I’m asking this in good faith (in case that doesn’t come through in text!), but how large was the study, and how diverse was the population? And how (if at all) does this account for genetics? I come from a long line of alcohol abusers who lived well in to their 90’s, along with everyone else on the family. I’ve read a lot about how the consensus now is that any drinking= dangerous, but these anecdotal non-examples from my life cause me to wonder. (For reference, these family members are white/wealthy, do these studies account for other risk factors like race and poverty?)
Thanks Darcy, that’s a completely fair and thoughtful question (and tone came across perfectly!).
The 2015 Stockwell meta-analysis I referenced pulled data from 87 separate studies, covering nearly 4 million participants across multiple countries. But like many large epidemiological studies, the population was skewed: largely Western, white, and middle-to-upper income. So yes, there are real limitations when it comes to diversity, both in terms of genetics and socioeconomic background.
As for genetics: you're right to raise it. We know genes influence alcohol metabolism (e.g., variants in ALDH2 common in East Asian populations can dramatically change risk profiles), but most of these large cohort studies don’t stratify by genotype. That means people like your family, genetically or culturally more resilient to alcohol's harms, can get washed out in the averages. At a population level, alcohol raises risk. But your personal risk? That could very well differ.
The other big piece is what researchers call "residual confounding." Your relatives were white, wealthy, and part of a long-lived family line - all factors that tend to predict longer lifespans, regardless of alcohol use. People with higher income and education levels tend to live longer, regardless of habits like drinking. So their lifespan may reflect their circumstances more than their alcohol intake.
But of course, risk isn’t destiny. If I had a dollar for every patient who told me, when I gently suggested quitting smoking, “But my uncle smoked two packs a day and lived to 90!”, I’d have enough for a very nice weekend away.
Outliers don’t invalidate the pattern; they just remind us that risk is personal, and our choices stack on top of whatever baseline we’re dealt.
Thanks again for reading and engaging so thoughtfully. I really appreciate it.
Excellent Article. Thank You for submitting and putting the facts out there. The oncology professionals have been very much in agreement with your research and conclusions. They put alcohol in the same group as tobacco and asbestos in terms of carcinogens. Pretty eye opening and scary. Thank You. Keep up the great work you do....
Thanks, Ron. I'm not sure the message has got out there. I think most people understand that very heavy drinking is associated with cirrhosis, but think that their level of 'moderate' drinking is far removed from any health risk.
Isn't it odd to think we have a multi-billion dollar industry in which we go out and willingly drink a carcinogen!
Yes it is. When I asked an oncologist why they haven't declared war on it, he responded, " we are still trying to get people to kick the smoking habit and get their weight under control. Taking away their booze is going to be a bigger nut to crack for the reasons you mentioned in your response. Thank You
Great column! I am a big fan of nuts and I stopped drinking alcohol several years ago based on advice from a doctor I trusted. I don’t miss it at all because I value my health. For my friends 55+ who are still drinking moderate amounts, does stopping now improve their odds or have decades of consumption locked them into a higher risk category?
You’re right. When you weigh up the pleasure of a drink vs potentially extra years of life, it shouldn’t bee too hard of a decision.
As for the effects of quitting alcohol later in life after years of drinking, It’s likely there will always be health benefits from cutting back or stopping. If we take a worst case scenario, someone who has drunk so heavily that they already have early liver cirrhosis, even they are likely to have markedly better health outcomes if they stop drinking.
Thanks, Mick. I love your comprehensive, detailed review. That was a lot of work.
As you suggest, I don’t think the message that there’s no safe level of alcohol consumption has really got out there. I suspect if we did a poll, most people would still think that drinking a little is good for us.
As I just said to someone else, it’s amazing to think we have a multi-billion dollar industry helping us voluntarily drink a proven carcinogen. The power of the dollar!
Ummm, no. This post should, instead, be titled: “What Ben Jones Got Wrong About Alcohol.”
Where in this article is the proof that “Decades of Alcohol Research [Are] Wrong”?
Extraordinary claims require extraordinary evidence. Where are the datasets that indicate this?
Where are any cites? Your only cite is Tim Stockwell, a Neo-Prohibitionist crusader who has long cherry-picked his data and been repeatedly discredited. See, for example, PMID 27588543:
Wow! That is quite an extraordinary rebuke for a professional journal!
His “sick-quitter” claim is easily disproven if you actually look at the data.
I challenge Dr. Jones to provide us with a couple of datasets where the typical 30% coronary heart disease event reductions seen in moderate alcohol consumers compared to non-drinkers disappears—or turns to statistically-significant harm—when prior drinkers (“sick-quitters”) are excluded from the dataset.
Heck, Dr. Jones, just provide us with a single such dataset! We’ll wait.
Many alcohol studies, in fact, have specifically segregated out former drinkers, and so, any “sick-quitters,” in their analyses. See, for example, the newer PMID 28331015, with almost 2,000,000 Brits followed for 6 years.
See its Figure 2. There the “non-drinkers” had 32% more heart attacks—and the “former drinkers” had 31% more—compared to the “moderate drinkers.” And with respect to cancer and other possible alcohol-related ailments, the non-drinkers had 24% higher All-Cause mortality, and the former drinkers 38% more, compared to the moderate drinkers. Care to explain, Dr. Jones? Do you want you readers to quit drinking and forgo moderate drinking’s CVD benefits—and have at least 24% higher net mortality? That’s from almost 2,000,000 people—excluding all “quitters."
Or see the study in Ann Epidemiol 2007;17:S3–S7 at:
It explains: “To address the issue of residual confounding by healthy lifestyle in drinkers [and the inclusion of any “sick-quitters” in the non-drinkers category], in a large prospective study we restricted analysis to only ‘‘healthy’’ men (who did not smoke, exercised, ate a good diet, and were not obese). Within this group, men who drank moderately had a relative risk for Coronary Heart Disease of 0.38 compared with [healthy] abstainers, providing further evidence to support the hypothesis that the inverse association of alcohol to CHD is causal, and not confounded by healthy lifestyle behaviors.”
Wow. That’s a coronary heart disease reduction of over 60%—when zero “sick-quitters” were included.
Or, put another way, among strictly-healthy men, those who forewent alcohol had 160% more CHD events than those who drank (i.e.1/0.38). Wow.
Care to explain, Dr. Jones?
For another study that explicitly separated out “former drinkers,” see PMID 21098615 at:
For the combined populations, the lifetime non-drinkers had 100% more coronary events compared to the drinkers who didn’t binge. Wow. The former-drinkers, which would include any “sick-quitters,” “only” had 57% more coronary events than the drinkers. So instead of pulling down the results of a united “non-drinker” category, it looks like the inclusion of former drinkers would have pulled them up.
See another dataset at PMID 15167206, “Changes in alcohol intake and risk of coronary heart disease and all-cause mortality in the MONICA/KORA-Augsburg cohort 1987-97” where: “The lowest all-cause mortality risk was observed among men drinking 20-40 g [of alcohol] per day (HRR 0.48).”
I.e. 50% lower all-cause mortality, including cancers, liver issues, and accidents, compared to lifelong non-drinkers. That’s with former drinkers excluded, so no “sick quitters.”
Wow, 100% higher All-Cause mortality in the teetotalers with no “sick quitters”! See it’s Tables 5 & 6, available free through ResearchGate if your institution doesn’t have access to it.
“We adopted several approaches to help limit potential bias involved in the study of never-drinkers (who might differ systematically from drinkers in ways that are difficult to measure, but which might be relevant to disease causation) and of ex-drinkers (a category that includes people who might have abstained from alcohol owing to poor health itself, as well as those who changed their habits to achieve a healthier lifestyle).”
Their results? Compared to infrequent drinkers (i.e drinkers, but at only 0.1-5 g/day of average alcohol, HR=1.00=reference), former-drinkers and never-drinkers had 24% & 17% more non-fatal coronary heart disease events, respectively. Those drinking 30-60 & >60 g/day alcohol, on the other hand, had 26% & 25% less CHD. See its Table 3. That means that the non-drinkers-excluding-any-sick-quitters had 58% more CHD events than those averaging 2+ drinks per day (i.e. 1.17/0.74). Wow.
For one on diabetes, see Table 2 of PMID 28748324 at:
They too separated out “current abstainers” (i.e. including any “sick-quitters”) from “lifetime abstainers.”
Using infrequent drinkers (<1 drink per week) as the Hazard Rate=1.00 reference, there was no statistically-significant benefit in future diabetes rates for Current Abstainers (i.e. any sick-quitters). But women who were Lifetime Abstainers had a 25% higher future diabetes rate. Men and women who drank 3 or 4 times per week, on the other hand, had 27% and 32% less future diabetes. Wow. That means that the non-drinkers-excluding-any-sick-quitters had about 80% more diabetes development than those drinking 3 or 4 times a week (i.e. 1.25/0.70). Wow, so high diabetes protection too.
See also PMID 20876712 and PMID 31568479.
So separating out former-drinkers from lifetime-teetotalers in these analyses doesn’t make modest-drinking’s health benefits disappear—if anything it amplifies them.
Now let’s see your datasets, Dr. Jones, indicating that all the huge CVD benefits from moderate drinking that we keep witnessing are due to including “sick-quitters” in with the non-drinkers, as you claim. We’ll wait.
Anyone telling you this is playing you for a lazy fool. It may be what you want to believe, but it’s simply spreading false medical information that’s going to hurt people. If you can’t win with scientific datasets, then you try to muddy up the water spreading “FUD,” fear, uncertainty, and doubt, I guess.
Please, Dr. Jones, examine all of the many datasets referenced above and then write an up-dated piece, or at least delete this one. You don’t want to kill your readers, do you? Your future credibility is on the line here.
