About That Correlation Between Boozing and Cancer in Ladies…

Remember that study that got big press recently that showed a link between drinking and cancer rates in women? U.K. researcher Naomi Allen told the Wash Post, "If you are regularly drinking even one drink per day, that's increasing your risk for cancer [since] there doesn't seem to be a threshold at which alcohol consumption is safe."

Over at Spiked, Patrick Basham and John Luik say bushwah:

The study is full of significant puzzles that suggest that its results are unreliable. For example, it reports that the incidence of all types of cancer studied in its non-drinking subjects was 5.7 per cent compared with 5.3 per cent for those subjects who had at least a drink a day, and up to 14 drinks a week.

In other words, not only was there no dose-response in terms of cancer risk, but teetotallers had a higher population incidence of cancer than those consuming up to 14 drinks a week!

Even those women in the study who drank the most (15 or more drinks a week) had a cancer incidence of 5.8 per cent, which is virtually identical to those who drank nothing. But this particular take-home message somehow escaped Allen's notice, and that of the media as well.

Bashma and Luik conclude, Allen's "public pronouncements, her recommendations to government, and the reports about her study in the media are certainly not supported by her results."

Whole thing here.

Comments

[ChicagoTom]

For example, it reports that the incidence of all types of cancer studied in its non-drinking subjects was 5.7 per cent compared with 5.3 per cent for those subjects who had at least a drink a day, and up to 14 drinks a week.

Haven't there been quite a few studies whose findings seemed to imply that having a moderate amount of alcohol daily (like a glass of wine a day etc) tend to have a positive health benefits.

[The Wine Commonsewer (You can]

I enjoy 28 drinks a week. Good thing I ain't a girl.

[kilroy]

My consumption of alcohol has had amazing positive health benefits for several people I would have killed otherwise.

[Episiarch]

A 6-pack is one drink, right?

[sage]

There are amazing benefits to be had from at least one Irish seven course meal* a week. Unfortunately these benefits are purely psychological.

*A six-pack of Guiness and a potato.

[domoarrigato]

That chick looks like a fun time!

[The Bearded Hobbit]

A 6-pack is one drink, right?

24 beers in a case, 24 hours in a day. Coincidence? I think not. . .

.. Hobbit

[sage]

That chick looks like a fun time!

Looks like she needs a ride home. Down some really straight stretches of road.

[Zeb]

so, is there anything in the study that justifies here assertion that there is no safe level of alcohol consumption, or did she simply ignore the results of the study?

[HammeredHead]

I enjoy 28 drinks a week. Good thing I ain't a girl.

Lightweight. Of course that may be 28 large glasses of wine and I bow to you sir...

My average is probably around 47 a week.

[Grandpa Whithers]

I'd do the chick in the pic.

[Grandpa Whithers]

Four large (12 oz) gasses of wine per day (sipped over the course of about eight hours) keeps me very healthy. BP: 110 over 68, Heart rate at rest: 54 bpm. Of couse it does help that I am trim.

[bigbigslacker]

So, I take it there are a few other people that find the forms at the doctor office don't go high enough on the drink count?

"More than 4 per day" might be technically accurate, but it does nothing to describe the shear volume I consume.

[ChicagoTom]

That chick looks like a fun time!

I thought it was Lindsay Lohan. Am I wrong ?

[John C. Randolph]

What ages are we talking about here?

I had a customer many years ago who was an oncologist, and he told me that the main reason we see so much cancer these days is that we live long enough to get cancers. Back when people routinely died by 60 or so, we didn't see it so much.

-jcr

[Neu Mejican]

Without reading the original study I have a Nitpick for the critique:

Fourth, the study looked at 21 types of cancer incidence. Of these, it found statistically significant associations between drinking and only four types of cancer. Moreover, these associations were barely significant.

So, four were statistically significant. (The construction "barely significant" is similar to "barely pregnant.") The fact that some were not is a red herring.

The analysis in the critique is too vague on too many points to judge the validity of its conclusions.

[Neu Mejican]

BTW,

For example, it reports that the incidence of all types of cancer studied in its non-drinking subjects was 5.7 per cent compared with 5.3 per cent for those subjects who had at least a drink a day, and up to 14 drinks a week.

The 17 types of cancer that were not significant may explain this...without harming the conclusions of the study at all.

Again, I haven't read the original study, but this does not seem to be a serious critique.

[Neu Mejican]

I see that one of the authors of the critique is from CATO...I hope that explains why this non-serious analysis was passed along here...

