Birth Order, Birth Weight and Asthma: how to assess mediation and the presence of Unmeasured Confounding

If the goal is to measure the causal relation between two variables when a third is involved and plays the role of mediator, it is essential to explicitly define the relational assumptions among these and others relevant variables. However if any of this assumptions are not met, estimates of mediated effects may be affected by bias. One example where this occurs is the widely discussed situation known as the Birth Weight paradox which arises when birth weight is the mediator of interest and unmeasured confounders affect the mediator-outcome relation. In this paper we will focus on a setting where such paradox might arise where birth order act as the exposure for childhood asthma (by age 1.5). After estimating the direct and indirect effects, we give a plausible graphical explanation of the empirical results and explore the magnitude of the causal relations using sensitivity analysis.