We write this letter in response to the Journal Club Roundtable discussion on our article, “Use of topiramate in pregnancy and risk of oral clefts.” We agree with the noted limitations and welcome the opportunity to briefly discuss recall and recall bias in the setting of our study.

Incomplete recall among the mothers of control infants, as discussed in the journal club, is always a concern in studies where exposure information is self-reported after the pregnancy outcome is known. During the period of data collection for our study, there was no information available to the public relating topiramate to oral clefts or other congenital malformations. Therefore, recall bias, if present, should have affected analyses on all malformations, inflating the estimated relative risks. The odds ratio of 1 for topiramate and major congenital malformations in the pooled data argues strongly against recall bias in the present study.

In addition, particularly given the specific queries about medication use that are included in the questionnaires by both the Boston University Slone Epidemiology Center Birth Defects Study and the Centers for Disease Control and Prevention’s National Birth Defects Prevention Study, we believe incomplete or erroneous recall of drug use is more likely in the case of a brief therapy episode aimed at symptom management than in the case of a long-term treatment of a serious chronic condition such as epilepsy or migraine.

In conclusion, we believe that recall bias in postmarketing case-control studies with self-reported exposure is a theoretical concern whose likelihood should be evaluated in light of the information that was available to the study participants at the time of data collection, the indications and patterns of use of the drugs of interest, and the rigor of the interview instrument. Sometimes, a null result like ours for major congenital malformations other than the malformation a priori hypothesized can empirically support the absence of recall bias.