I’ve Always Wanted To Know: A Guide to Drinking
Updated: Mar 22
(Note: Nearly all the literature on this topic is correlational, so we should be skeptical about causal explanations. Since that’s what we have we’ll do the best we can. Also, I’m the wrong kind of nerd for the topic, so don’t take this as serious health advice; when in doubt, talk to the other kind of doctor (M.D.) instead.)
There’s a lot of hearsay about drinking alcohol: what is a(n) (un)healthy amount, what are good habits, how to avoid hangovers, and so on. I have been passing along this hearsay my whole life without ever looking into it. So, I looked into it. Here’s what I found.
How much is “a drink”?
Is it unhealthy to drink alcohol?
Is it healthy to drink alcohol?
Ways to avoid getting too drunk?
What causes a hangover?
What about “mixing liquors”?
For the impatient reader, my conclusions are summarized at the end.
How much is “a drink”?
Did you know that there’s a unit of alcoholic beverage? It’s called the Standard Drink Unit (SDU). It’s not perfect, but it takes into account the extent to which (e.g.) having a pint of beer is similar to taking a shot of vodka.
There’s an equation for this. In the US, an SDU is defined as 14 grams of pure alcohol mass (the definition varies by country). For a volume V of a drink with a given ABV (“alcohol by volume”, what is usually called “alcohol percentage”), you can calculate
where D=0.789 g/mL is the mass density of alcohol. The number of SDUs in your drink is then
We can see now that 12 oz of a 5% ABV beer gives roughly 1 SDU: since 12 oz = 355 mL, we have V*(ABV)*D = (355 mL)*(0.05)*(0.789 g/mL) = 14 g. When I use the noun “drink” in what follows, you can mentally substitute 1 SDU if you want to be precise.
Is it unhealthy to drink alcohol?
Everyone knows that they shouldn’t drink excessively, but it’s less straightforward to say what “excessively” means. This is of course not surprising, because all negative health effects show up as increased risk of [X,Y,Z] rather than a guarantee of something bad. (It’s a spectrum.) Still, let’s see what the research suggests.
There seem to be three versions of the unhealthiness of drinking. The first is the risk of very bad acute effects on a particular night of your life. This includes things like drunk driving:
It varies a lot by country, but on average globally, alcohol was a factor in >20% of road fatalities in 2010 (in the US, >30%). Similarly it was a factor in >10% of all serious road injuries (see Figure 3.2). Alcohol is also a factor in sexual assaults, being present in ~50% of such cases. (I’m confused about the statistics on that, as these estimates seem to generalize data of college students to the general public, and also the legal definition of sexual assault in some cases includes drunk sex. This topic is beyond the scope of this post.) For homicides, the alcohol-influence rate is ~40-50%; for violent crimes, it’s ~15-20%. Probably we should conclude that, when you’re drunk, you’re much more likely to get yourself in trouble. This is not a surprise.
A second risk to your health is Alcohol Use Disorder, which is defined in the DSM-5 by its broad effects on your social life and psychological well-being. If you meet 2 of 11 warning criteria, you can be diagnosed with AUD; the criteria include things like
Spent a lot of time drinking? Or being sick or getting over the aftereffects?
Continued to drink even though it was causing trouble with your family or friends?
Given up or cut back on activities that were important or interesting to you, or gave you pleasure, in order to drink?
These effects are bad and are to be avoided. You can take various surveys online to see your risk of AUD, given your pattern of alcohol consumption. For example, 3 of 10 people who drink heavily at least one day a week would be diagnosed with AUD. From this line of argument we find a working definition of “heavy drinking”:
Heavy drinking for Men: More than 4 drinks on any day or 14 per week
Heavy drinking for Women: More than 3 drinks on any day or 7 per week
This is the accepted definition by both the NIH and the CDC. Only 2% of people who never drink heavily (by this definition) would be diagnosed with AUD.
Why care about AUD, aside from the personal and social effects? AUD is comorbid with really bad psychological disorders: schizophrenia (33.7%), major depressive disorder (28%), and personality disorders (50-70%). (High comorbidity means that the two disorders are likely to be simultaneously present.) This is all correlational, of course, since you can’t randomly assign people to abuse alcohol or have schizophrenia, so we can’t say which way the causal arrow might point; maybe psychologically unwell people drink more rather than heavy drinkers becoming psychologically unwell.
The other way drinking can lead to bad health outcomes is epidemiological: you can increase your risk of various major diseases. Much ink has been spilled on this topic and I can’t do justice to it, but most of what I’ve usually heard seems to be true. Heavy drinking (as defined above) leads to increased risk of various cancers (mouth, throat, liver, breast, ++), liver disease, and high blood pressure. Most of these risks disappear if you don’t drink heavily, but there are two important caveats. One is that even moderate drinking, defined by
Moderate drinking for Men: Less than 2 drinks per day
Moderate drinking for Women: Less than 1 drink per day
can increase cancer risk if other negative health factors (bad diet, smoking, etc.) are also present. Further, even a single drink per day may increase the risk of breast cancer in women. (Apparently though this can be mitigated by increased intake of folate? Interesting, but I’m very out of my depth on this.)
Is it healthy to drink alcohol?
Recently there was a highly-publicized study in the Lancet, which argued that even 1 drink per day is dangerous, and that the risks always outweigh any benefits; others aren’t convinced. In fact, there’s some evidence that the best amount of alcohol to drink is not zero. The most robust result seems to be the reduced risk of heart disease in moderate drinkers compared to zero-alcohol drinkers, which has been replicated in several studies with different populations. It doesn’t seem to matter what kind of alcohol you drink, which is a bit surprising if you thought (e.g.) that the real benefit comes from antioxidants in wine or something like that.
Speaking of: there have also been claims about the health benefits of specifically red wine on heart health. Apparently this idea comes from population-wide correlations. Consider the so-called French paradox: France consumes a lot of saturated fat but has unusually low rates of ischemic heart disease; why is that? Could it be due to how much wine they drink? This doesn’t sound that convincing at first look, as there are many differences in diet and lifestyle across cultures besides wine consumption. Given that studies see similar results for all forms of alcohol they study, rather than from wine in general, I’m skeptical. The hypothesis would be testable in experiment, but wine in particular isn’t very well-studied; the American Heart Association says
No direct comparison trials have been done to determine the specific effect of wine or other alcohol on the risk of developing heart disease or stroke.
This review article looks at the issue from many angles, but from what I can see, the benefits we see from wine consumption can be boiled down to (a) whatever benefits we get from moderate drinking in general (as above), plus (b) some unsurprising healthy intake of flavonoids and something called resveratrol, both of which you can get from grape juice.
Of course, there are potentially positive psychological or social effects from moderate drinking. It can be relaxing; it can smooth out awkward social situations; it can help with nervousness or anxiety. Enjoy responsibly.
Ways to avoid getting too drunk?
Folklore seems to hold that two things you do help you ‘come down’ if you are getting too drunk: drink water, or eat something (especially greasy food). Are these suggestions backed up by research?
For water: I couldn’t find any evidence that it will keep you from getting drunk (aside from the fact that drinking water distracts you from drinking more alcohol). There aren’t a lot of great sources on this. I’ll just say that the knowledge seems very common and the lack of evidence for these claims is telling. Water for sure helps with dehydration, which will make your hangover less severe later on (see next section), but it won’t make you sober.
(My personal anecdote is consistent with this: I spent a few months of my life living by the rule that for each alcoholic drink I consumed, I drank 12 oz of water (water first then alcohol). The effect, in my subjective experience, was not a diminution of my feeling of drunkness; instead, my hangover the next day was nearly eliminated, no matter what I drank the night before. It was great, and I only stopped because I had to pee really frequently and it was annoying.)
For food, the literature is pretty interesting. Start with some biology: alcohol is absorbed by both your stomach lining (slowly, and about 20% of the total) and your intestines (quickly, and about 80% of the total). If you speed up your “gastric emptying rate”–the speed at which stuff in your stomach moves to your small intestine–you speed up the overall rate of absorption, making you feel drunker at the peak; if you slow down gastric emptying, the alcohol is absorbed over a longer period and you feel less drunk. The time between finishing your drink and the peak of your Blood Alcohol Concentration (BAC) is apparently anywhere from 15 minutes to 2 hours.
It’s plausible that eating before drinking may slow the absorption of alcohol through your stomach lining, allowing the alcohol to be processed more slowly, making you less drunk. To get quantitative, there’s an app you can use called Virtual Bar to estimate your BAC given what you ate and drank on a given night; it also gives you an estimate of how long it will take you to sober up. Before looking at it, my expectation was that a lower peak BAC would imply a longer tail; if alcohol is processed more slowly, then it seems like it would take longer, given the same total quantity. But according to Virtual Bar, the peak is shifted but the overall length of the drunkness is roughly the same:
It’s a very cool app, but it’s missing some key information. It says it uses “the latest science” but in fact the model is based on a study with 14 participants eating Panda Express, Taco Bell, Pizza Hut, and In-N-Out Burger; let’s just say it may not scale more generally. Also, they explicitly point out that they ignore the effect of gastric emptying, which is critical given the biology above.
The rate of gastric emptying decreases when you eat, which would decrease peak drunkness. According to various sources, the rate is largest for carbohydrate-heavy foods and smallest for fats; this suggests that, if you want to decrease the rate of absorption of alcohol, you should indeed eat fatty (greasy) foods. This must be the source of the folk wisdom.
This all focused on alcohol absorption, from stomach and intestines, but not on the actual process of alcohol processing in the liver. On the latter point, there are studies suggesting that eating speeds up alcohol elimination, the opposite of the effect we see in the above graph; further, in this study there was no measured difference in the alcohol elimination rate between a high-protein, high-fat, or high-carb meal. They hypothesize that this is due to “hepatic blood flow and in the activity of alcohol-metabolizing enzymes”, which, you know, sounds fine to me (??). It’s explained in layman’s terms here:
…equally important is the fact that alcohol elimination rates are inversely proportional to alcohol concentration in the blood. Therefore the suppressed levels of alcohol due to food ingestion cause the body to eliminate the alcohol that is absorbed at a faster rate.
If we believe all of the above results, there are competing effects: eating may slow the absorption of alcohol in your stomach, as well as the rate of gastric emptying; yet it will increase the speed of alcohol elimination once it is absorbed. I admit that I’m unsure about these conclusions.
All of this assumes to eat before or during the time that you drink. Mostly everyone seems to agree that eating after drinking does nothing for you; the alcohol starts being processed very fast, and what you do with your stomach after that point won’t make a difference. This post about the alcohol elimination paper above, which suggests that eating after drinking will sober you up, seems to misunderstand the study.
What causes a hangover?
From the Mayo Clinic:
Alcohol causes your body to produce more urine. In turn, urinating more than usual can lead to dehydration — often indicated by thirst, dizziness and lightheadedness.
Alcohol triggers an inflammatory response from your immune system. Your immune system may trigger certain agents that commonly produce physical symptoms, such as an inability to concentrate, memory problems, decreased appetite and loss of interest in usual activities.
Alcohol irritates the lining of your stomach. Alcohol increases the production of stomach acid and delays stomach emptying. Any of these factors can cause abdominal pain, nausea or vomiting.
Alcohol can cause your blood sugar to fall. If your blood sugar dips too low, you may experience fatigue, weakness, shakiness, mood disturbances and even seizures.
Alcohol causes your blood vessels to expand, which can lead to headaches.
Alcohol can make you sleepy, but it prevents deeper stages of sleep and often causes awakening in the middle of the night. This may leave you groggy and tired.
Alcoholic beverages contain ingredients called congeners, which give many types of alcoholic beverages their flavor and can contribute to hangovers.
So, dehydration, headaches, sleepiness… sounds about right. Importantly, there seems to be no change in the likelihood that a hangover will occur regardless of what type of alcohol you drink.
Although I concluded above that drinking water during a night out won’t sober you up, it will help with hangovers. Since at least the 1930s, studies have shown that alcohol dehydrates you, and one reason is that it inhibits the uptake of an anti-diuretic hormone called vasopressin. This is also, I guess, the source of the myth about “breaking the seal”, the idea that if you pee once while drinking you’ll have to go a lot more after. By the time you have to pee that first time, the alcohol is likely starting to be absorbed and the uptake of vasopressin is being inhibited. From then on, you’re going to pee a lot. You will pee a lot whether or not you hold it that first time.
The Mayo Clinic list says that alcohol causes the blood vessels to expand, but I’d always been told that it causes contraction, which is why alcohol use leads to (e.g.) high blood pressure. Apparently both are true: during an evening of drinking, alcohol causes expansion of blood vessels, improving circulation and increasing your heart rate to ensure sufficient blood flow; later on (or the next morning), your blood vessels contract again, slowing your heart rate and leading to those morning-after headaches that are typical of a hangover.
