Log in

No account? Create an account
Not the first time reading through medical studies has driven me to drink - Jackdaws love my big sphinx of quartz [entries|archive|friends|userinfo]

[ userinfo | livejournal userinfo ]
[ archive | journal archive ]

Not the first time reading through medical studies has driven me to drink [Nov. 22nd, 2012|04:15 am]

So it's pretty well-known that small amounts of alcohol can actually be beneficial to your health, but I had never realized exactly how big the effect was before.

"Alcohol Dosing and Total Mortality in Men and Women" is one of the top studies in this area. It's a meta-analysis of 34 studies totaling about a million people. It finds very robustly that "moderate drinking" (up to 4 drinks/day in men and 2 drinks/day in women) decreases all-cause mortality about 17% to 18%. The "ideal" alcohol amount seems to be about 3 drinks/week, but it's much better to drink more than that than to drink less.

You can find it all divided up into graphs by age and sex here.

Some studies show even greater effects. For example, when you try to divide it out by kind of alcohol, a Danish study of about 20000 people age 30-70 finds that although all forms of alcohol are beneficial, wine has a relative risk of 0.51; that is, you're half as likely to die in any given year if you're a moderate wine drinker.

Just to put these in perspective, if these results are to be believed, not-drinking relative to drinking is almost as unhealthy as being obese, and not-drinking-wine relative to drinking wine is almost as unhealthy as being a light smoker.

Are these results to be believed? There have been no randomized trials, so it's all population studies. Those are vulnerable to confounders, and it's not hard to think of some. For example, people who literally drink zero alcohol might be weird for some reason. They might be sick people who have been warned off alcohol by their doctors. They might be former alcoholics who know they need to abstain or else they'll revert to their former ways. They might be poor people who can't afford alcohol (this was brought up, but does not conform with my experience of poor people).

But the studies consider these possibilities and feel they do not discredit their results. All the smart people exclude former alcoholics from their studies and control for social class, and this usually adjusts the 15-20 point result by like one or two points. There is also one study, very interesting, which measures people who change their drinking habits: people who adjust from moderate drinking to no drinking do indeed get higher mortality rates.

Why might this happen? According to one journal article:
Numerous mechanisms have been proposed to explain the benefit that light-to-moderate alcohol intake has on the heart, including an increase of high-density lipoprotein cholesterol, reduction in plasma viscosity and fibrinogen concentration, increase in fibrinolysis, decrease in platelet aggregation, improvement in endothelial function, reduction of inflammation, and promotion of antioxidant effects.

These effects seem by far strongest on people over the age of 40 or so. If you are younger than that, it is probably not a disaster if you don't drink. But it looks like older people should seriously consider it; the effect size is really impressive.

[EDIT: People are linking to this story as "not drinking is almost as unhealthy as smoking". That's not quite what I'm saying here; it seems to be compared to drinking wine in particular, and this is only if you believe the study with the largest effect size. I've also re-checked the mortality numbers for smoking and found various different numbers; in order to be conservative I've specified a light smoker.]

[EDIT2: My father quite rightly mentioned that percent change in mortality was a terrible statistic to use because it should be an absolute rather than a relative value. In absolute value, drinking in general gains you 2.2 years extra life, and drinking wine 4.7, according to this which I have actually not read in full because I need to go have Thanksgiving.]

Page 1 of 2
<<[1] [2] >>
[User Picture]From: maniakes
2012-11-22 09:53 am (UTC)
Moderate drinking is reputed to be an effective way to cope with stress (e.g. the concept of "having a drink to unwind"), which my personal experience seems to corroborate, and stress is a well-documented risk factor for a variety of health problems, especially heart disease.

If moderate drinking actually does serve as a destressor, I'd expect that to yield an indirect health benefit from displacing other, less healthy coping techniques (e.g. binging on junk food) in addition to the direct health benefits of reducing stress.
(Reply) (Thread)
[User Picture]From: philmophlegm
2012-11-22 10:05 am (UTC)
I don't drink.

(Reply) (Thread)
[User Picture]From: mountainkiss
2012-11-22 10:41 am (UTC)
Nothing decreases mortality.
(Reply) (Thread)
[User Picture]From: marycatelli
2012-11-22 04:39 pm (UTC)
One person, one death. And if modern science comes up with more and more ways to put it off, still the heat death of the universe is going to get us all.
(Reply) (Parent) (Thread)
From: (Anonymous)
2012-11-22 10:50 am (UTC)
Is there a breakdown by ethnicity? I'd expect populations who didn't have alcohol until recently (e.g. in America and Australia) to get very little benefit, and white people to gain most.
(Reply) (Thread)
[User Picture]From: squid314
2012-11-22 07:49 pm (UTC)
I'd expect that too, but I haven't seen a study on that.
(Reply) (Parent) (Thread)
[User Picture]From: st_rev
2012-11-22 11:34 am (UTC)
Seems like the expected result of a few thousand years of obligate alcohol consumption, at least in Eurasia.
(Reply) (Thread)
From: harpersprose
2012-11-22 09:54 pm (UTC)
Really? Population exposure to a toxin should result in increased tolerance, but thats distinct from benefit or dependence.
(Reply) (Parent) (Thread)
From: deiseach
2012-11-22 01:34 pm (UTC)
Guinness is good for you?

Does that apply to wine only, or do other drinks also have a beneficial effect if taken in moderation?
(Reply) (Thread)
[User Picture]From: squid314
2012-11-22 07:50 pm (UTC)
Although the study I link above finds only wine, several others have found smaller (but still impressive) effects for non-wine drinks.
(Reply) (Parent) (Thread) (Expand)
From: wallowinmaya
2012-11-22 01:41 pm (UTC)
Two thoughts.

