Will moderate drinkers live immeasurably?
Alexey Vodovozov, XX2 century
The lecture "To drink or not to drink", which I read as part of the "Skepticon", caused quite a heated discussion. Therefore, I will allow myself to repeat and expand what has been said in places. The topic is very relevant and debatable: are small doses of alcohol harmful or useful, as well as whether complete abstinence from alcohol is useful.
Both points of view have ardent supporters, approximately the same in strenuousness, and their attitude to C 2 H 5 OH is quite utilitarian. The former argue that moderate alcohol consumption is simply necessary for the prevention of a number of diseases, primarily from the heart. A glass of red wine a day, that's all. One of the favorite arguments: ethanol is a natural, normal metabolite of our body, the phrase about "endogenous alcohol" is often thrown around, and about the fact that we have a special enzyme system for a reason, sharpened only on C 2 H 5 OH.
The devil, as usually happens in such cases, lies in the details. Because it is not possible to find in our very rich biochemistry a single reaction, the result of which would be endogenous ethanol. Here intestinal symbionts really know how to drive it – about 3 grams per day from raw materials coming with food. And this is just described. At the same time, there is a rare condition – auto-intoxication syndrome, when saccharomycetes multiply too quickly in the colon and the production of ethyl alcohol is put on industrial rails; however, other options are possible. So from a modern point of view, ethanol is an exogenous substance, since it is produced in the external environment of the body, that is, in the lumen of the intestine.
And then why do we need alcohol dehydrogenases (ADH) and aldehyde dehydrogenases (ALDH) inside the cells? Firstly, to protect cells from ethanol, and secondly, to kill two birds with one stone, simultaneously "burning" (oxidizing) it in the mitochondria oven. When were calories superfluous? The opinion of the hosts and especially the hostesses is not taken into account by the body, it has its own reasons. In addition, there are other alcohols in us. For example, vitamin A in the form of retinol is very much a client of the ADH-ALDH pair. And other substrates are also processed there – products of lipid peroxidation, hydroxysteroids, that is, everything that really turns out to be in the internal environment of the body and is subject to oxidation.
That is, both sides are wrong on this issue. No, it is not our natural metabolite, as proponents of alkoprophylaxis of cardiovascular and other diseases claim, it is a toxic product supplied by symbiotic flora. She also knows how to distill the pectin of apples or pears into methanol, by the way. And no, absolute sobriety (sorry, Sobriety, from the second side of the barricades, this term is usually written with a capital letter) is impossible, every person "drinks" about 3 g of ethanol per day (sorry again, "alcoholic poison", we use the correct terminology of the sect of witnesses of "farting bacteria"), which is forcibly shoved into him by insidious microorganisms.
Now we turn to the doses. Moderate consumption is how much in grams? There is no definite answer to this question. Moreover, for quite a long time, studies have indicated how many glasses a particular cohort participant drinks. Then it turned out that the volume of the glass varies greatly in the understanding of different people. Someone has 150 ml, someone has 450. It took some kind of universal unit to determine the boundaries of consumption, which is extremely undesirable to cross. This is how the drink (or unit – in some countries) appeared. The range of values from 8 g (UK) to 20 g (Austria) depends on local characteristics, including traditional drinks, volumes and the system of measures.
For example, in the USA, where everything is tied to liquid ounces, a drink is equal to 14 g or 18 ml. Grams in milliliters are translated simply: divide them by 0.79 (ethanol density). If you need to convert milliliters into grams, on the contrary, multiply by 0.79, in general, pure arithmetic. One drink is contained in a 12-ounce can of beer with 5 volume percent ethanol, a 5-ounce glass of wine with 12 volume percent ethanol and 1.5 ounces (1 shot) of something whiskey-like.
In Russia, some documents flash a certain "standard dose" equal to 10 g of ethanol – for example, France, Germany, Italy, the Netherlands, Spain and Poland. A lot of noise was made in June 2017 by the news published by Izvestia that the Ministry of Health allegedly calculated safe and sometimes even recommended alcohol consumption rates. Check out the numbers:
The Ministry of Health then had to justify itself for a long time, including because the World Health Organization, which was also referred to in the article, adheres to the strategy "the less the better", that is, there are no "recommended" or "safe" doses, in the understanding of WHO, does not exist. But "heavy episodic drunkenness" there is – this is at least 60 g of ethanol (that is, 6 Russian standard doses) once, at one sitting during the last month.
Moderate doses or small – this is already determined in each specific study. On average, it turns out that the first ones are those who drink 3-4 drinks a day, although I can't imagine how you can call a moderate bottle of wine every day, and the second ones are 1-2 drinks a day. However, the most problematic group in "alcoholic" studies are non–drinkers, although, it would seem, what difficulties can there be with them.
But about everything in order. So, when supporters of "drinking for health" talk about the usefulness of moderate alcohol, they often refer to U-shaped graphs of the relationship between mortality and drinking. Something like this:
It is implied that in the area of small and moderate doses, a certain protective effect is observed, that is, the same glass of wine a day reduces the risk of dying from something, for example, from a myocardial infarction. But professional drinkers, as well as professional non-drinkers, are not lucky about the same. Sometimes there are less radical J-shaped charts, where non-drinkers are doing a little better, but still worse than moderate drinkers.
But if we conduct a competent meta-analysis of such studies, one very interesting thing will turn out. As an example, we will analyze a relatively recent systematic review of 2016 from a group of authors who specialize in alcohol meta-analyses. This is important, because if you homogenized mice yesterday, delved into T-reg receptors today, and tomorrow took up analyzing alcohol studies, then there is a risk of messing up in an unfamiliar topic. Tim Stockwell, Jinhui Zhao and Tanya Chikritzhs are experienced people, they have more than one systematic review on their conscience in exactly the area that interests us.
