Red Meat, Bacon And Cancer: The WHO Graded Its Confidence And The Papers Printed A Body Count

On 26 October 2015 the World Health Organization’s cancer agency put processed meat in the same category as tobacco smoking and asbestos, and a nation started looking at its breakfast like a suspect.

The classification was real. The panic was built afterwards, out of a relative number that almost nobody reading it was ever shown how to convert.

Processed meat has a small, genuine association with bowel cancer and a chemical mechanism with some weight behind it. Eaten at 2 rashers of bacon a day for an entire adult life, it shifts a person’s lifetime odds of bowel cancer by roughly 1 in 100. Fresh red meat’s case is thinner still, resting on human evidence the WHO’s own experts said could be down to chance, bias or confounding. What genuinely decides how bowel cancer goes for most people is age, family history, smoking, drink, body weight, and whether the screening kit goes back in the post.

The quick versions of this argument live in red meat health myths and carnivore dangers. This is the long one, with the receipts.

The WHO’s cancer groups grade certainty, and the size of the risk was never on the scale

In October 2015, 22 experts from 10 countries met in Lyon for the International Agency for Research on Cancer and assessed more than 800 studies of meat and cancer.

They put processed meat in Group 1, carcinogenic to humans, judging the evidence sufficient that eating it causes colorectal cancer, which is bowel cancer. They put red meat in Group 2A, probably carcinogenic, on limited evidence in humans backed by strong mechanistic evidence.

Limited has a precise meaning in IARC’s vocabulary. It means a link has been observed, but chance, bias or confounding could not be ruled out as the explanation. Every red-meat-causes-cancer headline since has been standing on a committee’s admission that the pattern might be an accident of who eats what.

IARC’s own question and answer page is blunt about the tobacco comparison. Sharing Group 1 with smoking and asbestos, it says, puts processed meat in their evidence class and says nothing about their danger, because the groups describe how strong the evidence is that something causes cancer and do not assess the level of risk. In 2019 the agency wrote the point into its own title, renaming the series the Monographs on the Identification of Carcinogenic Hazards to Humans. A hazard is something capable of doing harm. A risk is how likely that harm is at the dose you actually meet. The papers were handed a hazard and printed a risk.

The company each group keeps shows how much weight the label can carry. Group 1 holds alcoholic drinks and wood dust alongside tobacco and asbestos. Group 2A, red meat’s group, holds drinks taken hotter than 65°C and acrylamide, the compound that forms when toast and chips brown. Coffee sat in Group 2B, possibly carcinogenic, from 1991 until 2016, when a review of more than 1,000 studies moved it down to Group 3, not classifiable.

For decades the scale also had a Group 4, probably not carcinogenic to humans. In its entire history it admitted a single member, caprolactam, a raw material for nylon, and IARC then removed Group 4 altogether. Nothing you have ever eaten was in it.

Pork counts as red meat. Processed meat covers ham, sausages, corned beef, biltong, tinned meat and meat-based sauces. Poultry and fish were never assessed. The experts could not say whether a safe level exists, whether any group of people is at higher risk, or whether the way meat is cooked affects the risk. And the agency’s own answer to whether people should stop eating meat opens with six words: “Eating meat has known health benefits.”

It printed a body count too. The Global Burden of Disease project attributes about 34,000 cancer deaths a year worldwide to diets high in processed meat. The same estimates put tobacco at around 1 million cancer deaths a year, alcohol at 600,000 and air pollution at more than 200,000. Same group as tobacco, about a thirtieth of the toll.

2 rashers a day for life moves bowel cancer by about 1 person in 100

The figure that launched a thousand headlines was 18%. Pooling data from 10 studies, IARC estimated that every 50g of processed meat eaten daily raises the risk of bowel cancer by about 18%. Red meat got a vaguer version, 17% for every 100g a day, offered only on condition that the link was ever proven causal.

18% is a relative number. It says how much a risk grows and nothing about how big it was to start with. A second lottery ticket doubles your chances of winning.

50g of processed meat is about 2 rashers of bacon. Cancer Research UK puts the lifetime risk of bowel cancer in the UK at about 1 in 17 for men and 1 in 20 for women.

Out of every 1,000 British men, then, about 59 will be diagnosed with bowel cancer at some point in their lives. If every one of them ate 2 extra rashers of bacon every single day for the rest of their lives, and if the entire 18% were caused by the bacon and none of it by anything else those men do, the 59 would become about 69.

