News Release

International study questions health benefits of moderate drinking

Previous analyses did not account for effects of age and illness that may make abstainer groups appear to have higher death rates than drinkers

Peer-Reviewed Publication

University of California - San Francisco

The majority of studies suggesting that "moderate" drinking helps prevent heart disease may be flawed, according to an international research group.

In a new report, researchers from the U.S., Canada, and Australia analyze 54 studies that linked how much people drink with risk of premature death from all causes, including heart disease. Researchers from the University of Victoria in British Columbia and the University of California, San Francisco led the team.

The researchers investigated a suggestion put forth in the scientific community that many of the studies conducted so far on drinking and premature death made a consistent and serious error by including as "abstainers" people who had cut down or quit drinking due to declining health, frailty, medication use or disability. When such studies show a higher death rate for abstainers than for moderate drinkers, this result may reflect the poor health of some abstainers who recently quit drinking rather than indicating a protective effect for alcohol.

The team found only seven studies that included only long-term non-drinkers in the "abstainers" group. The results of the seven studies showed no reduction in risk of death among the moderate drinkers compared with abstainers. When the researchers combined the data from these studies, they showed that it was possible to perform new analyses that appeared to show a protective effect of moderate drinking--but only when they deliberately included the error of combining long-term abstainers with people who had cut down or quit drinking more recently.

The authors caution that their report, published online in advance of the May 2006 issue of Addiction Research and Theory, has not disproved the notion that light drinking is good for health, as too few error-free studies have been performed. They suggest, however, that the extent to which these benefits actually translate into longer life may have been exaggerated.

"The widely held belief that light or moderate drinking protects against coronary heart disease has had great influence on alcohol policy and clinical advice of doctors to their patients throughout the world," said Tim Stockwell, PhD, of the Centre for Addictions Research at the University of Victoria. "These findings suggest that caution should be exerted in recommending light drinking to abstainers because of the possibility that this result may be more apparent than real."

"We know that older people who are light drinkers are usually healthier than their non-drinking peers," said Kaye Fillmore, PhD, of the UCSF School of Nursing. "Our research suggests light drinking is a sign of good health, not necessarily its cause. Many people reduce their drinking as they get older for a variety of health reasons."

The authors emphasize that there is a need for more well-designed research in the future that assesses people's alcohol intake and abstinence more precisely as their drinking patterns change with age.


The research that the team analyzed consisted of prospective studies, including hundreds of thousands of individuals who were followed over a number of years. Most studies were from North America and Europe, some from the Caribbean and Asia. The majority of the results were published in the 1980s and 1990s, with one study published in 1974 and one as recently as 2004.

The team investigated this body of research using meta-analysis – a method that synthesizes the results from many different studies to determine if their results are in agreement, statistically significant and the degree to which there is variation in the results of the studies.

Light drinking was defined as having two drinks or fewer per day (or about 30mls of pure alcohol per day) and drinking at least once a month. Moderate drinking was 2 to 4 drinks per day. The effects of other factors that influence health and life expectancy were controlled as far as possible in analysis.

Two parallel analyses were carried out: one for 54 studies evaluating death from all causes including heart disease and the other for 35 studies evaluating heart disease deaths specifically. The two analyses showed similar results. Moderate drinking was associated with protection from premature death in the majority of studies -- those containing the "abstainer error." Protection was not found in the few studies without the error, e.g. where moderate drinkers were compared with long-term abstainers.

Previous meta-analyses that combined data from many studies have shown that moderate drinking appeared to have a protective effect against heart disease deaths and against premature death in general. These studies did not address the effect of the "abstainer error." "By ignoring this error, these meta-analyses perpetuated it," Fillmore said.

"In this analysis, every attempt was made to test the proposition of a health benefit more thoroughly than ever before in the past," Fillmore said. "Many other factors may account for the different findings across studies. Fully 57 characteristics of these studies were tested (e.g., smoking, health status, measurement characteristics). None were found to challenge the finding that the 'abstainer' error was responsible for alcohol's apparent protective effect."

Stockwell noted that other diseases and conditions have been found to have a protective effect as a result of the use of alcohol (in contrast to abstaining from alcohol). "It is critical that future research test whether faulty measurement may have contributed to their findings," he said. "Also, it is well to remember that the types of research usually used to find links between lifestyle factors (e.g. diet, exercise, medication use, in addition to alcohol use) and disease later in life have a high potential for error and on their own cannot prove causation."

The authors credit British heart specialist Gerry Shaper, MD, of the Royal Free and University College Medical School in London, England, for first proposing the possibility of an "abstainer error" in the design of prospective studies of the association between alcohol use and heart disease risks.

"This study was performed in the spirit of the tradition that all scientific findings are open to challenge," Fillmore said. "Competing hypotheses, even if unpopular, should be encouraged and tested to be sure that accepted beliefs about health are sound."


Co-authors in the research were William Kerr, PhD, of the Alcohol Research Group, Berkeley, CA; Tanya Chikritzhs, PhD, of the National Drug Research Institute, Perth, Australia, and Alan Bostrom, PhD, of UCSF.

The research was funded principally by the Australian Alcohol Education and Rehabilitation Foundation, an independent agency established by the Australian federal government with funds collected from beer taxes. Seed money for development of the work came from the UCSF School of Nursing and from NordAN in Copenhagen, Denmark.

UCSF is a leading university that consistently defines health care worldwide by conducting advanced biomedical research, educating graduate students in the life sciences, and providing complex patient care.

The University of Victoria is one of Canada's leading universities, known for dedication to excellence in teaching and for innovative and interdisciplinary approaches to research and scholarship.

Reporters who would like to arrange interviews should contact:

University of California, San Francisco
Janet Basu (415) 476-2557
Kaye M. Fillmore, PhD
UCSF School of Nursing
Tel: 510-233-8368

University of Victoria, Victoria, BC
Maria Lironi (250)721-6139
Tim Stockwell, PhD
Centre for Addictions Research of BC
Tel: 250-415-7376

Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of news releases posted to EurekAlert! by contributing institutions or for the use of any information through the EurekAlert system.