Gravar-mail: Assessment and Indirect Adjustment for Confounding by Smoking in Cohort Studies Using Relative Hazards Models