Sid, I’ve seen your comments and appreciate that you’ve taken the time to share your perspective. You’ve raised a lot of points and studies, and I want to make sure I take the time to consider them properly and respond with the depth they deserve. I’ll post a more detailed reply once I’ve had a chance to look at them.
Considerations about alcohol and health are not restricted to cardiovascular disease.
Bagnardi et al. showed that alcohol consumption is linked in a dose-dependent manner to a higher risk of cancers of the oral cavity, pharynx, oesophagus, larynx, stomach, colon, rectum, liver, female breast, and ovaries, even at typical moderate levels of drinking. They concluded: 'The analysis did not identify a threshold level of alcohol consumption below which no increased risk for cancer was evident.' - In other words, any amount of alcohol increases cancer risk.
Bagnardi followed up in 2013, specifically looking at light drinking. Their conclusion: "Alcohol increases the risk of cancer of the oral cavity and pharynx, oesophagus and breast even at low doses. Given the high proportion of light drinkers in the population and the high prevalence of these tumours, especially of breast cancer, even small increases in cancer risk are of great public health relevance."
Other 'J-shaped curve' associations with alcohol are also being challenged.
It was previously believed that moderate drinking protects against dementia. However, an observational cohort study of 550 adults over 30 years found a "dose-dependent association between alcohol consumption over 30 years of follow-up and hippocampal atrophy, as well as impaired white matter microstructure... There was no evidence that light drinking offers protection over abstinence in terms of brain structure or function." They concluded, "The finding that alcohol consumption in moderate amounts is linked to multiple markers of abnormal brain structure and cognitive function has significant potential public health implications for a large sector of the population. For example, in our sample, nearly half of the men and a quarter of the women were currently drinking in this range."
Of course, the ideal way to answer this question of whether alcohol protects against coronary heart disease is with a large prospective randomised controlled trial where participants are randomly assigned to low-level or no alcohol and then followed for many years to see how their rates of coronary heart disease differ. The MACH15 study was planned to do just this, but its funding was terminated.
Will a similar study be conducted? In the words of Carr et al. "Due to the growing number of studies suggesting increased disease risk, including cancer associated with alcohol use even at very low levels, the use of RCTs to study alcohol consumption is ethically questionable." - In other words, the risks of alcohol, even at low levels, are considered too high for it to be acceptable to use it for a clinical trial.
It’s worth repeating: even if light drinking did protect against coronary heart disease, the broader harms of alcohol are so well established that researchers no longer consider it ethical to test in a clinical trial.
And even if that protection were real, no credible clinician would advise drinking to prevent disease, which makes the entire argument moot.
You’ve accused me of misleading readers, of “ignoring the science,” even of “killing my readers.” Not only is that scientifically indefensible, it’s offensive.
In the article, I stated that alcohol carries health risks and should be approached with caution, not mistaken for a health tonic. That statement is irrefutably grounded in the scientific consensus. I stand by it.
If you’re genuinely concerned about public health, maybe start by modelling respectful dialogue. You're welcome here if you're ready to argue in good faith and back your claims with evidence. However, if you're looking for rhetorical gladiatorial combat, I'm not interested. If that's the case, you may feel more at home on Twitter/X.
To overcome the challenges of studies that depend on participants accurately reporting their alcohol consumption and the potential confounding effects of factors such as socioeconomic status and lifestyle, researchers have turned to Mendelian randomisation (MR) studies.
These studies divide participants into groups based on certain genes associated with drinking. For example, East Asians with certain ALDH2 gene variants have an unpleasant reaction to alcohol, so they drink less. You can then compare this group to other East Asians without the gene, who, on average, drink more.
Because we inherit these genes from our parents at conception, they are set long before we choose whether to smoke or go to college, and whether we live a comfortable middle-class life or in poverty, all of which are contributing factors to our health later in life. Researchers use different genes in different populations, and the analyses are restricted to single ethnic groups to avoid other confounding factors.
As a result, MR studies remove complications such as whether or not people are poor or get confused about how much they drink. They also have an advantage over observational studies in that they can be used to infer whether an exposure causes the outcome being studied rather than just being associated with it.
In one of the earliest, Holmes et al. found in a study of almost 262,000 participants that "individuals of European descent with a genetic predisposition to consume less alcohol had a reduced risk of coronary heart disease and ischaemic stroke, and lower levels of several established and emerging risk factors for cardiovascular disease. These findings suggest that reductions of alcohol consumption, even for light to moderate drinkers, may be beneficial for cardiovascular health. Our results therefore challenge the concept of a cardioprotective effect associated with light to moderate alcohol consumption reported in observational studies and suggest that this effect may have been due to residual confounding or selection bias."
More recently, Kember et al. (2024) compared a regular observational study and an MR study on the same group of almost 200,000 US veterans. The observational study again showed the U-shaped curve, but the MR analysis showed no link between alcohol use and coronary heart disease.
Their conclusion: "We conclude from these findings that the observational associations showing a U-shaped curve for the association of drinking level with CHD and T2D reflect confounding attributable to a variety of unmeasured factors. These include inaccurate or invalid reporting of alcohol consumption and lifestyle, genetic, and socioeconomic factors associated with moderate alcohol consumption." - In other words, the U-shape relationship has nothing to do with alcohol.
Also in 2024, Carr et al. performed a landmark in-depth analysis of 127 observational, case-control, and MR studies on alcohol and ischaemic heart disease using advanced Bayesian burden of proof techniques.
For the observational and case-control studies, the familiar J-curve was apparent; however, in many analyses, it lost statistical significance when study heterogeneity (differences in study methods) was taken into account.
That said, these studies did show protection with 'average' consumption levels. Pooling case-control and cohort data decreased the risk of ischaemic heart disease by around 4% which, according to the authors, "suggests that the association – on the basis of the available evidence – is weak." - Far from the 60% protection you suggested.
When they examined data from MR studies, they found no evidence of protection, although they requested MR studies using more genes and techniques better suited to detecting curved relationships.
Already in press when this was written was an MR study that did exactly that.
Kassaw looked at almost 280,000 white British adults. In a standard analysis, the J-shaped curve was apparent, though they found "the alcohol–mortality relationship differs between younger and older individuals and is strongly affected by adjustment to potential confounders and indicators of poor health, supporting the role of selection bias and reverse causality."
When they conducted a detailed MR analysis designed to detect non-linear (curved) relationships with 94 gene variants, they found "MR analyses did not provide any evidence to support the J-shaped association between alcohol intake and mortality suggested in the conventional observational analysis." They suggest, "...a reduction in the amount of alcohol consumed is likely to be beneficial regardless of the level of current intake."
Their final conclusion, "Genetic evidence strongly suggests that as alcohol consumption increases, there is a linear increase in the risk of premature death, including from specific causes such as CVD, cancer, and digestive illness, with no evidence for any protection by modest intakes... A re-evaluation of current public policies regarding drinking guidelines may be warranted."
Far from being discredited as you suggest, Stockwell's analysis is supported by study after study.
I believe there is evidence that for some people, notably older adults with or at risk of coronary heart disease specifically (not other forms of cardiovascular disease), light alcohol consumption may reduce their risk of cardiovascular disease and increase life expectancy. However, this is a decision to be made after taking into account all aspects of their health and lifestyle, i.e. after a conversation with their doctor, not advice to be given in an online article.
That said, I cannot find a single clinician advocating light drinking to prevent or manage cardiovascular disease, even those who accept the J-curve hypothesis. Day and Rudd (2019) are typical in their conclusions: "Observational and prospective studies have consistently shown a lower risk of cardiovascular and all-cause mortality in people with low levels of alcohol consumption when compared to abstainers (the ‘J’-shaped curve)... However, this evidence is contested, and overall, the detrimental effects of alcohol far outweigh the beneficial effects."
This advice has been the same for decades, regardless of the arguments swirling around. Marmot & Brunner back in 1971, whilst agreeing that moderate drinking may be protective against cardiovascular disease, concluded: "In conclusion, the balance of harm and benefit does not weigh in favour of making a recommendation to the public to drink in order to prevent coronary heart disease."
Sid, thanks for your comments. Let me be clear. This isn’t a place for performative condescension.
If you want to debate the science, I’m all for it. But the sarcastic jabs, the “we’ll wait” taunts, the smug one-liners, that’s not how serious people debate science. Your tone undermines your credibility far more than any dataset could.
One of the reasons people come to Substack is to get away from that kind of noise. If you have a case, make it with evidence and respect.
Now, to the substance.
Much of the evidence you provide is on biomarkers. These are helpful for understanding potential mechanisms, and some are used as surrogate endpoints in short clinical trials when you can't wait for long-term health outcomes. However, hard clinical end points - heart attacks, strokes, deaths - are what really matter, and we have plenty of data on these, so that's where I'll focus.
You dismiss Tim Stockwell, whose 2015 data I used for the graph in my article, as "a Neo-prohibitionist crusader who has long cherry-picked his data and been repeatedly discredited."
If your knowledge of the topic is as extensive as your tsunami of citations suggests, you'll know this is far from the truth. Stockwell was not the first to suggest that the J-shaped curve may be an artefact of study design. This debate has been ongoing for decades.
For example (I quote directly from the authors):
Wannamethee & Sharper - Neuroepidemiology 1998
"Non-drinkers, both ex-drinkers and lifelong teetotallers, have an increased prevalence of conditions likely to increase morbidity and mortality compared with occasional or light drinkers. In addition, regular light drinkers tend to have characteristics extremely advantageous to health [he's talking about them being largely well-educated, middle-class and white]. Whilst there is considerable evidence that alcohol intake at some level has a beneficial effect on CHD, the degree of protection claimed is almost certainly exaggerated by comparison with an inappropriate control group and by the limited adjustment procedures used to take into account the differing characteristics of the various alcohol intake groups."