For 1,000 British women the same sum turns 50 into about 59.

That is roughly 10 extra cases per 1,000 men and 9 per 1,000 women, from a daily habit held for decades, on the most generous reading of the evidence there is. The 931 men who ate their 2 rashers every morning of their adult lives and never got bowel cancer do not feature in any headline.

Cancer Research UK ran the same sum in 2015 and reached the same shape: about 56 cases per 1,000 among the people eating the least processed meat, and 66 among those eating the most.

Kevin McConway, then professor of applied statistics at the Open University, set it beside the thing it was being compared with. Of 100 lifelong non-smokers in the UK, about 1 gets lung cancer. Of 100 people smoking a pack a day, more than 20 do. A twentyfold jump and a move from 6 to 7 share a group, because the group was only ever measuring how confident the committee felt.

The relative number made the front pages because the absolute one sells nothing. “Bacon raises cancer risk by 18%” shifts papers. “2 rashers a day for life gives 1 extra person in 100 bowel cancer, if the link is causal” is the same fact, and it was never going to lead the news on a Monday morning.

You will also meet Cancer Research UK’s estimate that 13% of UK bowel cancers are caused by eating processed meat. It is built by taking that same relative risk, applying it across the whole population, and assuming every bit of the association is causal. 13% of a lifetime risk of 5 or 6 in 100 still comes to less than 1 person in 100.

Official advice treats the stakes as modest, whatever the headlines say. The UK government asks anyone eating more than 90g of red and processed meat a day to come down to 70g. In the national diet survey, 34% of adults who ate red or processed meat went over 70g a day, which leaves about two thirds of meat eaters already inside the limit. The government’s own line was never zero.

The relative risks all start life as a questionnaire

The relative risks come from observational cohorts: large groups of people who filled in a food questionnaire, often once, and were then followed for a decade or more to see who got cancer. Nobody has ever run the trial that would settle it, with one group assigned bacon for 30 years and another assigned none.

A food frequency questionnaire asks how often, across the past year, you ate each item on a printed list. The big American cohort in this field, the NIH-AARP study of around half a million people aged 50 to 71, estimated meat intake from a questionnaire given at the start, a 124-item form covering the previous 12 months. Its definition of red meat took in bacon, cold cuts, hot dogs, hamburgers and steak, and also the meat in pizza, lasagne and stew. A ribeye and a takeaway pepperoni pizza went into the same column.

The people who ticked the most red meat were a very particular slice of the population. In the researchers’ own description they were more likely to be current smokers, carried a higher body mass index, ate more calories, had less education, were less physically active and took fewer vitamin supplements. UK Biobank, one of Britain’s largest cohorts, collected its meat data from a touchscreen questionnaire at recruitment, then adjusted for education, waist size, smoking, alcohol and physical activity. Adjustment removes what was measured, to the accuracy it was measured, and nothing more.

That matters because of how small the signal is. Cancer Research UK’s own figures put bowel cancer risk 17% to 21% higher in current smokers than in people who have never smoked. That is the size of the processed meat effect, measured in cohorts where the heaviest meat eaters are the ones smoking more. How precise does a smoking adjustment made from a questionnaire have to be before an 18% signal is safe from it?

IARC understood the problem well enough to apply it once. Asked whether people should go vegetarian, its Q&A explains that the evaluation did not directly compare vegetarians with meat eaters, because the two groups differ in other ways besides meat. Every cohort it did rely on compared its heaviest meat eaters with its lightest: two groups that differ in other ways besides meat.

When Oxford did compare real British vegetarians with real British meat eaters, in the EPIC-Oxford cohort of 63,550 people, the vegetarians had 39% more bowel cancer in the cohort’s first cancer report, in 2009. The pooled follow-ups that came after found no difference in either direction. In Britain’s flagship study of vegetarians, going vegetarian bought no protection from bowel cancer at all.

Why food questionnaires cannot carry this weight is argued in full in nutrition science. The same move, taking a population whose whole way of life differs and crediting one food with the difference, is how the longevity myths were built as well, which is the subject of the blue zone hoax.

The randomised trials that tested the advice found nothing

Questionnaires can only raise suspicions. Randomised trials test them, and the eating pattern recommended against bowel cancer has been through several.

