The science of “ultra-processed” foods is misleading

Over the summer, a story circulated across news outlets claiming that eating plant-based burgers led to heart disease.

“New research,” the Washington Post reported in June, “found eating plant-derived foods that are ultra-processed — such as meat substitutes, fruit juices, and pastries — increases the risk of heart attacks and strokes.”

“Vegan fake meats linked to heart disease, early death,” the New York Post declared.

There was just one problem: The narrative was totally fake.

The claim emerged from a study on plant-based “ultra-processed” foods by a team of nutrition researchers at the University of São Paulo and Imperial College London. Using data from a sample of 118,397 people in the UK who had reported what they ate over at least two days, the paper found that increased consumption of ultra-processed plant foods was associated with higher rates of cardiovascular disease and premature death, while eating non-ultra-processed plants like fruits and vegetables was linked to better health outcomes.

But plant-based meats were virtually absent from the study: Just 0.2 percent of calories across the sample came from meat alternatives. The bulk of the plant-based ultra-processed calories instead came from what the authors describe as “industrialised packaged breads, pastries, buns, and cakes,” and “biscuits,” better known in the US as cookies — foods that have little to do with plant-based meats or other specialty vegan products. The new generation of vegan burgers, including Impossible and Beyond burgers, did not yet exist when the data was collected between 2009 and 2012.

“With such a small contribution, we can’t draw any meaningful conclusions about plant-based meat alternatives specifically,” University of São Paulo researcher Fernanda Rauber, lead author of the study, told me in an email.

That makes sense. Not many people, after all, regularly eat vegan meat alternatives. So why did the media focus on plant-based meats?

The answer is bigger than just one misreported study. It connects to deeper tensions within the science of “ultra-processed foods” (UPFs), a relatively recent category in nutrition research used to describe packaged foods with dubious-sounding ingredients not typically used in household kitchens. Robert F. Kennedy Jr., Donald Trump’s pick to lead US health policy, promises to crack down on ultra-processed foods and has called plant-based meats instruments of corporate control over our food system and humanity. And it’s not just RFK Jr. and his MAHA supporters. Sen. Bernie Sanders (I-VT), too, has recently called for regulating UPFs.

Last week, however, the scientific panel that advises the creation of the federal dietary guidelines concluded that there was limited evidence on UPFs’ health effects and that “few studies were designed and conducted well.”

The supposed danger of ultra-processed foods has resonated among the general public in the last several years, tapping into anxieties about industrial modernity and a sense that we’re being poisoned by big food companies. “It really responds to this feeling that a lot of consumers have, which is that the food industry is not protecting their health,” Aviva Musicus, science director for the health policy advocacy group Center for Science in the Public Interest, told me.

Consumers are right about that: The American food environment is unhealthy and disease-promoting, and the food industry bears much of the blame. But ultra-processed foods — a framework “so broad that it borders on useless,” as Oxford nutrition researcher Nicola Guess argued in the New York Times this week — does little to clarify the reasons why. Taken at face value, it could even steer consumers away from healthier, more planet-friendly plant-based foods.

What happened with that study — and why the “ultra-processed” concept is so confusing

Journalists have a responsibility to verify the facts of any research they cover. But the framing of that University of São Paulo–Imperial College study, and the promotional materials associated with it, might have made it easy for reporters to misunderstand what the research really found.

A news release from Imperial College London led with a photo of plant-based burgers, sausages, and meatballs, as one nutrition researcher not associated with the study pointed out at the time, and the first example the release mentions of ultra-processed plant foods is plant-based meat. “Many plant-based foods, including meat-free alternatives such as some sausages, burgers and nuggets, can be classified as ultra-processed foods (UPFs), despite often being marketed as healthy options,” the release reads. That’s neither a fair representation of the research nor of plant-based meat’s relatively small role in most diets.

The use of these examples, Rauber told me, “are technically correct because they do fall into the ultra-processed plant-based group. That said, these foods contributed very little to the overall calories in our study,” she acknowledged. “I probably wouldn’t have chosen that specific photo to illustrate the findings, since our study examined broader dietary patterns — comparing ultra-processed plant-based foods with their non-ultra-processed counterparts — not specific food categories. But press teams often need concrete examples for clarity, and we understand the media’s role in shaping how findings are presented.”

Things get weirder when you dig into how the study defined “ultra-processed” meat alternatives. Included on that list are tofu and tempeh, soybean-based foods that have been used in East and Southeast Asian cuisines for centuries. They bear little to no resemblance to products like Impossible and Beyond burgers.

This fact, more than anything else about the study, set off my BS detector. Ultra-processed foods researchers categorize foods according to the Nova classification, which consists of four tiers, going from least to most processed:

Depending on how you interpret these categories, tofu probably belongs in group three, and tempeh, which is just fermented soybeans, may belong in group one. Neither of them fit the ultra-processed category. Foods with added gluten, too, have been arbitrarily slotted into category four by the creators of the Nova classification, although gluten has a long history as a meat alternative (known as seitan) in East Asian cuisines. Not only can you use it in your home kitchen, but you can make it yourself from flour.

If you’re confused, don’t feel bad — some of the world’s top nutrition experts are, too. “You look at these papers, and it’s still very hard to pin down what the definition [of ultra-processed] really is,” Walter Willett, a professor of epidemiology and nutrition at Harvard, told me. It’s a concept prone to illogical free association, lumping together Cheetos with ultra-healthy fermented beans.

