Association analyses with age at onset of daily smoking in daily smokers were conducted using the statistical software QTDT (Version 2.4.5, http: sph.umich.edu/csg/abecasis/QTDT; [24]). Among daily smokers both total and within-family (orthogonal model) evidence for association was assessed using age at onset of daily smoking as a continuous variable. The orthogonal model is the generalization of the Fulker model [24], allowing for families of any size (with or without parental genotypes) and was used in a variance components framework. QTDT total association test evaluates total evidence for association and is more powerful than QTDT orthogonal test in absence of population stratification. Therefore we first evaluated the evidence of population stratification by comparing the between and within components of association using QTDT stratification test.
