Red Red Wine, Stay Close to Me
A 10,000-Year Experiment
Somewhere around 6000 BCE, in the hills of what is now the Republic of Georgia, someone figured out that crushed grapes left to sit long enough produced something extraordinary. Not just a beverage, but a companion to meals, a staple of religious ritual, a social lubricant, and — if the ancients were to be believed — a medicine. Hippocrates prescribed wine for fevers, digestive complaints, and wound care. The Romans carried it across their empire. Monasteries perfected it. And for millennia, moderate wine consumption was woven into the daily life and dietary fabric of cultures that, by many measures, produced some of the longest-lived and healthiest populations in human history.
Then, in 2018, The Lancet published a study from the Global Burden of Disease Alcohol Collaborators that declared: “The level of alcohol consumption that minimises health loss is zero.”
Not moderation. Not a glass with dinner. Zero.
That single sentence has shaped public health messaging, guidelines, and clinical conversations ever since. And it deserves to be examined carefully — because the story is far more nuanced, far more interesting, and far more scientifically honest than a bumper sticker.
The “Not a Drop” Doctrine: Where It Came From
The 2018 Global Burden of Disease (GBD) alcohol study was enormous — 694 data sources on alcohol consumption and 592 prospective and retrospective studies on 23 health outcomes. Its conclusion that “no safe level” of alcohol exists generated worldwide headlines and became the foundation for increasingly restrictive guidance from the World Health Organization and several national health authorities.
But the study itself acknowledged something important that the headlines did not: moderate alcohol consumption was associated with a lower risk of ischemic heart disease and diabetes. The protective signal was real in the data. The authors simply argued that the harms from cancer and other outcomes, when summed across all 23 conditions, outweighed those cardiovascular benefits.
This matters. The difference between “alcohol at any dose increases your net health risk when we sum cancers, injuries, tuberculosis, drowning, and liver disease together across the global population” and “a glass of wine with dinner harms your heart” is vast. But these two very different statements have been conflated in public messaging to the point where patients tell me they feel guilty about a single glass of Barolo with Sunday dinner.
That conflation is not supported by the evidence.
The J-Curve: Reports of Its Death Have Been Greatly Exaggerated
The “J-curve” — the observation that moderate drinkers appear to have lower cardiovascular risk than both heavy drinkers and abstainers — has been documented in epidemiological studies for over four decades. It has survived repeated attempts to kill it off.
The most comprehensive recent review is by Carr, Espinosa Dice, Gmel, and colleagues in Addiction (2026), which combined a scoping review of 56 meta-analyses of cohort studies with a systematic review of 20 Mendelian randomization studies. Their findings:
Critics of the J-curve have raised legitimate concerns. The “sick quitter” hypothesis argues that the abstainer reference group includes former heavy drinkers who quit due to illness, artificially inflating their mortality. The “healthy user bias” suggests that moderate drinkers tend to be healthier overall. These are real methodological issues.
But they have been addressed. Multiple studies have excluded former drinkers and sick individuals from the reference group, and the J-curve persists. A 2013 Australian review in the Medical Journal of Australia concluded that corrections for possible misclassification have been attempted in more recent studies, and none support misclassification as the sole explanation for the J-shaped curve.
The Mendelian Randomization Challenge
In fairness, Mendelian randomization (MR) studies — which use genetic variants as instruments to approximate the effect of a randomized trial — have complicated the picture. Most MR studies examining hard cardiovascular endpoints have reported null associations, meaning no clear protective effect of moderate drinking.
However, MR studies of alcohol have a known Achilles’ heel: the primary genetic instrument, the ADH1B rs1229984 variant, affects how rapidly ethanol is metabolized and is highly pleiotropic — meaning it affects many things besides alcohol intake. This makes it a problematic instrument for causal inference. MR studies also cannot model drinking pattern (wine with dinner versus binge-drinking on weekends), dose timing, or beverage type — all of which may matter enormously.
So Where Does the Curve Actually Turn?
