Selection of controls in case-control studies on maternal medication use and risk of birth defects

In case-control studies on teratogenic risks of maternal drug use during pregnancy, the use of normal or malformed controls may lead to recall-bias or selection bias. This can be avoided by using controls with a genetic disorder. However, researchers are hesitant to use these as controls because it is unknown whether their selection is independent of exposure status. The aim of this study is to investigate whether first trimester drug use among mothers of children with genetic disorders is representative for the "general pregnant population".

From a birth defects registry 565 mothers of infants with a genetic disorder born between 1998-2004 were selected (the "genetic population"). The first trimester exposure rate was calculated for prescription-only drugs as the number of exposed women per 100. By calculating the rate ratio (RR) and 95% CI, the exposure rates in the genetic population were compared with those in the "source population" obtained from a population-based prescription database and consisting of 10,870 mothers who gave birth to a child between 1998-2004.

The mean age at birth was 32.1 for the genetic population and 29.6 for the source population (p = .000). In the genetic population, a higher use was found for antimigraine medication (RR = 2.7, 95% CI = 1.0-7.8) and for ovulation stimulants (RR = 1.6, 95% CI = 1.0-2.6). After adjustment for maternal age, the difference in use of ovulation stimulants disappeared.

Except for antimigraine medication, first trimester drug use among mothers of infants with genetic disorders is representative for the general pregnant population.