Bacon, cancer, and the one number nobody printed
The 2025 “put warnings on bacon” story rests entirely on a confusion of relative and absolute risk. The real finding is 7 in 100 instead of 6. Here is how an honest statistic became a panic.
In October 2025, ten years to the month after the World Health Organization classified processed meat as a carcinogen, a group of scientists wrote to the UK health secretary demanding that bacon and ham carry cigarette-style cancer warnings. The Guardian's report led with the claim that inaction had “contributed to 54,000 cases of bowel cancer,” and quoted an expert reminding readers that the WHO files nitrite-cured meats “in the same carcinogenic category as tobacco and asbestos.” Around the world, the story bounced back in the familiar register: bacon is basically cigarettes; the full English is a death wish.
Here is the entire scientific finding underneath that coverage, stated honestly. If you eat little or no processed meat, your lifetime risk of bowel cancer is about 6 in 100. If you eat an extra 50 grams — a daily bacon sandwich — every single day, for the rest of your life, that risk rises to about 7 in 100. One extra case per hundred lifelong daily eaters. That is the whole thing. Everything else is framing.
It is worth sitting with those two grids, because almost all of the alarm evaporates in front of them. The science is real and I am not here to dispute it: processed meat genuinely does raise bowel-cancer risk, the evidence is solid, and eating less of it is sensible. But “one more case in a hundred, if you eat bacon every day for life” is a profoundly different sentence from the one most readers carried away in October 2025 — and the gap between them was manufactured almost entirely by a single, specific statistical sleight of hand.
The trick: a percentage with no baseline
The figure at the heart of the story is that processed meat raises bowel-cancer risk by 18% for each 50 grams eaten per day. That number is correct. It comes from a careful meta-analysis, and Cancer Research UK reports it on its own website. But 18% is a relative risk — a percentage of something — and a relative risk is meaningless until you know what it is a percentage of.
Eighteen percent of a small number is a smaller number. The baseline lifetime risk of bowel cancer is roughly 6%. Raising that by 18% gives roughly 7%. So the dreaded “18% increase” is the move from 6 in 100 to 7 in 100 — not a competing, scarier fact, but the very same fact wearing a more frightening outfit.
This is the engine of nearly every “everyday food causes cancer” story, and it works because the two framings feel like different magnitudes of danger when they are arithmetically identical. “Eighteen percent more cancer” invites you to imagine 18 people in 100, or a fifth of bacon-eaters struck down. The honest translation — one extra case per hundred lifelong daily eaters — invites you to shrug and order the sandwich slightly less often. A relative risk reported without its absolute baseline isn’t false. It is simply built to be misunderstood, and it usually is.
The second trick: “same category as tobacco”
The 2025 letter leaned hard on a second line: that the WHO places processed meat “in the same carcinogenic category as tobacco and asbestos.” This too is true, and it too misleads, because it trades on a confusion the WHO has explicitly tried to head off. The International Agency for Research on Cancer sorts substances by how sure we are that they can cause cancer at all — the strength of the evidence — not by how much cancer they cause. Group 1 means “we are certain this is a carcinogen.” It says nothing whatsoever about potency.
Tobacco and bacon share a shelf because the evidence that each causes cancer is conclusive — not because a bacon sandwich is as dangerous as a cigarette. IARC itself has said the classification describes “the strength of the scientific evidence… rather than assessing the level of risk.” Smoking raises a smoker’s lung-cancer risk by something like twenty-fold; processed meat raises bowel-cancer risk by 18%. Both are Group 1. They differ in effect by a factor of roughly a hundred. “Same category” is doing an enormous amount of dishonest work in a sentence where every individual word is accurate.
Where the journalism went wrong
None of this is a defence of processed meat, and it is not really a criticism of the scientists either — a campaign letter is allowed to advocate. The failure belongs to the journalism, and the Guardian’s write-up is a clean specimen of it. The piece led with the most alarming framing available — the raw case count, the tobacco comparison, the demand for cigarette-style warnings — and at no point did it translate the risk into the absolute terms that would let a reader judge it: the 6-in-100-to-7 figure that Cancer Research UK, the country’s own authority on this, publishes plainly.
That omission is the whole ballgame. A reader who is told “same category as tobacco, 54,000 cases, demand for warnings” and is not told “one extra case per hundred lifelong daily eaters” has been handed every ingredient of a panic and denied the single number that would defuse it. The figures were all accurate. The picture they assembled was false. And it was false in the most common way a true statistic can be: by reporting the relative risk and silently dropping the baseline.
The 54,000-cases figure deserves the same treatment. It is a national total accumulated across the entire UK population over years — a genuine public-health quantity, and a reasonable thing for campaigners to cite. But a big aggregate number tells an individual reader nothing about their own risk, and placed beside “warnings like cigarettes” it is plainly meant to be felt as personal danger rather than understood as population accounting. The honest companion sentence — “which works out to about one extra lifetime case per hundred people who eat bacon every day” — is precisely the sentence that never appears.
How to read the next one
This story will be back. Some food, some study, some round-numbered case count, some “same category as” comparison, and the same missing baseline. Three questions disarm almost all of it.
Eighteen percent of what? Whenever you meet a relative risk — “X% more cancer” — hunt for the absolute baseline before you feel anything. “18% more” on a 6% risk is a one-point move. The percentage is designed to be frightening; the baseline is what makes it intelligible.
What’s the dose, and is it mine? The risk here is for 50 grams every day for life — a sandwich daily, indefinitely. Most people eat far less, and the risk scales down accordingly. A finding quoted at its maximum dose is not a description of your occasional fry-up.
Does “same category” mean same danger? Almost never. Hazard classifications describe certainty, not severity. “In the same group as tobacco” is a statement about how good the evidence is, and tells you nothing about how big the risk is.
Eat a bit less processed meat if you like; the advice is reasonable and the science behind it is sound. But do it in proportion to the actual risk — one more case in a hundred, at a sandwich a day for a lifetime — and not in proportion to a headline that took an honest 18% and quietly forgot to mention what it was 18% of.