If you ask where the J-curve’s protective effect ends and harm begins, the answer depends entirely on which outcome you measure — and this distinction is critical to understanding why guidelines say what they say.
For ischemic heart disease specifically, the nadir of the J-curve sits at roughly 1–2 drinks per day, and the crossover to net cardiovascular harm does not occur until well above that — older meta-analyses placed it around 4–5 drinks per day. For myocardial infarction, the relationship appears to be log-linear and inverse — meaning higher consumption continues to be associated with lower MI risk even at levels that harm other systems.
But for all-cause mortality — which includes cancer, liver disease, and accidents — the picture is very different. The landmark Wood et al. analysis in The Lancet (2018), pooling 83 prospective studies and nearly 600,000 current drinkers, found that the lowest-risk threshold for all-cause mortality was approximately 100 grams of ethanol per week — about 7 U.S. standard drinks, or roughly one per day. Above that, risk climbed steadily. Their conclusion was explicit: “These data support limits for alcohol consumption that are lower than those recommended in most current guidelines.”
Not All Drinks Are Created Equal: Wine, Beer, and Spirits
One of the most critical and underappreciated problems in alcohol research is the lumping together of all alcoholic beverages as if they were interchangeable. They are not — biochemically, culturally, or in the way they are consumed.
A systematic review in Alcohol and Alcoholism (Arranz et al., 2012) examined human studies separating wine, beer, and spirits, and concluded that wine and beer — especially red wine — appear to confer greater cardiovascular protection than spirits, attributable in part to their polyphenol content. The Copenhagen City Heart Study, which followed over 13,000 individuals for 12 years, found that moderate wine intake, but not beer or spirits, was associated with lower cardiovascular and all-cause mortality.
But Here’s the Problem: Wine Drinkers Are Different People
This is where the confounders become devastating, and intellectual honesty demands we acknowledge them.
In the Copenhagen cohort, wine consumers were better educated and wealthier than beer and spirits drinkers. U.S. survey data consistently show that adults with college degrees and higher income prefer wine, while those with lower education and income gravitate toward beer and spirits. MR studies confirm this is not just correlation: higher educational attainment is causally linked to greater red wine consumption.
Wine drinkers also tend to eat differently. A recent dietary study found that beer, cider, and spirits are more often consumed outside of meals and are linked to lower overall diet quality, while wine is typically consumed with food as part of a more balanced dietary pattern.
Does this mean wine has no benefit beyond confounding? Not necessarily. But it means we should be far more cautious about the claims than we typically are. And it means the blanket statement that “no amount of any alcohol is safe” carries its own misleading assumption: that a glass of Barolo and a bottle of grain alcohol are biologically equivalent.
The French Paradox, Revisited
On November 17, 1991, CBS correspondent Morley Safer held up a glass of red wine on 60 Minutes and introduced America to the “French Paradox” — the observation that the French, despite eating a diet rich in saturated fat, had remarkably low rates of coronary heart disease. French cardiovascular researcher Serge Renaud attributed this largely to regular red wine consumption.
Red wine sales in U.S. supermarkets rose 44% in the four weeks following the broadcast. Some retailers began marketing wine as “health food.”
The science behind the paradox was partly correct and partly naive. Renaud published in The Lancet in 1992 linking moderate red wine to reduced platelet aggregation. But the premise itself — that saturated fat causes heart disease and the French were defying it — has been undermined by subsequent evidence. A 2009 systematic review found insufficient evidence for a causal link between saturated fat and coronary heart disease, making the “paradox” less paradoxical.
Still, the French Paradox planted an important idea: that context matters. The way the French consume wine — modest amounts, daily, with meals, as part of a broader dietary and social pattern — is fundamentally different from binge drinking. That distinction has proven durable even as the saturated fat hypothesis has weakened.
HDL: The Red Herring in the Glass
One of the most frequently cited mechanisms for alcohol’s supposed cardiovascular benefit is its ability to raise HDL cholesterol. And indeed, alcohol does raise HDL-C — reliably and robustly. The mechanism involves reduced activity of cholesteryl ester transfer protein (CETP) and increased phospholipid transfer protein (PLTP) activity, leading to accumulation of larger, cholesterol-enriched HDL2 particles.
