LAUSR

We rely on post-marketing approaches to define the risk of medications in pregnancy because information at the time of drug approval is limited. Most studies in pregnancy focus on a single or selected outcomes. However, women must balance the benefit of treatment against all possible adverse effects. Our objective was to apply and evaluate a tree-based scan statistic data mining method (TreeScan) as a safety surveillance approach that allows for simultaneous evaluation of a comprehensive range of adverse pregnancy outcomes, while preserving the overall false positive rate. We evaluated TreeScan with a cohort design and adjustment via propensity score techniques using two test cases: (1) opioids and neonatal opioid withdrawal syndrome, and (2) valproate and congenital malformations, implemented in pregnancy cohorts nested in the Medicaid Analytic eXtract (1/1/2000 - 12/31/2014) and IBM MarketScan Research Database (1/1/2003 - 9/30/2015). In both cases, we identified known safety concerns, with only one previously unreported alert at the preset statistical alerting threshold. This evaluation shows the promise of TreeScan-based approaches for systematic drug safety monitoring in pregnancy. A targeted screening approach followed by deeper investigation to refine understanding of potential signals will ensure pregnant women and their physicians have access to the best available evidence to inform treatment decisions.