[Joe_D]

It's unclear (though it seems unlikely) that the authors of this article seriously reviewed the data or submitted these questions to the authors of the study in question. I would expect, even if alcohol directly caused cancer, that there would be a negative correlation between the two when looking at a diverse and large population. Why? Young people drink more, old people get more cancer. So you *must* control for age when determining whether alcohol might cause cancer. Failing to do so will produce meaningless and sometimes contradictory data.

This is why we have a peer review system, and perhaps why the reviewers are scientists, not journalists. Or maybe I'm wrong and the article authors aren't hacks.

[Lush (aka Shtoner shcaired sht]

Bu' itsh worf it!

[jtuf]

I don't know. The more I drink, the healthier the women arround me look. Seriously though, every article I've seen on the topic says that a drink a day maximizes health. In related news, The Australian reproted to day the Australia's ministry of health reduced their recommended alcohol consumption to two drinks a day for men and women. The ministers thought of advocating t-totaling, but they didn't want the report completely ignored by the public. Boy, I love Australia.

[The Wine Commonsewer (You can]

Hobbit! That sounds a little like a Stephen Wrightism. 🙂

You can't just drink. You have to take care of yourself, too. Eat good food, get plenty of exercise, have a roll in the hay now and again, hug you kids.

correct for age

I'm too lazy to cite the source, but there is at least one serious study of people over 65 that shows that moderate drinkers have a much greater life expectancy than heavy drinkers or teetotalers.

[Alan Vanneman]

Say what you like, that Katie M-W knows how to party. Go, Katie, go!

[CharlesWT]

ChicagoTom:

"I thought it was Lindsay Lohan. Am I wrong ?"

No.

[CharlesWT]

http://www.mikepaulblog.com/blog/media/Lindsay%20Lohan%20Drunk%202.jpg

Hmmm, I think Radley Balko recently demonstrated the inadvisability of hot linking images on other people's web sites.

[libertarian democrat]

What NM said about that paragraph.

While I wonder about the study too, the critical article is crap. I'm sort of ashamed to see it here.

[libertarian democrat]

You have to be either entirely uneducated in research methods or an idiot to think a good portion of that article makes sense as any sort of real critique (whether or not a real critique is to be had) of the study.

I mean, really?

[Mr. Bayes]

Statistician in the house.

NM: If those 17 cancers had "insignificant" correlations, then they are very unlikely to make a difference to the overall rate, otherwise they wouldn't be "insignificant".

Also, NM, if you carry out 21 "significance" tests at a 5% level, you're going to expect some to be "significant" by chance. A 5% chance, as it happens, not uncoincidentally. So getting four "significant" out of 21 is not that big a deal.

Oh, and can you tell that I don't think much of the concept of "statistical significance", at least in terms of what real-world consequences it has? I really wish people -- I mean pretty much everyone, including and maybe especially academic researchers -- would stop using it. Only xkcd should be allowed.

The Spiked article is non-refereed yet makes as much sense as the refereed article, after all that. And why don't you (= all the people who are dumping on the Spiked article) criticise Ms. Allen for going way beyond what her study can support when she is talking to the journalists?

[libertarian democrat]

Also, NM, if you carry out 21 "significance" tests at a 5% level, you're going to expect some to be "significant" by chance. A 5% chance, as it happens, not uncoincidentally. So getting four "significant" out of 21 is not that big a deal.

5% chance would predict that approximately 1/21 would be significant by chance. I am certainly no expert at statistics, I am not sure the best test to use here, but 4/21 does seem like it means something.

Still not sure about the study itself, but I am critical of the criticism, which is what was posted here. If I saw the study described on another site I was reading, I would probably take issue with it as well.

Also, statistical significance may just mean that there is a relationship, but that matters. The effect size matters too, but the relationship is valuable in and of itself.

[If you are a real statistician]

Mightn't they have used controlled statistics that took the high amount of tests into consideration?

Is there anyway where The association with breast cancer, with by far the largest number of cases in the study (almost 22,000), was non-significant. Therefore, of the cancer-drinking correlations examined, virtually none was statistically significant. makes sense?

Also, who said they are at a p.05? It sounds like the ps are actually more stringent to me.

[Rene]

Check out the author who reviewed this study earlier for a better explanation:
http://junkfoodscience.blogspot.com/2009/03/in-vino-veritas-part-two.html
The study behind this news story came up dry and couldn't find any viable correlation between any cancer and alcohol. Not only was there no dose response and no link (relative risk) beyond chance or statistical error for a secondary epidemiological modeling of data, but MOST of the cancers showed LOWER risk associated with alcohol -- The news failed to report those!-- and no overall risk.

[wizard of oz books]

With many new announcement about the wizard of oz movies in the news, you might want to consider starting to obtain Wizard of Oz book series either as collectible or investment at RareOzBooks.com.