A study says junk food may drive a third of heart disease — as regulators fight over what 'junk' even means

Food & Farming 3 min 80 sources

New research links ultra-processed food to a large share of heart-disease deaths, just as the FDA and USDA struggle to define the category at all. The problem may be that our whole way of judging food looks at the wrong thing.

Key takeaways

A Canadian modelling study estimates ultra-processed foods may drive a quarter to a third of heart-disease cases — and they now make up over half the average UK and US diet.
US regulators are stuck fighting over how to even define "ultra-processed": by processing, by ingredients, or by nutrition — a decision that will shape labels and policy.
The deeper problem is that our food labels measure single nutrients, while the suspected harm may live in the processing and combination — which a parts-by-parts label can't see.

A modelling study out of Canada suggests that ultra-processed foods — UPFs — could be a “substantial and potentially preventable” driver of heart disease, tied to somewhere between a quarter and a third of cases [1]. If the estimate holds, cutting back could prevent thousands of deaths [1].

The catch is that these foods are not a niche indulgence. In the UK and US, more than half of the average diet is now ultra-processed [1]. For younger people and those in poorer areas, it can reach 80% [1]. “Ultra-processed” covers ready meals, breakfast cereals, protein bars, fizzy drinks, and fast food — foods rebuilt in a factory into a cheap, long-lasting, hard-to-stop-eating form [1]. As always with a modelling study, this is an estimate built on associations, not a controlled proof, and “UPF” is a broad, contested label. But the signal keeps showing up.

Nobody can agree what the word means

At the exact moment the health case is building, regulators can’t define the thing. The FDA and USDA — the two US food agencies — are in a three-way fight over how to draw the line around ultra-processed food [3]. Should it be defined by the processing it went through, by its ingredients, or by its nutrition — fat, sugar, salt? [3] Public-health groups want the agencies to move fast; industry groups argue the whole concept is too broad to be useful [3]. The definition is not academic: it will shape food labels, nutrition policy, and state laws [3].

The disagreement runs deeper than politics. Critics of the UPF idea say it “ignores nutritional composition and focuses primarily on processing” [15]. Supporters say that’s the point — that the processing itself, not any one nutrient, is what tracks with poor health [15]. In other words, the fight is over whether a food’s harm lives in its parts or in what was done to them.

A small study of the same problem

A Cambridge laboratory study this week showed the puzzle in miniature. Researchers tested 39 common sweeteners and found many of them directly change the growth of gut bacteria [47]. More striking: they logged over 100 cases where a sweetener behaved differently once it was combined with a medication, caffeine, or a flavouring — one pairing, a sweetener and a common antidepressant, sharply cut beneficial bacteria [47]. This was in a dish, not in people, so it proves nothing about your breakfast [47]. But it illustrates the thing regulators are stuck on: a substance judged safe on its own can act differently in company.

The wallet points the same way

There’s a quieter pressure worth noting. In Japan, this year’s food-price rises are hitting processed foods hardest — instant noodles, canned and frozen goods — with price increases logged across 5,780 processed products [22]. Around the world, the cheapest, most available, longest-lasting food is often the most processed. So the pull toward these foods isn’t only that they’re engineered to be moreish. It’s that they’re frequently what a tight budget can reach.

None of this means hunting for a villain ingredient on the back of a packet. If anything, the week’s news points the other way — that the label, and the way we’ve long read it, may be looking at the wrong thing entirely.

02 · Lesson · why it matters

The label reads the parts. The harm hides in the whole.

We judge food one ingredient at a time. The growing worry is that the danger isn't in any single line on the label — it's in what the processing does once everything is combined.

A harm with no culprit

Here is the strange thing about the ultra-processed food story. Scientists keep finding that a diet heavy in these foods tracks with worse health — and they keep failing to pin it on any one villain. The fat is often modest. The sugar is within bounds. The salt is nothing unusual. Read the nutrition label line by line and each number looks unremarkable.

That is exactly why the science has been so slow and so contested. We are built to look for a bad ingredient. This is not a story about a bad ingredient.

”Processed” is a verb, not a thing

Ultra-processed doesn’t name a substance. It names something that was done. Whole foods are broken down to their cheapest parts — starch, oil, protein powder, sugar — then rebuilt into something soft, shelf-stable, and engineered to be hard to stop eating. The fibre is stripped. The texture is manufactured. The flavour is added back.

If these foods do harm, the harm is a property of that form — the speed you eat it, what it displaces, how it behaves in the body all at once — not of any component you could circle on the ingredients list. The pieces can each be fine and the finished thing still not be.

A vocabulary built for parts

This is why regulators are stuck arguing over a definition. The whole apparatus we use to judge food — the nutrition label, the dietary guidelines, the rules — is built on parts. Grams of fat. Grams of sugar. Milligrams of salt. It is a parts list, and it is very good at parts.

But a parts list cannot describe a property that only exists in the combination. Asked to define “ultra-processed,” the agencies reach for their usual tools — ingredients, nutrients — and the tools don’t fit, because the thing they’re chasing isn’t an ingredient or a nutrient. It’s a pattern. Trying to capture it with a nutrient table is like trying to describe a melody by listing which notes it uses. You’ll get every note right and miss the song.

Safe alone is not safe together

You can watch the same trap in miniature. This week a lab tested dozens of sweeteners — each one long treated as safe in isolation — and found that in combination, with a medicine or with caffeine, some behaved in ways none of them showed alone.

That’s the crux of it. We test and approve things one at a time because one at a time is tractable. But almost nothing is consumed one at a time. The real world is the combination, and the combination has behaviours that no single part announces in advance.

The pattern is everywhere once you see it

Step back and this stops being about food. A great deal of the world is judged by its parts and lived as a whole.

A team of individually excellent people that works badly together. A portfolio of individually safe loans that sinks together when they all fail at once. A city of reasonable rules that traps you at their intersections. In each, you can audit every part, find nothing wrong, and still have a whole that behaves badly — because the behaviour was never in the parts. It was in how they touched. And an inventory, however careful, cannot see a relationship.

What the packet can’t tell you

You are inside this one, not above it. Half of what you ate this week was probably ultra-processed, and no amount of squinting at the label was going to reveal it, because the label was never measuring the thing that might matter.

The honest response isn’t to find a new ingredient to fear — that’s the same parts-thinking that missed the problem in the first place. It’s quieter than that: to notice the form of what you eat, not just its contents, and to hold your confident verdicts — this one’s bad, this one’s safe — a little more loosely. The scientist, the regulator, and you are all looking at the same packet with tools made for parts, at a thing that lives in the whole. None of us can quite see it yet. That’s worth knowing before the next study, or the next label, tells you it has the answer.

03 · Lab · your turn

Build the Cereal

Process a whole food step by step and watch a parts-based nutrition label barely register what the processing did.

04 · Hope · carry this

For years the harm in our food had no name and no shape, and so it hid in plain sight. That scientists and regulators are now arguing hard about how to see it is not a sign of confusion — it is what the slow, stubborn work of finally noticing something looks like from the inside.