ABSTRACT
Some questions remain on the relationship between smoking and bone health. Detailed analyses of the relationship between smoking and BMD are presented, essentially ruling out non-linear associations as an explanation for inconsistent results in the literature. INTRODUCTION: To provide comprehensive multiple regression and dose-response analyses of the association between smoking behavior variables and bone health as assessed by radiologically determined bone mineral density in NHANES III. METHODS: Analyzes of a representative cross-sectional survey of the noninstitutionalized population of the USA. Self-reported smoking behavior and bone mineral density of 14,510 participants were analyzed using survey design-based multiple linear regression modeling. Dose-response patterns were analyzed using restricted cubic spline regression. RESULTS: Femoral neck bone mineral density in current smokers was numerically lower than in never smokers, but this was not statistically significant after controlling for confounders. In former smokers, bone mineral density T scores were 0.064 units higher for every 10 years of abstinence, with little impact of confounder adjustment. Spline regression revealed no relevant non-linearity in the associations studied. CONCLUSIONS: Non-linearity is an unlikely explanation for inconsistent results in the literature on smoking and bone mineral density. Further and especially longitudinal studies of the complex relationship smoking with bone health would be particularly important given the still substantial prevalence of smoking in an aging global population.