Alcohol and cancer: What does a ‘500% increase’ in risk really mean?

Kevin Lomangino is the managing editor of HealthNewsReview.org. He tweets as @KLomangino.

[Editor’s note: This post has been updated. Scroll to the bottom to read the addendum.]

Last night, NBC Nightly News ran a story about the cancer risks related to alcohol consumption.

But instead of communicating those risks in a way that would educate and inform, NBC’s coverage was an example of misinformation and fear-mongering.

Consider this graphic touting what the reporter describes as a “staggering increased risk of 500%” for head and neck cancers from heavy drinking.

What’s truly staggering here is NBC’s failure to provide any context around these statistics. 500% is a big, scary number. But what does it really mean and how worried should viewers be? Without knowing how common these cancers are to begin with, it’s impossible to say. A 500% increase to a tiny number might still leave us with a very small risk.

That’s why we always urge journalists to report risks in absolute terms. We’ve even published an informative primer on the topic to help.

The same problem is evident with all of the other figures presented in the story. What does a 17% increase in colon cancer risk look like for women who drink a glass of wine a day?

And breast cancer? The story claims a 4% increase for women who are “light” drinkers. But “light” drinking is never even defined. Presumably this means less than one serving of alcohol per day (a level which the NBC report defines as “moderate” drinking).

So is that 4% increase enough to justify abstaining from even the odd glass of Cabernet? There’s nothing here that will help women make sense of the risks or balance them against the pleasure they might derive from an occasional cocktail.

Wisdom from crowds, but not MD-journalists

The NBC piece was produced by physician medical correspondent John Torres, MD — the latest in a long line of such MD-journalists at the network. But those MD credentials don’t seem to add much when it comes to reporting the news in a credible, nuanced fashion. We’ve cataloged many examples of problematic MD-journalist stories over the years at NBC and other networks.

The upshot: viewers who think an MD byline ensures the ultimate in accurate and balanced TV reporting should think again.

NBC certainly isn’t alone in its statistical inadequacy, by the way. Over at the New York Times, their coverage states that daily alcohol consumption “increases premenopausal breast cancer risk by 5 percent and postmenopausal risk by 9 percent.”

Unlike at NBC, the Times offers a mechanism for readers to push back against imbalanced stories — and a number of smart readers pointed out the problem with the Times’ risk descriptions:

“Please present the absolute figures to show the difference in risk of cancer. With these absolute figures we can also figure out the number of drinkers who should stop drinking to avoid one new case of cancer.”

“again please what is the increased risk/100,000 people or some other way to describe risk besides : “5% increased risk”. This does not give a very useful relatable measure.”

“And where is the discussion about the protective effects of alcohol against heart disease? Heart disease is far more common than many of the cancers mentioned here. When they write that risk is doubled it doesn’t necessarily mean much — if your risk is only half of 1%, doubling it doesn’t amount to very much. Doctors often only think about their area of specialty, and ignore related risks and protective factors for other illnesses.”

“I wish there was a way of expressing risk more clearly, that would help me decide which pleasures are really worth giving up.”

The relationship between alcohol and cancer is a complicated one, and we’ve encouraged reporters to capture that nuance by exploring the limits of observational studies and crafting appropriately cautious messages about such research.

We have a toolkit of resources that can help and we urge journalists to take advantage of it.

Addendum 11/10/17: 

Aaron E. Carroll, writing on the New York Times’ Upshot blog, has a nice breakdown of some of the numbers in the ASCO report and what they mean in absolute terms. He writes:

Let’s stipulate that there may be a correlation between light or moderate drinking and some cancers. We still don’t know if the relationship is causal, but let’s accept that there’s at least an association. For breast cancer — which is the cancer that seems to be garnering the most headlines — light drinking was associated with a relative risk of 1.04 in the announcement. Relative risk refers to the percentage change in one’s absolute (overall) risk as a result of some change in behavior. (And 1.04 is a 4 percent change from 1.0, which represents a baseline of no difference in risk between an experimental group and a control group.)

A 40-year-old woman has an absolute risk of 1.45 percent of developing breast cancer in the next 10 years. This announcement would argue that if she’s a light drinker, that risk would become 1.51 percent. This is an absolute risk increase of 0.06 percent. Using what’s known as the Number Needed to Harm, this could be interpreted such that if 1,667 40-year-old women became light drinkers, one additional person might develop breast cancer. The other 1,666 would see no difference.

