The objective of a recent swiss study was to estimate the health impact of alcohol consumption in terms of the number of deaths and person-years of life lost (PYLL) in Switzerland.
Three hypothetical situations were compared to the current situation. The study addressed the question of how many deaths and person-years of life lost would be avoided if (1) all individuals were abstainers; (2) all individuals were “low drinkers,” (defined as up to 40 g/d (about 3 ½ drinks) for men and up to 20 g/d (just under 2 drinks) for women); and (3) abstainers remained abstainers but heavier drinkers reduced their intake to “low levels.” Etiological fractions of alcohol-related health consequences by age and gender were applied to the officially reported numbers of deaths to estimate the total number of deaths and PYLL.
Results indicate that, depending upon the hypothetical situation, the estimated number of alcohol-related deaths varied between an overall net protective and an overall net causal effect: in situation (1), there would be 730 additional deaths in reference to the current situation (+1.2% of the total alcohol-related mortality in Switzerland), whereas in situation (2) there would be 3,460 fewer deaths (- 5.5%). Thus, the current situation in Switzerland would be preferable compared to a society in which no one drinks alcohol. A society, however, in which every individual drinks at low levels would be preferable to the current situation. The corresponding figures for PYLL are 23,596 vs. 29,229 (6.4% vs. 7.9% of total PYLL).
The authors conclude that the hypothetical situation decisively impacts on the number of deaths attributable to alcohol. In contrast, it has much less influence on the number of alcohol-related PYLL. Thus, mortality is at least a questionable indicator of alcohol-related health outcomes. The study also indicated that more lives and PYLL were saved in the low-risk situation than in the abstinence situation. They state that public health policies should not concentrate on the promotion of abstinence.
Professor Curtis Ellison comments that there are three important factors that should be considered in interpreting their results: they did not adjust for the pattern of drinking, they attributed “alcohol-related diseases” to low levels of consumption, and the upper limits of alcohol included in the “low-drinking” category were fairly high. Had the authors been able to adjust for pattern of drinking, they would undoubtedly have shown much stronger reductions in death from CHD and other diseases for regular drinking (as week-end bingers, who do not show beneficial health effects, were included in their low drinker group). Further, whereas many studies show some increase in “alcohol-related diseases” for drinkers reporting less than 40 g/d for men and less than 20 g/d for women, these increases are thought to be due to either under-reporting of actual intake or primarily binge drinking, as experimental and biologic studies generally do not show that such low levels of alcohol relate to cirrhosis, alcohol-related cancers, and other such conditions. Finally, up to 40 g/d for men and 20 g/d for women are considerably higher limits than advised in the US (up to 24 g/d for men and up to 12 g/d for women). “These factors suggest to me that regular, moderate intake of 1 to 2 drinks/day would surely lead to greater benefits in reducing mortality than those reported for situation (2) in this study.” states Ellison.
The most important message of this paper is that public health recommendations should focus on encouraging low levels of alcohol intake, not abstinence. As the authors state, “Given its positive effect on overall mortality, clearly regular low-level drinking, not abstention, should be the objective of public health policy in many countries.” Further, they point out that a message of moderation, rather than abstention, may be easier to communicate to heavy drinkers in attempting to get them to reduce their intake.
Source: Gutjahr E, Gmel G. Measuring the impact of some methodological options. Eur J Epidemiol 2005;20: 37-47.