What about “impurities in the alcohol”, which is what people keep telling me leads to a hangover. It’s why you drink good vodka and not $4 vodka, right? The story is apparently about the last element of the list, the “congeners”–“minor compounds other than ethanol that occur naturally in alcohol beverages as a result of distilling and fermenting processes”–which lead to worse hangovers. Congeners seem to make the severity of the hangover worse. Bourbon, which contains a lot of congeners, is associated with much worse hangovers than vodka, which has a lot less. It is believed that congener density roughly correlates with dark-colored liquor to light-colored liquor, which maps from bad-hangover to not-as-bad hangover.
This result is across different liquors; surprisingly little actual research has been done to make a direct comparison within a particular liquor. Yet, if congeners matter in this way then it also fits the common wisdom that cheap liquor leads to worse hangovers. Expensive liquor often touts being “triple distilled” or whatever, meaning impurities (i.e. congeners) would be filtered out.
What about “mixing”?
Beer before liquor: never been sicker. Liquor before beer: you’re in the clear. Grape or grain, but never the twain. Wine before beer will make you feel queer, but beer before wine and you’ll feel fine.
So liquor before beer and beer before wine, but not both wine and liquor at the same time. (See, I can rhyme too!) But is it true?
I give it a resounding “kind of, maybe.” The strongest evidence in support comes from the ever-scientific, high-quality news outlet Buzzfeed, who interviewed a faculty member at the University of Wisconsin medical school; he said
It’s about the order. When you drink something with lower alcohol first… you get used to drinking and getting drunk at a certain rate. When you switch to something harder, you tend to remain used to the earlier rate of drinking and intoxication — and as a result you take in the booze faster.
The truth of this claim hinges on some of those complicated claims about the effect on alcohol elimination on “hepatic blood flow” and “activity of alcohol-metabolizing enzymes” from before, which I can’t comment intelligently on. Taken at face value, this suggests moving from high-ABV to low-ABV drinks, which roughly means liquor before wine before beer..
Other ideas support the same conclusion: once you start to feel drunk, your inhibitions are harder to suppress, and you’re likely to keep drinking more than you should; at that time, if you’re drinking something strong, you’ll tend to overdo it, so you should instead end the night at low ABV. Once you’re drunk you are also more likely to lose track of how much you’ve already drank. Then once you’ve drank way too much, you’ll end up feeling terrible. Also, if you drink more than one type of alcohol you incur the effects of each; you can drink low-congener light beer all night but those couple of shots of high-congener bourbon will come after you in the morning.
The only semi-chemical argument against mixing drinks seems to be that putting a lot of disparate substances in your stomach at once is likely to upset your stomach and make you feel nauseated. I conclude from all of this:
Liquor before beer’s probably fine, as long as you make sure it’s beer after wine; ABV high to low, that’s the rule to mention; mixing’s no problem if you’re paying attention.
What about mixing alcohol with caffeine? The leading-order effect seems to be, again, that you’re more likely to drink more than you would have otherwise; this is because a stimulant like caffeine will mask the depressant qualities of the alcohol and you’re likely to stay out later or (excitedly) drink more drinks. The effect seems to be stronger when the stimulant is a diet soda rather than a regular one, for unknown reasons.
Physiologically though, it’s more complicated: depressants activate your parasympathetic nervous system, making you feel more relaxed; whereas stimulants activate your sympathetic nervous system, the “fight or flight” response. So the effects don’t “cancel out”; it’s more like pressing down on both the gas and the brake in your car, straining both systems.
I’ve been told many times that taking “uppers” (stimulants, e.g. caffeine) with “downers” (depressants, e.g. alcohol) will damage my heart. But both uppers and downers, taken in large quantity, harm the heart. Is it worse to mix? In a recent survey study in Reuters, alcohol and caffeine were both among the highest risk factors associated with atrial fibrillation, an irregular heart rhythm that’s a leading cause of stroke. However, this specific combination of alcohol+caffeine seems only to imply a risk that is the sum of the risks of the separate substances. (Tangentially, there is evidence that the specific combination alcohol+cocaine leads to production of cocaethylene, which is especially bad. For an opioid+caffeine, there may be an increase in the likelihood of opioid addiction.)
