Current issues of ACP Journal Club are published in Annals of Internal Medicine


Modest alcohol consumption decreased total mortality

ACP J Club. 1995 Mar-April;122:49. doi:10.7326/ACPJC-1995-122-2-049

Source Citation

Doll R, Peto R, Hall E, Wheatley K, Gray R. Mortality in relation to consumption of alcohol: 13 years' observations on male British doctors. BMJ. 1994 Oct 8;309:911-8



To assess the risk for death associated with alcohol consumption among British men.


Cohort study with 13-year follow-up.


Community-based study in the United Kingdom.


In 1978, 12 321 male physicians born between 1900 and 1930 responded to a questionnaire that contained questions on alcohol use.

Assessment of risk factors

Participants were assessed by age and smoking habits and were classified by the number of alcohol units (1 unit = 0.5 pints beer; 3.5 oz. wine; 1.8 oz. sherry or port; or 1 oz. spirit or liquor) they consumed in a week (none, undefined, 1 to 7, 8 to 14, 15 to 21, 22 to 28, 29 to 42, and ≥ 43 units).

Main outcome measures

Causes of death were obtained from death certificates and were grouped into 3 main categories: alcohol-augmented causes (those for which other studies had shown an association with alcohol: cirrhosis; alcoholism; injury; poisoning; and cancer of the liver, mouth, esophagus, larynx, and pharynx), ischemic heart disease, and other known causes.

Main results

Smoking and drinking were associated. Among men who did not drink, the proportion of those who did not smoke was 6 times higher than that for men who consumed ≥ 43 units of alcohol/wk (42% vs 7%). Death from alcohol-augmented causes was positively correlated with the amount of alcohol consumed per week. In participants who died of ischemic heart disease and other known causes, those who reported no alcohol consumption had a higher mortality rate than did those who drank 1 to 14 units of alcohol/wk. When all causes of death were combined and the rates were adjusted for smoking habit, men who consumed 8 to 14 units of alcohol/wk had the lowest mortality.


Modest amounts of alcohol protected against death from ischemic heart disease and against overall mortality. Consumption of > 21 units of alcohol/wk was associated with a progressive increase in mortality.

Sources of funding: Imperial Cancer Research Fund and Medical Research Council.

For article reprint: Professor R. Doll, Imperial Cancer Research Fund Cancer Studies Unit, Nuffield Department of Clinical Medicine, Radcliffe Infirmary, Oxford, OX2 6HE, England, UK. FAX 44-1865-5588-17.


Cigarette smoking caused half the deaths in male physician smokers

We already know that smoking is bad for health and that alcohol can be. What, then, do these 2 studies add to our knowledge?

Doll has been observing the consequences of smoking in a cohort of male British physicians for more than 40 years. The first reports from this study (1) were among the most important health research findings in this century; we know what we know about the risks of smoking largely because of Doll's pioneering work. In this article, Doll and colleagues describe the second 20 years of follow-up. Because the study is based on unusually long and complete follow-up, it adds to our understanding of how the duration of smoking affects risk.

The authors show that risk increases with duration of smoking. If smokers stopped smoking before middle age, their pattern of risk did not differ from that of nonsmokers. The benefits of quitting persisted throughout life; even elderly persons who quit had a lower risk than those who did not. As with other studies, the burden of illness from smoking was enormous. For example, two thirds of the deaths in middle-aged smokers were attributable to smoking. These observations remind us of the powerful effects of smoking on health and the value of quitting at any age.

Smoking did not necessarily cause all the deaths associated with it. A limitation of this cohort study is that smokers may have differed from nonsmokers in other ways that may be related to the risk for death; except for age, these other factors were not considered in the analyses. Also, information about the cause of death recorded in death certificates may have been affected by the attending physician's belief about the risks of smoking—for example, by attributing sudden death to coronary disease or calling a cancer of undetermined origin lung cancer. The findings, however, are consistent with those of other studies, and we should behave as if most of the 25 causes of death that were associated with smoking are at least partly caused by it and could have been avoided if the person had not smoked or had quit. Actual cause of death is, however, of secondary importance because all-cause mortality was strongly correlated with smoking.

As for alcohol, we know from many studies that it is harmful. Alcohol causes life-threatening disease of nearly every body system (liver, nervous system, gastrointestinal tract, heart, and others), accidents, and some cancers. Evidence also suggests, however, that moderate drinking may protect against cardiovascular disease (2), perhaps by raising high-density lipoprotein levels (3), decreasing platelet aggregation, and causing favorable changes in coagulation. This study addresses the net effect, how it relates to the alcohol dose, and how it can be explained.

The authors show, as others have (4-6), that in a "U-shaped" dose-response curve, alcohol consumption is related to death, with lower death rates occurring in men consuming 1 to 2 drinks/d than in abstainers or heavier drinkers. The authors also present evidence that the overall curve is the sum of at least 2 different dose-response curves, one for causes of death believed (from other studies) to be from alcohol and the second for cardiovascular disease. With increased alcohol consumption, the alcohol-related death rate increases, but the number of cardiovascular-related deaths decreases (at lower levels of drinking).

The higher death rates among abstainers observed in this and other studies are controversial. It is an unattractive finding for clinicians who have seen so many patients suffering from alcohol addiction and its complications. The observation is also consistent with findings in some alcohol drinkers who stop drinking when they become ill. Doll and colleagues present evidence against this explanation, showing that the association between alcohol and death was at least as strong in men without previous disease as in those with it.

The findings have subsequently been shown to be generalizable to women and older men but not younger men (7). On the basis of this and other information, however, clinicians probably should behave as if moderate alcohol intake is, on average, not just harmless but beneficial.

Robert H. Fletcher, MD, MSc
Harvard Medical SchoolBoston, Massachusetts, USA


1. Doll R, Hill AB. The mortality of doctors in relation to their smoking habits. A preliminaryreport. BMJ. 1954;1:1451-5.

2. Rimm EB, Giovannucci EL, Willett WC, et al. Prospective study of alcohol consumption and risk of coronary disease in men. Lancet. 1991;338:464-8.

3. Suh I, Shaten BJ, Cutler JA, Kuller LH. Alcohol use and mortality from coronary heart disease: the role of high-density lipoprotein cholesterol. The Multiple Risk Factor Intervention Trial Research Group. Ann Intern Med. 1992;116:881-7.

4. Marmot MG, Rose G, Shipley MJ, Thomas BJ. Alcohol and mortality: a U-shaped curve. Lancet. 1981;1:580-3.

5. Shaper AG, Wannamethee G, Walker M. Alcohol and mortality in British men: explaining the U-shaped curve.Lancet. 1988;2:1267-73.

6. Klatsky AL, Armstrong MA, Friedman GD. Alcohol and mortality. Ann Intern Med. 1992;117:646-54.

7. Thun MJ, Peto R, Lopez AD, et al. Alcohol consumption and mortality among middle-aged and elderly U.S. adults. N Engl J Med 1997;337:1705-14.