The authors were interested in the question: does moderate alcohol consumption reduce mortality rates. In the course of screening out of 2662 studies found in PubMed and Web of Science, only 87 remained. In the vast majority of cases, either the outcomes of diseases (including death) were not indicated, or the doses of alcohol consumed were not established. Also, letters to the editor, comments, case-control studies, in general, all garbage were thrown out of the review.
As a result, the sample turned out to be impressive – 3998626 people, among whom 367103 deaths were registered. And yes, indeed, after plotting the graph, the same J-shaped curve turned out.
Everything seems to be clear, as in many similar studies: abstinents, that is, a group of non-drinkers, had no additional risk of mortality, which is logical. In heavy drinkers (5-6 drinks a day) and abusers, it turned out to be increased, but for casual, few and moderate drinkers, the same protective effect of alcohol could be observed, they clearly had a reduced risk. It would seem that you can open champagne and go celebrate. But, as I have already said, the group of authors was experienced, so they began to analyze in detail who the groups of non-drinkers consist of. And we came across the now famous "teetotaler's mistake".
This term means the inclusion of "wrong" people in the group of abstinents. For example, alcoholics who have quit. If alcohol consumption over the past 1-5 years was taken into account in the cohort, and people in the more distant past played litrball professionally, then their health condition will leave the best, they will greatly spoil the mortality of the non-drinking group. Another option: a person does not drink only because a certain chronic disease does not allow him. And by chance, few and moderate drinkers are healthy, so they can afford it.
This is a known, but still unresolved problem. In general, alcohol is difficult: interventional studies in this area will not be missed by any ethics committee. Well, imagine, because it will take months, or even years, to specially give people certain doses of alcoholic beverages, preferably the same. What if someone gets drunk? This possibility cannot be excluded with regular intake of alcohol. And if we are talking about pregnant or nursing at all? So we have to be content with ready-made cohorts and quasi-experiments. To avoid mistakes, the old cohorts will have to be recruited anew, develop new questionnaires, correct existing ones, in general, it will take time and money, and neither one nor the other is superfluous for researchers. So we have to be content with what we have.
If we analyze the 87 studies included in the review more carefully, a terrible thing turns out: from 1980 to 2014, there was no teetotaler error in only 13 cases. In 41, abstinents were both those who had tied up and accidentally drinking, in 24 – those who had tied up, in 9 – only accidentally drinking.
If we look for a link between low alcohol consumption and mortality, it will turn out to be highly dependent on the level of normalization, that is, on the number of variables that could distort the results and which we took into account.
If you remove only covariates, the protective effect of small doses is very clearly visible. On a blobogram (the word is what ... forest plot or "herringbone" is much nicer), the confidence interval (a segment with serifs) does not cross the line of absence of differences and is located to the left of it. So, yes, to reduce the risk of mortality from common causes, you need to drink. A little bit, but to drink. Normalize the smoking groups. The effect is less pronounced, but still there. But from the moment of normalization by race error (some representatives of the Mongoloid race tolerate alcohol much worse than Caucasians), teetotaler error, emissions (falling out, extremely large and extremely small results), and then in general by all available parameters, the effect simply evaporates. "What we took for an orgasm for 20 years turned out to be asthma."
And now – terrible news for the advocates of Sobriety. If we begin to evaluate the quality of research and eventually select only those in which abstinents are only people who refuse alcohol for ideological reasons, we will not be able to statistically distinguish groups of low drinkers from non–drinkers. In terms of mortality, they will be the same. It is clearly seen that there are no differences between these groups in studies without teetotaler error (13 pcs.) and in studies without methodological errors at all (only 6 out of 2662).
The adjusted chart will remain J-shaped, but the line of absence of risk will shift down. As a result, everything will fall into place:
Moderate drinkers have equaled those who have quit, and casual drinkers – with few drinkers, there are still true abstinents trampling next to them somewhere else. And no one has a reduced risk. The authors consider the slight decrease observed in casual drinkers to be an artifact, most likely, rare alcohol consumption is just a marker of their lifestyle, which affects this indicator much more strongly, but for some reason is not taken into account.
The most valuable thing is that the authors of the meta-analysis did not just knock the ground out from under the feet of the two opposing sides, showing that there is no scientific basis for either "horror stories" from the adherents of Sobriety, or under the same glass of wine a day, but also formulated a couple of recommendations for the future. The first is that in "alcoholic" studies, it is necessary to separate those who are tied and accidentally drinking from ideological abstinents, otherwise we will again receive beautiful, but absolutely incorrect graphs. The second is that it is better to use casual drinkers as a control group, rather than abstinents for greater correctness of research. It is possible that over time there will be several dozen new studies, the authors of which will take into account the mistakes of their predecessors, and then we will see either confirmation of the correctness of Tim Stockwell's associates, or one of two.
So what's the end result? A fairly simple and logical conclusion: one should be equally skeptical of research on both the unambiguous benefits of alcohol and the equally unambiguous harm of even its smallest doses. To drink or not to drink is a personal choice of everyone, and it is highly desirable that this choice be responsible. And outrageous lies – both about dead neurons that fall into the toilet with urine by the thousands after taking any amount of alcohol, and about the protective effect of any ethanol-containing beverages – are irresponsible.
On the WHO Eurobureau website, you can take an AUDIT test to get an approximate idea of the degree of professionalism of your relationship with alcohol ;).
Portal "Eternal youth" http://vechnayamolodost.ru