Thuy N Thai, Judith C Maro, Joshua Brown, Stephan Schmidt, Sonja A Rasmussen, Almut G Winterstein
By scanning more than 300 infant circulatory-related outcomes, we identified unspecified hypotension as a potential safety signal among newborns whose mothers were exposed to promethazine-ondansetron combination rather than metoclopramide-ondansetron combination during the first trimester. This finding warrants further investigation and illustrates the utility of tree-based scan statistics as a hypothesis-generating tool for pregnancy medication safety research.
BACKGROUND: Pregnant women sometimes need antiemetic combination therapy to control severe nausea and vomiting during pregnancy. Because promethazine, ondansetron, and metoclopramide have QT-prolonging effects, combination use may potentiate arrhythmic effects in mothers and cause circulatory conditions in infants, with promethazine combinations expected to exert stronger effects than metoclopramide combinations.
OBJECTIVE: We implemented tree-based scan statistics to screen for signals of adverse circulatory-related conditions in newborn infants following maternal exposure to promethazine-ondansetron combination compared to metoclopramide-ondansetron combination.
METHODS: We used Merative® Marketscan® Commercial Claims 2005-19 and Medicaid Analytic eXtract data 2005-15 to identify pregnant women aged 12-55 years with a live birth. To define combination use, pregnant women had to fill a prescription for a second antiemetic during the active days' supply of the first antiemetic and then refill the first antiemetic during the active days' supply of the second antiemetic. Our outcomes included circulatory diseases, individually and clustered within a hierarchical tree structure. We adjusted for potential confounders using propensity score quartile stratification and used Poisson tree-based scan statistics to screen for signals (p < 0.05).
RESULTS: A total of 6865 and 4186 live birth deliveries with first-trimester exposure to promethazine-ondansetron combination or metoclopramide-ondansetron combination, respectively, met our inclusion criteria. After multiplicity adjustment for over 300 outcome clusters, we found three significant signals: unspecified hypotension (relative risk = 3.71, p < 0.001), chest pain (relative risk = 2.01, p < 0.001), and essential hypertension (relative risk = 1.76, p = 0.002) among infants with maternal promethazine-ondansetron combination exposure. Unspecified hypotension likely reflects neonatal conditions, whereas unspecified chest pain and essential hypertension may represent maternal conditions based on the claims origin.
CONCLUSIONS: By scanning more than 300 infant circulatory-related outcomes, we identified unspecified hypotension as a potential safety signal among newborns whose mothers were exposed to promethazine-ondansetron combination rather than metoclopramide-ondansetron combination during the first trimester. This finding warrants further investigation and illustrates the utility of tree-based scan statistics as a hypothesis-generating tool for pregnancy medication safety research.