Alcohol and cancer: should we all stop drinking?

This past week, stories on the front page of many newspapers reported that women who drink alcohol in any amount have a significantly greater risk of cancer, especially breast cancer. “Even moderate drinking affects women's cancer risk” read the headline in The Guardian (London).

Is this true, or is this another case where the media (and the scientists talking to the media) are over-hyping their results? The only way to find out is to read the article reporting these results, carefully, and look at their actual numbers, which I did. What I learned was that the lowest risk of cancer was found in women who drank 3-6 glasses of wine per week. Surprised? Read on.

The news stories said that the risk was greater whether you drank red wine, white wine, or any other kind of alcohol. The journal itself issued a press release titled “Million Women Study Shows Even Moderate Alcohol Consumption Associated with Increased Cancer Risk” and it also quoted an editorial appearing in the same issue, which said that “There is no level of alcohol consumption that can be considered safe.”

Wow. Scary stuff. My wife read the front-page story and decided that same day not to drink a glass of wine with dinner. But what is the real risk here? Should you stop drinking wine?

First, here’s how they reported risk: they computed the increase in relative risk for each additional 10 grams of alcohol per day, where 10 grams is approximately the alcohol contained in 1 glass (125 ml) of wine. (125 ml is exactly 1/6 of a bottle, so if you get 6 glasses from a bottle, you’re having a standard glass.)

The article reported a 12% increase in relative risk of breast cancer from each additional glass of wine per day. (That's true - but read on.) The study involved very large numbers, over 1 million women, which means that this increase was statistically quite significant. But they don’t report the absolute risk – only relative risk. Fortunately the article provides the raw numbers, so I’ve computed absolute risk here.

The women were divided into “non-drinkers”, who reported drinking less than one glass of wine per week (24% of the participants), light drinkers who drank 1-2 glasses per week (29% of the participants), and three other groups based on weekly consumption of alcohol, as I show in the table below. The number of cases of cancer was measured an average of 7.2 years later, so what the study really measured was the likelihood that a woman would get some type of cancer over a 7-year period. The average age of the women at the beginning of the study was 55.

Here are the key numbers from the study:

GroupNumber of womenTotal cancersBreast cancers
1-2 drinks/week371453 19307 7841
3-6 drinks 29389115183 6642
7-14 drinks24089412838 5672
>14 drinks67292 4031 1816

Now, here is what they didn’t report: absolute risk. As you can see from the table below, the difference in absolute risk of any type of cancer, or of breast cancer, is very, very small for light to moderate drinkers:

Absolute cancer risk in the JNCI study
               Any cancer(%)   breast cancer (%)
Non-drinkers 5.68 2.09
1-2 drinks/week 5.20 2.11
3-6 drinks 5.17 2.26
7-14 drinks 5.33 2.35
>14 drinks 5.99 2.70
Perhaps the first thing to notice is that light drinkers – 1-2 drinks per week – had a LOWER risk of cancer than non-drinkers. Thus it seems that eliminating all alcohol might not be a good idea. The authors of the study, though, had a different theory: they explained that some nondrinkers might have eliminated alcohol due to an existing medical problem, and therefore that this group might be less healthy than the others. Therefore they excluded this group from their analysis, and used the “light drinkers” (1-2 drinks per week) as the baseline group. The increased risk is entirely based on comparisons to this group.

Next, notice that the risk of breast cancer in women who drink 3-6 glasses of wine per week rises from 2.11% to 2.26%. That’s a 7% increase in relative risk, but the absolute risk is very small in either case. Also, the risk from all types of cancer actually decreases, from 5.20% to 5.17%, and in fact this figure (5.17%) represents the lowest overall cancer risk in the study. So it appears that the quote from the editorial above, that “there is no level of alcohol consumption that can be considered safe,” is simply not supported by the data. (The editorial was written by two scientists from NIH, M. Lauer and P. Sorlie.)

Third, let’s look at the risk for women who drink 7-14 glasses of wine per week. This is 1-2 glasses with dinner every night – still fairly moderate. As compared to the very light drinkers, risk of breast cancer increased from 2.11% to 2.35%. That’s an increase in relative risk of 11% – this is what the study reported – but an 11% increase sounds a lot more scary than an 0.24% increase. The overall cancer risk in this group increased by just 0.13%, which even in relative risk is only 2.5%. The reason for the smaller overall cancer risk is that the risk for some cancers actually went down with increasing alcohol consumption.

The article, while generally well written, does make a glaring overstatement in its conclusions, one that the editors and reviewers should never have let stand. It is this: “regular consumption of low to moderate amounts of alcohol by women increases the risk of cancers of the upper aerodigestive tract, rectum, liver, and breast.” That is not correct. The study showed an association between increasing alcohol consumption and cancer, but correlation does not necessarily imply causation. The authors should have written that increasing one’s consumption of alcohol is associated with increased risk – not that it “increases the risk.” They didn’t prove that, nor did they offer a mechanism that would explain it.

So: what the study really found was that the healthiest women drink 1-2 glasses of wine per week, or 3-6 glasses per week, depending on how you read the numbers. Emphasizing relative risk, as the authors and the journal editors did, overstates the real risk. The commentaries also conveniently neglected to point out that nondrinkers were excluded from the analysis.

As for me, I just bought a couple of bottles of wine yesterday. I think I'll open one tonight.


  1. I learned a long time ago that to enjoy food and drink one should completely avoid reading the epidemiology literature. For example, google "coffee linked to". It's a miracle I'm alive. And if you want to scare your wife further read her this one: PMID 15325026.

  2. Thank you Stephen. It's not like I was going to give up drinking, but at least I won't have to feel guilty about it. Besides, I live with Rafa. I need at least a daily glass of wine.

  3. Oops mispelled your name. And happy hour hasn't even started so there goes that excuse.

  4. My sister emailed me when this paper first hit the press. She wondered if she should give up having a glass of wine each night. I looked over the article and came to the same conclusions that you did. Their conclusions seem to be based on what would be sensational and get good press, not what the study actually showed. Thanks for your nice summary.


Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="">FoS</a> = FoS

Note: Only a member of this blog may post a comment.