February 20, 2007

Just How Many Drinks A Day Is Bad?

Is a glass or two of wine a day good for you?  You would think this would be an easy question to answer, but it's not, and that's because of this:

How many glasses of wine are in a bottle?

If you answered 4-5, continue reading.  Because guess what?  Apparently the answer is eight. No, I'm not kidding. Yes, they are serious.  Unplug your monitor and ram it into your skull as hard as you can. 


"Doctors don't know.  They pretend to know.  Because they have a rectal thermometer in their pocket.  As if it were an appeal to a higher authority."-- Lewis Black 

It comes down to this: medicine doesn't have all the answers, but in presenting their recommendations it sounds like they do.  And we get confused.

"Two drinks a day."  What's a drink?  Are the health benefits/risks of whisky and wine identical?  Then why lump them together?  And if the answer is, "well, we don't know enough yet" then why are you making recommendations with the authority of medical certainty?

The reason wine's benefits/risks seem confusing is that no one emphasizes the size of a "drink."

A "drink" is often defined as 10g alcohol-- that's 1/8 of a bottle of wine. 

Many reference guides, in an attempt to make things simple to understand (if you're drunk, maybe) use 10g/drink as a standard. There are 750ml in a bottle of wine. If the bottle is 13% alcohol by volume, then there is 98ml alcohol per bottle. Alcohol's specific gravity is .79, so there are 77g alcohol in a bottle.   That means that there are 7-8 "drinks" in a bottle of wine, which, if I may editorialize, is so preposterous as to hardly merit comment. 

Similarly, for 5% beer, there is 0.05 x355ml x 0.79= so 14g per 12 oz can.  Is a beer a drink and a half?

Alcohol content varies greatly among wines and beers:

Also, each wine has a different alcohol content-- 12.5% is the typical French ideal, and most wines are built (i.e. alcohol osmotically removed) to stay under 14% because the tariff increases above that.   There is a leeway of 1.5% in the listing, so 12.5% could be 11% or 14%.  That's a 2 "drink"/bottle difference.

In the past few years, and especially with California wines, Syrahs, Zinfandels, places with hot climates, the trend has been towards using higher Brix (sugar content) grapes.  (Riper means more sugar, which means more alcohol.)  About 55% of the sugar ferments to alcohol, and the common 25 Brix grapes convert to a 13.75% wine.  Plus you lose some water in the wine making process, so it may be even higher than that (15%).

To complicate things further, each policy group advocates different safe drinking levels that are nearly incomprehensible to the layman, unless you convert them to some common measure (here, I convert to grams):  The U.S. government says no more than 2 drinks per day-- but that's 14g each.  France says no more than 5  per day-- but it's 12g each.  Britain says 3-4 "units"-- at 8g each!  Do the British know this?  A British study (in Scots) found that people generally pour out two, not one, unit per drink.  It's no wonder people are confused.

A better conversion is this: there is 77g alcohol in a 13% bottle of wine.  That's equivalent to  almost a six pack.  Go.

Blood Alcohol Content: as accurate as a New York Times poll, but you can still go to jail.

Converting to grams as a reference for drinking is useful because it allows you to predict your BAC.  Here is how everyone tells you to calculate it:  If you drink 40g alcohol and weigh 70kg, your BAC will be .05% (40g/70000g).  Or, if you weigh 70kg, every 14g beer will raise your BAC by .02%  Or, every 1/4 bottle of wine raises the level .035%.  Isn't math fun?  I have seen countless "reference tables" using this method.

But the units of BAC are g alcohol/100ml water.  You're not all warer, are you?  You're about... 60% water?  So that 40g alcohol in really in 70kg x .6= 42kg water.  40g alcohol/42000ml water= .09%.  Congrats.  You're drunk.  Sort of.

In practice, those reference tables telling you your estimated BAC already incorporate the Widmark constant-- the percentage body water. It can range from 40-85% water.  The more water you have, the lower will be your BAC.  Women have less water, so their conversion runs lower (40-50%).  Muscle= more water; fat=less water. The problem, obviously, is while BAC calculations use a standard-- for example, my .6, above-- individuals can vary greatly.  Hence, lawyers.