The Women’s Health Initiative randomly assigned 48,835 postmenopausal American women either to carry on as usual or to follow the approved pattern: less fat, more vegetables, more fruit and more grain, with group sessions and self-monitoring to keep them on it. It did not reduce colorectal cancer.

The Polyp Prevention Trial took 2,079 people who had just had precancerous growths removed and put one group on a diet of 20% fat, 18g of fibre per 1,000 calories and 3.5 servings of fruit and vegetables per 1,000 calories. About four years on, the growths had come back in 39.7% of the diet group and 39.5% of the group left on their usual diet.

A trial that gave 1,429 people either a high-fibre or a low-fibre wheat bran supplement found the high-fibre version did not protect against the growths returning. A European trial of fibre supplements saw more of them come back in the fibre group.

Cancer Research UK still attributes 28% of UK bowel cancers to eating too little fibre, a figure built from the same kind of questionnaire data as the meat estimate. The questionnaires said fibre protects, and the trials found nothing. The questionnaires said meat harms, and nobody has ever run a trial built to test it.

What the evidence on meat-based eating does show is collected in carnivore science.

Processed meat and a steak are two separate arguments

Defences of meat that try to clear both with one argument lose the room at exactly this point. The case against processed meat is stronger, and it earns a straight answer.

Curing and smoking can form carcinogenic chemicals, including N-nitroso compounds and polycyclic aromatic hydrocarbons. Nitrite, the curing salt that keeps bacon pink and ham safe, is the starting point for the N-nitroso compounds, and human studies have tracked red meat driving their formation in the colon in a dose response. In rats given a chemical carcinogen, cured ham and hot dogs both promoted the precancerous lesions researchers use as an early marker of colon cancer.

“Uncured” bacon gives no escape. The word means the nitrite came from celery or another vegetable instead of manufactured sodium nitrite, and the nitrite molecule is identical whatever it came from. American labelling rules still oblige celery-cured bacon to call itself uncured and to state that no nitrates or nitrites were added.

So the processed case stands: a real mechanism, a Group 1 listing, and a size of about 1 extra person in 100 at 2 rashers a day for life. If one thing on a meat eater’s plate deserves keeping to the weekend, it is the cured stuff. If bacon stays on the menu, cook it properly and take it off before it blackens.

A steak cooked properly is a different case, and the evidence thins at every step. Red meat sits in 2A because the human evidence was judged limited, and what lifted it to probable was the mechanistic evidence. Much of that mechanism was worked out in rodents.

The leading suspect is haem, the iron-carrying molecule that makes red meat red. In the gut it drives the oxidation of fats into reactive aldehydes that are toxic to the normal cells lining the colon. The French team behind much of this work showed beef promoting precancerous lesions in rats that had first been given a carcinogen and were fed a low-calcium diet. Then they added calcium to the rats’ food, and the beef’s effect was completely suppressed. At the doses found in food, the same group later found haem drove the lesions while the charring compounds and nitroso compounds had no effect in their rats. The experiment that condemned the steak was run on animals eating it without the cheese.

Humans get their own version of the calcium result, from precisely the kind of study this page has spent several paragraphs doubting. In 2025 Oxford published the largest study yet of diet and bowel cancer, following more than 542,000 women from the Million Women Study for an average of 16 years across 97 foods and nutrients. The strongest protective signal was calcium, 17% lower risk for each extra 300mg a day, roughly a glass of milk, with dairy foods lining up behind it. The strongest harmful signal was alcohol, 15% higher per extra 20g a day, about a large glass of wine. Red and processed meat came in at 8% for each extra 30g a day. A genetic analysis in the same paper found genetically predicted milk drinking linked to lower bowel cancer risk, which sidesteps the questionnaire entirely.

By the rules the meat scare runs on, a large glass of wine a day carries nearly twice the risk of an extra 30g of meat, and a glass of milk moves the dial further the other way than either. Anyone who takes the 18% as gospel is bound by these numbers too. Steak with the cheese, then; the carnivore food list was never short of either.

Charring is a separate question again. Cooking any muscle meat at high heat, beef, pork, fish or poultry, forms heterocyclic amines, and fat dripping into flames produces polycyclic aromatic hydrocarbons. Both damage DNA in the lab and caused tumours in rodents, at doses the US National Cancer Institute puts at thousands of times what a person eats in a normal diet. Well-done grilled chicken sits on the same high-concentration list as well-done steak. IARC itself did not have enough data to conclude whether cooking method affects cancer risk.

