What You Need to Know About the Coffee Cancer Warning in California

Drinking coffee is associated with apparent health benefits, and no evidence of net harm—cancer or otherwise. The judicial standard applied in the California lawsuit distorts that view.
Demi DeHerrera/Unsplash

Coffee now causes cancer, but only if you buy it in California. That, in essence, is the message in a judgment rendered in a lawsuit filed in that state by a non-profit organization called the Council for Education and Research on Toxics (CERT). Roasting coffee beans at high temperature can produce a chemical called acrylamide, a potential human carcinogen. Coffee sellers in California must now post labels warning of this exposure, and the potential hazard. This ruling exposes coffees sellers—from Starbucks, to mom-and-pop operations—to massive liability and potential fines.


I defer the business implications of this ruling to others. I want to address the judge’s determination that the defendants “failed to show that [coffee] posed no risk” or that “consumption of coffee confers a benefit to human health.” After that, I will tell you a bit about acrylamide, and what I think you ought to do about your coffee habit.

But first, let’s consider what it would take to meet the judge’s standard of showing that coffee posed “no risk” of cancer. Actually, that bar is set far too high, as I’ll explain in a moment, so let’s do the opposite. Let’s imagine we had to show that coffee DOES increase cancer risk.

Let’s imagine that the cancer risk attached to coffee consumption is meaningful—perhaps one extra cancer diagnosis per 100,000 regular coffee drinkers per year. At the population level, considering how many of us drink coffee, that’s a non-trivial risk; it would be enough to matter. (This figure, one extra case in every 100,000 people exposed, is the standard under the state’s cancer warning law, Proposition 65.)

How would you prove such a risk? The ideal study would be carried out in people, of course, and randomly assign them to drink coffee routinely throughout the year, or to drink some corresponding “placebo,” whatever exactly a coffee placebo might be. The proof of coffee’s cancer-causing peril would be more cancer in the group randomly assigned to coffee than the group randomly assigned to the corresponding placebo, coffee-free beverage. We must, by the way, presume that the placebo beverage is free of any alternative, potentially carcinogenic components of its own.


Watch More From Munchies:

If we had enrolled 200,000 study participants into this trial, we would have 100,000 in each group. So, according to the risk estimate with which we started, there would be one extra case of cancer in the coffee-drinking group—assuming all other risk factors for cancer, from lifestyle to genes, were perfectly matched between the groups.

I will spare you the gory details of a statistics lesson, but suffice it to say that one extra cancer in a group of 100,000 people would be statistically invisible—well within the realm of chance variation. The analytical methods of statistics are put together specifically to distinguish between trivial variations easily attributable to chance—and differences large enough to imply something about cause and effect. This finding would, decisively, fall in the former range.

What if we doubled the size of the trial to 400,000 people for a year—or by extending the trial of 200,000 people out to two years—and ran the same analysis? Well, two extra cancer cases in 200,000 people is also statistically trivial. So, too, if we tripled, or quadrupled the size or length of the study.

We may comfortably leave it there, because to the best of my knowledge, there has never been a randomized intervention trial enrolling 800,000 people in the entire history of biomedical research. This would, in other words, be the largest, most expensive clinical trial in the history of the world—and would still fail to “prove” that coffee causes cancer.


But even so, this task is easy compared to proving that coffee does NOT cause cancer. Sadly for judges with such inclinations, proving lack of risk is not just hard; it’s impossible. A randomized trial of 10 million people might show that coffee doesn’t cause one extra cancer case per 5 million people per year, but could not possibly show that it doesn’t cause one extra cancer per 8 million people per year. I trust you get the idea; no trial, of any size or length, could truly prove the total absence of risk.

As for the proof of benefit the judge requested, there’s abundant evidence to suggest health benefits from coffee consumption—from the obvious enhancement of alertness, to the potential long-term benefits related to intake of antioxidants—but no decisive proof that coffee, per se, produces some particular health benefit.

Such attribution of effect to an isolated food, nutrient, or ingredient is among the constant challenges in nutritional epidemiology. People who drink more coffee likely drink less of something else. If there is a benefit associated, which is the cause? We almost never know for sure, and thus know a lot more about the general effects of dietary patterns than the specific effects of foods.

What we know about coffee drinking in general is that it is associated with apparent health benefits, and no evidence of net harm—cancer or otherwise. We can’t say for sure that coffee is conferring that benefit, but it looks that way from epidemiologic altitude. The judicial standard applied in the California case distorts that view.


What of the potential carcinogen, acrylamide? The compound is formed when foods containing both starches and the amino acid asparagine, are cooked at high temperatures. Coffee is on the list of foods containing acrylamide, but so are breads and crackers, canned black olives, prune juice, French fries and potato chips, and breakfast cereals. Why coffee was singled out in the California lawsuit is a mystery to me. Of note, there is no proven case of human cancer attributable to acrylamide. Rather, at doses massively higher than human exposures, acrylamide has been shown to produce tumors in rodent models.

I am not at all shy about indicting food industry elements for crimes against human health when I feel that is warranted. I am going the other way in this case because I think cancer warning labels on coffee will invite people to worry about a risk that is vanishingly remote at most, and quite possibly non-existent. Such distortions tend to encourage preoccupation with the trivial, and neglect of the important. At this level of punctilious labeling, sunlight should certainly carry a cancer warning label in California. If all outdoor activity ceased as a consequence, I rather doubt net health benefit would result.

The net effect of routine coffee consumption in common quantities is apparent health benefit, and absence of discernible harm. That does not prove the absence of the possibility of harm, because that standard just isn't achievable in medical research.

I will continue to drink my daily French roast with a bit of unsweetened almond milk. You do yours whatever way you like. But if you like coffee, keep drinking it. Any risks are trivial compared to many others you take daily.

But then again, there’s the risk that no one will sell it to you any more for fear of being sued. Among other things, America runs on litigation.

David L. Katz, MD, MPH, is the founding director of Yale University’s Yale-Griffin Prevention Research Center and founder/president of the True Health Initiative. He is the author of the forthcoming book The Truth About Food.