Despite the fact that prescription medications are commonly used by pregnant women, for most products and for new drugs in particular, there is typically little to no human safety information available to aid clinicians and patients in managing risk. As randomized clinical trials are usually not considered ethical to perform in pregnancy, observational epidemiologic studies are often the next best option to address human pregnancy exposure. An increasingly common approach to gathering human safety data is postmarketing pregnancy registries.
These pregnancy registries are initiated many times by agreement between the manufacturer and a regulatory agency as a postmarketing commitment or requirement following shortly after drug approval. Furthermore, because use of a new drug in pregnant women might be relatively rare, a pregnancy registry may be the only feasible method for gathering preliminary safety information as quickly as possible so that potential signals might be detected and clinical decision making can be better informed.
Pregnancy registries vary in design, but all involve collection of data on exposure to the drug of interest in pregnant women, and collection of outcome data for those pregnancies. The primary outcome of interest is typically major congenital anomalies; some registries also collect outcome data on fetal/infant growth, preterm delivery, pregnancy loss, specific neonatal outcomes, and postnatal longer term growth and development. The rates of these outcomes can be compared with general population reference rates, or rates occurring in a specific comparison group that might be more similar to the exposed women, for example, in terms of the underlying maternal condition being treated by the drug.
In addition to early information on a new drug, some of the major advantages of many pregnancy registry designs are the ability to collect information on the exposure and other pregnancy details before the mother knows what the outcome of her pregnancy will be; direct collection of exposure information from the mother herself, so that important factors such as drug and alcohol use, dose, and exact timing of exposure to the drug of interest; information on important other factors such as tobacco, alcohol, and multivitamin use.
The most challenging aspect of pregnancy registries is recruitment, for which registries largely depend on obstetric providers and other specialty physicians. Although low numbers of recruited pregnancies may be caused by limited use of a new drug, clearly most pregnancy registries enroll a very small fraction of all exposed pregnancies that are in existence. A second, related issue is that there may be bias in the self-selection of women who do find out about the registry and agree to participate, thus raising questions about the generalizability of the findings. A third issue is that many registries experience high rates of "lost to follow-up," in which outcome information is unobtainable from the health care provider or the pregnant woman – in some cases as high as 40%. There is also a concern about bias involved with the timing in gestation when a pregnancy enters a registry, such as the later in gestation a pregnancy is enrolled, the more likely that prenatal diagnosis, pregnancy loss, or other adverse outcomes have already occurred – thus making the enrollment essentially retrospective.
Another concern is that few registries have a concurrently enrolled group of unexposed women for purposes of comparison. Thus, their findings are commonly compared to external reference statistics which may not be the most appropriate. Finally, in some registries, the absence of information on individual dose and specific timing in gestation of exposure may preclude evaluating the biological plausibility of any registry findings. All of these issues can lead to long delays in accumulation of sufficient information to draw meaningful inferences, and potential concerns about interpretation of results.
How can awareness of pregnancy registries and more representative enrollment of exposed women be improved? A variety of methods are used to inform physicians and their patients about existing pregnancy registries for the purpose of encouraging referrals, including the Food and Drug Administration website, information in product labeling and on product websites, direct to provider or direct to consumer advertising, and in commonly used resources for clinicians such as this column and Reprotox, an information system developed by the Reproductive Toxicology Center. However, with the rapidly increasing number of registries, it is challenging for physicians to remain current on which medications are being monitored through a registry, what the criteria for enrollment are, and how a physician or patient can find out more. Pregnancy registry designs that are disease based – such as encompassing all medications used to treat a specific disease in pregnancy – help simplify the referral process by broadening the criteria for enrollment. Particularly for specialty physicians, this can ease the burden of identifying eligible women for enrollment.