For many years, this was considered a good thing. Higher HDL-C was “good cholesterol,” and anything that raised it should be protective. This reasoning had a fatal flaw.
As detailed in our HDL article, HDL-C as a biomarker tells us nothing about HDL functionality. The history of CETP inhibitors is the most dramatic proof: torcetrapib raised HDL-C by 72% and increased death — largely due to off-target effects including blood pressure elevation, aldosterone stimulation, and electrolyte disturbances that had nothing to do with HDL. Dalcetrapib raised HDL-C by 40% and showed zero cardiovascular benefit — halted for futility. Evacetrapib raised it a whopping 133% with no improvement in outcomes, again stopped for futility. In each case, the HDL-C number moved dramatically upward while the clinical needle did not budge — because raising the cholesterol cargo in the particle did nothing to improve its function. Niacin told the same story from a different angle: it raised HDL-C beautifully in the AIM-HIGH trial and was halted for futility, then the massive HPS2-THRIVE trial essentially killed it due to harm.
Dr. John Kastelein, the lipidologist who has arguably thought more deeply about CETP than anyone alive, has been vocal about the distinction between raising HDL-C and improving HDL function. In the context of alcohol, the concern is that ethanol-mediated CETP inhibition may produce HDL particles that are cholesterol-overloaded but functionally impaired — what some have called “dysfunctional HDL.”
The evidence on this specific question is mixed. One study of heavy drinkers found that their HDL2 particles actually enhanced cholesterol efflux from macrophages, suggesting preserved function. But chronic heavy alcohol feeding in animal models reduced HDL’s efflux capacity. And the ELSA-Brasil cohort warned that the association between excessive alcohol consumption and very high HDL-C levels may represent a “false signal” of cardiovascular protection.
Resveratrol: The Molecule That Couldn’t
No discussion of red wine and health is complete without addressing resveratrol, the polyphenol that launched a thousand supplement bottles. Found in grape skins, red wine, and certain berries, resveratrol has shown impressive effects in petri dishes and mice: anti-inflammatory, antioxidant, activator of sirtuins (longevity-associated proteins), and potential cardioprotectant.
The problem is getting it from the lab to the human body.
A landmark review of human intervention trials (Berman et al., Nutrients, 2017) documented the core issue: resveratrol has high absorption but extremely low bioavailability. It is rapidly metabolized by the gut and liver, with the majority excreted in urine within four hours. Plasma concentrations achieved through supplementation are orders of magnitude lower than the concentrations used in the in vitro experiments that generated the initial excitement.
As for clinical outcomes: a Tufts University analysis of a cohort study found no protective benefit associated with higher resveratrol levels — and in fact, those with the highest levels were at slightly greater risk. No benefit was seen against cancer or markers of inflammation, either. A systematic review concluded that the “magic bullet” hypothesis was challenged by clinical trials showing resveratrol does not substantially influence health status and mortality risk.
This does not mean polyphenols are irrelevant. Red wine contains hundreds of bioactive compounds — flavonoids, anthocyanins, tannins, quercetin — that may act synergistically in ways we do not yet understand. The error was isolating one molecule and declaring it the explanation. Biology is rarely that simple.
Blue Zones: When Centenarians Raise a Glass
Dan Buettner’s Blue Zones research identified five regions with unusually high concentrations of centenarians: Sardinia (Italy), Ikaria (Greece), Okinawa (Japan), Nicoya Peninsula (Costa Rica), and Loma Linda, California (Seventh-day Adventists). In four of these five populations, moderate, regular alcohol consumption is a consistent feature of daily life.
In Sardinia, the traditional drink is Cannonau, a red wine reported to contain two to three times the concentration of artery-friendly flavonoids found in other red wines. In Ikaria, survey data suggest over 80% of elderly residents drank red wine daily throughout their adult lives. In Okinawa, moderate intake of awamori, a traditional rice-based spirit, accompanies communal gatherings. Only the Seventh-day Adventists of Loma Linda abstain entirely.