Light drinking is a hallmark of better-educated, wealthy, healthy, often white populations who, overall, have better health outcomes than the rest of the population. Therefore, it's argued, light drinking is often a marker of being in a group with good health outcomes, rather than the cause of it. Go outside that group, and the J-shaped curve often doesn't hold up.
Sempos et al. - Alcohol Clinical & Experimental Research 2003
As an example, Sempos looked at the relationship between alcohol and mortality for African Americans. "No J-shaped curve was found... no beneficial effect appeared and mortality increased with increasing average consumption for more than one drink a day."
Andréasson - Alcohol Clinical & Experimental Research 2006
"Many studies report J- or U-shaped curves in describing the association between level of alcohol consumption and mortality. J-shaped curves may, however, be the result of complex associations between psychosocial and other confounding factors and health, rather than a simple causal association derived from the biological effects of alcohol."
There are others, but you get the gist. Stockwell did a large analysis to see if suspicions that the so-called J-shaped curve suggesting moderate alcohol protects against cardiovascular disease were correct. His results supported the conclusion that non-drinkers are a mixed bag of teetotalers and 'sick quitters' and that the J-shaped curve is an artefact of study design.
Has he been discredited? Far from it. Yes, you found a 'letter to the editor' shortly afterwards criticising the findings, but he has subsequently been extensively cited and has had a significant impact on the direction of research in the field.
Just one of many examples: "Therefore, although there is a strong body of evidence that shows a J-shaped association between alcohol intake and mortality, it seems likely that the link between low–moderate intake and improved survival is not causal, and there is no ‘safe’ threshold intake below which drinking gives benefit." - DJ Stott - Age and Ageing 2020.
Stockwell is not alone with this finding. Others have found the same thing, including Knott, who, using data from the Health Survey for England, found that when never drinkers were used as the comparator, a J-curve appeared - but only in men aged 50–64 and women over 65.
To test whether this pattern held up, the researchers repeated their analysis using five increasingly strict definitions of “occasional drinkers” as the reference group. The rationale was simple: if the J-curve reflects a genuine protective effect of moderate drinking, it should appear no matter who you compare it to. But it didn’t.
As the authors note, “Supplementary analyses found that most protective effects disappeared where calculated in comparison with various definitions of occasional drinkers.” - How you define the comparator group makes all the difference to the results.
While meta-analyses subsequent to the Stockwell paper continue to show a J-shaped curve, authors increasingly acknowledge that study design limitations may drive these findings. For example, in the BMJ 2021 paper by Ding et al., the authors state: "there is evidence that reductions in risk of all-cause mortality and subsequent events might have been overestimated due to the inclusion of former drinkers in the non-drinking reference group.... we cannot definitely determine the extent of this overestimation with very few studies that explicitly excluded former drinkers."
These analyses are complicated by the fact that cardiovascular disease is not a single entity; it's a constellation of conditions.
A Lancet analysis of nearly 600,000 individuals found that the relationship between alcohol and cardiovascular disease varies by subtype. While the study confirmed a J-shaped curve for myocardial infarction (heart attack), it reported a clear dose-dependent increase in risk for stroke, heart failure, and deaths from other cardiovascular causes. The authors concluded: “For cardiovascular disease subtypes other than myocardial infarction, there were no clear risk thresholds below which lower alcohol consumption stopped being associated with reduced disease risk. These data support limits for alcohol consumption that are lower than those recommended in most current guidelines.” - That's a confusing way of saying the less you drink, the lower the risk.
A cohort study of almost 108,000 individuals followed for 14 years showed a nonlinear association between alcohol consumption and atrial fibrillation. There was no protection at low levels of consumption and a 16% increased risk (HR 1.16) at 12g alcohol (1 drink) a day.
A meta-analysis including data on more than 360,000 individuals found that even low levels of drinking were associated with a 19% increased risk of hypertension in men (RR 1.19). In women, an increased risk wasn't seen until consumption reached 3-4 drinks per day, but no protection was seen in men or women.
In addition, the effect of alcohol varies by age, gender and geography.
A Lancet report from the Global Burden of Disease Study in 2022 shows that the level of drinking associated with the lowest health loss varies by age, gender and region. The authors say, "We provide clear evidence that the level of alcohol consumption that minimises health loss varies significantly across populations and remains zero or very close to zero for several population groups, particularly young adults." For some older adults, there appears to be a risk reduction associated with moderate alcohol consumption (0.1-1.9 standard drinks per day), likely in those adults at higher risk of MI. They suggest recommendations should vary by age and location, and that guidelines for younger adults should be reduced.
Ben, thank you for quickly responding back at length. But you largely ignored my points, instead going off in new directions (to which I will later respond as well).
1. But first let’s return to what you originally posted and what my comments actually were.
Your whole thesis on “What Decades of Alcohol Research Got Wrong” dealt with the classification of supposedly “sick-quitters.”
You wrote: “By lumping [former drinkers (aka the “sick quitters”)] in with lifelong abstainers, researchers unintentionally stacked the non-drinker group with people who already had higher health risks. That distorted the results, making light drinkers look healthier than they really were. … When researchers re-analysed the data and looked only at lifelong abstainers, the story changed dramatically. Once you remove high-risk former drinkers, the J-curve disappears. The more you drink, the greater your risk of premature death. The message is brutally clear: the more alcohol you drink, the higher your risk of dying early. There’s no U-shape. No protective effect from drinking lightly.”
I responded back that this was false. That when we exclude former drinkers from the non-drinkers category, instead, we get the same results. The benefits of moderate drinking don’t “disappear.” The story does not “change dramatically.” The “U-shape” remains. That it’s “brutally clear” that there is still a protective effect from drinking lightly. I provided 6 separate datasets to demonstrate this. Since you are British, let’s review the first one, with almost 2,000,000 of your fellow countrymen, and perhaps some of your patients are in it.
I asked you to explain the results in its Figure 2. You haven’t yet. That’s what I was hoping for in a return post.
There the “lifetime non-drinkers” had 32% more heart attacks—and the “former drinkers” had 31% more—compared to the “moderate drinkers.” Same amount. And with respect to cancer and other possible alcohol-related ailments, the non-drinkers—with any and all “sick-quitters” excluded— had 24% higher all-cause mortality compared to the moderate drinkers. And I provided 5 more similar datasets explicitly indicating the same thing—that excluding sick-quitters does not affect the U-shaped results.
I challenged you to provide a couple of datasets where the typical 30% coronary heart disease event reductions generally seen by moderate alcohol consumers, compared to non-drinkers, disappears—or turns to statistically-significant harm, as your graph claims—when prior drinkers (“sick-quitters”) are excluded from the dataset. My “we’ll wait” wasn’t meant to be rude, I’m sorry if it struck you that way; it was meant to provide emphasis that this is what it would it take to prove the fundamental assertion of your post. And it still is. And you still have not provided a single example. Please do.
Also, where did that second all-cause mortality dose-response curve that you published come from, the one that you purport indicates “The reality”? The numbers look incredible. At zero grams/day of alcohol consumed, the relative risk of dying from any cause is 0.5. At 20 g/day, a bit more than one glass of wine every day, the RR is about 1.2, or 240% the death rate of teetotalers. (150 ml/glass x 12% alcohol = 18 g.) At about two glasses per day, 40 g/d, the relative risk of dying is about 400% that of the lifetime-non-drinkers. At three glasses per day, the risk is about 600%! Where do such incredible data come from? Reality?
Now compare that with the results of the 2,000,000-Brits dataset. There, excluding former drinkers resulted in the teetotalers having 24% greater all-cause mortality than the moderate drinkers. That’s not the moderate drinkers having 400% greater mortality than the teetotalers. Please explain it to me.
There those consuming 20-40 g/d of alcohol had all-cause mortality of HRR=0.48 compared to those at 0 g/d, with former drinkers excluded (and so no “sick quitters”). According to your plot of reality, they should have instead had an HRR of about 3.00! Where are you getting such stuff?
2. I agree with you that “hard clinical end points - heart attacks, strokes, deaths - are what really matter.”
But you cannot just ignore the biochemistry, as you did. Alcohol raises HDL levels, improves the apolipoprotein profile, lowers fibrinogen levels, increases adiponectin, decreases HbA1c levels, decreases stress, and so, cortisol levels, and red wine is stock full of anthocyanins. There is a strong biological basis for the hard-outcome CHD and mortality results that we see keep seeing in all of the epidemiological datasets.
3. I also agree with you that Stockwell was not the first to suggest the “sick-quitter” angle or to cherry-pick his studies, so there was no need for all of your cites there. Indeed, my link to that remarkable rebuke by so many of his colleagues at https://www.jsad.com/doi/epdf/10.15288/jsad.2016.77.834?role=tab made it clear that his methods basically just duplicated earlier, similar ones:
“These same concerns were raised by Fillmore et al. (2006) and were subsequently discussed at length at an international symposium later in 2006 (that included Fillmore and 28 other scientists working in the field of alcohol and health; The Harms and Benefits of Moderate Drinking, 2006). The conclusion at that meeting was that the criteria used by Fillmore et al. had inappropriately excluded many solid studies. … Unfortunately, in this new article by Stockwell et al. (2016), the authors have used the same selection criteria that were demonstrated earlier to be invalid. [Also,] they do not acknowledge that the “errors” that had been raised in 2006 have been adjusted for in most recent reputable investigations without changing overall results.”
Stockwell is just the most vocal of the modern crusaders. And they are on a crusade. For many this is not about science and truth, this is simply about getting more people to drink less, or not at all. As your friend Mick Skolnick recently substacked: “Ethanol is clearly a poison … I am less concerned about making a “fair and balanced” presentation of the science than I am about persuading people to avoid, or at least reduce their consumption of alcohol.” (Well, at least he’s honest.) But a lot of people in this space are like that. I can appreciate that lots of people work at addiction research facilities and all they see are the, admittedly, many downsides of alcohol: alcoholism, auto accidents, cirrhosis, violence, fetal alcohol syndrome, mean dunks, etc. They have invested their entire lives in trying to battle the downsides and they just can’t accept the thought of any up-sides. And many other abstainers are just holier-than-thou/healthier-than-thou nags.
