Processed Meat and Bowel Cancer: What the 18% Means

Every year or two the same headline comes back around: bacon causes cancer, and the World Health Organization says so. The claim is not invented. Something specific happened in October 2015, and a real number sits underneath it. Most people repeating the headline have never seen that number, which is why the conversation swings between panic and dismissal with nothing useful in between. Here is what was decided, what was measured, and what it looks like once you put it on an absolute scale.

What IARC decided in 2015

The decision came from the International Agency for Research on Cancer, the WHO body whose job is to evaluate whether things cause cancer in humans. A working group of 22 experts from 10 countries, convened by the IARC Monographs Programme, considered more than 800 studies of cancer in humans — more than 700 of them providing data on red meat and more than 400 on processed meat IARC/WHO 2015. The evaluation was summarised for the literature in a two-page note in The Lancet Oncology, which reports no original data of its own Bouvard 2015.

Processed meat was classified in Group 1, carcinogenic to humans, on what IARC called sufficient evidence in humans that eating it causes colorectal cancer. Unprocessed red meat went into Group 2A, probably carcinogenic to humans, on limited evidence in humans plus strong mechanistic evidence IARC/WHO 2015.

The definitions are narrower than the headlines. Processed meat, in IARC's wording, is meat transformed through salting, curing, fermentation, smoking or other processes to enhance flavour or improve preservation; the examples IARC gives are hot dogs (frankfurters), ham, sausages, corned beef, biltong or beef jerky, canned meat, and meat-based preparations and sauces. Red meat is all types of mammalian muscle meat, such as beef, veal, pork, lamb, mutton, horse and goat IARC/WHO 2015. Bacon is not in IARC's example list, though it does sit inside the processed-meat category used by the meta-analysis reporting the same magnitude, defined there as ham, bacon, sausages, cured or preserved meats Chan 2011.

The cancer at the centre of the evaluation is colorectal — bowel cancer. IARC also reports that an association with stomach cancer was seen for processed meat, but that the evidence is not conclusive; for red meat, the strongest though still limited evidence is for colorectal cancer, with evidence of links to pancreatic and prostate cancer as well IARC/WHO 2015. IARC is also explicit about what it could not settle: there is not enough information to say whether higher or lower risks attach to any particular type of red or processed meat, and whether different preservation methods contribute, and how much, is unknown IARC/WHO 2015. So the classification is a statement about the category, not a verdict on any given label.

The number under the headline

The figure that circulated with the announcement is 18%. IARC's own question-and-answer document states that an analysis of data from 10 studies estimated that every 50 gram portion of processed meat eaten daily increases the risk of colorectal cancer by about 18% IARC/WHO 2015. A published meta-analysis of prospective studies reports the same magnitude independently: a summary relative risk of 1.18 (95% confidence interval 1.10 to 1.28) for every 50 g a day of processed meat, from 9 studies and 10,863 cases Chan 2011. IARC does not name the source of its own figure, so the honest description is that two documents state the same number, not that one is derived from the other.

That meta-analysis is worth describing accurately, because it is not a trial. Its highest-versus-lowest analyses pooled 24 prospective studies — 2 case-cohort, 3 nested case-control and 19 cohort studies — while the dose-response estimates rest on smaller subsets. It reported no evidence of small-study or publication bias Chan 2011. Two further details from the same paper rarely survive into the headlines. For combined red and processed meat, risk rose approximately linearly only up to about 140 g a day, where the curve approaches a plateau; the paper fitted no non-linear model to processed meat alone, so the per-50-gram figure carries no demonstrated ceiling. And for rectal cancer specifically, no significant dose-response association was observed Chan 2011.

The other phrase that gets dropped is "per day". The estimate describes an amount eaten daily, over years — a bacon breakfast on Sunday mornings is a different exposure from a ham sandwich every weekday lunch, and only one of them resembles what the figure was built to describe. As for what 50 grams looks like on a plate, none of the sources cited here gives portion weights, so the only reliable reference is the weight printed on whatever you actually buy.