Buettner has noted that in all five Blue Zones, “they drink a little bit every day, and it adds to their quality of life.”
Beer Potomania and Acetaldehyde: Different Poisons, Different Pathways
When patients ask me whether all forms of alcohol are equally harmful, I think it’s important to distinguish between the universal toxin and the beverage-specific hazards.
Acetaldehyde: The Universal Problem
All ethanol — regardless of source — is metabolized by the same two-step pathway: alcohol dehydrogenase (ADH) converts ethanol to acetaldehyde, and aldehyde dehydrogenase (ALDH2) clears acetaldehyde to harmless acetate. Acetaldehyde is the villain in this story. It is toxic, mutagenic, and classified as a Group 1 carcinogen — the same category as tobacco smoke and asbestos.
Approximately 40% of East Asian individuals carry the ALDH2*2 allele, which produces a low-activity form of the enzyme. When these individuals drink, acetaldehyde accumulates rapidly — causing the characteristic facial flushing but also conferring a significantly increased risk of esophageal and upper GI cancers. This is one of the clearest examples of how genetics modifies alcohol risk.
The key point: acetaldehyde production is an unavoidable consequence of ethanol metabolism. Wine, beer, spirits — all produce it. This is the primary mechanism behind alcohol’s cancer risk, and it applies regardless of beverage type.
Beer Potomania: A Peculiar Syndrome
Beer potomania is an entirely different beast. It is a form of dilutional hyponatremia — dangerously low sodium levels — that occurs in chronic heavy beer drinkers who consume large volumes while eating very little. Beer is mostly water with negligible solute (minimal sodium and protein), so massive intake overwhelms the kidney’s ability to excrete free water. The resulting hyponatremia can be severe and, paradoxically, dangerous to correct rapidly due to the risk of osmotic demyelination syndrome.
Beer potomania is unique to beer — you don’t see it with wine or spirits, because people don’t typically consume those in the same massive volumes. It is rare, and it occurs in the setting of chronic heavy use combined with malnutrition. But it illustrates an important principle: the vehicle matters, not just the ethanol.
Hangovers: Does the Morning After Mean Anything?
Anyone who has ever compared the aftermath of a bourbon bender to a vodka evening knows intuitively that not all hangovers are created equal. Science confirms the intuition, at least partially.
In a well-designed controlled crossover study, Rohsenow and colleagues (2010) gave 95 heavy drinkers either bourbon or vodka on separate nights, matched for alcohol content, with a placebo soda on a third night. Bourbon produced significantly worse hangovers. The explanation: bourbon contains 37 times the congeners found in vodka — compounds like acetaldehyde, acetone, tannins, and furfural produced during fermentation and aging.
Red wines, similarly, contain more congeners than white wines. Dark spirits (bourbon, scotch, brandy) contain more than clear spirits (vodka, gin).
But here’s the important finding: cognitive impairment the next day was identical regardless of beverage type. Both bourbon and vodka drinkers performed equally poorly on tasks the morning after. The hangover felt worse with bourbon, but the functional damage was the same.
Stanley Tucci, Sicily, and the Italian Liter
While watching the Sicily episode of Stanley Tucci’s CNN series Searching for Italy, I heard him remark that the Italian government once recommended one liter of wine per day — not to support the wine industry, but for health promotion.
I tried to verify this, and the honest answer is: the specific “one liter” government recommendation is difficult to trace to an official document. Italy’s first formal dietary guidelines were published in 1986, and subsequent editions in 1997 and 2003 have progressively lowered alcohol recommendations. Current Italian guidance, consistent with the European Union framework, suggests no more than two alcohol units per day for men and one for women.
What Tucci likely captured is an older cultural reality rather than an official policy. In post-war rural Italy, wine was a daily staple — consumed with every meal, considered part of nutrition, and provided to workers as a caloric supplement. The idea that wine in generous quantity was normal and healthful was simply the prevailing cultural assumption, one that predated formal dietary guidance by centuries.
