Secondary Data Analyses

Secondary analyses were performed to further explore our primary findings. First, individuals were divided into slow and normal metabolizers using a cut-off of ≤0.85 on the metabolism metric as previously described (Chen et al., 2014). This cut-off captures the lowest quartile of metabolizers, and this dichotomous variable was used in logistic regression models of smoking behaviors. Second, since the majority of the COGA sample was recruited from families at high-risk for alcoholism, the primary analyses examining the continuous metabolism metric and smoking milestones were repeated with the covariate of lifetime DSM-IV alcoholism dependence. Third, after observing an association between the metabolism metric and the time to first cigarette dichotomous variable (>5 and ≤5 minutes), the 4 level variable of time to first cigarette after waking (>60, 31–60, 6–30, ≤5 minutes) was also investigated in cumulative logistic regression models. These analyses were performed to assess whether the continuous metabolism metric predicted response across the four ordinal categories.