What 18% does to an absolute risk

An 18% increase in what? A relative risk multiplies a baseline and tells you nothing on its own, so it is worth reaching for a study that reports absolute numbers. The European Prospective Investigation into Cancer and Nutrition followed 478,040 men and women across 10 European countries for a mean of 4.8 years, documenting 1,329 incident colorectal cancers, and published the absolute arithmetic: within its study population, the risk of developing colorectal cancer within 10 years for a subject aged 50 was 1.71% in the highest category of red-meat intake and 1.28% in the lowest Norat 2005. That is a between-category contrast in one cohort rather than the 18%-per-50-grams figure, and it is red and processed meat combined — but it does show the shape of the thing. On an absolute scale, differences of this kind are fractions of a percentage point over a decade, not doublings.

Baselines themselves vary. In the GLOBOCAN 2020 estimates — descriptive statistics for 36 cancers in 185 countries, not a study of causation — colorectal cancer was the third most commonly diagnosed cancer worldwide, at 10.0% of new cases, and the second leading cause of cancer death, at 9.4% Sung 2021. The same report's abstract gives an incidence gap of two- to three-fold between transitioned and transitioning countries, but that figure is for overall cancer incidence across all sites combined, not for colorectal cancer specifically Sung 2021.

Age is built into how screening bodies write their advice. The US Preventive Services Task Force recommends colorectal cancer screening for asymptomatic average-risk adults with a Grade A recommendation from 50 to 75 years, a Grade B recommendation from 45 to 49, and a Grade C — selectively offer — from 76 to 85 USPSTF 2021. That statement is explicitly scoped: it excludes people with a prior colorectal cancer diagnosis, adenomatous polyps, inflammatory bowel disease or a genetic predisposition syndrome, and it is a US recommendation only USPSTF 2021. If you want a starting age, the one that applies to you is the one your own programme publishes.

Why Group 1 is not a danger ranking

Group 1 also contains tobacco smoking and asbestos, and that company is the whole origin of the "bacon is as bad as smoking" line. IARC addresses it head-on: processed meat has been classified in the same category as causes of cancer such as tobacco smoking and asbestos, but this does not mean that they are all equally dangerous, because the classifications describe the strength of the scientific evidence about an agent being a cause of cancer rather than assessing the level of risk IARC/WHO 2015.

The same document supplies the comparison in deaths. Citing Global Burden of Disease Project estimates, it puts about 34,000 cancer deaths a year worldwide as attributable to diets high in processed meat, against about 1 million a year from tobacco smoking, 600,000 from alcohol consumption and more than 200,000 from air pollution IARC/WHO 2015. A confident finding of a small effect and a confident finding of a large one land in the same box. IARC adds one caution against reading its position as permissive: risk increases with the amount of meat consumed, and the data available did not permit a conclusion about whether a safe level exists IARC/WHO 2015.

What might be driving it

IARC's explanatory material names several candidate compounds rather than one settled explanation. Meat contains haem iron; carcinogenic chemicals that form during meat processing include N-nitroso compounds and polycyclic aromatic hydrocarbons; and cooking red or processed meat produces heterocyclic aromatic amines along with other chemicals including polycyclic aromatic hydrocarbons, which are also found in other foods and in air pollution. Some of these are known or suspected carcinogens, but IARC states that despite this knowledge it is not yet fully understood how cancer risk is increased by red meat or processed meat IARC/WHO 2015.

Two things follow. Polycyclic aromatic hydrocarbons appear on both routes — processing and high-temperature cooking — so they are not a cooking-only phenomenon. And these compounds should not be read as working-group conclusions about meat: the General Remarks of IARC's Monographs Volume 114 state that, for haem iron, heterocyclic aromatic amines, N-nitroso compounds and polycyclic aromatic hydrocarbons, the working group did not specifically evaluate these agents in relation to meat consumption IARC/WHO 2015. On cooking method, IARC says high temperatures or direct contact with a flame or hot surface produce more of certain carcinogenic chemicals, but that there were not enough data to reach a conclusion about whether the way meat is cooked affects cancer risk IARC/WHO 2015. Mechanism is why the association is considered biologically credible. It is not itself the evidence of harm.

