I Drank a Known Carcinogen... For Science / by F. Perry Wilson

An Annals of Internal Medicine Study suggests that very hot tea may increase the risk of esophageal cancer.

I admitted something to myself this week that I think maybe I always knew but was afraid to say.  I don’t like really large medical studies.

This moment of self-reflection occurred to me as I was reading this article, appearing in the Annals of Internal Medicine, linking Hot Tea Drinking with Esophageal Cancer.

Slide1.JPG

We’ll get to the study in a minute, but first a bit of background. I was somewhat surprised to find that the hot tea – esophageal cancer literature is actually fairly robust.

These studies come out every year or so, and they tend to get a bit of media play because you know, cancer.

Slide2.JPG

Is there biologic plausibility here? A bit actually, thermal damage to tissue, especially repeated damage over time may cause cancer.

But how hot are we talking here?

The World Health Organization considers drinks above 65 degrees Celcius, 149 degrees Fahrenheit to be potentially carcinogenic. I have no idea how hot that is in a drink, actually. So, for the benefit of all of you out there, I drank a known carcinogen:

OK back to the study.

Researchers in China used data from the HUGE China Kadoorie Biobank, a prospective study of more than 500,000 people.

At baseline, individuals were asked a slew of dietary questions, including if they drank tea, what kind, how often, and how hot.  This last was a self reported metric ranging from “room temperature” to “burning hot”. Is there no Iced Tea in China? Business opportunity.

Anyway, the topline results are that, after adjustment, burning hot tea drinkers who were men had a rate of esophageal cancer that was around 1.5 times higher than those who didn’t drink tea. This controlled for both smoking and alcohol-intake, well described esophageal cancer risk factors. There was even a bit of a dose-response with hotter tea leading to higher cancer rates.

Data 1.png

This paints a coherent picture, hence the news articles I’m sure we’ll be seeing, but when I looked closer at the data, not everything added up.

First – look at the raw rates of esophageal cancer.

Data 2.png

This graph shows you the incidence of esophageal cancer across the various tea categories. Not too impressive right? In fact, the highest incidence is in the people who don’t drink tea very much – that’s the reference group in the adjusted analysis.

In other words, an unadjusted analysis suggests that drinking tea is protective.

OK, I thought, there must be a factor that is linked to esophageal cancer that makes you less likely to drink tea. What could it be?

I’ll spare you the sleuthing, except to say that it involved a series of emails with the primary authors.

The authors were nice enough to do some extra analyses and told me that this was due to stratification by region.

What this means is that there are areas in China where hot tea drinking is low, but esophageal cancer rates are high. But if you look within each area, you find that hot tea is associated with more esophageal cancer, even though the cancer rate in hot tea drinkers from Area A may be less than the cancer rate in warm tea drinkers in Area B, you only compare people to their Area compatriots.

You can see the effects of region-based stratification in this graph.

Data 3.png

It’s pretty profound. We’re seeing the effect size flip from protective to harmful just by stratifying by location. This is a fine example of a statistical phenomenon called Simpson’s Paradox, which I’ve discussed before.

Confused yet? So am I. And it brings me back to my first point. I don’t like really huge studies. They have too much statistical power. They allow us to discover things that might be statistically significant, but have no real meaning.

That’s why you can end up with a study like this which, at first glance, suggests that hot beverages are associated with esophageal cancer, but on deeper inspection raises more questions than it answers. Why is the link only seen in men? Why is it so dependent on the region the patient lives in?

Maybe hot beverages are a carcinogen. Fine. But even if they are, they aren’t that bad. The results of this study imply that you would need to stop 1800 men from drinking burning hot tea to prevent one case of esophageal cancer. I mean, why not, I guess? Like give your tea a minute. Let it chill out. Reasonable advice.

But I can’t help the feeling that this study is too large for the question and that’s why we’re left trying to parse a lot of significant findings that don’t hang together that well. In the meantime, you’ve got enough to worry about without taking the temperature of your tea every day. But if you do end up scared, don’t forget this delicious alternative.

Problem Solved.

Problem Solved.