But wait: Breathalyzers. It's measuring the alcohol content of your breath, not blood.  What's the ratio of alcohol in breath to blood?  2400:1?  2100:1?  Generally, breathayzers are calibrated to underread your alcohol level, by about 10%.  So even though most humans run 2400:1, it is calibrated at 2100:1.  But don't try to argue "individual variability" of a breathalyzer in court: 2100:1 is part of the statute, and thus your reading is your sentence.   But remember, liquid to gas transitions are described by Henry's coefficient: heating a substance (e.g. alcohol) puts more in the air (breath); cooling the air (breath) makes the substance stay in liquid (blood).  So before you blow into the machine, hyperventilate and roll in the snow.

The point here is that you-- and guidelines-- cannot predict your BAC based on how many "drinks" you had, because there are so many confounding variables. 

Note that BAC doesn't tell you how drunk you are-- tolerance might mean you're an effective Lisp programmer at .1%, or you're beer goggling at .02%. At a given weight, higher percentage body fat= more drunk. Also, food delays absorption. Finally, some people metabolize alcohol faster than others; the old rule "a drink an hour" is based on the assumption that you metabolize 10g alcohol per hour (or your BAC falls by .01%/hr)-- but in you it may be 20g/hr (e.g. a daily drinker), or 5g/hr (e.g. young woman rarely drinks, on Tylenol) etc.

But legal driving limit is usually .08%.  And 50% of the time, .4% is death, so there's that.

Health Benefits of Wine?  Or No? 

So since the term "drink" is uselessly vague, in reviewing the literature on wine and beer's effects, I'll do my best to convert to grams of alcohol.  Just remember that a bottle of wine is 77g, and 12oz 5% beer is 14g. 

Cholesterol, triglycerides, coronary artery disease: about half a bottle of wine, but at least 20g/d, raises HDL,;decreases TG, CRP, fibrinogen, and decreases risk of CAD.

Generally, moderate alcohol consumption (say, 30-40g/d) is associated with decreases in mortality.  This is hypothesized to be related to a) its HDL raising effect; b) its reduction of pro-inflammatory proteins CRP and fibrinogen (i.e. it's anti-inflammatory.) 

One of the studies, in Nature,  that popularized "moderate consumption" was this: 40g/d (from beer) for men, 30g/d for women, reduced inflammatory markers C-reactive peptide (35%) and fibrinogen (12%),  increased HDL (10%), with no change in TG or liver enzymes.  after 3 weeks of drinking.   The study called this "four glasses" but a better way of understanding it is three beer cans or  half a bottle of wine.  Also: BAC 1 hour after drinks was 10mmol/l.  Yes, mmol.  Sigh.  46g/mol: BAL .046%

A prospective study confirmed the "well-known" relationship between alcohol consumption and HDL, which rose from 40 to 50 with >30g/d alcohol.

A German study of 7000+ people found HDL rose, and fibrinogen decreased,  for women who drank 10-20g/d and men >30g/d.

A Danish study found an interesting relationship: women who drank at least once per week had lower risk of CAD than abstainers; but drinking more often did not promote the effect.  But for men, daily drinking (more than less frequent drinking) was associated with the lowest risk. 


Oxidative Stress: doesn't ethanol cause lipid peroxidation (free radicals?) Answer:  you're not drinking ethanol, you're drinking wine--which probably increases antioxidant capacity. 

This is how you get plaques: free radicals in your diet (e.g. cooked fat) promote LDL oxidation, which goes on to promote arterial plaque formation.  Free radical scavengers, such as Vitamin E, would lessen this effect-- but are consequently reduced.  Importantly, the LDL from a meal is more susceptible to oxidation than normally circulating (fasting) LDL.

Alcohol promotes oxidation in test tubes.  So why wouldn't it do so in people?  For example, a careful study controlled for many confounding variables that are associated with high or low alcohol intake-- such as smoking, vitamins, exercise, etc-- and found that the more alcohol consumed, the higher the oxidized LDL, with no change in HDL.  Where did the protective effect go?  One possibility jumps to mind: median consumption was 6g/d; and the above studies found the relationship with the higher "doses."  And you need to be a regular drinker: 96 hours after a single dose of wine there was no effect on LDL. Surely I've made this up?  No: 300ml red wine (better than 300ml white wine) inhibited oxidation (e.g. LDL oxidation).  The likely explanation is that even though alcohol can cause oxidative stress, wine-- and it's constituents (polyphenols, resveratrol, etc) may overwhelm this effect.  But you have to drink enough (>300ml) so that it overwhelms alcohol's effects (but not so much your wife leaves you.)

Additionally, wine's beneficial effects in preventing oxidative stress may be enhanced when you have more oxidative stress to begin with.  Take the easy case of eating a fatty meal.  The LDLs that result from this meal are  more likely to be oxidized than the normal  fasting LDLs  in circulation.  Drinking 400ml of wine with a meal made these post-meal LDLs more resistant to oxidation than even the existing LDL, and maintained the Vitamin E levels.  And in case you're a rat, in rats who were force fed a high cholesterol diet, wine reduced the cholesterol levels and improved antioxidant parameters. 

Not just meal related oxidative stress: 1/3 bottle of red wine a day for two months in people who just had angioplasty substantially increased antioxidant reactivity and decreased oxidative damage.  There is a logic to this: the lower your CRP, the better is your natural antioxidant capacity, and wine lowers CRP proportionally more if it is already high.  A glass of wine (or one espresso- how do you like that!)  was equivalent to an orange or 200g spinach in antioxidant capacity.


Homocysteine (which causes coronary plaques)?   Maybe it goes up a little, but that might not matter, especially if you're drinking wine.

42 men got to drink half a bottle of WHITE wine a day for a month: lower oxidation products (and coincident increase in free radical scavengers and HDL), but also increased homocysteine. 

A prospective study found that after 6 weeks of 30g/day of wine/beer/spirits, homocysteine levels were higher than in controls.  Folate levels were also lower (except in beer-- because beer has about 30ug folate/beer and0.1ug vitamin B6/beer.)  Folate and B12 are cofactors in the conversion (methylation) of homocysteine which is then broken down (sulphyrated) with vitamin B6 as a cofactor; so low folate/B12= high homocysteine.  Similarly, in chronic alcoholics homocysteine was much higher-- but less so with beer.

And again, but with 40g/d drinking wine and spirits for three weeks, homocysteine went up 9%.  Beer had no effect.  But B6 went up with all drinks (more with beer).  Not only does B6 facilitate homocysteine degradation, it is also an independent inverse risk factor for cardiovascular disease.

But perhaps amounts are relevant: in another prospective study, 1/2 bottle/d of red wine for two weeks had no effect on homocysteine, while doing the expected increase of HDL and antioxidant capacity. 

A study using pig coronary arteries found that while homcysteine impaired endothelial cell relaxation, red wine negated this adverse effect. 

It appears that homocysteine goes up, but that doesn't translate to any increased cardiovascular risk because of some beneficial effects of the wine, which may include B6, antioxidants, increased HDL and increased antioxidant capacity.

Blood Pressure?  Answer: No serious effect below a bottle of wine a day. 

German study (above) finds <80g/d associated with <2 mmHg increase; >80g/d associated with  4-6mm Hg increases.  American Idol makes mine go up more.

Much of the negative data on blood pressure is perplexingly inaccurate. By "perpelxingly" I mean that the errors could not have simply been oversights, could they?  People are lumped together, as are quantities and types of alcohol, giving misleading results.  For example, in an article entitled, "Alcohol is Bad For Blood Pressure"-- seriously, that's the title of the scientific article-- the authors state:

Since then, large-scale prospective studies from Japan (6) and the US(7) have indicated that the risk of hypertension increases twofold with alcohol intake of 30–50 g/day or more. 

Hmm.  "Increases twofold."  I'm not sure what article they read, but reference 7 pretty clearly says the opposite:

Our principal finding was the association between the consumption of low to moderate amounts of ethanol (up to 3 drinks per day) and either the incidence of hypertension or increase in blood pressure levels in blacks. In white men, there was no evidence of an increase in systolic or diastolic blood pressure over time at this level of consumption. Similarly, for most beverages, a low to moderate intake of alcohol was not associated with a higher incidence of hypertension in white men and with an increased incidence in black men.

And later:

the observation that low amounts of alcohol intake may not increase blood pressure in most race-gender strata could lead to a more tolerant view of the consumption of alcohol in small amounts...

Black men who drank heavily had double the incidence of hypertension (defined as a jump to > 140/90): 15% vs. to 30% in drinkers.  But I should add that the risk was relevant only in black men who drank beer or spirits; only 8 out of 250+ drank any wine at all.

Thus, blood pressure is minimally affected by wine, and even beer or spirits, if other variables are controlled.  There is a negative effect of beer and spirits in blacks that needs to be explored, as does the effect of wine in blacks. 


Pancreatitic disease: How many drinks before you're in trouble? Answer: >30% of your daily calories from alcohol if you poor nutrition; or  >1 bottle wine/day for 25 years, especially if you eat like a pig.   Smoking=death.  (But you knew that.)

You'd be amazed at how hard this simple question was to answer.

As an aside, almost every study done in 2005-2007 on alcohol and pancreatic disease was done in Japan or China.  I'm sure there's a reason for this, but for the life of me I can't tell you what it is.  And if someone is able to explain to me how the Japanese and Chinese physiologies are generalizable to everyone, I'd like to hear it; but that's what happens.

The main problem with the studies is that risks of pancreatitis are associated with an arbitrary cut off that does not reflect the actual toxicity of alcohol.  For example, a study found that >2 drinks/d, compared to <2 drinks/d, was significantly associated with pancreatic necrosis.  So we're all going to die? The problem is that this association was either/or, not calibrated to amount.  For example, what if those who had the necrosis all drank more than 10drinks/d?  It would still be true that the risk was higher at >2 drinks/d.  So why 2/d as the cutoff? "The cutoff of two drinks per day was selected based on animal studies which have shown that the equivalent consumption of two drinks per day in rats results in measurable change in pancreatic histology and physiology(13)."  So, of course, I looked up (13): in rats who received 12%, and worse with 36% of their calories from alcohol increased pancreatic protein hypersecretion, starting the road to pancreatitis.  If you eat 2000 calories a day, then this would be equivalent to a little more than 1 bottle of wine/d.

A Japanese study found that the traditional rates of pancreatitis among alcoholics-- 2-5%-- may be low: they find that  9-17% of people who drank >150g alcohol/d developed alcoholic pancreatitis.  The alcoholic pancreatitis patients began drinking at a younger age (18), drank for 20 years) and drank 180g/d alcohol.  Additionally, they cite other studies where meat and lipid may be co-factors.

An interesting study found the risk of acute pancreatitis may be increased in the first day of withdrawal of drinking; these drinkers had drank an average of 700g/week (400-900g), and 3600/two months.  Alcohol suppresses inflammation, so this may be a rebound inflammatory response.

A Chinese study found that smoking, high meat and heavy drinking was associated with pancreatic cancer.  Heavy drinking was ">20 cup-years;" basically, 11g/d for 20 years, or 22 g/d for 10 years, etc.  The article did not address the hihger rates od ALDH2*2 allele of aldehyde dehydrogenase in the Chinese, which slows the metabolism of aldehyde (and allows it to build up-- see below.)

Another Japanese study (come on) found risk increased 10 fold for >100g/d, and >30 years of drinking.

Alcohol alone is not a risk factor for pancreatic ductal adenocarcimona, which is most closely associated with smoking.  Alcohol may indice pancreatitis and diabetes, which are themselves risk factors.  Also, acetaldehyde, an intermediary metabolite of alcohol which is ordinarily quickly metabolized to acetic acid, is procarcinogenic;  heavy drinkers with cancer, vs. alcoholics without cancer, had higher salivary aldehyde levels due to fast metabolism of alcohol to aldehyde. (I SPECULATE that binge drinking, and frequent exposure to acetaldehyde (read: hangovers) is more dangerous than low but daily drinking.)

Finally, you should know that many studies describing the risks of alcohol are not able to control well for smoking, which is a major risk factor.  Consider that 60% of chronic pancreatitis cases are smokers; but 80% of alcoholic chronic pancreatitis cases are smokers. And high BMI is a risk.

