Alcohol
Ethanol — the most widely used psychoactive drug
The harms of alcohol are well established and dose-dependent — cancer, hypertension, atrial fibrillation, liver disease, and disrupted sleep. Confidence is high that these are causal. The belief that moderate drinking protects the heart is not supported by the methods best able to test it, and confidence is high that this belief is mistaken. Absolute risk at low intake is small, and the decision is a personal one.
How confident is BioSignal?
Our overall position, and how sure we are of it across each dimension — including where we are not sure at all.
High confidence
The harms of alcohol are well established and dose-dependent — cancer, hypertension, atrial fibrillation, liver disease, and disrupted sleep. Confidence is high that these are causal. The belief that moderate drinking protects the heart is not supported by the methods best able to test it, and confidence is high that this belief is mistaken. Absolute risk at low intake is small, and the decision is a personal one.
Well-supported by consistent, high-quality evidence.
Biological Role
Ethanol has no nutritional role. Its metabolite acetaldehyde is directly genotoxic — it damages DNA — which is the mechanistic basis of the cancer link, and it is not in dispute.
Human Evidence
Enormous. Randomised trials for blood pressure, Mendelian randomisation for causality, and consistent cohort evidence for cancer across multiple sites.
Harm Evidence
Causal links to at least seven cancers, hypertension, atrial fibrillation, and liver disease are well established, and the relationships are dose-dependent.
Cardioprotection Claim
The J-curve is substantially explained by confounding. Genetic (Mendelian randomisation) studies — which are not vulnerable to sick-quitter bias — do not support a protective effect.
Safety Confidence
There is no established threshold below which cancer risk is absent, though absolute risk at low intake is small. Heavy intake carries substantial, well-characterised harm.
Research Activity
Active, and moving — largely in the direction of revising guidance downward as the confounding in older studies has been unpicked.
Where the evidence stands today
How mature the science is, what kinds of evidence exist, and — the part nobody else prints — what is still missing.
8/9
steps proven in humans
Evidence-rich
1 step not demonstrated: Large human RCTs
Guideline / regulatory support
ProvenGroup 1 carcinogen; national guidance tightening.
Clinical outcomes
ProvenCausal links to cancer, hypertension, AF, and liver disease established.
Large human RCTs
Not shownA trial of moderate drinking with hard endpoints was attempted and abandoned.
Small human outcome trials
ProvenRandomised reduction trials lower blood pressure and AF recurrence.
Human safety data
ProvenAmong the best-characterised harm profiles in medicine.
Human biomarker / pharmacology
ProvenDose-dependent effects on blood pressure, liver enzymes, sleep architecture.
Animal
ProvenExtensive.
Cell / in vitro
ProvenDNA damage well characterised.
Mechanistic plausibility
ProvenAcetaldehyde is directly genotoxic.
The bottom line
What we know, what we think, what we don't know — and what would change our mind.
What We Know
Alcohol is a Group 1 carcinogen — the same category as tobacco and asbestos, which describes the strength of the evidence, not the size of the risk. It causally raises blood pressure, causes atrial fibrillation and liver disease, fragments sleep, and worsens sleep apnoea.
What We Think
The 'moderate drinking is good for your heart' story is very likely an artefact. The abstainer groups in those studies contained people who had stopped drinking BECAUSE they were unwell, which made drinkers look healthier by comparison. Genetic studies that sidestep this problem find no protection.
What We Don't Know
Exactly where risk begins for each outcome, and how much of the residual harm at low intake is real versus measurement error. What we can say is that no level has been established as risk-free, particularly for breast cancer.
Active Research
Refining the dose-response at low intakes, understanding individual variation (notably ALDH2 variants, common in East Asian populations, which make drinking substantially more harmful), and evaluating reduction interventions.
What Could Change Our Mind
A randomised trial of moderate drinking with hard cardiovascular endpoints would settle the J-curve. One was attempted and abandoned, and it is unlikely to be repeated — so this may remain a question answered by genetics rather than by trial.
The biggest myth
“Moderate drinking protects the heart”
This is the most consequential misconception in the field, and it deserves a careful explanation rather than a flat contradiction. The J-shaped curve came from observational studies in which the 'non-drinker' comparison group included former drinkers who had quit due to ill health — the sick-quitter effect — plus people who abstain for reasons correlated with poorer health. Mendelian randomisation studies, which use genetic variants affecting alcohol metabolism and are therefore not vulnerable to this bias, find no cardioprotective effect and a linear increase in blood pressure and stroke risk.
Ask BioSignal
Still have a question about alcohol?
Answers are retrieved from this record and the rest of the knowledge graph — never generated.
Why people take alcohol
Popularity is not evidence — but it is not stupid either. This explains the interest on its own terms.
Almost everyone has heard that a glass of red wine is good for the heart, and almost no one has heard why that turned out to be an artefact of how the studies were built. This record exists to explain that carefully and without condescension — and then to leave the decision where it belongs, with the reader.
If you're here because…
Jump straight to the part of the evidence that answers your question.
Approval, safety and regulatory
What it is approved for, who should be careful, what remains unknown — and the limits of what this evidence can tell you.
Regulatory status
Not a medicine. A regulated consumer product
Availability
Widely available
Sport (WADA)
Prohibited in certain sports in competition — confirm against the current list
Known safety profile
Alcohol's harms are dose-dependent and well characterised. The important point for most readers is not the extremes but the middle: harms accumulate at intakes many people consider unremarkable.
Common issues
- Disrupted sleep architecture and worsened sleep apnoea
- Raised blood pressure
- Weight gain (7 kcal/g, and it disinhibits eating)
- Reflux
- Impaired recovery from exercise and reduced muscle protein synthesis
Use caution if
- People who are pregnant or trying to conceive (no safe level established)
- People with liver disease, hepatitis, or MASLD
- People with atrial fibrillation or hypertension
- People with obstructive sleep apnoea (alcohol relaxes the airway)
- People with a personal or family history of alcohol use disorder
- People with ALDH2 variants (flushing response) — substantially raised oesophageal cancer risk
- People taking medications metabolised by the liver, or sedatives
Long-term unknowns
The precise dose-response at low intakes remains debated, and it is the region most people actually live in — which is unfortunate, because it is where the guidance matters most.
Limits of this evidence
Most alcohol evidence is observational, because randomising people to drink is neither practical nor ethical. That is precisely why Mendelian randomisation matters here: it is the best available tool for separating what alcohol DOES from what drinkers ARE, and it is what turned the cardioprotection story around.
Full regulatory and sport detail
- Regulatory approval
- Not a medicine. Alcohol is a regulated consumer product, not a therapeutic agent.
- Approved indication
- None.
- Research chemical
- N/A.
- Sport (WADA)
- Prohibited in competition in certain sports — confirm against the current list.
- Publication note
- Re-confirm national drinking guidance at publication; it has been actively revised downward.
Demand, separated from evidence
Every claim people make about this, counted against what the evidence actually showed.
BioSignal evaluated 10 popular claims about alcohol.
Here is where each one landed — including the claims of harm, where “not established” is reassuring rather than damning.
- Supported by evidence
- 5
- Not established
- 5
Popularity is not evidence. This is simply a count of every claim BioSignal evaluated on this page, sorted by what the evidence actually showed — the full reasoning behind each verdict is in the Evidence Review below.
Supported by evidence
- Alcohol causes cancer
- Alcohol raises blood pressure
- Alcohol causes atrial fibrillation
- Alcohol lowers testosterone
- Alcohol causes fatty liver and liver disease
Not established
- Moderate drinking protects the heart
- Red wine is healthy because of resveratrol
- A nightcap helps you sleep
- There is a safe level of drinking
- Everyone metabolises alcohol the same way
The evidence review
Every claim with the reasoning behind its verdict, the doses actually studied, where scientists agree and disagree, and the questions still open.
What people claim
Every popular claim, with BioSignal’s verdict and how confident we are in it. The verdicts are always visible; open any claim to read the evidence behind it.
Alcohol causes cancerSupportedHigh confidence
Alcohol is classified as a Group 1 carcinogen and is causally linked to cancers of the breast, colon and rectum, oesophagus, liver, mouth, throat, and larynx. Acetaldehyde, its main metabolite, damages DNA directly. Breast cancer risk in particular rises from relatively low levels of intake.
Red wine is healthy because of resveratrolNot establishedHigh confidence
The resveratrol in wine is present in quantities far too small to produce the effects seen in laboratory studies — you would need to drink an impossible amount. Whatever red wine's merits, they are not a reason to drink it for health.
Alcohol raises blood pressureSupportedHigh confidence
Dose-dependent, causal, and — importantly — largely reversible. Randomised reduction trials show blood pressure falls when intake falls. This makes alcohol one of the more actionable contributors to hypertension.
Alcohol causes atrial fibrillationSupportedModerate confidence
Both binge drinking ('holiday heart') and sustained regular intake raise the risk of atrial fibrillation, and reducing intake reduces recurrence in people who already have it.
A nightcap helps you sleepNot establishedHigh confidence
Alcohol is sedating, so it shortens the time to fall asleep — which is why the belief persists and why it feels true. But it then fragments sleep in the second half of the night, suppresses REM, and relaxes the upper airway, making snoring and sleep apnoea worse. It trades sleep quality for sleep speed.
Alcohol lowers testosteroneSupportedModerate confidence
Heavy and chronic intake suppresses testosterone, both directly and via its effects on sleep and the liver. It is one of the reversible causes of low testosterone that should be addressed before considering therapy.
Alcohol causes fatty liver and liver diseaseSupportedHigh confidence
Well established and dose-dependent, from reversible fatty liver through hepatitis to cirrhosis. Alcohol and metabolic (MASLD) drivers frequently coexist and compound one another.
There is a safe level of drinkingNot establishedModerate confidence
For cancer specifically, no threshold below which risk is absent has been established, and risk appears to rise from low intakes. That said, absolute risk at low intake is small, and 'no established safe level' is not the same as 'any amount is dangerous'. Both halves of that sentence matter, and stating only one of them would be a distortion.
Everyone metabolises alcohol the same wayNot establishedModerate confidence
They do not. Variants in the ALDH2 gene — common in East Asian populations — impair the clearance of acetaldehyde, causing flushing and substantially increasing the risk of oesophageal cancer for the same intake. Individual risk genuinely differs.
Doses used in human studies
What was actually given to participants in the research. These are descriptions of studies, not recommendations.
US guidance
- Intervention
- Standard drinks
- Dose
- ≤ 2/day for men, ≤ 1/day for women
- Duration
- Ongoing
- Outcome
- Risk-limiting threshold (not a target)
- Notes
- A US standard drink is ~14 g of pure alcohol. Guidance is a ceiling, not a recommendation to drink
UK guidance
- Intervention
- Units
- Dose
- ≤ 14 units/week, spread over 3+ days
- Duration
- Ongoing
- Outcome
- Low-risk drinking threshold
- Notes
- A UK unit is 8 g of pure alcohol — different from a US standard drink
Canadian guidance (2023)
- Intervention
- Standard drinks
- Dose
- Risk rises from ~3 drinks/week; 'no amount is safe'
- Duration
- Ongoing
- Outcome
- Continuum-of-risk framing
- Notes
- Notably stricter than older guidance — reflecting the reassessment of the cardioprotection evidence
Adults with hypertension or atrial fibrillation
- Intervention
- Reducing intake
- Dose
- Any reduction
- Duration
- Ongoing
- Outcome
- Blood pressure; AF recurrence
- Notes
- Randomised evidence: reduction lowers blood pressure and reduces AF recurrence. This is actionable
Where scientists agree — and don’t
Agreed
- Alcohol is a Group 1 carcinogen, causally linked to at least seven cancers.
- It causally and dose-dependently raises blood pressure, and reduction lowers it.
- It causes atrial fibrillation and liver disease.
- It fragments sleep and worsens obstructive sleep apnoea.
- The observational cardioprotection signal is substantially confounded.
Debated
- Whether any residual cardiovascular benefit exists at very low intakes.
- Exactly where the dose-response for each cancer begins.
- How guidance should be communicated without either alarmism or false reassurance.
Unknown
- Whether a genuinely risk-free level exists for any outcome.
- How much individual genetic variation should change personal guidance.
- The long-term effect of small reductions across a population.
What remains unknown
- Is there any level of intake at which cancer risk is genuinely not elevated?
- Does any cardiovascular benefit survive once sick-quitter and abstainer bias are fully removed?
- How much should ALDH2 status and other genetic variation change individual advice?
- Why has the cardioprotection belief proved so durable even as the evidence has moved?
Questions people actually ask
Isn't a glass of red wine good for my heart?
This is the most common thing people believe about alcohol, and it is worth explaining rather than simply denying — because the belief was reasonable given the evidence available at the time. Studies did consistently show moderate drinkers had less heart disease than non-drinkers. The problem was the comparison group: the 'non-drinkers' included a lot of people who had STOPPED drinking because they were already ill, which made the drinkers look healthier than they were. When researchers use genetic methods that avoid this trap, the protective effect largely vanishes — while blood pressure and stroke risk go up. The resveratrol in red wine, meanwhile, is present in amounts far too small to matter.
Alcohol is a Group 1 carcinogen — the same as asbestos. Is one drink really that dangerous?
No, and this is an important distinction that the classification itself often obscures. Group 1 describes how CERTAIN we are that something causes cancer — not how much cancer it causes. We are as certain about alcohol as we are about asbestos; the magnitude of risk from one drink is far, far smaller. The honest framing is that risk rises with dose, no threshold has been established as safe, and the absolute risk at low intake is small.
Does alcohol help me sleep?
It helps you fall asleep and then damages the sleep itself — which is exactly why the belief survives, because the part you notice is the part that works. Alcohol is a sedative, so you drop off faster. But as it is metabolised it fragments the second half of the night, suppresses REM sleep, and relaxes the muscles of your upper airway, which worsens snoring and obstructive sleep apnoea. You will often wake feeling unrested despite having slept, in duration, quite a lot.
How much is too much?
Guidance differs by country, which tells you something in itself: the US suggests no more than two drinks a day for men and one for women, the UK no more than 14 units a week, and Canada's 2023 guidance is far stricter, describing risk as rising from around three drinks a week. These are ceilings, not targets — no guideline anywhere recommends that a non-drinker start drinking.
I don't want to stop drinking. What's the most useful thing to know?
That the relationship is dose-dependent, which means reducing genuinely helps even if you don't quit — this is not an all-or-nothing decision. Cutting back measurably lowers blood pressure, reduces atrial fibrillation recurrence, improves sleep, and lowers cancer risk. Drinking is a legitimate choice, and BioSignal's job is to make the trade-off legible, not to make it for you.
Practical takeaways
- Alcohol is a Group 1 carcinogen, causally linked to at least seven cancers — including breast cancer, from low intakes.
- The 'good for your heart' belief came from studies whose non-drinker groups included people who quit because they were ill. Genetic studies find no protection.
- It raises blood pressure and causes atrial fibrillation — and both improve when you drink less. That is genuinely actionable.
- A nightcap speeds sleep onset and then wrecks the rest of the night, and it makes sleep apnoea worse.
- Absolute risk at low intake is small. No established safe level is not the same as 'any amount is dangerous', and this is your decision to make.
How it works
The mechanism comes last on purpose. A compelling explanation of how something might work is the easiest part of the story to tell, and the part most likely to outlive the evidence for it.
Ethanol has no nutritional role. Its metabolite acetaldehyde is directly genotoxic — it damages DNA — which is the mechanistic basis of the cancer link, and it is not in dispute.
How we found out
The J-Curve Era
Observational studies consistently found moderate drinkers had less cardiovascular disease than non-drinkers, producing the 'moderate drinking is protective' consensus that shaped guidance and public belief for decades.
The Confounding Is Identified
Researchers recognised that abstainer groups were contaminated with former drinkers who had quit due to illness — the sick-quitter effect — which artificially made drinkers look healthier.
Genetic Methods Change The Answer
Mendelian randomisation studies, using genetic variants that affect alcohol metabolism and are immune to sick-quitter bias, found no cardioprotective effect and a linear increase in blood pressure and stroke risk.
Guidance Tightens
National guidance began moving downward — most notably Canada's 2023 guidance, which describes risk as rising from a few drinks per week — while public belief in cardioprotection has proved considerably more durable than the evidence for it.
References
Verified sources. BioSignal does not print a citation it has not checked.
References for this record are being verified and will be published with the next review. BioSignal does not print citations it has not checked.
Version history
Version 1.0
Initial record (Editorial Expansion, Batch 2), authored to the Creatine benchmark. Calibration: Established maturity, HIGH confidence in the harms AND high confidence that the cardioprotection belief is mistaken. Tone was deliberately held to explanation rather than moralising: the absolute risks at low intake are small, this is a personal choice, and a record that lectures informs nobody. Review cadence: Annually.
Continue exploring
Three places this record leads next.