May 17 2018

California Rules Coffee a Carcinogen

California Superior Court Judge Elihu Berle ruled, in accordance with Proposition 65 law, that coffee is a carcinogen and requires a warning label in the state of California. This ruling, however, is not in accord with science and rational medicine.

What went wrong are all the things that can go wrong when trying to assess health risk.

The ruling is based on the presence of acrylamide in coffee. There is evidence that acrylamide is a potential risk factor for certain cancers. This is based mostly on animal and in vitro studies.

How scientists determine that a substance is potentially a cancer risk is a little complex, since we cannot do direct clinical studies. It is unethical to expose a subject to a potential carcinogen to see if they get cancer. So we start with studies on cells and animals.


These types of pre-clinical studies, however, can only tell us about hazard, not risk. A hazardous substance is something which can theoretically cause harm depending on exposure, while risk is actual harm caused by exposure in a certain population to a certain dose over a certain period of time.

Think of it this way – sharks are hazardous, but a shark in a tank poses very little risk. Swimming with sharks in that tank, however, poses a high risk.

So how do we determine the risk to humans of a substance which is potentially hazardous based on pre-clinical studies? For that we rely on epidemiology and observational studies. If we, for example, ask people about their coffee drinking habits, and then track them over time for the development of cancer, we can then make comparisons and see if drinking coffee correlates with increased cancer risk.

There are many shortcomings from this type of data, however, specifically the potential for confounding factors. Drinking more coffee may correlate with more smoking, or consumption of other hazardous products, for example. You can try to control for possible confounding factors, but not the ones you are not aware of. So this limits the utility of such data.

Even still, it is possible to build a robust base of evidence clearly showing risk. If the observational studies all triangulate to only one plausible cause, then we can be highly confident of that interpretation of the correlations. Smoking, for example, correlates with risk of lung cancer every possible way you can look at the data. This is a solid causal link.

The same solid link does not exist for acrylamide in coffee. Therefore coffee is deemed only a probable cause of cancer by the IARC, which is a notoriously cautious body that errs on the side of calling substances “probable” carcinogens. Essentially, if some hazard shows up in pre-clinical studies, but there is no clinical evidence of actual risk, they still call it “probable.”

California law then takes the IARC “probable” and turns that into “coffee causes cancer.” In each step the precautionary principle ratchets up the warning, until we go from no clinical evidence of actual risk, to a required warning label that it does cause cancer.

This approach, however, is not without its own risk, turning the precautionary principle on its head.

As I reported previously on SBM, a 2013 study found that 72% of random ingredients from a cookbook had published data showing that they increased the risk of some cancer. What is the public supposed to do with this information?

There are two real risks here, the first of which is what is called alarm fatigue. If you set your threshold for warning alarms too low, then the alarms will always be going off, and people will learn to ignore them.

Do you think many people will stop drinking coffee because of the warning? I wonder what percentage will give up their daily coffee habit based on this decision. Most people will likely (and correctly) file this away as a manifestation of the overly cautious CYA nanny state and ignore the warnings. But this makes it more likely they will ignore more meaningful warnings about actual risk that they should be listening to.

If everything causes cancer, then perhaps nothing really does, and you should ignore all such warnings. The warnings themselves become useless.

There is also the opposite risk, that some people, in the sincere desire to be healthy or out of anxiety, will try to heed all such warnings. This can lead to what some are calling “orthorexia” (not currently a recognized clinical term), in which people excessively restrict their diet in the hopes of avoiding risky food.

If 72% or so of foods or ingredients can be linked to cancer, and you try to avoid them all, then what will you eat? What are the risks of such a highly restrictive diet? I strongly suspect that if you followed all the warnings about potential risks of foods, there would be net harm from poor nutrition.

There is also an opportunity cost to spending so much time and effort carefully vetting everything you eat for any possible risk. That time could be better spend on more fruitful healthy activities.

There is also an industry of self-help advice turning these public warning into marketable fearmongering. They have essentially been “weaponized” for either marketing or to promote various ideological agendas. The anti-GMO campaign is perhaps the best example.

The state should be trying to mitigate these counterproductive forces, not aid them. If anything they should be a mediating force, putting together panels of experts to create sober and useful bottom-line advice for the public, advice that people can plausibly used, and designed to minimize unintended consequences.

California’s approach, I think, is harmful. In an endeavor to protect the health of their citizens I think they are harming health. They are also potentially undercutting faith in experts an expertise itself, which has further downstream negative consequences.

This is the precautionary principle out of all control.

No responses yet