Carroll doesn’t address the 500% increase in head and neck cancer risk mentioned by NBC, but statistics from the American Cancer Society (ACS) can be used to clarify that relationship. According to the ACS, a 40-year-old person has a 0.1% chance of developing cancer of the oral cavity/pharynx during his or her lifetime. If that person is a heavy drinker, that risk might presumably increase to 0.6% — a 500% relative increase and 0.5% absolute increase in risk. This is not to suggest that this increased risk is not important or shouldn’t be addressed with preventive messages — merely that the magnitude of the risk may be perceived much differently (and accurately) when the absolute numbers are used.

You might also like

Comments (8)

We Welcome Comments. But please note: We will delete comments left by anyone who doesn’t leave an actual first and last name and an actual email address.

We will delete comments that include personal attacks, unfounded allegations, unverified facts, product pitches, or profanity. We will also end any thread of repetitive comments. Comments should primarily discuss the quality (or lack thereof) in journalism or other media messages about health and medicine. This is not intended to be a forum for definitive discussions about medicine or science. Nor is it a forum to share your personal story about a disease or treatment -- your comment must relate to media messages about health care. If your comment doesn't adhere to these policies, we won't post it. Questions? Please see more on our comments policy.

Charles L Carter

November 8, 2017 at 4:50 pm

All of your comments, as usual, are on the money. But you left me hanging! Now I’ll have to look up those absolute risks myself!
I didn’t see the news reports, but yesterday in another discussion of H & N cancer, another commenter noted those results about alcohol. I refreshed my memory on UpToDate.com where a very similar risk attributed to alcohol was provided, citing a 1999 study. To their credit the authors, chosen for their expertise, acknowledge difficulty teasing apart effects of alcohol and tobacco; but one must dig a little to find absolutes which are given in national totals, not rates/100,000. That’s a resource intended for practicing MD’s. Maybe their editors could use a copy of your guidance.

Reply

David Fox

November 9, 2017 at 9:47 am

It would be helpful to direct us to some examples of good reporting on this topic.

Reply

    Kevin Lomangino

    November 9, 2017 at 12:16 pm

    I have been in contact with the ASCO report authors in an attempt to nail down what this means in numbers that would be meaningful. Unfortunately, without success thus far. I will update if/when I have something useful to share.

    Kevin Lomangino
    Managing Editor

    Reply

David Davoust

November 10, 2017 at 2:18 pm

Thank you for the article. I appreciate that you were not addressing the actual topic, but just the manner of reporting. However, I think the other think this article misses is calling for media to list the actual study being referenced preferably with a way to see the study. Also, how does this compare against past studies? What were the controls? Is alcohol a risk factor or just another component and the risk is actually other lifestyle?

Reply

Joanne Ahrendt

November 11, 2017 at 6:50 pm

Well I am one of the lucky 0.01 % & am 5 months out of treatment for Throat Cancer and all I have to say is if you can cut the risk do it because the treatment is still basic and very brutal & your life will change in ways you can’t even begin to imagine. I never thought it would happen to me but it did.

Reply

Bruce Solomon

November 13, 2017 at 4:11 am

A change from 0.1 to 0.5 is a 400% increase, not 500%.

Reply

    Kevin Lomangino

    November 13, 2017 at 7:51 am

    Correct, and updated. Thanks for the close read.

    Kevin Lomangino
    Managing Editor

    Reply

George Henderson

November 13, 2017 at 5:07 pm

Alcohol intake is always under-reported in surveys; various estimates from studies that double checked put real intakes 2-3x higher than those reported.
This is easy to verify, take total alcohol sales from Govt figures (most accurately in Canada), divide by adult population, then compare with reported intakes in any piece of nutrition epidemiology available. I guarantee it won’t add up.
Alcoholism by its very nature if often not reported at all.
Add to this the problem that factory and agricultural workers, exposed to most carcinogens, may be heavier drinkers than say clerical workers or academics. This is also an issue with meat-and-cancer epidemiology, because I’m pretty sure vegans aren’t working in tyre plants.
Of course alcohol is pretty certainly carcinogenic at a high enough intake, and may well be beneficial at lower intakes, but what these intakes are, isn’t that easy to learn.

Reply