee linked to a food company or being reimbursed for speaking at an industry conference. This McCarthyite rummaging for tenuous links between scientists and the world’s largest industry only makes sense if those scientists are corrupt, greedy or dishonest. If they are knowingly promoting food that kills people, as van Tulleken implies, they could even be described as evil. His lawyers won’t let him say this, and it is almost certainly untrue, but if we are meant to disregard hundreds of studies because the authors have links to the food industry, that is the only conclusion to draw. These hacks must be routinely fiddling the figures to produce results that favour their paymasters and the peer review system must be completely useless.

Is it really worth smearing academics and damaging trust in science to this extent? Is it responsible behaviour for a doctor to tell people that most of the food they eat could make their jaws shrink, give them cancer, diabetes and dementia, make them infertile and cause “penile shrinkage”? Is it sensible to tell people that they have no willpower and are helpless victims of a borderline conspiracy involving everyone from the FDA to Cancer Research UK? Isn’t there enough misinformation and quackery around without bestselling books spreading alarm about “chemicals” in food? 

I am clearly in the minority. I have yet to see a bad review of this book, and I’m sure van Tulleken’s disciples will dismiss this one as being paid for by Big Sugar. My copy of Ultra-Processed People is plastered with warm words from minor celebrities. Chris Packham calls it “incendiary and infuriating”. He is half-right.

Enjoying The Critic online? It's even better in print

Try five issues of Britain’s newest magazine for £10

Critic magazine cover