Patterns and correlates of tobacco use among young adults at college and university

This thesis describes college and university students' smoking behaviours and examines whether socioenvironmental and personal characteristics experienced during adolescence are differentially associated with their smoking participation. Results show more college students than university students currently smoke (37% and 21 % respectively) and more began smoking prior to post-secondary school (93% and 84% respectively). Early age of onset of alcohol use increased the odds of current smoking (main effect model, OR = 8.56 CI = 6.47, 11.33), especially for university students (interaction effect model, b = 2.35 CI = 7.50, 14.64). Lower levels of high school connectedness were associated with increased odds of current smoking but for university students only (interaction effect model, b = -0.15 CI = 0.84, 0.88). While limitations associated with convenience sampling and low response rate exist, this is the first Canadian study to examine college and university students separately. I t reveals that tobacco control programming needs to differ for college and university students, and early alcohol prevention and school engagement programs for adolescents may influence tobacco use. Given that both educational pathway and use of tobacco are associated with SES, future research may consider examining in more detail, SES-related socioenvironmental variables.

Associations Between Personal Cancer History and Lung Cancer Risk

The study aim was to investigate the relationship between factors related to personal cancer history and lung cancer risk as well as assess their predictive utility. Characteristics of interest included the number, anatomical site(s), and age of onset of previous cancer(s). Data from the Prostate, Lung, Colorectal and Ovarian Screening (PLCO) Cancer Screening Trial (N = 154,901) and National Lung Screening Trial (N = 53,452) were analysed. Logistic regression models were used to assess the relationships between each variable of interest and 6-year lung cancer risk. Predictive utility was assessed through changes in area-under-the-curve (AUC) after substitution into the PLCOall2014 lung cancer risk prediction model. Previous lung, uterine and oral cancers were strongly and significantly associated with elevated 6-year lung cancer risk after controlling for confounders. None of these refined measures of personal cancer history offered more predictive utility than the simple (yes/no) measure already included in the PLCOall2014 model.