Intended for healthcare professionals


The cardioprotective effects of moderate alcohol consumption

BMJ 1996; 312 doi: (Published 11 May 1996) Cite this as: BMJ 1996;312:1179
  1. Ian R White
  1. Lecturer in medical statistics Medical Statistics Unit, London School of Hygiene and Tropical Medicine, London WC1E 7HT

    No real evidence exists that wine is better than beer or spirits

    The recent British government report Sensible Drinking1 followed a scientific consensus in recognising the cardioprotective effect of moderate alcohol consumption in men aged over 40 and postmenopausal women. Two points of controversy are the government's consequent effective increase in the sensible drinking levels2 and whether specific alcoholic beverages carry more benefit than others.

    Developed countries with higher wine consumption tend to have lower mortality from coronary heart disease3 4; the association is weaker or absent for beer and spirits. This contrast has led to the hypothesis that wine has a special cardioprotective effect. The finding, however, may be an artefact resulting from higher total alcohol consumption in wine drinking countries than in those where beer and spirits are drunk more often. Comparisons between countries are often misleading—they also suggest that the effect of alcohol consumption on male mortality from coronary heart disease is three times that of smoking3—and should be viewed with great caution.

    The recent paper by Rimm et al provides an overview of 10 cohort studies which related coronary heart disease in individuals to their consumptions of particular beverages.5 The results are remarkably contradictory. For example, consumption of spirits was associated with significantly increased mortality from coronary heart disease in the Copenhagen city heart study and a significantly decreased incidence of coronary heart disease in the health professionals follow up study. This evidence does not point to a greater cardioprotective effect of any specific beverage. Rather, the consistency of the cardioprotective effect of alcohol consumption across populations with different beverage preferences6 suggests that alcohol per se is the major factor. The different socioeconomic or behavioural characteristics of wine, beer, and spirits drinkers in different countries may explain the diversity of these results. An alternative explanation lies in the cultural drinking patterns associated with different beverages—for example, spirits might be consumed more in “binges” in some European countries and more regularly in the United States.5

    The French paradox refers to the finding that French consumption of animal fat is near to the average of other developed countries while French mortality from coronary heart disease is only one third of the average. Comparison between countries seemed to show that the high French consumption of wine could be an explanation,4 but the results of Rimm et al now contradict this. Nevertheless, alcohol consumption per se is an implausible explanation of a national threefold risk reduction since in individuals its effect is only a 25-50% reduction in mortality from coronary heart disease.6

    The recent paper by Hein et al proposed an alternative explanation for the French paradox.7 In Copenhagen the incidence of coronary heart disease was inversely related to alcohol consumption for men in the top fifth of the low density lipoprotein cholesterol distribution (above 5.25 mmol/l) but was only weakly related in others. Thus the incidence of coronary heart disease was about three times higher in non-drinking men with high low density lipoprotein cholesterol concentrations than in all other men. This result can, however, explain only a small part of the lower French mortality from coronary heart disease, since non-drinkers are in general a small minority (10% in the United Kingdom). It may also be a chance finding: many other studies with data on low density lipoprotein cholesterol have related coronary heart disease to alcohol consumption. None of these studies has reported differences in this relation between low density lipoprotein subgroups and this may mean that there were no differences. These studies should be reanalysed to confirm or refute the findings of Hein et al.

    Alcohol consumption is believed to reduce the incidence of coronary heart disease partly by increasing concentrations of high density lipoprotein cholesterol. The study by Paunio et al in this issue (p 1200) further investigates the inverse relation between mortality from coronary heart disease and high density lipoprotein cholesterol concentrations in male Finnish smokers and supports a previous finding of an upturn in the highest group.8 Put differently, men with high density lipoprotein cholesterol concentrations of 1.62-1.75 mmol/l have one third the mortality from coronary heart disease of men in adjoining high density lipoprotein groups, but on the basis of only four deaths. It would be unwise to place much credence on this result without assessments of significance9 in this and other studies.

    The report Sensible Drinking based its recommendations for both men and women on the statement, “The evidence we have studied for men of all ages identifies a band of minimal mortality associated with a weekly consumption of between about 7 and 28 units a week.”1 This summary of the scientific evidence may disproportionately reflect two studies10 11 published in 1994. Other studies,12 a preliminary quantitative overview,13 and a more extensive overview currently in progress suggest that the minimum mortality occurs below 21 units per week for men. This casts serious doubt on the appropriateness of the upwards revision of the sensible limits. Reliable conclusions about alcohol epidemiology require pooling of comparable analyses from all available cohort studies.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    View Abstract