• Alcohol consumption;
  • drinking patterns;
  • IHD mortality

The growing awareness that alcohol may be both good and bad for the heart has become a major concern in epidemiological alcohol research. A crucial issue is what public health implications can be drawn, given these intricate research findings; more precisely, is alcohol consumption positive or negative for the overall health status of populations when all effects are factored in? This body of research also has a number of methodological issues which are brought to a head, in particular whether we should trust individual or aggregate-level data, as they typically show different results; numerous individual-level studies show a lower ischaemic heart disease (IHD) risk for moderate drinkers than abstainers, whereas aggregate-level analyses seldom identify any protective effects [1].

The study by Kerr and colleagues [2] acknowledges all these complexities and offers one of the more rigorous aggregate-level analyses conducted so far of the link between alcohol and IHD mortality. Using different forms of time–series techniques on a very large data set from the United States, the major message of the study is clear: the overall impact of alcohol on IHD is negative from a public health perspective, i.e. higher alcohol consumption in the United States leads to more additional IHD deaths than may be ‘saved’. Given the strong analytical design, I believe that this is a very important finding, in line with studies showing that protective effects become smaller the higher the quality of the research design in individual-level studies [3]. It also accords with the idea that the protective effect is, to a large degree, the result of moderate drinkers displaying a better health status and thus is not related causally to their drinking [4].

So far, the message is probably good news for those favouring a restrictive alcohol policy and bad news for others, e.g. the alcohol industry. However, when the analysis is taken one step further by introducing beverage-specific effects as a way to disentangle effects of good and bad drinking patterns, a more obscure picture emerges, with spirits being bad and beer and wine being good for the heart. This is interpreted by the authors as a result of different drinking patterns, with more heavy drinking patterns among spirits drinkers and more protective drinking patterns among beer and wine drinkers.

However, what empirical evidence is there to support these interpretations? It sounds plausible that spirits drinking in the United States is a marker for less favourable drinking patterns, but are wine and in particular beer drinkers mainly moderate drinkers? From a non-US perspective the idea that beer consumption is associated strongly with moderate drinking patterns is surprising, and other evidence to support this would give more credibility to this thought-provoking result.

The authors are very careful with respect to policy implications from the beverage-specific findings and adhere basically to a recommendation to reduce overall drinking. However, it is not likely that every reader will withdraw from recommending policies favouring beer and wine for public health reasons. For example, it is easy to imagine that the brewing industry will disseminate research suggesting a positive health impact of more beer drinking, and avoid focusing on the risk that this will raise overall drinking and thus IHD mortality (and other harm rates) [5]. In fact, the beverage-specific findings open up many different interpretations concerning public health implications, and the meaning of these findings needs to be elaborated in future studies. Hopefully, they will not conceal the very important message of this paper: there is an overall negative impact of alcohol on IHD mortality in the United States.


  1. Top of page
  2. Declaration of interest
  3. References