Asked why tofu and tempeh were classified this way, Rauber said the dietary questionnaire filled out by people in the dataset grouped together tofu, tempeh, and soya mince, also known as textured vegetable protein (a UPF, but one that’s a perfectly reasonable source of protein and fiber made after the fat has been removed from soybeans in the production of soybean oil).

“While plain tofu itself might not be considered ultra-processed, we observed that many options available on the market at the time of data collection contained natural flavourings, thickeners like guar gum, and other ingredients that align with the Nova definition of UPF,” she wrote. That’s true of some flavored tofus — though the addition of an ingredient like guar gum wouldn’t much impact their nutritional properties. Added sugar, however, definitely would — but sugar is not an ultra-processed ingredient, according to the Nova classification, unless it comes in the form of something like high-fructose corn syrup, which is.

For the most part, the UPF category targets ingredients that have only come into use with modern food science and industrial technology. Without a doubt, many foods that meet the ultra-processed criteria are bad for us, and we’re better off eating mostly unprocessed or minimally processed foods. Processed meat is classified as a carcinogen by the World Health Organization because of the specific harms of that type of processing. UPFs like Twinkies and Oreos are unhealthy because they’ve been processed in a way that strips important nutrients and adds super tasty, health-damaging components like sugar — things that are well-established in nutrition science, without reference to the concept of ultra-processing.

The relevant question about a novel scientific concept is not whether it happens to correlate with stuff we already know is true, but whether it adds something genuinely new to our knowledge, without also being wrong about a bunch of other things, as New York University environmental scientist Matthew Hayek pointed out to me. UPF, at least so far, doesn’t seem to clear that bar — it casts a net that manages to be overbroad while excluding some unhealthy forms of processing that have been around longer.

Meanwhile, the ultra-processed framework has needlessly cast aspersions on foods that are perfectly fine (like store-brought 100 percent whole-grain bread with some added gluten — generally still a better choice than less processed white bread) and that can make it easier to enjoy unprocessed whole foods (like MSG, another ingredient I use at home). On the Nova scale, “homemade soup is a 1 unless you use a bouillon cube, in which case it catapults to a 4,” Washington Post food columnist Tamar Haspel wrote earlier this year.

These arbitrary categorizations can make it harder to make informed comparisons between foods. “Some of the plant-based alternatives to meat are quite a bit healthier, it looks like, than the actual beef or pork that people are consuming. It’s a big step in a healthier direction, a huge step in reducing greenhouse gas emissions,” Willett said, citing meat’s high carbon footprint.

Beyond Meat, which has recently switched the fat source in many of its products from coconut and canola oils to avocado oil, fares particularly well against beef, with much lower saturated fat, lots of protein, iron, and even a bit of fiber. Would it be better to eat an unprocessed source of plant protein, like beans? Sure. “Minimally processed foods are almost always the optimal choice,” Willett said.

I wouldn’t eat Beyond burgers every day, much like it would not be a good idea to eat a beef burger every day. But there’s no reason to be afraid of them.

What this means in the real world

Having said all that: I get it. It feels intuitive to think there is something fundamentally not right about ultra-processed foods. I can understand why people would be freaked out by a vegan burger that looks and tastes like meat. I shudder at the junk that was normal for kids to eat when I was growing up — Gushers, Fruit Rollups, Coke — and think: That is not food. (Though someone might have said the same thing the first time sugarcane was processed into granulated sugar, and they’d have a point.)

It makes sense to have humility about how much we have yet to learn about the impacts of the sweeping changes to our diets that have taken place over the last century. We do need more research into how specific food additives might contribute to specific health outcomes, like impacts to our microbiomes, an area not yet well understood. “Emerging evidence suggests some of them might harm health, particularly through gut microbiome disruption, inflammation, and even DNA damage,” Rauber said.

If UPF were a more intellectually modest concept, it might have more analytic value. But much of the UPF literature has committed itself to the untenable position that whatever it classifies as ultra-processed is automatically an inferior choice, even a dangerous one. Meanwhile, people in the real world are making real food choices under all sorts of constraints, and it would make no sense to tell them that they should avoid unsweetened soy milk just because it contains a thickener.

Yet that’s what another, more recent UPF study, with some of the same authors as the University of São Paulo–Imperial College paper, suggests doing. “Pescatarians, vegetarians, and vegans were more likely to include plant-based milk and meat alternatives in their diet,” the study concludes, a finding that the authors find “concerning.” They argue that “it is, therefore, important that urgently needed policies that address food system sustainability” — like encouraging a transition to more plant-forward diets — “also promote rebalancing diets towards minimally processed foods away from UPFs.”

This kind of rigidity only makes it harder to make healthier, more sustainable, more humane food choices free of animal products. “Soy milk is almost for sure, in the long run, going to be healthier than cow milk,” Willett said.

The breadth and ambiguity of the campaign against “ultra-processed” foods make it vulnerable to sloppy thinking and manipulation by pseudoscience purveyors like RFK Jr. Combine that with a political climate in which multiple red states have banned cell-cultivated meat and meat producers seize every opportunity to thwart plant-based competitors, and you can imagine how plant-based meats could be targeted by an unprincipled, politicized application of ultra-processed food research.

Vegans and the products associated with them make an easy punching bag — for everyone from RFK Jr. to universities chasing media coverage to news outlets seeking reader eyeballs — because they make people feel bad about eating meat. It’s easier to write off meat alternatives as weird and synthetic than it is to reckon with the environmental and ethical degradation of animal agriculture. But the vilification of these foods, as ever, is not based in well-founded fears about their health effects. It’s really just about the vibes.