An improved aetiologic fraction for alcohol-caused mortality.

A method is proposed for estimation of alcohol-caused mortality in which the aetiologic fraction of deaths is derived from an exposure contrast between the 'unsafe' drinker and 'safe' drinker; not between the drinker and the nondrinker. The method is consistent with the objectives of public health intervention in the alcohol education field. The 'safe' drinker is the reference exposure category and aetiologic fractions are obtained for the single categories of a multiple-level exposure scale, using separate relative risks for the 'unsafe' drinker and the nondrinker. Preliminary application of the method to Australian data yields an estimate of the number of deaths caused by misuse of alcohol that is larger than those obtained under the old conceptual model. Thus, the public health importance of unsafe alcohol consumption has been underrepresented.