Addressing an Illinois temperance society in 1842, Abraham Lincoln said something about "intoxicating liquor" that probably got a frosty reception. "It is true that... many were greatly injured by it," the future president noted. "But none seemed to think the injury arose from the use of a bad thing but from the abuse of a very good thing."

America has always had trouble deciding whether alcohol is a bad thing or a good thing. Millions who remember Prohibition, when all alcoholic beverages were illegal, now witness a constant stream of advertisements from producers of alcoholic beverages encouraging people to drink. Despite alcohol's popularity today, however, many still consider abstinence a virtue. Certainly heavy drinking and alcoholism deserve deep concern for the terrible toll they take on alcohol abusers and society in general. But worry about the dangers of abuse often leads to emotional denials that alcohol could have any medical benefits. Such denials ignore a growing body of evidence indicating that moderate alcohol intake wards off certain cardiovascular (circulatory system) conditions, most notably heart attacks and ischemic strokes (those caused by blocked blood vessels). A few studies even show protection against dementia, which can be related to cardiovascular problems.

The Alcohol Effect
A DISCUSSION OF moderate drinking requires a working definition of "moderate." Simple definitions of light, moderate or heavy are somewhat arbitrary, but a consensus in the medical literature puts the upper limit for moderate drinking at two standard-size drinks a day [see box on opposite page]. Studies show that drinking above that level can be harmful to overall health, although sex, age and other factors lower and raise the boundary for individuals.

The main medical benefit of reasonable alcohol use seems to be a lowering of the risk for coronary heart disease (CHD), which results from the buildup of atherosclerosis (fatty plaque) in the arteries that bring blood to the heart.

Atherosclerosis restricts blood flow to the heart and can promote the formation of vessel-blocking clots. It can thereby cause angina (chest discomfort resulting from low oxygen levels in the heart muscles), heart attack (the death of heart tissue that occurs when a blood clot or narrowing of the arteries prevents blood from reaching the heart) and death, often without warning. The condition usually starts at a young age but takes decades to blossom into overt CHD. The most common form of heart disease in developed countries, CHD causes about 60 percent of deaths from cardiovascular ills and about 25 percent of all deaths in those nations.

Pathologists uncovered the first clues to the value of alcohol in the early 1900s, noting that the large arteries of people who died of alcoholic liver cirrhosis seemed remarkably "clean"--that is, free of atherosclerosis. One explanatory hypothesis assumed that alcohol was a nebulous solvent, essentially dissolving the buildup in the arteries; another explanation held that heavier drinkers died before their atherosclerosis had a chance to develop. Neither idea truly explained drinkers' unblocked arteries, however.

A more telling hint emerged in the late 1960s, when Gary D. Friedman of the Kaiser Permanente Medical Center in Oakland, Calif., came up with a novel idea: use computers to unearth unknown predictors of heart attacks. The power of computing could first identify healthy people who had risk factors similar to heart attack victims. Such factors include cigarette smoking, high blood pressure, diabetes, elevated levels of low-density-lipoprotein (LDL, or "bad") cholesterol, low levels of high-density-lipoprotein (HDL, or "good") cholesterol, male gender, and a family history of CHD. Friedman then searched for predictors of heart attacks by comparing the patients and the newly found controls in hundreds of ways--for example, their exercise and dietary habits and their respective levels of various blood compounds. The computers spit out a surprising discovery: abstinence from alcohol was associated with a higher risk of heart attack.

Various studies had missed the connection because they neglected to examine alcohol use as a behavior separate from smoking. We now know that because drinkers often also use cigarettes, the negative impact of smoking was masking the beneficial effect of alcohol. In 1974 my Kaiser Permanente colleagues Friedman and Abraham B. Siegelaub and I were the first, to our knowledge, to publish an examination of moderate drinking in the absence of smoking. We saw a clear connection between alcohol consumption and a decreased risk of heart attack.