The fix costs nothing and improves dinner anyway: a hot pan and a hard sear, with the meat kept out of the flames and off the heat before the edges turn to carbon. The method is in the perfect steak.

The 2019 review told adults to carry on, and its critics agreed with its numbers

At the end of September 2019 the Annals of Internal Medicine published a guideline built on five reviews of existing research from NutriRECS, a panel of 14 people from seven countries co-led by Dalhousie University’s Bradley Johnston. They graded the evidence with GRADE, a system its critics said was better suited to drug trials.

They found what everyone else had found. Cutting red meat by three servings a week was associated with about 7 fewer cancer deaths per 1,000 people over a lifetime, and the certainty behind even that figure was low. Their recommendation was that adults continue current red and processed meat consumption, a weak recommendation on low-certainty evidence.

The reaction was immediate. Calls for retraction arrived before the papers were even published, a very settled way to treat a piece of science. Frank Hu, then chairman of nutrition at the Harvard T.H. Chan School of Public Health, made the most revealing complaint of the lot: “There was no breakthrough. It just confirmed previous findings.”

He was right, and it is the most useful sentence the row produced. Both sides were reading the same cohorts and the same small numbers. The fight was over what to call them.

The charge that still arrives whenever the review is mentioned landed days later. The New York Times reported that Johnston had been senior author of a similar 2016 study in the same journal challenging sugar guidelines, paid for by the International Life Sciences Institute, an industry group whose members have included McDonald’s, Coca-Cola, PepsiCo and Cargill, one of North America’s largest beef processors. On the meat work he had indicated no conflicts within the past three years.

Johnston’s reply was that the ILSI money arrived in 2015, and the disclosure form only asked about the past three years. He said the red meat research took no food industry funding. The journal’s editor in chief said it would have published the study even if the past relationship had been disclosed.

On the last day of 2019 the journal published a correction. Johnston had also left off a grant of $76,863 from Texas A&M AgriLife Research, for a separate review of saturated fat, and that one did fall inside the three years. AgriLife’s beef research that year was half paid for by beef industry groups.

At its strongest, with the worst assumed about his motives, the charge still moves nothing, because the numbers were never his. The cohorts and the relative risks were the same ones IARC and every guideline committee had relied on, and the chair of nutrition at Harvard confirmed it in the week of publication. A conflict of interest is a reason to check somebody’s arithmetic. This arithmetic had been done by the field long before Johnston touched it, and the complaint from Harvard was that his came out the same.

Money runs through nutrition in more than one direction. Who gains when a population is told to eat less meat is the subject of meat restriction, and the committee rooms where the original dietary guidelines were written are opened up in nutrition corruption.

England eats less red meat, and its young adults get more bowel cancer

Britain’s national diet survey tracked meat eating from 2008 to 2019. Average meat intake fell from 103.7g to 86.3g a person a day, with red meat down 13.7g and processed meat down 7g, while white meat rose by 3.2g.

Across roughly the same decade, bowel cancer in younger adults in England climbed 3.6% a year, the fourth-fastest rise of the countries studied and faster than anywhere else in Europe. Rates in 50- to 74-year-olds in England fell, and the rise in the young was steeper in women than in men.

Held to the standard this page has applied to everyone else, the trend proves nothing. It is two lines moving in one country at the same time, the weakest evidence in the book. The meat figures are self-reported, the study’s authors note that food supply data is more equivocal and can even suggest an increase, and people born between 1980 and 1999 were the highest meat consumers in the survey. Today’s 30- and 40-somethings are among them. And England’s rise was not matched in Scotland, Wales or Northern Ireland.

The country spent a decade eating less red meat, and bowel cancer in young English adults climbed straight through it. Whatever is driving that climb, eating less red meat did not halt it.

The most promising lead so far points somewhere else entirely. In 2025 a team at the University of California San Diego, funded in part by Cancer Research UK, sequenced 981 bowel cancer genomes from 11 countries. They found the mark of colibactin, a toxin made by certain strains of E. coli in the gut, 3.3 times more often in patients diagnosed under 40 than in those diagnosed after 70. The researchers date the exposure to the first 10 years of life. Whatever is behind the rise in young adults, the strongest new evidence says it reached them as children, decades before any of them filled in a food questionnaire.

