The age at onset for habitual smoking among daily smokers was analyzed using Cox proportional hazards regression analysis (coxph) using the R statistical analysis package. This analysis can incorporate a clustered sandwich estimator to account for the familial correlation among observations. The age at onset measures for non-daily smokers were right-censored at each subject's age at their most recent assessment. The first principal component from the stratification analysis (pc1), age at interview, gender, maternal/paternal daily smoking and maternal/paternal regular drinking were included as covariates. The violation of the proportional hazards assumption was tested with non-zero slope of Schoenfeld residuals versus time using the survival analysis package in R.