How load-bearing the evidence is

The evidence cited here is observational: prospective cohorts of what people reported eating, pooled into summary estimates Chan 2011. That design supports association, not proof of cause in any individual, and it carries the standing caveat of nutrition cohorts — residual confounding by whatever else travels with the exposure — which the studies address by adjustment rather than by removal. One guideline panel weighed exactly that. Working from five de novo systematic reviews and GRADE certainty ratings, with 14 members from 7 countries including 3 community members, the NutriRECS consortium rated the certainty of the evidence low and issued weak recommendations that adults continue their current unprocessed red meat consumption and their current processed meat consumption Johnston 2019. That is a judgement about how to act under acknowledged uncertainty, not a new measurement, and it does not dispute that the association exists.

Independent cohorts do point the same way. The UK Biobank analysis of diet and colorectal cancer, following 475,581 adults aged 40 to 69 who reported their diet on a short food-frequency questionnaire, recorded 2,609 cases over an average of 5.7 years. Participants reporting an average of 76 g a day of red and processed meat had a 20% higher risk of colorectal cancer (95% confidence interval 4 to 37) than those reporting 21 g a day, and the per-increment estimate was a hazard ratio of 1.17 (1.04 to 1.32) per 50 g a day Bradbury 2020. Two qualifications matter: the exposure is red and processed meat combined, so it is not a like-for-like replication of the processed-meat-only figure, and the authors note that participants meeting the UK government recommendation of no more than about 90 g a day still showed elevated risk Bradbury 2020. The same analysis found no association with fish, poultry, cheese, fruit, vegetables, tea or coffee, a 14% lower risk (2 to 24) in the highest fifth of fibre intake from bread and breakfast cereals, and an 8% higher risk (4 to 12) per 10 g a day of alcohol Bradbury 2020.

The EPIC cohort corroborates it, and the detail is worth getting right. Its highest-versus-lowest contrast — more than 160 g a day against less than 20 g a day of red and processed meat — gave a hazard ratio of 1.35 with a 95% confidence interval of 0.96 to 1.88, which crosses the null, although the trend across categories was significant at P = .03. The continuous estimate was already significant before any correction, at 1.25 (1.09 to 1.41) per 100 g a day, rising to 1.55 (1.19 to 2.02) after correction for dietary measurement error using a calibration substudy of 36,994 subjects Norat 2005. The finding did not depend on the correction; the correction strengthened it. That single calibration is also the clearest illustration in this reference set of how imprecise self-reported diet can pull an estimate toward the null — one analysis, not a general law.

The same paper reported an inverse association with fish intake: 0.69 (0.54 to 0.88) for more than 80 g a day versus less than 10 g a day, and 0.70 (0.57 to 0.87) per 100 g a day before calibration, 0.46 (0.27 to 0.77) after. Its authors conclude that their data confirm a positive association with red and processed meat and support an inverse association with fish Norat 2005. That is observational too, and the paper makes no claim about swapping one food for the other. If you are curious what you would actually be eating instead, our breakdown of the macronutrient profiles of fish and shellfish is a reasonable starting point.

The most recent pooling in this reference set is larger than the one behind the headline. A 2025 meta-analysis of 60 prospective studies reported, for the highest versus the lowest consumption categories, a hazard ratio of 1.21 (1.14 to 1.28) for processed meat and colorectal cancer, 1.13 (1.07 to 1.20) for colon cancer, 1.17 (1.05 to 1.30) for rectal cancer, and 1.15 (1.10 to 1.21) for red meat and colorectal cancer. Its authors conclude that high consumption of red and processed meats is significantly associated with increased risk of colorectal, colon and rectal cancers, and that the findings reinforce dietary recommendations to limit intake Ungvari 2025. Those are categorical contrasts, not per-50-gram or per-100-gram increments, so they cannot be substituted for the 18% figure — they are a different quantity answering a related question, and they land in the same direction.

Where the guidance lands

IARC does not tell anyone what to eat, and says so: it evaluates the evidence on the causes of cancer but does not make health recommendations, which are the responsibility of national governments and WHO. Its Q&A adds that individuals concerned about cancer could consider reducing their consumption of red meat or processed meat until updated guidelines specific to cancer are developed, and that eating meat has known health benefits IARC/WHO 2015.

