A related commentary discusses this research. One challenge to researchers examining the risk of pharmacological treatment during pregnancy is the complexity of the context within which treatment is prescribed. Without inclusion of a comparison group who are untreated but have the same mental disorder, how do we know if an outcome is related to exposure to treatment or the illness for which it is prescribed?
In this study, the comparison group comprised pregnant women with both bipolar and unipolar mood disorders who were not exposed to lithium for their main analyses. By including women with a depression diagnosis, they potentially capture data from those with lesser illness severity and most likely different overall risk profiles than those prescribed lithium throughout their pregnancy. Therefore, this comparison group is unlikely to enable investigators to account for the effects of illness or comorbidities on outcomes. In the sensitivity analyses a comparison was made with controls with bipolar disorder alone. When the authors did this comparison, one of the main findings—an increased risk of neonatal re-admission in those exposed to lithium—was no longer significant.
It is important for future research to examine longer-term risks for exposed children, as for a woman to weigh the risks and benefits of treatment during pregnancy, knowledge of any longer-term risks for her child is an important part of that decision.