Drugs during pregnancy: an issue of risk classification and information to prescribers

The Swedish system for the classification of fetal risk of drugs was the first of its kind and was implemented in 1978. Drugs for use in pregnant women are classified in 4 general categories--A to D. The US Food and Drug Administration (FDA) introduced a system in 1979 also using the letters A to D, together with an X category. However, the definitions differ considerably between the FDA system and the Swedish system, resulting in a very different allocation of drugs to the respective categories. In the Swedish system, category A includes drugs that have been extensively used and/or for which there are reliable clinical data indicating no evidence of disturbance of the reproductive process. Category B includes drugs for which data from pregnant women are insufficient for making any solid estimation of human teratogenic risk, and classification is therefore based on animal data, with allocation to 3 subgroups. For products in category C, the pharmacological action of the drug may have undesirable effects on the human fetus or newborn infant. Finally, category D contains drugs for which human data indicate an increased incidence of malformations. The categorisation statement is always followed by a short explanatory text. In contrast to the FDA system, the Swedish system has been well accepted, as judged by an interview study including 934 physicians and pharmacists. We believe that much of the American dissatisfaction may be a consequence of shortcomings in the category definitions of the FDA system. The FDA system requires an unrealistically high quality of data, e.g. the availability of controlled studies in pregnant women that fail to demonstrate a risk to the fetus are needed for a drug to be assigned to category A. Consequently, the majority of drugs on the US market are allocated to category C, interpreted as 'risk cannot be ruled out'. The distribution of drugs into the various categories is thus very different between the Swedish and FDA systems. We think that the issue of this debate reflects a fundamental problem related to public health information: how should a large, compounded, changing and difficult to evaluate databank be organised before it is made available to professionals and secondarily to lay people?