The constant publication of clinical trials may give the impression that for every question, there is an answer. The reality, however, is that many clinical questions do not lend themselves to investigation in a randomized, controlled fashion. Whether assisted reproductive technology causes an increased rate of birth defects is one such question. Although for some time there has been a known association between birth defects and infertility treatments, it has been impossible to know whether infertility treatments actually cause birth defects, or whether this association is actually due to the underlying parental factors that necessitate infertility treatment in the first place. Moreover, it has also remained unclear whether specific types of fertility treatments, such as in vitro fertilization, (IVF) or intracytoplasmic sperm injection, (ICSI), differentially increase birth defect risk. How to find the answer?
True randomization remains elusive, but in “Reproductive Technologies and the Risk of Birth Defects,” Davies et al of the Robinson Institute in Australia present data from the largest registry to-date, further untangling some of these complicated associations. In this population-wide cohort study, the authors compared over 6000 births from assisted reproductive technology with some 300,000 unassisted conceptions to assess associated rates of birth defects. The following types of births were compared: births from each mode of infertility treatment, births as a result of unassisted conception among women who required assisted reproductive technology previously, births to women with a history of infertility who had not received reproductive technology and births to women with neither a history of infertility nor treatment.
What did they find? The risk of a birth defect appeared significantly higher in pregnancies conceived using either in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI), as compared with unassisted births, with an adjusted odds ratio of 1.28. Various types of congenital defects were assessed for their association with assisted reproduction. Assisted conception was associated with an increased rate of any defect and multiple defects, as well as with cardiovascular, urogenital, gastrointestinal abnormalities, and cerebral palsy. It should be noted that there were also differences, such as in age, race, and medical conditions including diabetes and pregnancy-induced hypertension, between mothers who required assisted reproduction and those who didn’t; after adjustment for such factors IVF was no longer associated with a higher rate of birth defects, although there was still an increased risk (albeit attenuated) associated with ICSI.
What to make of these data? In the absence of randomization, these results must be interpreted with caution, as potential confounders remain. In this specific case, paternal data were unavailable, raising the question of whether the residual risk that remained was related to parental factors, rather than to the ICSI itself. Furthermore, an increased risk of birth defects was also found among women with prior infertility without assisted reproductive technology, again suggesting that parental factors may be contributing, as opposed to the technology. Moreover, it is important to note that the absolute magnitude of the risk increases seen with assisted reproductive technologies was modest. The risk of any birth defect was 8.3% for those who received assisted technology, versus 5.8% for those who conceived without assistance.
That said, patient counseling should likely note the presence of a persistent association between IVF and birth defects, despite controlling for maternal factors known to cause birth defects. More importantly, as far as identification of problems that are “actionable,” the possibility of a more elevated risk of ICSI warrants further investigation, as this may be an approach couples ultimately choose when clearly needed for conception.
Ultimately, these data don’t deliver answers that are as definitive as we would like. However, given the virtual impossibility of randomized trials with this particular cohort, these may be the best answers we will ever get.