Reporting Risk

October 27, 2015
Posted by Jay Livingston

Thoughts of Mrs. S, my sister-in-law’s mother, rarely cross my mind. My brother’s wedding was the only time I ever met her. But as I read yesterday’s headlines about bacon, sausage, hot dogs, etc.  – “Processed Meat Causes Cancer” – I recalled the one salient fact about her that I knew: For lunch every day of her adult life, Mrs. S ate a hot dog. She died at age 86, having outlived most women in her cohort.

I’m using anecdotal evidence here not to refute the scientific reporting from the International Agency for Cancer Research and W.H.O. I’m not James Inhofe using a snowball in late February to demonstrate that global warming is a hoax. But those headlines do raise the problem of how to report scientific findings.

To many people, the word “cause” implies a nearly certain relationship. Gravity causes the apple to fall. Every time. The stronger the gravitational force, the harder they fall. So “Hot dogs cause cancer” means that if you eat hot dogs, you’ll get cancer. The more hot dogs you eat, the sooner and more severe the cancer.

Some headlines used the more cautious “linked to” instead of “cause.” But that too gets it wrong. It suggests that we have mere correlation with no sure cause.

The accurate headlines talked about risk. Or as the New York Times cautiously put it, “Meat Is Linked to Higher Cancer Risk.”

(Click on the image for a larger and perhaps crisper view of the bacon.)

But what is risk?  The way I explain it to intro students early in the semester is this: Risk is a way of talking about the individual as though she were a lot of people. The research shows that eating a lot of processed food increases the risk of cancer by about 18%. The overall risk of colorectal cancer in the US is about 50 per 1000. That is, if there were 1000 of you, 50 of would get cancer. Now if those thousand yous scarfed down the bacon and hot dogs, 59 yous out of the 1000 would get cancer. Your raised your risk 5rom 50 to 59 per 1000, or from 5.0% to 5.1%

Of course there’s only one of you. Either you get cancer or you don’t. You don’t get fifty one-thousandths of cancer or fifty-nine one-thousandths of cancer.

The headlines were also misleading in another way. The IARC report was not about a newly discovered increase in risk. It was about certainty. Their review of the existing research led them to put processed meats in their highest category of certainty: “causes cancer.” Non-processed red meat was in the next category: “probably causes cancer.” That increased risk for processed meats – from 5.0% to 5.1% – may have been small relative to other things you might do, like smoking cigarettes or working with asbestos, but the IARC was now sure that sausage and bacon caused that risk to increase.    

Maybe the headline should have been:      
Scientists Now Sure That Processed Meats
Cause a Small Increase in the Risk of Cancer
As for Mrs. S, she died of lung cancer. It was the cigarettes* that finally did her in, not the hot dogs – or the Ho Hos, brownies, and chocolate candy that usually followed.

* Cigarettes have long been in the category of causes that we’re certain about. The raising of  processed meats to that same category led some news sources to the utterly wrong interpretation.
“Just two rashers of bacon a day raises your risk of cancer: Health chiefs put processed meat at same level as cigarettes” said the Daily Mail

The Guardian was worse: “Processed meats rank alongside smoking as cancer causes – WHO”

Technically defensible but totally misleading.

Other headlines around the Internet were just plain wrong. To cite one of many,
“World Health Organization: Bacon, sausage as bad a cigarettes” (a Raleigh, NC radio station).

Nonsense. Cigarettes are far more harmful. They increase the risk of lung cancer not by 18% but by 2500%.

No comments: