select search filters
roundups & rapid reactions
before the headlines
Fiona fox's blog

background comments ahead of this afternoon’s IARC Evaluation of the carcinogenicity of Coffee, Maté and Very Hot Beverages (IARC Monographs volume 116)

The World Health Organization’s International Agency for Research on Cancer (IARC) is expected to make an announcement later on the carcinogenicity of coffee and other hot drinks.


Prof. Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“IARC conclusions are published in a series called “Monographs on the Evaluation of Carcinogenic Risks in Humans”, but technically that’s a misnomer, as I’ll explain later. The key thing about this classification system is that it is based solely on strength of evidence that there is some increase in the chance of cancer in people exposed to the thing in question, and not on how much the chance increases.

“There are well over 100 different agents in IARC’s “Group 1” in its classification. These are things where IARC thinks there is sufficient evidence that they really do increase the chance of cancer. But some may increase it by a lot, some by a small amount. Tobacco smoking and eating processed meat are both in Group 1. Of 100 lifetime non-smokers in the UK, around 1 will get lung cancer. For 100 smokers of a pack a day, more than 20 will get lung cancer. That’s a huge increase in risk. Again in the UK, about 6 people in every 100 will get bowel cancer in their lives. According to IARC data, if these 100 people in the UK start eating an extra 50g of processed meat a day, then 7 of them will get bowel cancer. OK, an increase, but only one more death and that’s nothing like the effect of cigarette smoking. So why are smoking and processed meat eating both in the same IARC group? Just because in both cases IARC is convinced that the evidence of some kind of effect, big or small, is quite strong. There are plenty of other agents in IARC Group 1 besides smoking and processed meat eating. Many of them are chemicals that most people (including me) won’t have heard of, but others are more familiar, such as burning coal at home, or working as a painter. Some of these have a big effect on the chance of cancer, others don’t.

“Likewise, before today’s new announcement, IARC had put drinking hot Maté into their Group 2A, which means that (in 1991 when they last reported on coffee, tea and Maté) they considered it “Probably carcinogenic to humans”. But all that means is that the evidence that it causes cancer was a bit weaker than for the things in Group 1. Again Group 2A contains a huge range of things, for which, if they do actually cause cancer, the increase in the chance of cancer might be big or it might be quite small. The group contains some scary-sounding things like glyphosate weedkiller, but plenty of others, including some kinds of shiftworking, or working as a hairdresser or barber, are familiar. Again you have to remember that the increased chance of cancer may be small, and in fact IARC did not consider that there is decisive evidence that there is any increased chance at all (or they’d have put it in Group 1).

“Before today’s new announcement, IARC had drinking coffee (in relation to bladder cancer) in their Group 2B, “Possibly carcinogenic to humans”. This means that the evidence that it causes cancer was considered even weaker than for group 2A, but again I have to emphasise that this tells you nothing about the size of the increased chance of bladder cancer in humans (if indeed the risk is increased at all). Meanwhile, drinking Maté that’s not hot is in Group 3, which means “Not classifiable as to its carcinogenicity to humans”. IARC had investigated it before, but had not found enough evidence that it did (or did not) cause cancer.

“IARC use the two words ‘hazard’ and ‘risk’ to explain what they do in these classifications, but they are using both words in a technical sense. They explain it in a Q&A document at They use the word ‘hazard’ to refer to the amount or weight of evidence that an agent is capable of causing human cancers, and it’s this ‘hazard’ that their classification is based on. They use ‘risk’ to refer to the chance that cancer will occur in a person exposed to the agent, and they say explicitly that their classification does not measure ‘risk’ in this sense, despite the title of the monograph series.”


Prof. Paul Pharoah, Professor of Cancer Epidemiology, University of Cambridge, said:

“I am not sure about how IARC works, but the Monographs explicitly differentiate the terms hazard and risk (which to me can be synonyms).

“IARC states that a cancer ‘hazard’ is an agent that is capable of causing cancer under some circumstances, while a cancer ‘risk’ is an estimate of the carcinogenic effects expected from exposure to a cancer hazard.

“What this means is that they are using the term hazard to mean a risk factor that is “causally associated” with risk of cancer. The risk is then a measure of the strength of association of that risk factor with cancer. The risk (measured as the probability of cancer occurring over a specific time or absolute risk) can be very small even for a causal risk factor (or hazard).

“In other words, just because a factor is deemed a cancer hazard does not mean that the absolute risk associated with that hazard is large or important.

“For example: cigarette smoke is a hazard for lung (and other) cancer(s) that confers a fairly high risk.  X-rays are a hazard for breast (and other) cancers(s) but the risk is very small. So both cigarette smoke and X-rays are classified as cancer hazards, but the importance of them as risk factors is not the same.

“IARC classify hazard into groups, but those groups are about the level of evidence that the putative risk factor is a hazard, not the magnitude of the risk conferred. In the examples above, both X-rays and cigarette smoke could be classified as a Group 1 hazard. Putting them in the same group does not mean they are equally important or relevant to individuals.

“I don’t know what this afternoon’s announcement will be, but IF a substance were classified as a group 1 cancer hazard (like processed meat was last year), it would not mean that it was as important as smoking as a cause of cancer. Far from it, the risk (if any) associated with that substance might be very small.

“Another way to exemplify the Group 1 hazards not being the same is: riding a motorbike in the Isle of Man TT races and driving a car on the M1 are both risk factors that are causally associated with fatal accidents. They would both be classed as a Group 1 hazard for fatality as we are fairly certain that both can be fatal. However, it would be very wrong to consider that the risk of the two activities is in any way similar or equally important.

“Jumping out of an aeroplane with a parachute is a hazard for a fatal accident (people sometimes die after doing this when the parachute fails).  It would be a Group 1 hazard for fatality. Jumping out of a plane without a parachute is also a Group 1 hazard for fatality.  The hazards are clearly not equivalent even though they are in the same group.”


Declared interests

Prof. Kevin McConway: “I have no conflicts of interest.”

Prof. Paul Pharoah: “I have no conflicts to declare.”

in this section

filter RoundUps by year

search by tag