Including prevalent users in pregnancy medication studies

Determine principled analytic methods for including prevalent users—pregnant individuals who have been using pharmacotherapies for chronic conditions prior to pregnancy—in studies evaluating treatment effects during pregnancy, ensuring that the approach is appropriate for perinatal pharmacoepidemiology and addresses the specific methodological challenges posed by prevalent user bias.

Background

The paper argues that mixing new and prevalent users in observational studies of pharmacotherapies for chronic conditions in pregnancy can introduce prevalent user bias, potentially overestimating benefits and underestimating harms. While the authors advocate for anchoring study designs on clinical decision points (e.g., prenatal care initiation) to mitigate time-related biases, they note that guidance on how to appropriately include prevalent users remains lacking.

The authors illustrate target trial emulations for antidepressant and antihypertensive treatments during pregnancy and emphasize that pragmatic concerns often require analyses involving individuals already on treatment before pregnancy. Despite methodological advances in target trial emulation and treatment decision designs, the specific strategies for incorporating prevalent users in a way that preserves valid causal inference in perinatal settings remain unresolved.