Since then, dozens of investigations in men and women of several racial groups in various countries have correlated previous alcohol use with current health. These studies have firmly established that nondrinkers develop both fatal and nonfatal CHD more often than do light to moderate drinkers. In 2000 Giovanni Corrao of the University of Milan-Bicocca in Italy, Kari Poikolainen of the Järvenpää Addiction Hospital in Finland and their colleagues combined the results of 28 previously published investigations on the relation between alcohol intake and CHD. In this meta-analysis, they found that the risk of developing CHD went down as the amount of alcohol consumed daily went up from zero to 25 grams. At 25 grams--the amount of alcohol in about two standard drinks--an individual's risk of a major CHD event, either heart attack or death, was 20 percent lower than for someone who did not drink at all.

Recent data about alcohol protecting against death from CHD are even more impressive. At a meeting of the American Heart Association in November 2002 my Kaiser Permanente colleagues Friedman, Mary Anne Armstrong and Harald Kipp and I discussed an updated analysis of 128,934 patients who had checkups between 1978 and 1985, with 16,539 of them dying between 1978 and 1998. CHD was responsible for 3,001 of those deaths. We discovered that those who had one or two alcoholic drinks a day had a 32 percent lower risk of dying from CHD than abstainers did.

The possible mechanisms by which alcohol has such an apparently profound effect on cardiovascular health primarily involve cholesterol levels and blood clotting. Blood lipids, or fats, play a central role in CHD. Numerous studies show that moderate drinkers have 10 to 20 percent higher levels of heart-protecting HDL cholesterol. And people with higher HDL levels, also known to be increased by exercise and some medications, have a lower risk of CHD.

That lower risk stems from HDL's ability to usher LDL cholesterol back to the liver for recycling or elimination, among other effects. Less cholesterol then builds up in the walls of blood vessels, and so less atherosclerotic plaque forms. Alcohol has a greater influence on a different HDL subspecies (HDL3) than on the type increased by exercise (HDL2), although both types are protective. (The biochemical pathways in the liver that could account for alcohol's ability to raise HDL levels remain incompletely known; it is thought that alcohol affects liver enzymes involved in the production of HDL.) Three separate analyses aimed at determining specific contributions of alcohol all suggest that the higher HDL levels of drinkers are responsible for about half of the lowered CHD risk.

Alcohol may also disrupt the complex biochemical cascade behind blood clotting, which can cause heart attacks when it occurs inappropriately, such as over atherosclerotic regions in coronary arteries. Blood platelets, cellular components of clots, may become less "sticky" in the presence of alcohol and therefore less prone to clumping, although data on this question remain ambiguous. A 1984 study by Raffaele Landolfiand Manfred Steiner of Brown University's Memorial Hospital revealed that alcohol intake increases the level of prostacyclin, which interferes with clotting, relative to the level of thromboxane, which promotes clotting. Walter E. Laug of the University of Southern California Keck School of Medicine showed that alcohol raises levels of plasminogen activator, a clot-dissolving enzyme. Finally, several studies suggest that alcohol lowers levels of another promoter of blood clots, fibrinogen.

Overall, alcohol's anticlotting capacity is not as well established as its HDL effect, and some effects, such as platelet clumping, may be reversed by heavy or binge drinking. Nevertheless, anticlotting appears to have a role in the lower risk for heart attacks enjoyed by moderate drinkers. In addition, studies have shown a beneficial effect on CHD risk in people who have far fewer than two drinks a day--say, three or four drinks a week. Anticlotting could be a major factor in the protection accorded by alcohol in these small amounts, which seem insufficient to affect HDL levels greatly.

Although alcohol reduces heart disease risk mainly by raising HDL levels and reducing clotting, it probably acts in other ways as well. Moderate drinking may lessen CHD risk indirectly by decreasing the risk of type 2 (adult-onset) diabetes, which is a powerful predictor of CHD. This benefit appears to be related to enhanced insulin sensitivity, which promotes proper glucose usage. (Heavy drinking, however, has been connected to higher blood glucose levels, a marker for future diabetes.) Evidence is also growing that inflammation contributes to CHD, and alcohol's anti-CHD power may be related to an anti-inflammatory action on the endothelial tissue that lines blood vessels.

