Using observational cohort data for studying drug effects on pregnancy outcome--methodological considerations.

Clinical data are urgently needed to specify the risk and safety of drug use during pregnancy. For several reasons pregnant women are usually excluded from clinical studies. Therefore, observational data are the main source of knowledge, cohort studies as well as case-control studies. Disadvantages of cohort studies based on observational data have been repeatedly discussed. However, being involved in individual risk characterisation of pregnant women it is the experience of clinical teratologists that even reports on small cohorts should not be disregarded if no other data are available. The recently published "Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement" underlines the value of observational data and provides a checklist regarding the most important inherent methodological problems. Our article describes how Teratology Information Services (TIS) document and evaluate their observations on pregnant women exposed to various drugs and discusses methodological problems and--considering the STROBE statement--how these could be addressed.