BACKGROUND: Upon making a new treatment option available, particularly to women of childbearing potential, its safety to mother, fetus, and infant is of key interest to regulators, manufacturers, patients and their families, and the public. Different study approaches are available but may differ in their ability to provide this needed information.
OBJECTIVES: To evaluate different approaches to pregnancy safety studies among women with multiple sclerosis (MS), provide information on optimal study designs, and evaluate the robustness of reported outcomes.
METHODS: A comprehensive literature review was performed using medical literature databases complemented with Internet research and limited to articles published in English from January 1993-October 2015.
RESULTS: A total of 51 studies were identified: 10 industry-sponsored therapy-specific pregnancy exposure registries, 15 studies using three disease-specific pregnancy registries, 8 surveillance programs, 7 miscellaneous prospective cohort studies, 4 retrospective chart reviews, 6 retrospective studies using purpose built databases, and 1 population based database study. No meta-analyses were identified. Details of study design were often not reported fully, particularly on planned study size and its justification. Investigated outcomes and their definitions varied across studies. Several studies failed to reach recruitment targets. Most studies had few drug-exposed pregnancies, data on infant health were very scarce, and less than half had internal comparator groups. Regardless of design or source population, even many years after initial marketing authorization, no study was able to produce robust relative risk estimates to reliably inform the use of the investigated drug in pregnancy.
CONCLUSIONS: Pregnancy safety studies in MS have produced limited informative outcomes data. Alternatives worth exploring for future studies include larger disease-specific prospective registries that can evaluate pregnancies exposed to specific drugs, retrospective database studies using large population-based or claims data with mother-child linkage, and multidatabase studies that combine data sources using meta-analytic methods for increased precision.