Before accepting alcohol's benefits, an epidemiologist attempts to locate hidden factors possibly at work. For instance, could lifelong abstainers differ from drinkers in psychological traits, dietary habits, physical exercise habits or other ways that might account for their higher CHD risk without the need to invoke the absence of alcohol? Were such traits to explain away alcohol's apparent protection, they would need to be present in both sexes, various countries and several racial groups. Considering that no such traits have been identified, the simpler and more plausible explanation is that light to moderate alcohol drinking does indeed enhance cardiovascular health.

In fact, the available evidence satisfies most standard epidemiological criteria for establishing a causal relation. The numerous studies examining light and moderate alcohol intake and health reach consistent conclusions. The prospective studies that exist have the correct temporal sequence--that is, individuals' habits of interest are identified, after which their health is monitored over the long term, and alcohol users have different health profiles than nondrinkers do. The positives associated with alcohol can be attributed to biologically plausible mechanisms. Alcohol offers specific enhancement of cardiovascular health, not general protection against all illness. And alcohol's effect can be identified independent of known "confounders," other alcohol-related factors that could be responsible for a subject's cardiovascular condition.

The 30 percent reduction in risk is, perhaps surprisingly to some, less convincing evidence than the arguments above, because a strong unknown confounder could still account for the connection. To take an extreme example, consider a hypothetical set of genes that confers on the possessor 60 percent less CHD risk and causes a strong predisposition toward liking moderate amounts of alcohol. The independent consequences of the genes could appear causally linked. In fact, however, no such confounder is known or likely.

Because heavy drinking is not more protective than lighter drinking, this absence of a clear dose-response relation is also a weakness. Nevertheless, the collected data make a strong case for the cardiac benefits of controlled drinking. I should note, however, that the kind of study considered to be the gold standard in human research--a prospective randomized blinded clinical trial--has not yet been done. Such a study might, for example, engage a large pool of nondrinkers, half of whom, chosen at random and without the knowledge of the researchers, would commence a moderate drinking regimen, while the other half remained abstainers. The two groups would be followed for years in a search for eventual differences in cardiovascular disease and heart-related deaths.

To Drink or Not to Drink
MOST PEOPLE DRINK for reasons other than alcohol's health benefits, and many of them are already using alcohol in amounts that appear to promote cardiovascular health. But the accumulated research on alcohol's positive effects presents a challenge to physicians. On one hand, mild to moderate drinking seems better for heart health than abstinence for select people. On the other hand, heavy drinking is clearly dangerous. It can contribute to noncardiovascular conditions such as liver cirrhosis, pancreatitis, certain cancers and degenerative neurological disorders, and it plays a part in great numbers of accidents, homicides and suicides, as well as in fetal alcohol syndrome. (No conclusive evidence links light to moderate drinking to any of these problems, but underreporting of drinking amount by some heavy drinkers clouds the issue for several conditions.)

Heavy drinking also contributes to cardiovascular disorders. Too much alcohol raises the risk of alcoholic cardiomyopathy, in which the heart muscle becomes too weak to pump efficiently; high blood pressure (itself a risk factor for CHD, stroke, heart failure and kidney failure); and hemorrhagic stroke, in which blood vessels rupture in or on the surface of the brain. Alcohol overindulgence is also related to "holiday heart syndrome," an electrical signal disturbance that disrupts the heart rhythm. The name refers to its increased frequency around particular holidays during which people engage in binge drinking.

Heart failure is a common consequence of various cardiovascular diseases. Usually defined as the inability of the heart to meet the blood supply needs of the body, heart failure is associated with CHD about 60 percent of the time and with other conditions (such as high blood pressure, heart valve disease, and other heart muscle diseases) in the other 40 percent. A recent report from our group showed that heart failure associated with CHD was substantially less likely in alcohol drinkers of any amount, whereas heart failure associated with other cardiac problems was unrelated to moderate drinking but more likely to occur in heavy alcohol drinkers.

