Letters to the Editor

Ultra-Processed Foods: When the Letters Page Is More Interesting Than the Paper

Does anyone still remember the Lancet’s Ultra-Processed Food series? At the time, it caused some minor media attention - and the authors dismissed most of the criticism. As one of the authors - Professor van Tulleken - said on BBC Radio 4 Today (transcript from the BBC):

"I should say these comments that are coming in, there were about 15 comments submitted through the Science Media Centre, itself a food industry-funded press office, of those, we did an analysis of these yesterday when we had an author's meeting, about a quarter of them were strongly negative. Of the strongly negative comments, every single person, every single scientist had an institutional or individual conflict of interest with an ultra-processed food company. 75% of good comments were from independent scientists. As you say, it's in The Lancet, it's been peer reviewed, this is sanctioned by WHO and UNICEF."

That is of course one approach to deal with factual criticism (which can be found here). Unfortunately, there was not enough time for the presenters to challenge this statement - even though there are many reasons why reducing scientific debates to actual or perceived conflicts is misleading.

But what happens when the criticism actually makes it into the journal itself?

Letters to the Editor

Letters to the editor are not just the prerogative of Sir Herbert Gussett, but are time-honoured way of scientific debate - especially for topics that are still controversial. And despite claims to the contrary, the idea behind the dangers of ultra-processed foods is far from mainstream and still rejected by many scientists. It was therefore not surprising that there were several Letters to the Editor to the Lancet - and in this week’s issue ofThe Lancet, no fewer than five letters, and the response by the authors, appeared.

For anyone still interested in the discussion - and especially anyone who actually cares about public health and not just fashionable topics - these letters, and the responses were interesting.

Dietary assessment

My main criticism of ultra-processed foods - and nutrition research in general - is the difficulty estimating intake reliably. I have raised this in my letter - including my concern that many studies, such as the various Harvard studies, rely on methods that were never designed nor validated for ultra-processed foods.

The response makes a number of fairly unsubstantiated claims:

However, exposure misclassification is likely non‑differential in regard to outcomes, and therefore bias associations towards the null.

Sounds convincing, doesn’t it? After all, they cite Walter Willet’s seminal textbook on Nutritional Epidemiology as evidence. But is it?

The claim essentially states that the error introduced by the limitations of estimating ultra-processed food intake is not linked to the outcomes - and therefore can be treated like random noise. Annoying, but if anything, only making the estimated links between ultra-processsed food intake and health weaker.

There is just a tiny problem: there are no data that support that claim. We simply don’t know whether the measurement error is random or not. And for one nutrient found in many ultra-processed foods, it is definitely not the case.

The villain sugar

It has been known for some time that people underreport sugar intake - unsurprisingly, no-one likes to admit guilty pleasures such as desserts or snacks. But people with a higher BMI tend to underreport sugar intake much more than those with a lower BMI - resulting in the very nice but unfortunately wrong impression that sugar consumption leads to weight loss. In reality, higher sugar-intake is unfortunately associated with a higher BMI - and any claim that “exposure misclassification is likely non‑differential in regard to outcomes” would be wrong.

Association between sucrose intake (self-reported - bottom; objectively measured, top) and odds of being obese. Data adapted from Kuhnle et al. (2015).

For sugars, we have biomarkers and can compare self-reported with actual intake. But we don’t have the same for ultra-processed food. So how can one make such a claim?

It would be nice to see actual data that supports such claims - and perhaps it will come eventually. Until then, I remain very sceptical.

Silencing critics?

There is another aspect of the ultra-processed food discussion that makes me very sceptical of the idea: the attempt to silence critics. Obviously, no-one likes to be criticised or told they are wrong - but only very few people demand zero tolerance for critics and use a guilty by association approach. The approach completely ignores the fact that scientific arguments should be based on facts - and nothing else.

Several of the letters comment on this approach - and Dr David Ludwig (Harvard Medial School)

I disagree with those who argue that researchers should stay within their area of expertise. Regarding important public health issues, scientific inquiry and advocacy can and must co-integrate. By dismissing scientific criticism and concern for unintended consequences, the Series does disservice to both inquiry and advocacy.

Unfortunately, the authors do not engage with this comment - but they do reply to similar concerns raised by FoodDrinkEurope, the food industry confederation in the European Union:

Science requires intellectual openness, robust evidence, and the willingness to engage with counter-arguments. Although we share the view that funding and conflicts of interest should be transparently communicated, excluding researchers for their funding sources or banning entire sectors from public fora is not only unscientific, but contrary to good public health policy. Doing so would narrow the knowledge base, precisely when it is needed.

Listening to different opinions, engaging with counter-arguments and not excluding researchers simply because of their association - is this really an outrageous suggestion? The authors of the UPF Lancet papers think so:

Based on decades of evidence from food, tobacco, alcohol, and fossil fuel research, we question the credibility of industry-funded science when commercial imperatives conflict with public health goals.

Comparing nutrition research to tobacco has become a very popular approach, completely ignoring the fact that there are huge difference between tobacco (one can easily live without) and food (very difficult to live without). And it does not explain why very factual criticism - for example regarding statistical methodology - should be ignored.

Transparency alone is insufficient. The empirical literature shows that disclosure does not neutralise bias, nor prevent the strategic use of funding to manufacture doubt, delay regulation, and frame debate in industry-favourable terms.

It is impossible to neutralise bias - but one could argue that a reputation as activist or influencer might create a similar bias to research funding.

Silencing the opposition is of course nothing new - and at least the ultra-processed food proponents don’t suggest that we should have tribunals or that undesirables should be sent for reeducation. Nevertheless: considering such an approach is rather concerning.

Apropos transparency: It is worth noting that the Lancet requires declaration of all competing interests - not just financial ones related to industry. Intellectual and advocacy commitments matter too. It would be good if this would scrutiny would be applied symmetrically.