First they said coffee protects against cancer. Read what happened next
While working on the “coffee” entry for the 10th anniversary edition of my book, The Healthiest Foods On Earth, I had an interesting discussion with a colleague about the nature of coffee research.
There’s been a ton of coffee research, and almost all of it has been very positive. Coffee drinkers have less risk of a baker’s dozen of conditions that aren’t exactly on your Christmas wish list—Alzheimer’s, gout, diabetes, pancreatic cancer. Compared to a matched group of non-coffee drinkers, coffee drinkers have a distinct advantage in the health risks sweepstakes.
Even the World Health Organization has gotten on the coffee bandwagon.
In 1991, it had warned that coffee might be a carcinogen. But in a rare reversal of position, it recently announced (June, 2016) that regular consumption of coffee could actually protect against at least two types of cancer.(1)
Now understand something. I love—repeat love—coffee and drink a lot of it, so I’m always happy to hear about another study confirming coffee’s benefits.
But I may have fallen into a trap, and it’s a trap that many people fall into when reading research, especially research which confirms what they already believe. I’ve pointed out this trap many times, which I thought would protect me from falling into it.
I was wrong.
At this point, I need to credit my esteemed colleague Dr. Dean Raffelock, who first pointed out the flaw in these studies. Follow the argument below, and you’ll soon see what that flaw is.
You see, we now know that there’s a genetic component to how people metabolize coffee. There are slow metabolizers and fast metabolizers, and which group you fall into has a lot to do with how coffee affects you. Slow metabolizers are likely to be kept awake by the caffeine, to have their blood pressure go up (and stay up) and to get the jitters (essentially high levels of cortisol at work). Fast metabolizers have none of these problems and enjoy coffee immensely.
The positive studies on coffee were almost all epidemiological, not clinical. What that means is that scientists look at large groups of people and watch what they do and take measurements. (A clinical study, on the other hand, is an actual experiment where you manipulate variables—like what drug is given, or how much sleep is allowed,. An observational study—which is what most of the coffee studies are—simply tries to find an essentially homogenous group of people (i.e. similar medical history, weight, age, etc) and compares those of them that drink coffee with those of them that don’t.
What a study like this does not do is find a homogenous group of people, break them up into two groups, and tell one of them to drink five cups a day and the other to abstain.
Do you see the difference?
The coffee studies essentially ask people to identify themselves as either someone who drinks coffee or someone who doesn’t. The study then follows the subjects for as many years as possible and at the end of the follow-up, compares the two groups on a whole bunch of medical metrics. The coffee drinkers generally have a better outcome.
But what Dr. Raffelock wisely pointed out is that on one crucial dimension, the two groups– those who drink coffee and those who don’t–are not the same.
“The coffee drinkers who would generally volunteer for a study on coffee drinking would be a biased (self-selected) population because they already possess a metabolism better able to process coffee in healthier ways and would naturally have less tendency toward developing cancer or other diseases from it” he wrote to me in an email.
“Most likely they have liver enzymes better at processing coffee just like some people have livers that process alcohol or other chemicals better than other people. So the coffee drinkers in this study could be a population that is already biased toward suggesting favorable results because they are already know they tolerate coffee well.”
This is the kind of analysis most of us forget to do when looking at epidemiological research. We’d all be wise to remember the mantra of observational studies—correlation is not causation.
In the case of coffee, it’s clearly true that coffee drinkers have a health advantage. But it’s ALSO true that those health advantages only accrue to people who do well with coffee.
Moral of the story: If you hate coffee because it makes you feel like shit, don’t start drinking it for the health benefits.
Metabolically, you’re unlikely to get the same results as people who tolerate coffee just fine.