Given the potential dangers of alcohol, how can individuals and their physicians make the decision as to whether to include alcoholic beverages in their lives and, if so, in what amounts? The ability to predict accurately an individual's risk of a drinking problem would be a great boon; the least disputed possible consequence of moderate drinking is problem drinking. Individual risk can be approximated using family and personal histories of alcohol-related problems or conditions, such as liver disease or, of course, alcoholism. Even when known factors are taken into account, however, unpredictable events late in life may result in deleterious drinking changes.

Exactly because of these dangers, public health concerns about alcohol until recently have been appropriately focused solely on the reduction of the terrible social and medical consequences of heavy drinking. And the correlation between total alcohol consumption in society and alcohol-related problems has been used to justify pushes for abstinence. Ultimately, however, a more complex message is necessary. Merely recommending abstinence is inappropriate health advice to people such as established light drinkers at high risk of CHD and at low risk of alcohol-related problems--which describes a large proportion of the population. Of course, the most important steps for this group are proper diet and exercise; effective treatment of obesity, diabetes, high blood pressure and high cholesterol; and avoidance of tobacco. But there is a place on that list of beneficial activities for light drinking. Most light to moderate drinkers are already imbibing the optimal amount of alcohol for cardiovascular benefit, and they should continue doing what they are doing.

Abstainers should never be indiscriminately advised to drink for health; most have excellent reasons for not drinking. Yet there are exceptions. One case is the person with CHD who "goes clean"--quits smoking, switches to a spartan diet, starts exercising and, with good intentions, gives up the habit of a nightly bottle of beer or glass of wine. This self-imposed prohibition should be repealed. In addition, a number of infrequent drinkers might think about increasing their alcohol intake to one standard drink daily, especially men older than 40 and women older than 50 at high risk of CHD and low risk of alcohol-related problems. But women also have to consider one possible drawback of alcohol: several studies link heavy drinking--and a few even link light drinking--to an increased risk of breast cancer, a less common condition than heart disease in postmenopausal women but certainly quite a serious one. For young women, who are generally at low short-term risk of CHD and therefore may not benefit greatly from alcohol's positive cardiovascular effects, this possible breast cancer link looms larger in estimating the overall risks and benefits of alcohol. Many public health officials recommend that women limit their intake to one drink a day.

The only clear-cut message regarding alcohol and health, then, is that all heavy drinkers should reduce or abstain, as should anyone with a special risk related to alcohol, such as a family or personal history of alcoholism or preexisting liver disease. Beyond that, however, the potential risks and benefits of alcohol are best evaluated on a case-by-case basis. Cardiovascular surgeon Roger R. Ecker and I constructed an algorithm that can help health practitioners and their patients decide how much--if any--alcohol is right for a given individual [see box on opposite page].

In short, health professionals should provide balanced, objective guidelines regarding their patients' use of alcohol, and such advice needs to be tailored to each person. I believe that it is possible to define a clear, safe limit for alcohol consumption that would offer a probable benefit to a select segment of the population. The ancient Greeks urged "moderation in all things." Three decades of research shows that this adage is particularly appropriate when it comes to alcohol.

ARTHUR L. KLATSKY is a senior consultant in cardiology and an adjunct investigator at the division of research at the Kaiser Permanente Medical Center in Oakland, Calif. A graduate of Harvard Medical School, he headed the medical center's division of cardiology from 1978 to 1994 and directed its coronary care unit from 1968 to 1990. Since 1977 he has been principal investigator of a series of studies of the link between drinking alcoholic beverages and health. His 1974 Annals of Internal Medicine article [see More to Explore, on page 29] was the first published epidemiological report of an inverse relation between alcohol drinking and coronary disease; it was cited in 1995 by the National Institute on Alcohol Abuse and Alcoholism as one of 16 seminal articles in alcohol research. His most recent honor was the Morris Collen Lifetime Achievement Research Award by Kaiser Permanente in 2004. Klatsky has completed six marathons and in 1990 climbed Mount Kilimanjaro.