The quantification of alcohol-caused morbidity and mortality in Australia: a critique

One of the more controversial conclusions of the Commonwealth Department of Community Services and Health's report on The quantification of drug caused morbidity and mortality in Australia 1988 is that alcohol "causes" about 5% of all deaths in Australia. Against the background of this conclusion, this paper reviews current concepts of what constitutes a causal relationship and how the existence of causation is diagnosed in a nonexperimental setting. The conclusion of the report appears to rest on the assumption that all differences in disease rates between users and non-users of alcohol (or other drugs) are causal in nature--an assumption which is tantamount to equating statistical association with causation. Moreover, the estimate of alcohol-caused deaths, derived from the summation of alcohol-caused deaths from a large number of medical conditions, is at considerable variance with an estimate of total alcohol-caused deaths computed directly from total death rates. The latter estimate actually indicates that alcohol prevents more deaths than it causes in the population as a whole, a conclusion that is compatible with the findings of several recent large cohort studies, from which it is in fact derived. The discrepancy between the two estimates casts doubt on the validity of the assumptions underlying the methodology that has been applied.