Exposed or not exposed? Exploring exposure classification in studies using administrative data to investigate outcomes following medication use during pregnancy

Purpose: The aim of this systematic review was to examine and compare differences in the way medication exposures are classified in studies using linked administrative data to investigate outcomes following medication use during pregnancy. This was undertaken with a focus on studies investigating specific neonatal outcomes following prenatal exposure to selective serotonin reuptake inhibitors (SSRIs). Methods: We searched Medline and Embase to identify studies that used linked administrative data to investigate specific neonatal outcomes (congenital malformations, birth weight, gestational age) following prenatal exposure to SSRIs. Results: Key factors such as dose, duration and timing of exposure were inconsistently addressed in the studies identified. In addition, there was a great deal of variability in the way medication exposures were classified and how women who stop taking their medication before or during early pregnancy are handled in analyses. Furthermore, there are issues in assuming how and when women who receive a dispensing for a medication actually take it during pregnancy. This creates a great deal of uncertainty around medication exposure during pregnancy in studies using linked administrative data, potentially resulting in biased risk estimates. Conclusions: There is a need for greater focus on determining the most effective and accurate way of using linked administrative data to investigate outcomes following medication use during pregnancy in an effort to minimise potential biases.