Key points are not available for this paper at this time.
Pregnant women and newborns are historically underrepresented in clinical trials, creating critical gaps in evidence on the safety and effectiveness of medications used during pregnancy. Real-world data (RWD) sources offer a promising avenue to address these gaps. To fully realize this potential, it is essential to link maternal and infant records accurately within and across diverse datasets. High-quality mother--infant linkage enables the robust evaluation of maternal medication use and its short- and long-term effects on both maternal and infant health. However, linking maternal and infant healthcare data introduces complex methodological and practical challenges. Achieving accurate linkage is often hindered by factors such as inconsistent personal identifiers, discrepancies in insurance coverage between mother and infant, data incompleteness, algorithmic accuracy, and strict data privacy regulations. Commonly used proxies for linkage (e.g. shared address or healthcare provider) may also be unreliable and can introduce misclassification or duplication. This commentary synthesizes current knowledge on mother--infant data linkage in RWD. It also systematically outlines the key challenges, emerging opportunities, and strategic directions to improve linkage quality and address privacy concerns in the identification of mother--infant dyads to support rigorous pharmacoepidemiologic research on maternal and infant health outcomes. By improving linkage methods and leveraging innovative approaches such as tokenization and validated algorithms, researchers can enhance the reliability of real-world evidence on maternal and infant health outcomes, including long-term follow-up across diverse data sources. Advancing these methodological frontiers is essential to generate evidence that supports safer and more informed treatment decisions for pregnant women and their children.
Kaplan et al. (Tue,) studied this question.