Active Surveillance of the Safety of Antipsychotic Medications in Pregnancy

Mental health conditions are common among pregnant and postpartum women as well as non-pregnant women of reproductive age. While many patients require pharmacologic treatment, the safety of psychotropic medications in pregnancy is an area with large evidence gaps. Historically, spontaneous reports, pregnancy exposure registries, and case-control designs were the main approaches used to evaluate the safety of psychotropic medications in pregnancy, all of which have well-known limitations. In recent years, the field has gained much expertise with the conduct of cohort studies nested in large healthcare utilization data. Across all these designs, studies tend to focus on a single or a few selected adverse pregnancy outcomes, and they are performed at a single time point many years after the drug has entered the market and has been used by many pregnant women. To avoid unnecessary exposure of mother and fetus to harmful medications or to avoid women being unnecessarily deprived of treatments for psychiatric disorders when no harmful effects exist, a new and systematic approach is needed to generate timely evidence on the safety of psychotropic medications in pregnancy with respect to all relevant maternal and fetal outcomes. We will develop and implement a TreeScan based approach to conduct active surveillance of antipsychotic medication safety in pregnancy. TreeScan is a novel method for drug safety surveillance, which scans hierarchical trees of specific outcomes as well as groups of clinically related outcomes for associations with the treatment of interest, while accounting for multiple testing of correlated hypotheses. We will first develop hierarchical trees for congenital malformations, maternal and other neonatal outcomes based on shared underlying disease processes, and modify the TreeScan approach to accommodate the unique challenges of drug safety evaluation in the context of pregnancy. Use of hierarchical trees increases power to detect clinically related outcomes, which would not be feasible by evaluating individual diagnoses only. We will then implement TreeScan to evaluate the risks of a broad range of adverse maternal and neonatal outcomes associated with antipsychotic medications. Antipsychotics are the mainstay of treatment for women with schizophrenia and bipolar disorder, but little is known regarding their safety profile in pregnancy, especially for the newer antipsychotics. We will use nationwide cohorts of over 3.5 million publicly and privately insured pregnancies in the US, nested in healthcare utilization databases that contain rich information on confounders. In the final aim, we will extend the approach to conduct near real-time prospective, sequential surveillance of newly approved antipsychotic and other psychotropic medications in order to detect potential safety signals as early as possible after approval. By providing the necessary information for healthcare providers to make evidence-based prescribing decisions and to counsel women about the use of specific antipsychotics during pregnancy, the proposed studies will have an immediate public health impact within the field of perinatal psychiatry.