To the Editor: We wish to comment on the interpretations offered in the article by David J. DeWit, Ph.D., et al. (1). The association between an early age at first alcoholic drink and a greater risk for alcoholism or heavy drinking has been reported in at least nine previous studies, but there is a considerable lack of agreement regarding how such results should be interpreted.

One hypothesis is that early drinking is a direct risk factor for the development of alcoholism. Dr. DeWit et al. interpreted their results as representing a causal process, viz., "The results demonstrate that prevention programs that are effective in delaying alcohol use until age 15 or 16 or later will avert substantial alcohol-related harm in later life" (p. 749). An alternative hypothesis is that early drinking and alcoholism are both manifestations of a general vulnerability to problem behaviors but are not causally related. On the basis of this model, interventions designed simply to delay alcohol use would not modify the underlying vulnerability that predisposes at-risk individuals to problem outcomes, such as alcoholism.

Distinguishing between these hypotheses is not possible with the use of epidemiological data from adult samples (as in the study by Dr. DeWit et al.) but requires additional measures, prospective longitudinal research designs, or additional, theoretically imposed structure. We employed the latter strategy in a twin study of age at first alcoholic drink and the risk for alcoholism among more than 2,100 adult twin pairs (2). Unaffected co-twins of twins with alcohol dependence began drinking earlier than the twins from pairs in which neither twin had a diagnosis of alcohol dependence. The results of twin pair analyses suggested that the association between early drinking and later alcohol dependence is due primarily to familial characteristics that probably reflect both shared environmental and genetic factors. We interpreted these results as consistent with the "shared vulnerability" hypothesis: that early drinking and alcoholism are both manifestations of familial vulnerability to problematic alcohol use.

In sum, we believe caution should be exercised before interpreting correlational findings as causal. We concur with Grant and Dawson (3), who described "an urgent need to integrate epidemiological and etiologic research…to ascertain if it is the delay in alcohol use or, more likely, other associated factors that account for the inverse relation between age at first drink and the risk of lifetime alcohol use disorders" (p. 109).