Big data apprOaches fOr Safe Therapeutics in Healthy Pregnancies (BOOST-HP) - PROJECT SUMMARY/ABSTRACT In the US, pregnant patients use 4 medications on average, and 70% use at least one. Yet, most drugs lack conclusive evidence about safety during pregnancy: of 290 new FDA labels approved between 2010 to 2019, 90% contain no human data on the risks or benefits for pregnant patients. With current evidence generation systems, the mean time for evidence development in pregnancy has been estimated at 27 years, which is too long. Current evidence generation relies largely on observational studies, typically prompted by signals from animal studies or extrapolation from known pharmacological pathways, which may miss pregnancy-specific context. Insufficient attention is also given to identifying causal mechanisms in vulnerable sub-populations at greatest risk. Building on our prior work in data-mining in FDA’s Sentinel System and CDC’s Vaccine Safety Datalink, conduct of pharmacoepidemiologic studies to evaluate prenatal medication safety, and pilot work with special focus on drug scans in pregnancy, we will implement a three-stage novel reverse translational framework to accelerate evidence generation that will use data-mining (“scans”) to identify new exposure- outcome associations, triage signals, and then formally evaluate top prioritized signals. To accomplish our goals, we will use our infrastructure developed for drug evaluations in pregnancy, including curated billing records from the NIH Collaboratory’s Distributed Research Network and the national Medicaid Information System, representing a broad cross-section of privately and publicly insured pregnant patients and their offspring. Our specific aims are: (Aim 1) To scan for associations between (1a) pregnancy loss and antecedent prenatal exposures on the individual drug, chemical and therapeutic class level; and (1b) the 50 most prevalent drugs in pregnancy with incomplete information on teratogenic risk and a broad selection of live birth adverse outcomes; and (1c) to prioritize signals via expert panel review. (Aim 2) To employ careful pharmacoepidemiologic designs to evaluate the two top prioritized signals involving (2a) pregnancy loss, and (2b) an adverse livebirth outcome. To control for confounding and measurement biases, these studies will employ previously validated measures, which are further enhanced via linkage to fetal death and birth certificate data for a cohort subsample to evaluate unmeasured confounding and conduct probabilistic sensitivity analyses on outcome and exposure misclassification. Big data apprOaches fOr Safe Therapeutics in Healthy Pregnancies (BOOST-HP) will offer an innovative advancement in evidence generation by evaluating numerous exposures and outcomes simultaneously. Our long-term goal is to build a reusable, scalable approach and infrastructure to accelerate evidence generation on the safety and effectiveness of medication use during pregnancy. By leveraging data-mining methodologies successfully deployed in public health surveillance along with infrastructure used by multiple federal government agencies, we will focus research efforts on novel, high-priority signals that pose the greatest risk to healthy pregnancies.