In the last 10 years, the FDA in the US cracked down on beverage producers with alcoholic products containing caffeine (noteably Four Loco was one of these drinks). People usually associate this with, I don’t know, possibly some evidence that these combinations are unsafe. Actually, their stance is merely that the combination has not been shown to be safe, and so it is considered unsafe by default. This is in accordance with the FDA stance, of guilty until proven innocent; in the same way that a new cancer treatment can’t be sold until it’s passed rigorous trials, Four Loco was taken off the market pending further study of the effects. The official FDA statement about the risks:
Based on its review of the published scientific literature, FDA believes that there are insufficient publicly available data and information relating to the safety of alcoholic beverages with added caffeine to form the basis of a consensus among qualified experts that the beverages at issue in these Warning Letters are safe.
The strongest positive statement they make is about “risky behaviors”:
FDA is concerned about the safety of alcoholic beverages with added caffeine because published peer-reviewed studies suggest that the consumption of beverages containing added caffeine and alcohol is associated with risky behaviors that may lead to hazardous and life-threatening situations. Caffeine appears to mask some of the sensory cues an individual might normally rely on to determine his/her level of intoxication. The scientific literature suggests that individuals drinking such beverages may consume more alcohol per drinking occasion, a situation that is particularly dangerous for naïve drinkers.
No statement at all about any particular risk at a biological or chemical level; only an expectation of increased “risky behaviors”. I basically conclude that the stories people tell about biological effects, like alcohol+caffeine harming your heart, are overblown. If there were any indication that they were true, the FDA would have every incentive to point them out.
A Standard Drink Unit corresponds roughly to 12 oz of a 5% ABV beer or 1.5 oz of a 40% ABV liquor. For any drink you can calculate it using the equations in the first section. (It’s fun to keep track!)
There are serious risks from heavy drinking. Seemingly the best way out of all of these risk factors is to stick to non-heavy drinking (less than 4/day and 14/week for men, or less than 3/day and 7/week for women).
If your drinking is non-heavy, you’re unlikely to get drunk and hurt yourself, or to suffer from AUD or negative health outcomes (exception is maybe breast cancer).
This conclusion depends on avoidance of other bad habits like smoking or a bad diet, as this can exacerbate the negative effects of alcohol.
There is some evidence that a small amount of drinking, 1-2 drinks per day, may decrease the risk of heart disease; this effect doesn’t seem to depend on the type of alcohol you drink.
There’s no indication that drinking water makes you less drunk, though it helps with hangovers.
There is some evidence that food helps if you eat it before or while you drink (not after), but from the research I’ve seen the effect is actually not so clear.
Alcohol is absorbed slowly through stomach lining (~20% of the total) and quickly through intestines (~80%). Eating reduces the rate of alcohol absorption and gastric emptying (movement of stomach contents to intestines), which both tend to decrease and delay peak drunkness; the effect is strongest for fatty foods and weakest for carby foods.
But once the alcohol is absorbed, it is processed by the liver and the actual rate of alcohol elimination increases if you eat, a competing effect. Because of this I don’t have a clear picture about whether the net effect of eating is getting more or less drunk.
At leading order, all types of alcohol have similar physical effects: slowing of heart rate, irritation of the stomach, expanding of blood vessels (leading to contraction later on), sleepiness, etc.
Alcohol has diuretic properties, which is why you become dehydrated and why you pee a lot while drinking (though “breaking the seal” isn’t really a thing).
The likelihood of a hangover seems to not depend on what you’re drinking, but the severity may correlate with the density of “congeners” (impurities) in the alcohol. Congener density is, roughly speaking, high for cheap and dark-colored liquors, and low for expensive (pure, filtered) and light-colored liquors.
Mixing different types of liquors, or mixing alcohol with caffeine, seem to have mostly behavioral (rather than adverse biological) effects.
Best to start with high-ABV when you’re in your right mind, and move to only low-ABV once you start to feel the effects. This maybe also tricks your body into processing the alcohol more slowly.
Any physical risks from alcohol+caffeine seem equivalent to the risks of the two separately; I see no evidence for unique combination effects, e.g. additional harm to your heart. Injesting caffeine does hide the depressive effects of alcohol, making you more energetic and more likely to drink more / stay out later than you would otherwise.