When the analyses were conducted in the subgroup of never-smokers, no significant associations were detected, probably due to the small sample size. Ordinal regression models confirmed the results by providing practically similar findings in all analyses. No collinearity problem was found when VIFs were determined.