Alcohol and the reduction of mortality
Alcohol and the reduction of mortality
Sir Richard Doll, Clinical Trial Service Unit & Epidemiological Studies Unit Radcliffe Infirmary, Oxford
Evidence to suggest that drinking alcohol might have a beneficial effect on the risk of ischaemic heart disease and, to a less extent, on some other major vascular diseases has accumulated slowly over the past six decades . Very little attention was paid to it at first, because of the clear evidence of harm from heavy consumption and the belief that alcohol was generally bad for the heart because large amounts caused cardiomyopathy and contributed to the prevalence of hypertension. Acceptance of a beneficial effect, moreover, complicated public health policy, which, based on the evidence of harm to the drinker and of harm that the drinker might afflict on others, was universally aimed at reducing consumption overall. It was not surprising, therefore, that the accumulating evidence that non-drinkers experienced a higher mortality from ischaemic heart disease than moderate drinkers tended to be dismissed as an artefact, due to the inclusion among non-drinkers of people who had given up alcohol on medical advice, because they had heart disease , or to the habit of drinking alcohol being confounded with some other activity that was really responsible for the beneficial effect (such as, for example, the consumption of a healthier diet or more physical activity) . In fact, the excess among non-drinkers proved to be present in people who had been lifelong non-drinkers and in those who were free of all previous vascular disease when first observed , while confounding with other factors was found to diminish the benefit of alcohol rather than the reverse , . Experiments, moreover, showed that alcohol brought about changes in the constituents of the blood that would be expected to reduce the risk of vascular disease, increasing the high-density lipoproteins , reducing the tendency of platelets to aggregate , decreasing the fibrinogen , and possibly having a beneficial effect on fibrinolysis . The beneficial effect on the risk of vascular disease in the middle-aged and elderly is, moreover, so substantial in developed countries (c.30%) that at low to moderate levels of consumption it outweighs the risk of harm and reduces the total mortality from all causes .
Just what level of consumption is optimal is, however, still far from clear. Two papers based on the personal experience of male medical doctors, whose mortality had been observed over a period of 10 to 13 years after they had reported their drinking habits have given somewhat different results. In a British study5 the minimum mortality from all causes was associated with a consumption of 8-14 drinks a week, was almost equally low with 15-21 drinks a week, and exceeded the mortality of non-drinkers only when the reported consumption exceeded 42 a week. In an American study6, the minimum mortality was associated with the consumption of 2-4 drinks a week and rose almost to the level in non-drinkers with a consumption of 7-13 a week. Several factors may have contributed to the apparently different results. One is a difference in the size of standard drinks which are postulated to contain 8 g ethanol in Britain and 12 g in the US. Others are differences in the adequacy of the standardization for smoking and in the age distribution of the doctors studied, changes in drinking habits in the course of observation, and, to some small extent, possibly chance. Similar differences have been observed in many other studies , but doctors have seemed of special interest, as it might have been hoped that they would report their habits more accurately than the general public who, experience has shown, substantially understate the amount they drink, as judged by figures for the amount sold , . The American authors note that other health professionals have been found to provide reliable reports of alcohol use6, but there has been so much disapproval of the use of alcohol by religious groups in the US that it would not be surprising if American doctors tended to understate their alcohol consumption.
Some conclusions are clear. First, some reduction in total mortality in middle and old age can be expected with consumption of one alcoholic drink a day, and some increase can be expected with the consumption of six or more a day. Secondly, some benefit can also be expected by women, as is shown, for example, by the study of US nurses ; but the benefit is less than in men because alcohol, even in quite small amounts, increases the risk of cancer of the breast . Thirdly, despite the claims that have been made for a specific effect of wine, the totality of the evidence shows that all forms of alcohol have similar effects , as is to be expected from the experimental findings with ethanol. In so far as there has been greater benefit reported when wine is the principal type of alcoholic beverage, this may be because wine drinkers tend to spread their consumption out through the week, rather than concentrating it over one or two days. Fourthly, the balance of risk and benefit will not be the same in all countries and will shift sharply away from benefit at any level, in communities in which vascular disease is rare and the risk of physical violence is high18.
Much, however, remains to be clarified. Beneficial effects may not be limited to the risk of vascular disease, for mortality has been found to be reduced in many studies for a broad group of miscellaneous conditions excluding vascular disease, cancer, and trauma5,6. But the principal question, the optimum level of consumption for the general public, may take many years to decide, owing to changes in an individual's drinking habits over time and the unreliability of selfreported consumption, both of which will distort the effects observed in cohorts like those on which the present estimates have been based.