Age, family and the kit in the post matter far more than the bacon

Bowel cancer is overwhelmingly a disease of age: more than 9 in 10 cases are diagnosed in people over 50, and more than 4 in 10 in people aged 75 and over.

Family comes next. A parent, brother or sister with bowel cancer raises your risk, and some families carry inherited conditions that change the odds entirely: people with Lynch syndrome face a lifetime risk of up to 80%, and almost everyone with familial adenomatous polyposis develops bowel cancer by their 40s unless it is treated. Families with either are screened earlier and differently, so tell your GP. Long-standing ulcerative colitis or Crohn’s disease in the colon raises the risk as well, which is why people with them are offered regular colonoscopies.

Of the everyday factors, Cancer Research UK attributes 11% of UK bowel cancers to excess weight, 7% to smoking, 6% to alcohol and 5% to too little physical activity. Excess weight leads that list, and the metabolic side of cancer is an argument of its own, made in ketosis and cancer.

Cook meat until it browns and stop before it burns. If cured meat is a daily fixture, that is the one dietary habit with a real file on it, and the file says about 1 in 100.

And send back the kit. The NHS posts a free home test, the faecal immunochemical test or FIT, every 2 years to everyone aged 50 to 74 in England, Scotland and Wales; Northern Ireland currently starts at 60. It takes a small sample, collected with a stick, sealed in a tube and posted back in the envelope provided. In England, anyone 75 or over can request a kit on 0800 707 6060.

The English programme finds around 100 cancers a week, yet in the latest year only 56.2% of 54-year-olds took part, against 73.5% of people aged 70 to 74. When bowel cancer is diagnosed at the earliest stage, more than 9 in 10 people survive it. In the US, screening is recommended from 45.

Whatever your age, don’t wait for the post if you notice blood in your poo, a change in bowel habit that sticks, weight loss you can’t explain, pain or a lump in your tummy, or tiredness with no obvious cause. Book the GP appointment.

The bacon headline has been recycled ever since 2015 without finding a single tumour. The kit on the doormat finds around 100 a week, and it never needed a relative risk to sound important.

Sources

• WHO/IARC, Q&A on the carcinogenicity of red and processed meat (2015)
• IARC press release 240, 26 October 2015
• IARC Monographs Volume 114, Red Meat and Processed Meat
• IARC press release 244, coffee and very hot beverages (2016)
• IARC, 2019 revised Preamble, Samet et al. on the renamed Monographs, JNCI 2020 and CAREX Canada on the removal of Group 4
• IARC Volume 71, caprolactam (Group 4)
• Cancer Research UK, bowel cancer statistics and risk factors
• Cancer Research UK on meat portion sizes (2019)
• Kevin McConway, Open University, via the Science Media Centre (2015)
• Cancer Research UK’s 2015 absolute-risk figures
• Sinha et al., NIH-AARP, Archives of Internal Medicine 2009
• Bradbury et al., UK Biobank, International Journal of Epidemiology 2020
• Key et al., EPIC-Oxford, AJCN 2009 and the 2014 update, AJCN
• Beresford et al., Women’s Health Initiative, JAMA 2006
• Schatzkin et al., Polyp Prevention Trial, NEJM 2000
• Alberts et al., wheat bran fibre trial, NEJM 2000
• Bonithon-Kopp et al., fibre supplement trial, via Cochrane
• Pierre et al., beef, haem and calcium in rats, British Journal of Nutrition
• Bastide et al., haem iron and colon carcinogenesis, Cancer Research 2015
• US National Cancer Institute, chemicals in meat cooked at high temperatures
• Papier et al., Million Women Study, Nature Communications 2025 and Oxford summary
• Washington Post on “uncured” bacon, via Press Herald
• AFP on the NutriRECS findings, reaction and Frank Hu, the disclosure row, New York Times, Johnston’s response, the journal’s correction and the Washington Post on AgriLife (2020)
• Stewart et al., UK meat consumption trends, Lancet Planetary Health 2021
• Sung et al., early-onset colorectal cancer, Lancet Oncology 2024 and Cancer Research UK on England
• UC San Diego on the colibactin study, Nature 2025
• NHS England, bowel screening figures, July 2026 and Cancer Research UK on screening across the UK
• NHS England on survival at the earliest stage (2025) and Bowel Cancer UK on age