Initial comparisons between Contacted, Deceased, and To Be Located subjects revealed significant (FDR-adjusted p ≤ 0.001) differences with respect to demographics, self-rated general health, alcohol involvement, psychiatric comorbidity, and whether they were a relative of current prospective study participants (Table 1). Comparing subjects whose whereabouts were still unknown versus located individuals (i.e., confirmed alive or deceased) revealed that the To Be Located subjects were younger and were less likely to be a relative of prospective participant, to be of European ancestry, or to have ever been married (all with FDR-adjusted p < 0.001) at baseline. In addition, To Be Located subjects had an earlier onset of DSM IV alcohol dependence, a greater number of other substance use disorders (in particular, marijuana and cocaine), and a higher lifetime prevalence of antisocial personality disorder (ASPD) at baseline (all with FDR-adjusted p < 0.001; Table 1). After controlling for baseline age, the differences in marital status, age of onset of DSM IV alcohol dependence, marijuana dependence, number of non-alcohol substance dependence disorders, and ASPD were no longer significant. As anticipated (Hypothesis 1), having
