The rationale for comparing maternal and paternal smoking associations with offspring bone mass was that there is likely to be residual confounding in these relationships from unmeasured GSK126 price factors. Differing distributions of unmeasured confounders in the complete case and multiply imputed datasets
could explain the difference between associations seen. Since there were differing educational distributions between the complete case and multiply imputed datasets and we found that parental smoking associations in the complete case differed between strata of parental education levels despite adjusting for all observed confounders, it seems that residual confounding is a possible explanation. Another possible reason for the difference is violation of the multiple selleckchem imputation assumption that the missing data mechanisms can be explained by other observed variables. However, we verified that missingness in each of the variables with missing data was strongly associated with other observed variables and included a number of predictors of missingness in prediction equations to impute missing
data. We therefore expect the multiply imputed datasets to be more representative of the study population and analyses based on these data more accurate. A limitation to our study was the self-report of smoking by the mothers and fathers. Maternal smoking could be affected by reporting bias since mothers may be aware of Adenosine the harmful effects of smoking and less likely to respond affirmatively. Nevertheless, where both the mother and father provided information about the father’s smoking status, there was agreement in 94.5% of couples. The study benefitted from its large size, the ability to control for a number of potential confounders and the ability to compare associations of bone outcomes with both maternal and paternal exposures
to assess the level of residual confounding. Conclusions Our study has found positive associations of maternal smoking during pregnancy with offspring total body and spinal bone mass in girls, with minimal evidence for any associations in boys, and our multivariable analyses and parental comparisons suggest that these associations are largely driven by familial characteristics related to childhood adiposity and unlikely to be due to intrauterine mechanisms. Although our findings do not demonstrate negative effects of maternal smoking in pregnancy on offspring bone mass, its known adverse effects for mothers and offspring health mean than women should be encouraged not to smoke.