The results of multinomial logistic and logistic regression analyses predicting latent classes indicative of severity of nicotine withdrawal from other measures of smoking behavior and psychopathology, in the adult and adolescent cohorts respectively, are depicted in tables 2 and 3. Every variable that we considered was significantly associated with increased risk for severe nicotine withdrawal, with the exceptions of smoking persistence and ADHD in the adolescent cohort. Severe nicotine withdrawal was most highly associated with HSI levels (adults: OR=3.5–127.5; adolescents: OR=2.5–18.3), impairment (adults: OR=13.5–74.6; adolescents: OR= 7.7), difficulty quitting (adults: OR= 4.4–11.8; adolescents: OR=3.8–5.3), and seeking treatment for smoking cessation (adults: OR=4.5–11.0; boys: 4.3). The relationship between difficulty quitting and severe nicotine withdrawal was significantly stronger in the boys (OR=5.3) compared to the girls (OR=3.8). In adjusted multivariate models, all significant univariate predictors remained except for heavy smoking, conduct disorder, and alcohol problems in the adolescents. However, in the adults, smoking persistence only remained associated with moderate nicotine withdrawal and not severe, while major depression only remained associated with increased risk for severe nicotine withdrawal.
