Clinical trial emulation can identify new opportunities to enhance the regulation of drug safety in pregnancy

From the perspective of most regulatory agencies, it is usually unethical to perform interventional clinical trials on pregnant people. While this policy recognizes the vulnerability of an expectant mother and unborn child, it has created a public health emergency for millions of pregnant patients through a dearth of robust safety data for many common drugs. To address this problem, we harnessed an enterprise collection of 2.8M electronic health records (EHRs) originally collected from routine primary care, leveraging the data linkage between mothers and their babies to create a surrogate for randomized, controlled drug trials in this population. To demonstrate the feasibility of our clinical trial emulation platform to stimulate new hypotheses for post-market drug surveillance, we identified 1,054 drugs historically prescribed to pregnant patients and developed a medication history-wide association study and follow-up evidence synthesis platform—leveraging expert clinician review and real-world data analysis—to test the effects of maternal exposure to these drugs on the incidence of neurodevelopmental defects in their children. Our results replicate known teratogenic risks and existing knowledge on drug structure-related teratogenic risks. Herein, we highlight 5 common drug classes that we believe warrant further assessment of their safety in pregnancy. We also discuss our efforts to develop a discovery-to-regulatory framework that could allow for pragmatic translation of our results to enhanced regulatory policy. Collectively, our work presents a simple approach to evaluating the utility of EHRs in guiding new regulatory review programs focused on improving the delicate equipoise of accuracy and ethics inherent to assessing drug safety in an extremely vulnerable patient population.