4. Ben, you provide cites on arrhythmias and hypertension and young adults. I never made any claims about arrhythmias or hypertension or young adults. Alcohol isn’t a wonder-drug. And at high levels it obviously generates adverse CVD outcomes. And its biomarker indications wouldn’t suggest helping arrhythmias or hypertension (which, technically, is a biomarker itself).
But an honest analysis of the data reveals that modest alcohol consumption really does substantially reduce hard adverse cardiovascular outcomes. It’s because of the magnitude of the effect (about 30%, whole-population) times the magnitude of the CVD problem (our number-1 killer, whole-population) that being an honest analyst is crucial. That is why it is important to acknowledge that in the 2,000,000-Brit study, similar to all the others, without any “sick-quitters,” they found that:
“Non-drinking was associated with an increased risk of unstable angina (hazard ratio 1.33), myocardial infarction (1.32), unheralded coronary death (1.56), heart failure (1.24), ischaemic stroke (1.12), peripheral arterial disease (1.22), and abdominal aortic aneurysm (1.32) compared with moderate drinking.”
I agree with you that: “there will never be a one-size-fits-all answer.” But we shouldn’t throw the baby out with the bathwater. Older people who don’t have significant atherosclerosis (say they have Coronary Artery Calcium scores of zero or near-zero,) won’t receive any health benefit from alcohol. If that’s 50% of the older population, though, that means that the other 50% who do have atherosclerosis will see twice the benefit as the whole-population numbers. When we edit out those who won’t benefit from the denominator, all those 25% whole-population MACE reductions indicated above then turn into 50% MACE reductions. 50%! Wow. Just by enjoying a daily beer or glass of wine or two. So why shouldn’t physicians actively prescribe alcohol to those who have atherosclerosis? If there is no reason to believe that they would abuse it (and these aren’t frat boys), then I believe that it is medically unethical not to.
Okay, I think that I’ve responded to all of your new cites and points in your Part 1. If I haven’t, let me know. Please return the courtesy to me here. Thanks. I’m currently working on your Part 2, including on why Mendelian randomization studies do not validly apply here, and on Part 3
Before I moved on to consider your Parts #2 & #3, I did a little deeper dive into your cites in Part #1.
Imagine my surprise with your cite of the BMJ 2021 paper by Ding et al. to supposedly support your position that when former drinkers are excluded from CVD-event and mortality analyses, that “the story change[s] dramatically,” “the J-curve disappears,” and “the more you drink, the greater your risk of premature death”!
Ding looked at alcohol consumption in secondary prevention, contributing two new datasets, plus additionally provided a meta-analysis on secondary prevention. Please review it again at:
As in the six datasets that I provided, Ding segregated out the “former drinkers” and any sick-quitters from the “never drinkers” in both new datasets. What did they find? Take a look at their Figure 1. They found no statistically-significant difference between the “former drinkers” and the “never drinkers” in either all-cause mortality, cardiovascular mortality, or cardiovascular events. Moreover in the larger UK Biobank dataset, with the former drinkers excluded, they found the same-old/same-old J-shaped dose-response curve. (The same was true of the HSE/SHeSs dataset, but technically it did not reach statistical significance there likely because of its substantially smaller size.) In the UK Biobank dataset, even the “High-level drinkers” had a statistically-significant 29% lower cardiovascular event rate than the teetotalers. The “Medium-level drinkers” had a statistically-significant 31% lower cardiovascular event rate, 37% lower cardiovascular mortality, and 29% lower all-cause mortality. Wow. And Ding’s meta-analysis in Figure 2 shows the typical old “J-curves,” with reduced risk compared to teetotalers out passed 60 alcohol g/day, or an average of 3 glasses of wine or 4 cans of beer a day.
Do you really think that this cite supports your substack post? It’s a seventh dataset.
And you think it is ethical to counsel your secondary-prevention patients against drinking?
Wouldn’t that raise their all-cause mortality by 25-30%?
Sid, you’ve clearly put time and energy into your response, and I recognise that you’ve read widely. But this exchange has moved well beyond reasonable scientific disagreement.
I initially engaged with your comments, despite their offensive tone, accusations, and sarcasm, because I believed readers would benefit from seeing the science discussed openly, even when opinions differ. But it’s now clear that you're not acting in good faith. You’ve repeatedly misrepresented what I’ve written, including your latest suggestion that I cited Ding et al to support the elimination of the J-shaped curve, when I clearly used it to show that even those who accept the J-shape relationship still acknowledge concerns about residual confounding and study design. That was in direct response to your claim that Stockwell has been “discredited.”
What you’ve presented isn’t a scientific rebuttal - it’s a performance. You’ve ignored nuance, reframed questions to suit your narrative, and treated disagreement as a contest to be won. This may feel like a game of gotcha to you, but I take the responsibility of public health writing seriously. I write for readers trying to make better choices, not for people trying to prove they’re the smartest in the room.
You asked whether I think it’s ethical to advise secondary prevention patients against drinking. Your reframing exemplifies the problem here. The central question - the one I addressed in the article - is not whether we forbid alcohol, but whether we should encourage moderate drinking for heart health. And the evidence simply doesn’t support that.
Say I tell a patient a glass of wine each evening might reduce their heart attack risk. What happens if they later develop breast cancer or have a stroke - conditions whose risk is increased even by low levels of alcohol? Have I helped, or harmed?
As I said in the article, and as most public health agencies now state, alcohol is not a health food. It’s a risk to be managed, not a medicine to be prescribed.
I won’t be continuing this conversation. The thread will be closed.
Their “Conclusions: Favourable changes in several cardiovascular biomarkers (higher levels of high density lipoprotein cholesterol and adiponectin and lower levels of fibrinogen) provide indirect pathophysiological support for a protective effect of moderate alcohol use on coronary heart disease.”
(You do believe in randomized controlled trials, don’t you Dr. Jones? Or maybe you’d just rather ignore the whole biological literature? Please cite for us all of your RCT evidence that alcohol makes CVD biomarkers worse. We’ll wait.)
You might also google PMID 16129796, 28264492, 11976563, 27705886, 23425242, & 23596141.
So alcohol’s widely-proven beneficial cardiovascular-event effects are exactly what we would expect based on biochemistry. It’s called “science.” To claim otherwise is just spreading medical misinformation.
And, yes, there is also lots of scientific evidence that the anthocyanins in red wine provide an extra biochemical health-boost to the alcohol. The only difference between red wine and white wine is the anthocyanin colorant molecules from the red grape varieties (plus, perhaps some oak aging). These are the same anthocyanin molecules that give color to cherries or blueberries or apple skins.
On red wine specifically, see: PMID 23408240, 37403999, 22205309, 23894543, 39458427, 38847707, 22205309, 22955728, 22999066, 16095600, 20522276, 36289599, 31597344, 23319811, and 26152229.
For anthocyanins, specifically, see PMID 23319811 (myocardial infarction hazard ratio=0.68), 26152229 (all-cause mortality HR=0.66), 21926181, 22810991, 24061071, 10477529, 22914551, 25578927, 24406274, 35082171, 36563462, and 37760077.
Surely, Dr. Jones, you do appreciate biochemistry enough to consider that a drug—alcohol—that significantly positively effects multiple heart-protective biological pathways, could be beneficial when administered at appropriate dosages?
I’m asking this in good faith (in case that doesn’t come through in text!), but how large was the study, and how diverse was the population? And how (if at all) does this account for genetics? I come from a long line of alcohol abusers who lived well in to their 90’s, along with everyone else on the family. I’ve read a lot about how the consensus now is that any drinking= dangerous, but these anecdotal non-examples from my life cause me to wonder. (For reference, these family members are white/wealthy, do these studies account for other risk factors like race and poverty?)
Thanks Darcy, that’s a completely fair and thoughtful question (and tone came across perfectly!).
The 2015 Stockwell meta-analysis I referenced pulled data from 87 separate studies, covering nearly 4 million participants across multiple countries. But like many large epidemiological studies, the population was skewed: largely Western, white, and middle-to-upper income. So yes, there are real limitations when it comes to diversity, both in terms of genetics and socioeconomic background.
As for genetics: you're right to raise it. We know genes influence alcohol metabolism (e.g., variants in ALDH2 common in East Asian populations can dramatically change risk profiles), but most of these large cohort studies don’t stratify by genotype. That means people like your family, genetically or culturally more resilient to alcohol's harms, can get washed out in the averages. At a population level, alcohol raises risk. But your personal risk? That could very well differ.
The other big piece is what researchers call "residual confounding." Your relatives were white, wealthy, and part of a long-lived family line - all factors that tend to predict longer lifespans, regardless of alcohol use. People with higher income and education levels tend to live longer, regardless of habits like drinking. So their lifespan may reflect their circumstances more than their alcohol intake.
But of course, risk isn’t destiny. If I had a dollar for every patient who told me, when I gently suggested quitting smoking, “But my uncle smoked two packs a day and lived to 90!”, I’d have enough for a very nice weekend away.
Outliers don’t invalidate the pattern; they just remind us that risk is personal, and our choices stack on top of whatever baseline we’re dealt.
Thanks again for reading and engaging so thoughtfully. I really appreciate it.
Tim Stockwell is a Neo-prohibitionist crusader who has long cherry-picked his data and been repeatedly discredited. See, for example, PMID 27588543:
https://www.jsad.com/doi/epdf/10.15288/jsad.2016.77.834?role=tab
Wow. That is quite an extraordinary rebuke in a professional journal!
