In addition to the results presented in Tables 1–3, we also evaluated the impact of smoke-free bar and restaurant policy on changes in smoking status. Analyses (not shown) indicated that after adjusting for sociodemographic characteristics, smoke-free policy was not significantly associated with smoking remission among all smokers (OR = 0.95; 95% CI = 0.73,1.24) nor among public drinking smokers (OR = 0.93; 95% CI = 0.68,1.26). Similarly, smoke-free policy did not predict smoking onset among all non-smokers (OR =1.14; 95% CI = 0.86, 1.50) nor among the subset of public drinking non-smokers (OR=1.44; 95% CI = 0.93, 2.25). When evaluating whether changes in smoking status explained the effects of smoke-free legislation on AUD transitions, we found that adding smoking status transition dummy variables to our regression models (smoking onset, smoking remission, and non-smoking, with persistent smoking as the reference group) had no impact on the effect size (ΔOR<0.02) or significance of the main effects of smoke-free legislation on AUD transitions in the sample overall nor among the subset of public drinkers, results inconsistent with mediation.