First of all, not drinking indicates an anti-social personality. If you can't stand other people you'll go out less and thus probably drink less, since most people drink mainly in social settings. And if you don't like alcohol you'll go out less and have fewer friends. It cuts both ways. And we know that having few friends and no social life is very bad for your health.

Also, I guess many folks begin to drink less and less when they try other drugs and learn that there are more interesting options. (Just go to a typical goa-party. Many people there are taking amphetamine, MDMA or LSD but aren't drinking alcohol.)
And drug use is of course bad for your health, for all kinds of reasons.

(Reply) (Thread)
[User Picture]From: naath
2012-11-22 02:49 pm (UTC)
If your social milieu "having friends" with "drinking" then I think you must live in a rather nasty corner of the world. I do drink alcohol, both alone and with friends; but I have friends who do not for various reasons and... gosh, they have friends!
(Reply) (Parent) (Thread) (Expand)
From: madprime.org
2012-11-22 03:01 pm (UTC)

Why is wine special?

While I'm partial to believing moderate alcohol consumption has an effect on mortality through a reduction in heart attacks, I've found the idea that "wine has some other special chemicals" hard to believe.

It leaves me brainstorming, "what other reasons might wine seem special?"

- Social class? You mention studies control for social class in the context of alcohol quantities. Surely they also tried to control for alcohol type... But maybe it's still too difficult to control for the constellation of behaviors associated with "the type of people who drink wine".

- Is wine harder to judge portion size? Maybe people reporting drinking habits for wine are underestimating their intake, and thus getting different apparent effect sizes.

- Self-fulfilling prophecy? Once people think wine is healthy, healthy people drink wine - it may be impossible to get an unbiased assessment now.

I also wonder how the U-curves break down by social class: I wonder if lower socioeconomic classes are more likely to suffer from conditions that would contribute to the righthand rise in the U-curve (e.g. hepatitis C or malnutrition). Maybe you saw something about this in your readings.
(Reply) (Thread)
[User Picture]From: squid314
2012-11-22 07:51 pm (UTC)

Re: Why is wine special?

A lot of people talk about reservatrol re:wine, although I haven't investigated fully.
(Reply) (Parent) (Thread)
[User Picture]From: gwern branwen
2012-11-22 04:05 pm (UTC)
In the same vein of cheap mortality reductions is baby aspirin.
(Reply) (Thread)
From: (Anonymous)
2012-11-22 05:54 pm (UTC)


These effects seem by far strongest on people over the age of 40 or so. If you are younger than that, it is probably not a disaster if you don't drink. But it looks like older people should seriously consider it; the effect size is really impressive.

Actually the meta-analysis you cite says that for young people any amount of alcohol is harmful. This is not surprising if you look at the correlations with particular causes of mortality. Alcohol is only correlated with a reduction in heart disease. For all other causes, it is correlated with more death, including accidents, the principal cause of death of young people. Thus, if you believe that switching works and you believe you can switch, the optimal mortality curve is from switching. But the putative effect on diabetes is evidence for long-term effects.

The graphs in the Danish study of types of alcohol claim that beer and spirits no effect (beyond, perhaps, not being a teetotaler). But I think other studies say that type of alcohol does not matter.

Correlation: Some people say that only four medical effects have been established with purely correlational evidence: lung cancer, skin cancer, HPV, and that alcohol causes cancer. Some people say only the first two. If your standards are low enough to accept positive effects of alcohol, how many other effects must you accept?
(Reply) (Thread)
[User Picture]From: James Babcock
2012-11-22 06:07 pm (UTC)
Huh. I brushed off those studied before, because they were observational and it seemed like there were still more confounders. That effect size, however, is too big for confounding by weak correlations and similar bullshit.

As for where the real source of the difference is, my first instinct is to look at ethanol's metabolites, particularly acetic acid. But there are *lots* of biochemical effects which could be big deals, so I really don't know.
(Reply) (Thread)
[User Picture]From: Andrs Kovcs
2012-11-22 06:19 pm (UTC)
Providing a bit of contrasting opinion, I remember reading this article quite a while ago:


(Reply) (Thread)
[User Picture]From: nancylebov
2012-11-22 07:08 pm (UTC)
It's possible that for people who don't drink because they don't like it (I'm one of those-- most of the time alcohol tastes bad to me, and I've never noticed it making me feel better), alcohol wouldn't improve their longevity-- preference is indicating something about health effects.

My tolerance for the taste of small amounts of alcohol has gone up with the years, but it's still way below average.
(Reply) (Thread)
[User Picture]From: Julia Wise
2012-11-23 01:19 pm (UTC)
Looks like women need to wait until 55 and men until 35 before drinking starts lowering our death rate instead of raising it. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC117446/figure/F4/

Most of the studies don't control for class, which is sad.
(Reply) (Thread)
[User Picture]From: lnr
2012-11-26 12:59 pm (UTC)
Having looked at the first of those studies in detail it looks like you need to be over 55 if you're a woman (35 for men) before drinking becomes non-harmful. Once you remember that a single large glass of typical modern wine can contain around 3 units - the recommended "best" amount to drink in a week for older women - it really doesn't sound like much of an endorsement for what is normally considered to be moderate drinking at all! In fact I'd call it very *light* drinking.
(Reply) (Thread)
Page 1 of 2
<<[1] [2] >>