I have consolidated my responses in a single reply to another of your comments.
So interesting, thanks for this response!
Excellent Article. Thank You for submitting and putting the facts out there. The oncology professionals have been very much in agreement with your research and conclusions. They put alcohol in the same group as tobacco and asbestos in terms of carcinogens. Pretty eye opening and scary. Thank You. Keep up the great work you do....
Thanks, Ron. I'm not sure the message has got out there. I think most people understand that very heavy drinking is associated with cirrhosis, but think that their level of 'moderate' drinking is far removed from any health risk.
Isn't it odd to think we have a multi-billion dollar industry in which we go out and willingly drink a carcinogen!
Yes it is. When I asked an oncologist why they haven't declared war on it, he responded, " we are still trying to get people to kick the smoking habit and get their weight under control. Taking away their booze is going to be a bigger nut to crack for the reasons you mentioned in your response. Thank You
Great column! I am a big fan of nuts and I stopped drinking alcohol several years ago based on advice from a doctor I trusted. I don’t miss it at all because I value my health. For my friends 55+ who are still drinking moderate amounts, does stopping now improve their odds or have decades of consumption locked them into a higher risk category?
You’re right. When you weigh up the pleasure of a drink vs potentially extra years of life, it shouldn’t bee too hard of a decision.
As for the effects of quitting alcohol later in life after years of drinking, It’s likely there will always be health benefits from cutting back or stopping. If we take a worst case scenario, someone who has drunk so heavily that they already have early liver cirrhosis, even they are likely to have markedly better health outcomes if they stop drinking.
Great information!!
Thanks, Kimberly! I’m so glad you found it helpful.
Great article, Ben. I've just updated my Post to include a prominent link to it:
https://drmick.substack.com/p/alcohol-whats-the-fuss
Thanks, Mick. I love your comprehensive, detailed review. That was a lot of work.
As you suggest, I don’t think the message that there’s no safe level of alcohol consumption has really got out there. I suspect if we did a poll, most people would still think that drinking a little is good for us.
As I just said to someone else, it’s amazing to think we have a multi-billion dollar industry helping us voluntarily drink a proven carcinogen. The power of the dollar!
Very nice article and great explanation about how the reference group of non-drinkers was comprised of a mix of low and high risk populations.
Thanks you. Isn’t it fascinating how we can have a blindspot for such an obvious methodological flaw for decades. I wonder what else we’re not seeing.
Ummm, no. This post should, instead, be titled: “What Ben Jones Got Wrong About Alcohol.”
Where in this article is the proof that “Decades of Alcohol Research [Are] Wrong”?
Extraordinary claims require extraordinary evidence. Where are the datasets that indicate this?
Where are any cites? Your only cite is Tim Stockwell, a Neo-Prohibitionist crusader who has long cherry-picked his data and been repeatedly discredited. See, for example, PMID 27588543:
https://www.jsad.com/doi/epdf/10.15288/jsad.2016.77.834?role=tab
Wow! That is quite an extraordinary rebuke for a professional journal!
His “sick-quitter” claim is easily disproven if you actually look at the data.
I challenge Dr. Jones to provide us with a couple of datasets where the typical 30% coronary heart disease event reductions seen in moderate alcohol consumers compared to non-drinkers disappears—or turns to statistically-significant harm—when prior drinkers (“sick-quitters”) are excluded from the dataset.
Heck, Dr. Jones, just provide us with a single such dataset! We’ll wait.
(My 30% CHD number above, along with 13% lower all-cause mortality, comes from PMID 21343207, an older meta-analysis of 84 observational studies at: https://www.bmj.com/content/bmj/342/bmj.d671.full.pdf )
Many alcohol studies, in fact, have specifically segregated out former drinkers, and so, any “sick-quitters,” in their analyses. See, for example, the newer PMID 28331015, with almost 2,000,000 Brits followed for 6 years.
https://www.bmj.com/content/bmj/356/bmj.j909.full.pdf
See its Figure 2. There the “non-drinkers” had 32% more heart attacks—and the “former drinkers” had 31% more—compared to the “moderate drinkers.” And with respect to cancer and other possible alcohol-related ailments, the non-drinkers had 24% higher All-Cause mortality, and the former drinkers 38% more, compared to the moderate drinkers. Care to explain, Dr. Jones? Do you want you readers to quit drinking and forgo moderate drinking’s CVD benefits—and have at least 24% higher net mortality? That’s from almost 2,000,000 people—excluding all “quitters."
Or see the study in Ann Epidemiol 2007;17:S3–S7 at:
https://www.sciencedirect.com/science/article/abs/pii/S104727970700004X
It explains: “To address the issue of residual confounding by healthy lifestyle in drinkers [and the inclusion of any “sick-quitters” in the non-drinkers category], in a large prospective study we restricted analysis to only ‘‘healthy’’ men (who did not smoke, exercised, ate a good diet, and were not obese). Within this group, men who drank moderately had a relative risk for Coronary Heart Disease of 0.38 compared with [healthy] abstainers, providing further evidence to support the hypothesis that the inverse association of alcohol to CHD is causal, and not confounded by healthy lifestyle behaviors.”
Wow. That’s a coronary heart disease reduction of over 60%—when zero “sick-quitters” were included.
Or, put another way, among strictly-healthy men, those who forewent alcohol had 160% more CHD events than those who drank (i.e.1/0.38). Wow.
Care to explain, Dr. Jones?
For another study that explicitly separated out “former drinkers,” see PMID 21098615 at:
https://www.bmj.com/content/bmj/341/bmj.c6077.full.pdf
For the combined populations, the lifetime non-drinkers had 100% more coronary events compared to the drinkers who didn’t binge. Wow. The former-drinkers, which would include any “sick-quitters,” “only” had 57% more coronary events than the drinkers. So instead of pulling down the results of a united “non-drinker” category, it looks like the inclusion of former drinkers would have pulled them up.
See another dataset at PMID 15167206, “Changes in alcohol intake and risk of coronary heart disease and all-cause mortality in the MONICA/KORA-Augsburg cohort 1987-97” where: “The lowest all-cause mortality risk was observed among men drinking 20-40 g [of alcohol] per day (HRR 0.48).”
I.e. 50% lower all-cause mortality, including cancers, liver issues, and accidents, compared to lifelong non-drinkers. That’s with former drinkers excluded, so no “sick quitters.”
Wow, 100% higher All-Cause mortality in the teetotalers with no “sick quitters”! See it’s Tables 5 & 6, available free through ResearchGate if your institution doesn’t have access to it.
See another dataset, PMID 29844013, at:
https://www.bmj.com/content/bmj/361/bmj.k934.full.pdf
“We adopted several approaches to help limit potential bias involved in the study of never-drinkers (who might differ systematically from drinkers in ways that are difficult to measure, but which might be relevant to disease causation) and of ex-drinkers (a category that includes people who might have abstained from alcohol owing to poor health itself, as well as those who changed their habits to achieve a healthier lifestyle).”
Their results? Compared to infrequent drinkers (i.e drinkers, but at only 0.1-5 g/day of average alcohol, HR=1.00=reference), former-drinkers and never-drinkers had 24% & 17% more non-fatal coronary heart disease events, respectively. Those drinking 30-60 & >60 g/day alcohol, on the other hand, had 26% & 25% less CHD. See its Table 3. That means that the non-drinkers-excluding-any-sick-quitters had 58% more CHD events than those averaging 2+ drinks per day (i.e. 1.17/0.74). Wow.
For one on diabetes, see Table 2 of PMID 28748324 at:
https://link.springer.com/article/10.1007/s00125-017-4359-3
They too separated out “current abstainers” (i.e. including any “sick-quitters”) from “lifetime abstainers.”
Using infrequent drinkers (<1 drink per week) as the Hazard Rate=1.00 reference, there was no statistically-significant benefit in future diabetes rates for Current Abstainers (i.e. any sick-quitters). But women who were Lifetime Abstainers had a 25% higher future diabetes rate. Men and women who drank 3 or 4 times per week, on the other hand, had 27% and 32% less future diabetes. Wow. That means that the non-drinkers-excluding-any-sick-quitters had about 80% more diabetes development than those drinking 3 or 4 times a week (i.e. 1.25/0.70). Wow, so high diabetes protection too.
See also PMID 20876712 and PMID 31568479.
So separating out former-drinkers from lifetime-teetotalers in these analyses doesn’t make modest-drinking’s health benefits disappear—if anything it amplifies them.
Now let’s see your datasets, Dr. Jones, indicating that all the huge CVD benefits from moderate drinking that we keep witnessing are due to including “sick-quitters” in with the non-drinkers, as you claim. We’ll wait.
Anyone telling you this is playing you for a lazy fool. It may be what you want to believe, but it’s simply spreading false medical information that’s going to hurt people. If you can’t win with scientific datasets, then you try to muddy up the water spreading “FUD,” fear, uncertainty, and doubt, I guess.
Please, Dr. Jones, examine all of the many datasets referenced above and then write an up-dated piece, or at least delete this one. You don’t want to kill your readers, do you? Your future credibility is on the line here.
Sid, I’ve seen your comments and appreciate that you’ve taken the time to share your perspective. You’ve raised a lot of points and studies, and I want to make sure I take the time to consider them properly and respond with the depth they deserve. I’ll post a more detailed reply once I’ve had a chance to look at them.
Thanks so much. No hurry. I’d much rather that you take the time that you need to make a responsible response. Happy analyzing!
Part 3
Considerations about alcohol and health are not restricted to cardiovascular disease.
Bagnardi et al. showed that alcohol consumption is linked in a dose-dependent manner to a higher risk of cancers of the oral cavity, pharynx, oesophagus, larynx, stomach, colon, rectum, liver, female breast, and ovaries, even at typical moderate levels of drinking. They concluded: 'The analysis did not identify a threshold level of alcohol consumption below which no increased risk for cancer was evident.' - In other words, any amount of alcohol increases cancer risk.
