---

title: "Alcohol and Health: The Collapse of the J-Curve" abstract: "For two decades, the 'J-curve' hypothesis — that moderate alcohol consumption is protective against heart disease — shaped dietary guidelines and public perception. New evidence, including reanalysis of the key studies and large Mendelian randomization studies, suggests the apparent protective effect was largely a statistical artifact caused by the 'sick quitter' problem. Any amount of alcohol increases the risk of several cancers, and for most people, the net health effect of even moderate drinking is likely neutral to negative." topic: health author: nonacademicresearch.org Editorial date: 2026-05-09 license: CC-BY-4.0

Alcohol and Health: The Collapse of the J-Curve

Abstract

For two decades, the "J-curve" hypothesis — that moderate alcohol consumption is protective against heart disease — shaped dietary guidelines and public perception. New evidence, including reanalysis of the key studies and large Mendelian randomization studies, suggests the apparent protective effect was largely a statistical artifact caused by the "sick quitter" problem. Any amount of alcohol increases the risk of several cancers, and for most people, the net health effect of even moderate drinking is likely neutral to negative.

Background

The J-curve refers to the observed statistical shape of the relationship between alcohol consumption and mortality: abstainers appeared to have higher mortality than light or moderate drinkers, with risk then rising steeply at higher consumption levels. This pattern, reported in dozens of observational studies from the 1980s onward, was widely interpreted as evidence that moderate drinking — one to two drinks per day — was cardioprotective, reducing coronary heart disease risk through mechanisms like HDL cholesterol elevation.

The problem with this interpretation was methodological. It took a remarkably long time for the field to fully reckon with it.

The Evidence

The "sick quitter" problem invalidates the comparison. The abstainer reference group in alcohol studies is contaminated by "sick quitters" — former drinkers who stopped drinking because of illness. When former drinkers are pooled with lifetime abstainers in the non-drinking comparison group, the comparison group has systematically worse health than moderate drinkers, not because abstaining is harmful, but because it includes people who stopped drinking due to pre-existing disease. Fillmore and colleagues (2006, Addiction Research and Theory) conducted a systematic review specifically examining studies that distinguished lifetime abstainers from former drinkers and found that the apparent protective effect of moderate alcohol largely or entirely disappeared.

Mendelian randomization studies find no protective effect. Mendelian randomization uses genetic variants associated with alcohol metabolism as "natural instruments" — people who carry certain variants of the ALDH2 or ADH1B gene drink less by biology, providing a cleaner test of alcohol's causal effects. Studies using this approach — including a large 2018 analysis in The Lancet by Wood et al. covering 83 prospective studies and 599,912 participants — found no evidence that moderate alcohol consumption was cardioprotective. The 2018 Lancet study concluded that the "safest level of drinking is none."

Alcohol is a Group 1 carcinogen. The International Agency for Research on Cancer classifies alcohol as a definitive human carcinogen. Alcohol is causally linked to cancers of the mouth, pharynx, larynx, esophagus, liver, colorectum, and breast. For breast cancer specifically, risk begins to increase at low consumption levels (one drink per day), and no safe threshold has been identified. The World Health Organization and many national cancer agencies updated guidance in the 2020s to state explicitly that no amount of alcohol is safe from a cancer risk perspective.

Cardiovascular benefits are disputed and may be population-specific. Some researchers maintain that alcohol's effects on HDL cholesterol and platelet aggregation could provide genuine cardiovascular benefits in certain populations. Ronksley et al. (2011, BMJ) found protective effects in meta-analysis specifically for cardiovascular mortality. But these analyses predate the widespread use of Mendelian randomization methods and may reflect the same confounding problems identified in earlier work. If the cardiovascular benefit is real, it appears to be offset at the population level by cancer risk and other harms.

Guidelines have begun to shift. Canada's 2023 Canada's Guidance on Alcohol and Health was the first national guidance document to explicitly characterize low consumption as "low risk" rather than safe, and to state that no amount of alcohol is risk-free. The 2025 US Dietary Guidelines Advisory Committee similarly removed the prior neutral framing of moderate drinking, concluding that the evidence no longer supports characterizing 1–2 drinks per day as health-neutral.

Counterarguments

The French paradox and Mediterranean diet data complicate the picture. Lower cardiovascular mortality in Mediterranean countries, where moderate wine consumption is common alongside other healthy lifestyle factors, has been used to argue for wine's benefits. But this pattern is almost certainly explained by the diet as a whole — the Mediterranean diet report included in this library discusses this evidence in detail.

Individual genetic variation may matter. People with certain genetic profiles may metabolize alcohol differently, and the aggregate population finding of no benefit obscures possible subgroup effects.

Social and psychological benefits are real. Moderate drinking in social contexts is associated with wellbeing, social bonding, and reduced anxiety. These are not trivial. A full accounting of alcohol's health effects must include context — which is one reason blanket prohibition messaging has historically failed.

What We Can Conclude

The J-curve's apparent protective effect of moderate alcohol was largely a methodological artifact. Mendelian randomization studies — the most causally rigorous approach available — find no cardiovascular benefit from moderate drinking that is not explained by confounding. Alcohol is a carcinogen with no established safe threshold for cancer risk. The appropriate public health conclusion is that any benefits of moderate drinking are smaller and less certain than a generation of dietary guidance suggested, and that cancer risk applies even at low consumption levels. This does not mean one drink occasionally causes serious harm, but the claim that moderate drinking is good for you has not survived careful scrutiny.

References

• Fillmore, K. M., Kerr, W. C., Stockwell, T., et al. (2006). Moderate alcohol use and reduced mortality risk: Systematic error in prospective studies and new hypotheses. Addiction Research and Theory, 14(2), 101–132.
• Wood, A. M., Kaptoge, S., Butterworth, A. S., et al. (2018). Risk thresholds for alcohol consumption: Combined analysis of individual-participant data for 599 912 current drinkers in 83 prospective studies. The Lancet, 391(10129), 1513–1523.
• Ronksley, P. E., Brien, S. E., Turner, B. J., Mukamal, K. J., & Ghali, W. A. (2011). Association of alcohol consumption with selected cardiovascular disease outcomes: A systematic review and meta-analysis. BMJ, 342, d671.
• International Agency for Research on Cancer. (2010). Alcohol consumption and ethyl carbamate. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 96.
• Canadian Centre on Substance Use and Addiction. (2023). Canada's guidance on alcohol and health. CCSA.