Field studies versus database studies on the risks and benefits of medication use during pregnancy: Distinct pieces of the same puzzle.

Over the past two decades, findings on medication use during pregnancy have been accumulating from observational data. Generally, field studies with prospective recruitment of subjects have better outcome ascertainment, and more control on the longitudinal collection of data, but have lower sample sizes and thus they often lack statistical power to detect increased risks for rare events such as major congenital malformations. In addition, given the rarity of specific drug exposures in a population, even relatively common outcomes, such as low birth weight, may become rare in combination with the specific exposure. On the other hand, administrative databases usually provide larger samples and thus increased statistical power, decrease the probability of selection and recall bias, but often have missing data on potential confounders. Hence, debate amongst researchers, regulators and public health officials has been ongoing with regard to the most appropriate study populations for perinatal epidemiologic research. With this commentary, we aim to highlight the importance of both study populations, which can make complementary and crucial contributions to the iterative determination of causality as well as discuss basic epidemiologic principles that need to be applied in the field of perinatal pharmacoepidemiology for the purpose of causality assessment. This is relevant at present given that the United States Food and Drug Administration (US FDA) has modified their medication label requirements, especially given the international importance of these modifications.