Home | Business News | Browse by Publication | C | Contemporary Drug Problems

Alcohol consumption and all-cause mortality in the United States, 1950-2002.

Article Excerpt
This study addresses the relationship between population drinking and all-cause mortality in the United States during the postwar period. The work can be seen as a continuation and empirical broadening of a series of similar studies (see below). The relationship at issue is of interest for three main reasons, partly related to each other:

* All-cause mortality is often regarded as an indicator of the health status of the population, and its relationship to the drinking level in society is therefore of particular public health concern.

* Numerous individual-level studies suggest a J-shaped risk function between alcohol and mortality. Generally this is thought to be the result of a beneficial cardio-protective effect of light to moderate drinking and an elevated risk of excessive drinking on a large number of outcomes, including injuries, liver cirrhosis, and certain forms of cancer. There is no need to detail the methodological problems associated with the studies suggesting a J-shaped curve, since the potential bias arising from selection effects (Shaper 1990) and misclassification of drinkers (Fillmore, Kerr. Stockwell, Chikritzhs & Bostrom 2006) is well-known by now. If we assume that the J-shaped curve is real, it is of great interest to know its implications on the population level. However, in practice it is impossible to predict the outcome on the population level of changes in aggregate consumption. This is because such a prediction would depend critically on several factors where precise knowledge is lacking, above all the shape of the risk curve and the distribution and dynamics of the consumption distribution (see Skog 1996 for a more detailed discussion). A more straightforward way to elucidate the outcome on the population level is thus to analyze directly how mortality responds to changes in aggregate consumption.

* It is generally recognized that the health impact of a given volume of alcohol depends on the way in which it is consumed, that is, on the prevailing alcohol culture and its drinking patterns. Additional aggregate studies on the link between alcohol and mortality thus increase our knowledge of the cultural variations in the health effect of alcohol. Such studies can thus also be seen as supplementary to the drinking pattern score literature (Rehm et al. 2004).

By now there is a fairly large number of comparable studies addressing the relationship between population drinking and all-cause mortality. As can be seen from the overview in Table 1, the studies cover a wide spectrum across time--including the late 19th century as well as the postwar period--and space, covering Russia, various parts of Europe and Canada. All studies except the one by Her and Rehm (1998) apply by and large the same protocol and model specification, and are thus methodologically comparable. In the Her and Rehm study a somewhat different method is used, and the countries under study are fairly heterogeneous with respect to drinking patterns (e.g., Italy vs. Poland). However, differences in the reliability...