Public health consequences of the J-curve hypothesis of alcohol problems.

This paper addresses some issues related to the hypothesis that the risk for premature death is a J-shaped function of alcohol intake. The first part of the paper contains a discussion of the individual level epidemiological evidence. It is argued that the evidence is not yet good enough to allow precise statements about "safe limits", nor does it allow a precise location of the "optimum" consumption level. Measurement errors and confounding variables not yet controlled for remain substantial problems. In particular, future studies need to control for social integration, as this factor may affect both health status and alcohol consumption. The second part of the paper discusses the population level relationship. It is argued that, typically, what is optimum for an individual is too much for a population. Unintended side effects of major public health importance should be expected in this area. In the last section, a plea is made for aggregate level studies as a way of addressing the public health side of the issue.