Diet and alcohol: in animal models of alcohol induced pancreatic disease, & calories due to alcohol is the measure.  For example, one mouse model uses 24%, and the mice had BAL 100mM (.46%). Most animal modesl use about 30%.  One study disputed the high protein/high fat risk of pancreatitis by finding that humans with pancreatic or liver disease took 50% of their calories as alcohol, and the worst cases had the highest percentage intake. A study in Mexico found high overall caloric intake (4110 vs. 2250 in healthy controls)  was the risk factor, but dividing the average daily alcohol (124g=868 cal) by calories (4110) gives you 21% calories from alcohol.

The type of alcohol here is not described. Was it red wine?  Vodka?  Beer?  You decide.

So there are two prongs: high caloric intake, especially from fats and protein, and consequent high BMI, along with alcohol (>100g/d, conservatively;) or poor nutritional intake with higher proportion of alcohol calories (>30%).  With both, smoking is a profound risk factor, especially for cancer.

Stop smoking. 



Resveratrol (a type of estrogen (DES)) is a polyphenol contained in wine (and fruits, grapes, etc.,)  that is itself anti-inflammatory and antithrombogenic (it's a COX1-- COX2?-- inhibitor), as well as possibly being neuroprotective. It probably is an anti-flu drug. It can possibly prolong life span through SIRT1 (which is how calorie restriction prolongs life.)    Resveratrol is one possible explanation for why the French can eat fried butter sandwiches with a bottle of wine and still tell their grandkids about it.

There is no accepted dose.  A bottle of red contains about 1mg, unless you're drinking muscadine wine (Florida grapes, some ports, etc.) It appears to have no toxicities.

I bring it up here only to tell you that as much as I think resveratrol is super and all, it oxidizes very quickly after the bottle is opened.  So drink fast.


There are 7 calories/gram alcohol.  So each bottle of wine has about 550 calories.  Each light beer is 110 calories.  There are about 50 calories in a shot of whisky.

Summary And Conclusions: 

Disclaimer: I'm not recommending anything to anyone, I'm not your doctor, results may vary, substantial penalty for early withdrawal (HA!).  Don't drink if you have GI disease.  Or if you drive.  Or if you're on medications.  Or if you're an idiot.  Especially if you're an idiot.

But it appears to me that 30-40g (1/3- 1/2 bottle) of wine alcohol a day is fine.  Enjoy it.  (Unfortunately, I'm a whisky guy.)  It seems to work best if you drink it with food.  Everyone else should just mellow the hell out.  This unprioritized rigidity, this obsession, with "health" and "prevention" is idiotic and counterproductive.  Today I cooked my family bacon and eggs.  BACON.  Take that, AMA's beliefs.

Some caveats:  most of the association studies, above, do their best to control for confounding factors, but sometimes this is impossible.  As a basic generalization, the person who drinks 1/2 bottle wine with dinner is likely to have a very different life than one who drinks 4 beers/day after work, notwithstanding the obvious confounding variable of alcohol with/without food.  So it may be impossible to say that wine, itself, is what is beneficial.

Despite this-- and why this is relevant to a psychiatry blog-- the error is to assume that one is "the type of person who drinks wine, and so would have lower risks" vs. "the type of person who drinks beers, and so would have higher risks."   It may be more accurate to consider that if one chooses to become the person who drinks wine with dinner instead of beer after work, a variety of other factors may also change.  As a simple example: beer at a bar is conducive to smoking, wine at home isn't.  Beer after work every day may be sabotaging your family life; a choice to switch to wine at dinner may improve things at home.  &c., &c.

This is important.  It is the thesis of this blog: nothing matters more than your will.  Even if wine and beer are themselves of no consequence to one's health, the lifestyle that follows with the conscious choice to drink either one is of consequence.  Every choice you make influences your identity, and not the other way around; the sooner you accept this, the sooner you can become the person you want to be. You get to pick who you are.  Go pick. 


(State laws prevent me from receiving donations of wine (or whisky.)  My drink is Balvenie 15 year.  It's about $65. Just saying.)

See also: Is This The Real Secret of Wine's Health Benefits?