Whether or not there was ever an official “one liter” recommendation, the broader point stands: a population that consumed wine daily in generous quantities for generations did not collapse under the weight of an alcohol epidemic. This doesn’t mean a liter a day is safe — modern evidence clearly suggests otherwise. But it does challenge the absolutist position that any alcohol in any amount is incompatible with health.
Apo A1 Milano: When Very Low HDL Is Protective
Here is an irony worth sitting with. In a small Italian town called Limone sul Garda, a genetic variant called Apo A1 Milano produces HDL-C levels in the teens — levels that would terrify any clinician looking at a standard lipid panel. Yet this population has one of the highest concentrations of centenarians in Italy and remarkably low rates of cardiovascular disease. Their HDL is so efficient at reverse cholesterol transport that the particles don’t need to carry much cholesterol.
Meanwhile, we have patients with HDL-C levels of 120 mg/dL whose HDL is functionally impaired — overloaded with cholesterol, unable to efficiently return it to the liver. These patients often have significant atherosclerosis despite numbers that look “great” on paper.
The alcohol-HDL connection fits neatly into this framework. Alcohol raises the number on the lab report. Whether it improves the function of the particle is a different question — and the answer appears to depend on dose, pattern, and individual genetics.
So What Do We Actually Know?
The Bottom Line
What the evidence supports:
- Moderate alcohol consumption — particularly wine, and particularly with meals — is consistently associated with lower cardiovascular risk in observational studies. The J-curve for ischemic heart disease is robust and has survived repeated attempts at methodological debunking.
- The 2018 Lancet GBD study’s “no safe level” conclusion was driven by the sum of all health outcomes, not by a finding that moderate alcohol harms the heart. The cardiovascular signal in their own data was protective.
- Wine drinkers appear healthier than beer or spirits drinkers, but much of this difference is confounded by socioeconomic status, education, diet quality, and drinking patterns.
- HDL-C elevation from alcohol is a poor argument for benefit, given what we now know about HDL functionality from the CETP inhibitor era.
- Resveratrol is not the explanation. Bioavailability is too low and human outcomes data are absent.
- Blue Zone populations that drink moderately and regularly with meals are among the longest-lived on earth, though this correlation does not prove causation.
- Acetaldehyde is a universal carcinogenic byproduct of all ethanol metabolism. This risk is real and dose-dependent.
What the evidence does not support:
- The absolutist position that “no amount of alcohol is safe” as applied to individual cardiovascular risk. This was a population-level modeling conclusion, not a clinical recommendation.
- The idea that doctors should advise non-drinkers to start drinking for health benefits. There is no evidence to support this, and we should not recommend it.
- The notion that all alcoholic beverages are biologically equivalent. They are not — in composition, in how they are consumed, in the populations who consume them, or in the confounders they carry.
What we should tell patients:
- If you don’t drink, there is no reason to start for cardiovascular benefit.
- If you drink moderately — especially wine, with meals, without bingeing — the current evidence does not suggest you are harming your heart, and you should not feel guilty about it.
- Heavy drinking is unambiguously harmful. The dose makes the poison.
- Beverage type, drinking pattern, food context, and individual genetics all matter. A blanket “not a drop” policy ignores a century of epidemiology and millennia of human experience.
- Talk to your doctor. Your risk calculus depends on your personal and family history, your medications, your cancer risk, and a dozen other variables that a headline cannot capture.
Questions to Ask Your Doctor
- Given my personal and family health history, is moderate alcohol consumption appropriate for me?
- Am I at increased risk for cancers where alcohol is a known risk factor (breast, esophageal, liver, colorectal)?
- Do any of my medications interact with alcohol?
- How does my HDL-C level factor into my overall risk — and should I worry about whether it’s functional?
- What does “moderate” actually mean for me — is it one drink? Two? With food?
— CardioAdvocate.com
