(304) Methodologic Considerations for Emulating Target Trial for Discontinuation of Antidepressant Treatment in Pregnancy

Publication Number: 978

Background: Emulation of a target randomized controlled trial (RCT) is a framework for estimating causal effects in observational data, when conducting an intervention study is not possible. This is a particularly relevant approach when evaluating the safety of medication use in pregnancy, as pregnant people are routinely excluded from RCTs. Recent work has further developed this framework for initiation of medications during pregnancy. Additional work is needed to extend these findings to treatment decisions for medications in use prior to pregnancy, particularly medications with an uncertain safety profile for the developing fetus, such as antidepressants (AD).

Objectives: To describe important considerations for specifying a target trial for AD discontinuation during pregnancy.

Methods: Key study design considerations for emulating a target trial for discontinuation of AD during pregnancy are described, including eligibility criteria, treatment strategy, start of follow up, and adherence to the assigned treatment strategy. To illustrate the impact of variations in these study design decisions, we will use data from the Norwegian Mother, Father and Child Cohort Study (1999-2008) linked to other national health registries and national school test scores in mathematics and reading. Eligible women self-reported use of AD medications in the 6 months before pregnancy. We will define 2 treatment strategies, initiated before pregnancy or within 2 weeks of the first prenatal care visit: 1) continue AD treatment, or 2) discontinue AD treatment. We will conduct an intention to treat analysis to estimate the effect of the initial treatment decision, and a per protocol analysis to estimate the effect of continuing to follow the initial treatment assignment. Follow up will begin at the initial treatment decision.

Results: 1092 women reported use of AD medication before or during pregnancy, of which 945 initiated treatment prior to pregnancy and 147 initiated during pregnancy. Of 945 initiating prior to pregnancy, 162 (17%) continued through delivery, while 378 (40%) discontinued before conception, 294 (31%) discontinued in the 1st trimester, 61 (6%) discontinued in the 2nd trimester, and 50 (5%) reported stopping and starting treatment. Initiation in pregnancy, later discontinuation, and starting/stopping, were associated with higher depressive symptoms. Planned analyses will evaluate associations between treatment decisions and school test performance in offspring.

Conclusions: Studies on AD safety often mix new and prevalent users, limiting their utility for clinical decision support. We show that defining a target trial in this sample is not straightforward, and planned analyses will demonstrate the sensitivity of results to study design decisions.