The findings reported here have limitations. First, this study focused on individuals of European Ancestry because the metabolism metric was optimized for this population (Bloom et al., 2011). Second, the precise timing of length of transitions between smoking behaviors could not be examined in these analyses because the smoking questions did not assess age of onset. Third, the majority of the COGA participants were from families at high risk for alcoholism and rates of DSM-IV alcohol dependence are high in this sample, which may affect the generalizability of the findings. Secondary analyses that include DSM-IV alcohol dependence as a covariate and replication of the primary findings in a community based recruitment sample (COGEND), however, support the conclusion that the findings are not specific to a high-risk population.