The bodies that do issue numbers are stricter than the classification alone implies. The World Cancer Research Fund and American Institute for Cancer Research, in the Cancer Prevention Recommendations of their 2018 expert report, advise eating no more than moderate amounts of red meat and little, if any, processed meat — with red meat limited to no more than about three portions a week, roughly 350 to 500 g of cooked weight per week, and very little, if any, processed meat. Their evidence grading mirrors IARC's: convincing evidence that processed meat is a cause of colorectal cancer, and red meat probably a cause WCRF 2018. Note the units: that red-meat allowance is per week, not per day.

The 2011 pooled meta-analysis reaches the same practical place, its authors writing that the overall evidence of prospective studies supports limiting red and processed meat consumption as one of the dietary recommendations for the prevention of colorectal cancer Chan 2011. So the disagreement in this reference set is narrow: it is between guidance that says limit and a guideline panel that, rating certainty low, suggested continuing current intake Johnston 2019. Nobody in it argues the association is absent.

Practically, the frequency question is the one the evidence speaks to most directly: the figures above describe amounts eaten daily and sustained, not a portion eaten now and then. If you would rather work from a whole-diet template than a single-food rule, our assessment of the Mediterranean pattern describes an approach that treats cured meat as occasional rather than staple. And if you are in the age range your own screening programme covers, that programme's published advice — not an age taken off a page like this one — is the thing to act on.

Frequently asked questions

Is processed meat really as dangerous as smoking?

No, and IARC says so directly: processed meat has been classified in the same category as causes of cancer such as tobacco smoking and asbestos, but its Q&A states this does not mean they are all equally dangerous, because the classifications describe the strength of the scientific evidence rather than the level of risk IARC/WHO 2015. The same document cites Global Burden of Disease Project estimates of about 34,000 cancer deaths a year worldwide attributable to diets high in processed meat, against about 1 million a year from tobacco smoking IARC/WHO 2015.

How much processed meat is 50 grams?

None of the sources cited here gives portion weights, so the honest reference is the label on whatever you buy. IARC's examples of processed meat are hot dogs (frankfurters), ham, sausages, corned beef, biltong or beef jerky, canned meat, and meat-based preparations and sauces, and it adds that there is not enough information to say whether higher or lower cancer risks attach to any particular type IARC/WHO 2015. The 18% figure was expressed per 50 grams eaten daily, so it describes a sustained habit rather than an occasional portion IARC/WHO 2015.

Does the classification cover fresh red meat too?

Separately, and less firmly. IARC placed unprocessed red meat in Group 2A, probably carcinogenic to humans, on limited evidence in humans plus strong mechanistic evidence IARC/WHO 2015. Its Q&A adds that if the red-meat association were proven to be causal, data from the same studies suggest risk could increase by 17% for every 100 gram portion eaten daily — a different unit from the processed-meat figure IARC/WHO 2015. The published meta-analysis reports a matching 1.17 (95% confidence interval 1.05 to 1.31) per 100 grams a day Chan 2011.

Does how I cook it change anything?

IARC says cooking at high temperatures, or with the food in direct contact with a flame or a hot surface, produces more of certain carcinogenic chemicals including polycyclic aromatic hydrocarbons and heterocyclic aromatic amines — but that there were not enough data for the working group to reach a conclusion about whether the way meat is cooked affects the risk of cancer IARC/WHO 2015. The 18% estimate came from studies of how much processed meat people reported eating, not how they cooked it Chan 2011.

Should I give up processed meat entirely?

The guidance bodies split on the practical answer. The World Cancer Research Fund and American Institute for Cancer Research recommend eating no more than moderate amounts of red meat and little, if any, processed meat WCRF 2018. The NutriRECS panel, working from five de novo systematic reviews and GRADE certainty ratings, issued weak recommendations on low-certainty evidence that adults continue their current processed and unprocessed red meat consumption Johnston 2019. IARC itself makes no dietary recommendation, but says people concerned about cancer could consider reducing their intake, and that eating meat has known health benefits IARC/WHO 2015.

References