[https://pmc.ncbi.nlm.nih.gov/articles/PMC6705703/]
Bagnardi followed up in 2013, specifically looking at light drinking. Their conclusion: "Alcohol increases the risk of cancer of the oral cavity and pharynx, oesophagus and breast even at low doses. Given the high proportion of light drinkers in the population and the high prevalence of these tumours, especially of breast cancer, even small increases in cancer risk are of great public health relevance."
[https://www.annalsofoncology.org/article/S0923-7534(19)36858-9/fulltext]
Other 'J-shaped curve' associations with alcohol are also being challenged.
It was previously believed that moderate drinking protects against dementia. However, an observational cohort study of 550 adults over 30 years found a "dose-dependent association between alcohol consumption over 30 years of follow-up and hippocampal atrophy, as well as impaired white matter microstructure... There was no evidence that light drinking offers protection over abstinence in terms of brain structure or function." They concluded, "The finding that alcohol consumption in moderate amounts is linked to multiple markers of abnormal brain structure and cognitive function has significant potential public health implications for a large sector of the population. For example, in our sample, nearly half of the men and a quarter of the women were currently drinking in this range."
[https://www.bmj.com/content/357/bmj.j2353]
Of course, the ideal way to answer this question of whether alcohol protects against coronary heart disease is with a large prospective randomised controlled trial where participants are randomly assigned to low-level or no alcohol and then followed for many years to see how their rates of coronary heart disease differ. The MACH15 study was planned to do just this, but its funding was terminated.
Will a similar study be conducted? In the words of Carr et al. "Due to the growing number of studies suggesting increased disease risk, including cancer associated with alcohol use even at very low levels, the use of RCTs to study alcohol consumption is ethically questionable." - In other words, the risks of alcohol, even at low levels, are considered too high for it to be acceptable to use it for a clinical trial.
[https://www.nature.com/articles/s41467-024-47632-7]
It’s worth repeating: even if light drinking did protect against coronary heart disease, the broader harms of alcohol are so well established that researchers no longer consider it ethical to test in a clinical trial.
And even if that protection were real, no credible clinician would advise drinking to prevent disease, which makes the entire argument moot.
You’ve accused me of misleading readers, of “ignoring the science,” even of “killing my readers.” Not only is that scientifically indefensible, it’s offensive.
In the article, I stated that alcohol carries health risks and should be approached with caution, not mistaken for a health tonic. That statement is irrefutably grounded in the scientific consensus. I stand by it.
If you’re genuinely concerned about public health, maybe start by modelling respectful dialogue. You're welcome here if you're ready to argue in good faith and back your claims with evidence. However, if you're looking for rhetorical gladiatorial combat, I'm not interested. If that's the case, you may feel more at home on Twitter/X.
Part 2
To overcome the challenges of studies that depend on participants accurately reporting their alcohol consumption and the potential confounding effects of factors such as socioeconomic status and lifestyle, researchers have turned to Mendelian randomisation (MR) studies.
These studies divide participants into groups based on certain genes associated with drinking. For example, East Asians with certain ALDH2 gene variants have an unpleasant reaction to alcohol, so they drink less. You can then compare this group to other East Asians without the gene, who, on average, drink more.
Because we inherit these genes from our parents at conception, they are set long before we choose whether to smoke or go to college, and whether we live a comfortable middle-class life or in poverty, all of which are contributing factors to our health later in life. Researchers use different genes in different populations, and the analyses are restricted to single ethnic groups to avoid other confounding factors.
As a result, MR studies remove complications such as whether or not people are poor or get confused about how much they drink. They also have an advantage over observational studies in that they can be used to infer whether an exposure causes the outcome being studied rather than just being associated with it.
In one of the earliest, Holmes et al. found in a study of almost 262,000 participants that "individuals of European descent with a genetic predisposition to consume less alcohol had a reduced risk of coronary heart disease and ischaemic stroke, and lower levels of several established and emerging risk factors for cardiovascular disease. These findings suggest that reductions of alcohol consumption, even for light to moderate drinkers, may be beneficial for cardiovascular health. Our results therefore challenge the concept of a cardioprotective effect associated with light to moderate alcohol consumption reported in observational studies and suggest that this effect may have been due to residual confounding or selection bias."
[https://www.bmj.com/content/349/bmj.g4164]
Multiple MR studies have now been completed.
More recently, Kember et al. (2024) compared a regular observational study and an MR study on the same group of almost 200,000 US veterans. The observational study again showed the U-shaped curve, but the MR analysis showed no link between alcohol use and coronary heart disease.
Their conclusion: "We conclude from these findings that the observational associations showing a U-shaped curve for the association of drinking level with CHD and T2D reflect confounding attributable to a variety of unmeasured factors. These include inaccurate or invalid reporting of alcohol consumption and lifestyle, genetic, and socioeconomic factors associated with moderate alcohol consumption." - In other words, the U-shape relationship has nothing to do with alcohol.
[https://doi.org/10.1111/acer.15445]
Also in 2024, Carr et al. performed a landmark in-depth analysis of 127 observational, case-control, and MR studies on alcohol and ischaemic heart disease using advanced Bayesian burden of proof techniques.
For the observational and case-control studies, the familiar J-curve was apparent; however, in many analyses, it lost statistical significance when study heterogeneity (differences in study methods) was taken into account.
That said, these studies did show protection with 'average' consumption levels. Pooling case-control and cohort data decreased the risk of ischaemic heart disease by around 4% which, according to the authors, "suggests that the association – on the basis of the available evidence – is weak." - Far from the 60% protection you suggested.
When they examined data from MR studies, they found no evidence of protection, although they requested MR studies using more genes and techniques better suited to detecting curved relationships.
[https://www.nature.com/articles/s41467-024-47632-7]
Already in press when this was written was an MR study that did exactly that.
Kassaw looked at almost 280,000 white British adults. In a standard analysis, the J-shaped curve was apparent, though they found "the alcohol–mortality relationship differs between younger and older individuals and is strongly affected by adjustment to potential confounders and indicators of poor health, supporting the role of selection bias and reverse causality."
When they conducted a detailed MR analysis designed to detect non-linear (curved) relationships with 94 gene variants, they found "MR analyses did not provide any evidence to support the J-shaped association between alcohol intake and mortality suggested in the conventional observational analysis." They suggest, "...a reduction in the amount of alcohol consumed is likely to be beneficial regardless of the level of current intake."
Their final conclusion, "Genetic evidence strongly suggests that as alcohol consumption increases, there is a linear increase in the risk of premature death, including from specific causes such as CVD, cancer, and digestive illness, with no evidence for any protection by modest intakes... A re-evaluation of current public policies regarding drinking guidelines may be warranted."
[https://doi.org/10.1093/ije/dyae046]
Far from being discredited as you suggest, Stockwell's analysis is supported by study after study.
I believe there is evidence that for some people, notably older adults with or at risk of coronary heart disease specifically (not other forms of cardiovascular disease), light alcohol consumption may reduce their risk of cardiovascular disease and increase life expectancy. However, this is a decision to be made after taking into account all aspects of their health and lifestyle, i.e. after a conversation with their doctor, not advice to be given in an online article.
That said, I cannot find a single clinician advocating light drinking to prevent or manage cardiovascular disease, even those who accept the J-curve hypothesis. Day and Rudd (2019) are typical in their conclusions: "Observational and prospective studies have consistently shown a lower risk of cardiovascular and all-cause mortality in people with low levels of alcohol consumption when compared to abstainers (the ‘J’-shaped curve)... However, this evidence is contested, and overall, the detrimental effects of alcohol far outweigh the beneficial effects."
[https://onlinelibrary.wiley.com/doi/10.1111/add.14703]
This advice has been the same for decades, regardless of the arguments swirling around. Marmot & Brunner back in 1971, whilst agreeing that moderate drinking may be protective against cardiovascular disease, concluded: "In conclusion, the balance of harm and benefit does not weigh in favour of making a recommendation to the public to drink in order to prevent coronary heart disease."
[https://www.bmj.com/content/bmj/303/6802/565.full.pdf]
Part 1
Sid, thanks for your comments. Let me be clear. This isn’t a place for performative condescension.
If you want to debate the science, I’m all for it. But the sarcastic jabs, the “we’ll wait” taunts, the smug one-liners, that’s not how serious people debate science. Your tone undermines your credibility far more than any dataset could.
One of the reasons people come to Substack is to get away from that kind of noise. If you have a case, make it with evidence and respect.
Now, to the substance.
Much of the evidence you provide is on biomarkers. These are helpful for understanding potential mechanisms, and some are used as surrogate endpoints in short clinical trials when you can't wait for long-term health outcomes. However, hard clinical end points - heart attacks, strokes, deaths - are what really matter, and we have plenty of data on these, so that's where I'll focus.
You dismiss Tim Stockwell, whose 2015 data I used for the graph in my article, as "a Neo-prohibitionist crusader who has long cherry-picked his data and been repeatedly discredited."
If your knowledge of the topic is as extensive as your tsunami of citations suggests, you'll know this is far from the truth. Stockwell was not the first to suggest that the J-shaped curve may be an artefact of study design. This debate has been ongoing for decades.
For example (I quote directly from the authors):
Wannamethee & Sharper - Neuroepidemiology 1998
"Non-drinkers, both ex-drinkers and lifelong teetotallers, have an increased prevalence of conditions likely to increase morbidity and mortality compared with occasional or light drinkers. In addition, regular light drinkers tend to have characteristics extremely advantageous to health [he's talking about them being largely well-educated, middle-class and white]. Whilst there is considerable evidence that alcohol intake at some level has a beneficial effect on CHD, the degree of protection claimed is almost certainly exaggerated by comparison with an inappropriate control group and by the limited adjustment procedures used to take into account the differing characteristics of the various alcohol intake groups."
