Big risks in small groups: The difference between epidemiology and counselling


  • An earlier version of this paper was presented as the Thomas H. Shepard Lecture to the Organization of Teratology Information Specialists on June 24, 2006.


Congenital anomalies do not occur in all babies born after a teratogenic exposure. Whether a given exposure is teratogenic depends on the chemical nature and physical properties of the agent, the dose and route of exposure, when in pregnancy the exposure occurs, and genetic and other factors that affect susceptibility. Teratogenic birth defects are inherently multifactorial. Absolute risk, relative risk, and population attributable risk provide useful but different information regarding teratogenic effects. Statistical significance and clinical significance also are important considerations, but they may not be concordant. Demonstrating a teratogenic effect is easier if it is sought in a subgroup of patients in whom the effect is likely to be particularly prominent. The ability to detect a significant risk is, therefore, generally increased by subgroup analysis of epidemiology studies, but the greater the number of analyses performed, the higher the probability of finding associations that reach nominal statistical significance by chance alone. This problem is well recognized, but it is difficult to solve. The only compelling evidence for the reality of an association between maternal exposure to an agent during pregnancy and teratogenic effects in the children is replication of the findings in independent studies, but this is hard to obtain. As a consequence, there are very few exposures for which the available information is sufficient to make evidence-based recommendations regarding the clinical management of teratogenic risks. It is important to admit these limitations and to learn more about exposures that cause birth defects and how to prevent them. Birth Defects Research (Part A) 2009. © 2009 Wiley-Liss, Inc.