Background: Evidence of a biologically plausible association between maternal smoking during pregnancy (MSP) and the risk of depression is discounted by null findings from two sibling studies. However, valid causal inference from sibling studies is subject to challenges inherent to human studies of MSP and biases particular to this design. We addressed these challenges in the first sibling study of MSP and depression conducted among adults past the peak age for the onset of depression, utilizing a prospectively collected and biologically validated measure of MSP and accounting for non-shared as well as mediating factors.
Methods: We fit GEE binomial regression models to correct for dependence in the risk of depression across pregnancies of the same mother. We also fit marginal structural models (MSM) to estimate the controlled direct effect of MSP on depression that is not mediated by the offspring's smoking status. Both models allow the estimation of within- and between-sibling risk ratios.
Results: The adjusted within-sibling risk ratios (RRW) from both models (GEE: RRW = 1.97, CI 1.16-3.32; MSM: RRW = 2.08, CI 1.04-4.17) evinced an independent association between MSP and risk of depression. The overall effects from a standard model evinced lower associations (GEE: RRT = 1.12, CI 0.98-1.28; MSM: RRT = 1.18, CI 1.01-1.37).
Conclusions: Based on within-sibling information free of unmeasured shared confounders and accounting for a range of unshared factors, we found an effect of MSP on the offspring's risk of depression. Our findings, should they be replicated in future studies, highlight the importance of considering challenges inherent to human studies of MSP and affective disorders.
Keywords: Barker hypothesis; DOHaD; depression; maternal smoking during pregnancy.