[https://pubmed.ncbi.nlm.nih.gov/9778595/]
Light drinking is a hallmark of better-educated, wealthy, healthy, often white populations who, overall, have better health outcomes than the rest of the population. Therefore, it's argued, light drinking is often a marker of being in a group with good health outcomes, rather than the cause of it. Go outside that group, and the J-shaped curve often doesn't hold up.
Sempos et al. - Alcohol Clinical & Experimental Research 2003
As an example, Sempos looked at the relationship between alcohol and mortality for African Americans. "No J-shaped curve was found... no beneficial effect appeared and mortality increased with increasing average consumption for more than one drink a day."
[https://pubmed.ncbi.nlm.nih.gov/12544011/]
Andréasson - Alcohol Clinical & Experimental Research 2006
"Many studies report J- or U-shaped curves in describing the association between level of alcohol consumption and mortality. J-shaped curves may, however, be the result of complex associations between psychosocial and other confounding factors and health, rather than a simple causal association derived from the biological effects of alcohol."
[https://pubmed.ncbi.nlm.nih.gov/9799961/]
There are others, but you get the gist. Stockwell did a large analysis to see if suspicions that the so-called J-shaped curve suggesting moderate alcohol protects against cardiovascular disease were correct. His results supported the conclusion that non-drinkers are a mixed bag of teetotalers and 'sick quitters' and that the J-shaped curve is an artefact of study design.
Has he been discredited? Far from it. Yes, you found a 'letter to the editor' shortly afterwards criticising the findings, but he has subsequently been extensively cited and has had a significant impact on the direction of research in the field.
Just one of many examples: "Therefore, although there is a strong body of evidence that shows a J-shaped association between alcohol intake and mortality, it seems likely that the link between low–moderate intake and improved survival is not causal, and there is no ‘safe’ threshold intake below which drinking gives benefit." - DJ Stott - Age and Ageing 2020.
Stockwell is not alone with this finding. Others have found the same thing, including Knott, who, using data from the Health Survey for England, found that when never drinkers were used as the comparator, a J-curve appeared - but only in men aged 50–64 and women over 65.
To test whether this pattern held up, the researchers repeated their analysis using five increasingly strict definitions of “occasional drinkers” as the reference group. The rationale was simple: if the J-curve reflects a genuine protective effect of moderate drinking, it should appear no matter who you compare it to. But it didn’t.
As the authors note, “Supplementary analyses found that most protective effects disappeared where calculated in comparison with various definitions of occasional drinkers.” - How you define the comparator group makes all the difference to the results.
[https://www.bmj.com/content/350/bmj.h384]
While meta-analyses subsequent to the Stockwell paper continue to show a J-shaped curve, authors increasingly acknowledge that study design limitations may drive these findings. For example, in the BMJ 2021 paper by Ding et al., the authors state: "there is evidence that reductions in risk of all-cause mortality and subsequent events might have been overestimated due to the inclusion of former drinkers in the non-drinking reference group.... we cannot definitely determine the extent of this overestimation with very few studies that explicitly excluded former drinkers."
[https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-021-02040-2]
These analyses are complicated by the fact that cardiovascular disease is not a single entity; it's a constellation of conditions.
A Lancet analysis of nearly 600,000 individuals found that the relationship between alcohol and cardiovascular disease varies by subtype. While the study confirmed a J-shaped curve for myocardial infarction (heart attack), it reported a clear dose-dependent increase in risk for stroke, heart failure, and deaths from other cardiovascular causes. The authors concluded: “For cardiovascular disease subtypes other than myocardial infarction, there were no clear risk thresholds below which lower alcohol consumption stopped being associated with reduced disease risk. These data support limits for alcohol consumption that are lower than those recommended in most current guidelines.” - That's a confusing way of saying the less you drink, the lower the risk.
[https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(18)30134-X/fulltext]
A cohort study of almost 108,000 individuals followed for 14 years showed a nonlinear association between alcohol consumption and atrial fibrillation. There was no protection at low levels of consumption and a 16% increased risk (HR 1.16) at 12g alcohol (1 drink) a day.
[https://academic.oup.com/eurheartj/article/42/12/1170/6090248]
A meta-analysis including data on more than 360,000 individuals found that even low levels of drinking were associated with a 19% increased risk of hypertension in men (RR 1.19). In women, an increased risk wasn't seen until consumption reached 3-4 drinks per day, but no protection was seen in men or women.
[https://www.ahajournals.org/doi/full/10.1161/JAHA.117.008202]
In addition, the effect of alcohol varies by age, gender and geography.
A Lancet report from the Global Burden of Disease Study in 2022 shows that the level of drinking associated with the lowest health loss varies by age, gender and region. The authors say, "We provide clear evidence that the level of alcohol consumption that minimises health loss varies significantly across populations and remains zero or very close to zero for several population groups, particularly young adults." For some older adults, there appears to be a risk reduction associated with moderate alcohol consumption (0.1-1.9 standard drinks per day), likely in those adults at higher risk of MI. They suggest recommendations should vary by age and location, and that guidelines for younger adults should be reduced.
[https://www.thelancet.com/journals/lancet/article/PIIS0140-6736(22)00847-9/fulltext]
Therefore, there will never be a one-size-fits-all answer.
Response to Part #1
Ben, thank you for quickly responding back at length. But you largely ignored my points, instead going off in new directions (to which I will later respond as well).
1. But first let’s return to what you originally posted and what my comments actually were.
Your whole thesis on “What Decades of Alcohol Research Got Wrong” dealt with the classification of supposedly “sick-quitters.”
You wrote: “By lumping [former drinkers (aka the “sick quitters”)] in with lifelong abstainers, researchers unintentionally stacked the non-drinker group with people who already had higher health risks. That distorted the results, making light drinkers look healthier than they really were. … When researchers re-analysed the data and looked only at lifelong abstainers, the story changed dramatically. Once you remove high-risk former drinkers, the J-curve disappears. The more you drink, the greater your risk of premature death. The message is brutally clear: the more alcohol you drink, the higher your risk of dying early. There’s no U-shape. No protective effect from drinking lightly.”
I responded back that this was false. That when we exclude former drinkers from the non-drinkers category, instead, we get the same results. The benefits of moderate drinking don’t “disappear.” The story does not “change dramatically.” The “U-shape” remains. That it’s “brutally clear” that there is still a protective effect from drinking lightly. I provided 6 separate datasets to demonstrate this. Since you are British, let’s review the first one, with almost 2,000,000 of your fellow countrymen, and perhaps some of your patients are in it.
https://www.bmj.com/content/bmj/356/bmj.j909.full.pdf
I asked you to explain the results in its Figure 2. You haven’t yet. That’s what I was hoping for in a return post.
There the “lifetime non-drinkers” had 32% more heart attacks—and the “former drinkers” had 31% more—compared to the “moderate drinkers.” Same amount. And with respect to cancer and other possible alcohol-related ailments, the non-drinkers—with any and all “sick-quitters” excluded— had 24% higher all-cause mortality compared to the moderate drinkers. And I provided 5 more similar datasets explicitly indicating the same thing—that excluding sick-quitters does not affect the U-shaped results.
I challenged you to provide a couple of datasets where the typical 30% coronary heart disease event reductions generally seen by moderate alcohol consumers, compared to non-drinkers, disappears—or turns to statistically-significant harm, as your graph claims—when prior drinkers (“sick-quitters”) are excluded from the dataset. My “we’ll wait” wasn’t meant to be rude, I’m sorry if it struck you that way; it was meant to provide emphasis that this is what it would it take to prove the fundamental assertion of your post. And it still is. And you still have not provided a single example. Please do.
Extraordinary claims require extraordinary evidence.
Also, where did that second all-cause mortality dose-response curve that you published come from, the one that you purport indicates “The reality”? The numbers look incredible. At zero grams/day of alcohol consumed, the relative risk of dying from any cause is 0.5. At 20 g/day, a bit more than one glass of wine every day, the RR is about 1.2, or 240% the death rate of teetotalers. (150 ml/glass x 12% alcohol = 18 g.) At about two glasses per day, 40 g/d, the relative risk of dying is about 400% that of the lifetime-non-drinkers. At three glasses per day, the risk is about 600%! Where do such incredible data come from? Reality?
Now compare that with the results of the 2,000,000-Brits dataset. There, excluding former drinkers resulted in the teetotalers having 24% greater all-cause mortality than the moderate drinkers. That’s not the moderate drinkers having 400% greater mortality than the teetotalers. Please explain it to me.
Or see the MONICA/KORA-Augsburg cohort dataset, PMID 15167206, https://pubmed.ncbi.nlm.nih.gov/15167206/
There those consuming 20-40 g/d of alcohol had all-cause mortality of HRR=0.48 compared to those at 0 g/d, with former drinkers excluded (and so no “sick quitters”). According to your plot of reality, they should have instead had an HRR of about 3.00! Where are you getting such stuff?
2. I agree with you that “hard clinical end points - heart attacks, strokes, deaths - are what really matter.”
But you cannot just ignore the biochemistry, as you did. Alcohol raises HDL levels, improves the apolipoprotein profile, lowers fibrinogen levels, increases adiponectin, decreases HbA1c levels, decreases stress, and so, cortisol levels, and red wine is stock full of anthocyanins. There is a strong biological basis for the hard-outcome CHD and mortality results that we see keep seeing in all of the epidemiological datasets.
