Methodologic Problems Encountered in Investigating the Possible Teratogenic Effects of Drugs

Abstract

In evaluating the effect of drug usage on fetal development, the desired design — that of a randomized double blind trial — is not practicable. The large number of pregnancies required to identify a sufficient number with a condition which can be treated by different medications is a sufficient deterrent, even if all other obstacles can be overcome. Consequently, knowledge in this area must be derived not from controlled experimental studies but by careful observations of phenomena as they occur. Such uncontrolled observational studies may, at best, identify associations between drug usage and outcome of pregnancy, but rarely can they establish a definite cause-and-effect relationship. Even when the effect was explosive and resulted in a large increase in the incidence of a rare condition, as for example in the thalidomide tragedy, it required careful observations and ingenious deduction before it was possible to infer with reasonable certainty the drug that was responsible for the malformation. The problem is much more difficult in less dramatic situations, when we attempt to evaluate deleterious effects of relatively common drugs which are used in day-by-day obstetrical practice. The fact that the drug is in common use is proof enough that its injurious effect on the fetus, if it exists, is not of the explosive variety or it would have long been detected. Instead, the effect is more subtle and of a lower order of magnitude. Its presence can be suspected by an increased probability of an unsuccessful outcome of pregnancy which manifests itself in higher rates of abortion, stillbirth, neonatal death and/or congenital malformation. Ascertaining the increased rates and correlating them with drug usage can be accomplished only by very careful and painstaking study.