3. I also agree with you that Stockwell was not the first to suggest the “sick-quitter” angle or to cherry-pick his studies, so there was no need for all of your cites there. Indeed, my link to that remarkable rebuke by so many of his colleagues at https://www.jsad.com/doi/epdf/10.15288/jsad.2016.77.834?role=tab made it clear that his methods basically just duplicated earlier, similar ones:
“These same concerns were raised by Fillmore et al. (2006) and were subsequently discussed at length at an international symposium later in 2006 (that included Fillmore and 28 other scientists working in the field of alcohol and health; The Harms and Benefits of Moderate Drinking, 2006). The conclusion at that meeting was that the criteria used by Fillmore et al. had inappropriately excluded many solid studies. … Unfortunately, in this new article by Stockwell et al. (2016), the authors have used the same selection criteria that were demonstrated earlier to be invalid. [Also,] they do not acknowledge that the “errors” that had been raised in 2006 have been adjusted for in most recent reputable investigations without changing overall results.”
Stockwell is just the most vocal of the modern crusaders. And they are on a crusade. For many this is not about science and truth, this is simply about getting more people to drink less, or not at all. As your friend Mick Skolnick recently substacked: “Ethanol is clearly a poison … I am less concerned about making a “fair and balanced” presentation of the science than I am about persuading people to avoid, or at least reduce their consumption of alcohol.” (Well, at least he’s honest.) But a lot of people in this space are like that. I can appreciate that lots of people work at addiction research facilities and all they see are the, admittedly, many downsides of alcohol: alcoholism, auto accidents, cirrhosis, violence, fetal alcohol syndrome, mean dunks, etc. They have invested their entire lives in trying to battle the downsides and they just can’t accept the thought of any up-sides. And many other abstainers are just holier-than-thou/healthier-than-thou nags.
4. Ben, you provide cites on arrhythmias and hypertension and young adults. I never made any claims about arrhythmias or hypertension or young adults. Alcohol isn’t a wonder-drug. And at high levels it obviously generates adverse CVD outcomes. And its biomarker indications wouldn’t suggest helping arrhythmias or hypertension (which, technically, is a biomarker itself).
But an honest analysis of the data reveals that modest alcohol consumption really does substantially reduce hard adverse cardiovascular outcomes. It’s because of the magnitude of the effect (about 30%, whole-population) times the magnitude of the CVD problem (our number-1 killer, whole-population) that being an honest analyst is crucial. That is why it is important to acknowledge that in the 2,000,000-Brit study, similar to all the others, without any “sick-quitters,” they found that:
“Non-drinking was associated with an increased risk of unstable angina (hazard ratio 1.33), myocardial infarction (1.32), unheralded coronary death (1.56), heart failure (1.24), ischaemic stroke (1.12), peripheral arterial disease (1.22), and abdominal aortic aneurysm (1.32) compared with moderate drinking.”
I agree with you that: “there will never be a one-size-fits-all answer.” But we shouldn’t throw the baby out with the bathwater. Older people who don’t have significant atherosclerosis (say they have Coronary Artery Calcium scores of zero or near-zero,) won’t receive any health benefit from alcohol. If that’s 50% of the older population, though, that means that the other 50% who do have atherosclerosis will see twice the benefit as the whole-population numbers. When we edit out those who won’t benefit from the denominator, all those 25% whole-population MACE reductions indicated above then turn into 50% MACE reductions. 50%! Wow. Just by enjoying a daily beer or glass of wine or two. So why shouldn’t physicians actively prescribe alcohol to those who have atherosclerosis? If there is no reason to believe that they would abuse it (and these aren’t frat boys), then I believe that it is medically unethical not to.
Okay, I think that I’ve responded to all of your new cites and points in your Part 1. If I haven’t, let me know. Please return the courtesy to me here. Thanks. I’m currently working on your Part 2, including on why Mendelian randomization studies do not validly apply here, and on Part 3
P.S.
Before I moved on to consider your Parts #2 & #3, I did a little deeper dive into your cites in Part #1.
Imagine my surprise with your cite of the BMJ 2021 paper by Ding et al. to supposedly support your position that when former drinkers are excluded from CVD-event and mortality analyses, that “the story change[s] dramatically,” “the J-curve disappears,” and “the more you drink, the greater your risk of premature death”!
Ding looked at alcohol consumption in secondary prevention, contributing two new datasets, plus additionally provided a meta-analysis on secondary prevention. Please review it again at:
https://bmcmedicine.biomedcentral.com/articles/10.1186/s12916-021-02040-2
As in the six datasets that I provided, Ding segregated out the “former drinkers” and any sick-quitters from the “never drinkers” in both new datasets. What did they find? Take a look at their Figure 1. They found no statistically-significant difference between the “former drinkers” and the “never drinkers” in either all-cause mortality, cardiovascular mortality, or cardiovascular events. Moreover in the larger UK Biobank dataset, with the former drinkers excluded, they found the same-old/same-old J-shaped dose-response curve. (The same was true of the HSE/SHeSs dataset, but technically it did not reach statistical significance there likely because of its substantially smaller size.) In the UK Biobank dataset, even the “High-level drinkers” had a statistically-significant 29% lower cardiovascular event rate than the teetotalers. The “Medium-level drinkers” had a statistically-significant 31% lower cardiovascular event rate, 37% lower cardiovascular mortality, and 29% lower all-cause mortality. Wow. And Ding’s meta-analysis in Figure 2 shows the typical old “J-curves,” with reduced risk compared to teetotalers out passed 60 alcohol g/day, or an average of 3 glasses of wine or 4 cans of beer a day.
Do you really think that this cite supports your substack post? It’s a seventh dataset.
And you think it is ethical to counsel your secondary-prevention patients against drinking?
Wouldn’t that raise their all-cause mortality by 25-30%?
Sid, you’ve clearly put time and energy into your response, and I recognise that you’ve read widely. But this exchange has moved well beyond reasonable scientific disagreement.
I initially engaged with your comments, despite their offensive tone, accusations, and sarcasm, because I believed readers would benefit from seeing the science discussed openly, even when opinions differ. But it’s now clear that you're not acting in good faith. You’ve repeatedly misrepresented what I’ve written, including your latest suggestion that I cited Ding et al to support the elimination of the J-shaped curve, when I clearly used it to show that even those who accept the J-shape relationship still acknowledge concerns about residual confounding and study design. That was in direct response to your claim that Stockwell has been “discredited.”
What you’ve presented isn’t a scientific rebuttal - it’s a performance. You’ve ignored nuance, reframed questions to suit your narrative, and treated disagreement as a contest to be won. This may feel like a game of gotcha to you, but I take the responsibility of public health writing seriously. I write for readers trying to make better choices, not for people trying to prove they’re the smartest in the room.
You asked whether I think it’s ethical to advise secondary prevention patients against drinking. Your reframing exemplifies the problem here. The central question - the one I addressed in the article - is not whether we forbid alcohol, but whether we should encourage moderate drinking for heart health. And the evidence simply doesn’t support that.
Say I tell a patient a glass of wine each evening might reduce their heart attack risk. What happens if they later develop breast cancer or have a stroke - conditions whose risk is increased even by low levels of alcohol? Have I helped, or harmed?
As I said in the article, and as most public health agencies now state, alcohol is not a health food. It’s a risk to be managed, not a medicine to be prescribed.
I won’t be continuing this conversation. The thread will be closed.
P.S. on mechanisms.
It is well-proven, scientifically, that alcohol, qua alcohol:
- raises HDL levels (i.e. “good” cholesterol levels, which lowers atherosclerosis);
- improves the apolipoprotein profile (which lowers cardiovascular disease risk);
- lowers fibrinogen levels (which reduces heart attack & stroke thromboses);
- increases adiponectin (which is anti-inflammatory, anti-atherogenic, and helps insulin sensitivity);
- decreases HbA1c levels (which reduces diabetes risk);
- decreases stress, and so, cortisol levels (which lowers CVD risk);
- and red wine’s anthocyanins also lower CVD risk.
See, for example, PMID 21343206, which meta-analysed 44 randomized controlled feeding trials on 13 different cardiovascular disease biomarkers:
https://www.bmj.com/content/bmj/342/bmj.d636.full.pdf
Their “Conclusions: Favourable changes in several cardiovascular biomarkers (higher levels of high density lipoprotein cholesterol and adiponectin and lower levels of fibrinogen) provide indirect pathophysiological support for a protective effect of moderate alcohol use on coronary heart disease.”
(You do believe in randomized controlled trials, don’t you Dr. Jones? Or maybe you’d just rather ignore the whole biological literature? Please cite for us all of your RCT evidence that alcohol makes CVD biomarkers worse. We’ll wait.)
You might also google PMID 16129796, 28264492, 11976563, 27705886, 23425242, & 23596141.
So alcohol’s widely-proven beneficial cardiovascular-event effects are exactly what we would expect based on biochemistry. It’s called “science.” To claim otherwise is just spreading medical misinformation.
And, yes, there is also lots of scientific evidence that the anthocyanins in red wine provide an extra biochemical health-boost to the alcohol. The only difference between red wine and white wine is the anthocyanin colorant molecules from the red grape varieties (plus, perhaps some oak aging). These are the same anthocyanin molecules that give color to cherries or blueberries or apple skins.
On red wine specifically, see: PMID 23408240, 37403999, 22205309, 23894543, 39458427, 38847707, 22205309, 22955728, 22999066, 16095600, 20522276, 36289599, 31597344, 23319811, and 26152229.
For anthocyanins, specifically, see PMID 23319811 (myocardial infarction hazard ratio=0.68), 26152229 (all-cause mortality HR=0.66), 21926181, 22810991, 24061071, 10477529, 22914551, 25578927, 24406274, 35082171, 36563462, and 37760077.
Surely, Dr. Jones, you do appreciate biochemistry enough to consider that a drug—alcohol—that significantly positively effects multiple heart-protective biological pathways, could be beneficial when administered at appropriate dosages?
I have consolidated my responses in a single reply to another of your comments.