Ondansetron for nausea and vomiting in pregnancy: re-evaluating the teratogenic risk

Nausea and vomiting in pregnancy (NVP) affects up to 80% of pregnant women with more severe hyperemesis gravidarum (HG) accounting for 1.5%. ¹ HG is associated with worse pregnancy outcome; however, the impact of NVP and HG on hospital admission and psychological wellbeing is substantial with 18% of women reporting post-traumatic stress and some women expressing a desire to end their pregnancy as a consequence of NVP/HG. ²

Antiemetic medication is the cornerstone for the treatment of NVP. Ondansetron is a 5-HT antagonist that was initially marketed to treat chemotherapy-related nausea and vomiting. Its superior efficacy meant that it was quickly adopted as a first-line antiemetic for other causes of nausea and vomiting, including NVP. It is commonly used ‘off-label’ to treat NVP/HG, in particular for inpatient management, where it is frequently prescribed. ³

In August 2019, the European Medicines Agency (EMA) published the minutes of a meeting from the Pharmacovigilance Risk Assessment Committee (PRAC), ⁴ where it was recommended that all manufacturers of ondansetron containing medicines were to update their product literature to state that:

Ondansetron is suspected to cause orofacial malformations when administered during the first trimester of pregnancy.

These recommendations are based on two studies which together included over 160,000 first trimester exposures to ondansetron.^5,6 Huybrechts identified an increased risk of cleft lip and palate (CLP), but no increased risk of cardiac defects, ⁵ while Zambelli-Weiner identified an increased risk of cardiac defects; however, this study has limitations including possible selection bias and exposure/outcome misclassification as indicated by a high rate of cardiac defects observed in comparison to the background rate. ⁶ While Huybrechts study appears to be methodologically robust, the issue with the study is not the result (a small increase in associated CLP with ondansetron exposure) rather the interpretation of risk. The background risk of CLP is small (accounting for 11 per 10,000 pregnancies); therefore, a small increase in relative risk means that the absolute risk from ondansetron exposure during the first trimester is approximately 14 per 10,000 pregnancies. This equates to an extra three cases of CLP per 10,000 exposed pregnancies. Arguably, one of the consequences of a very large sample size is the statistical amplification of a small signal, allowing it to reach statistical significance.

There are a number of issues with the EMA PRAC recommendations, which seem disproportionate to the small increase in absolute risk. Firstly, the wording of the recommendation is inaccurate: cohort studies are not able to establish causation. Ondansetron has not been shown to ‘cause’ orofacial malformations, rather ondansetron has been ‘associated with’ a small increase in orofacial malformations. Although a subtle difference in terms of language, the implications are very different. Secondly, although the EMA PRAC are correct that there are ‘conflicting results’ with regard to cardiac malformations in that two studies do not agree, the studies are very different. The more robust study, which represents the best evidence to date, found no statistically significant increased risk of cardiac defects overall after adjusting the results for a large number of co-variates (aRR 1.01, 95% CI: 0.92–1.12). By stating that there are ‘conflicting results’, the EMA PRAC imply equal weight to the two studies, whereas the weight of evidence supports no such increase in cardiovascular risk.

Thirdly, the statement that ‘ondansetron should not be used in the first trimester’ is the most problematic since many health care professionals will take this recommendation at face value. The EMA is, after all, a respected European regulatory authority. Why is this statement so strong? Do the EMA believe ondansetron to be a major teratogen despite evidence to the contrary? Undoubtedly, there are risks and benefits to be considered for all medications taken in pregnancy. Ondansetron comes with a small risk, which women may be willing to take if it means that they can eat, drink and function normally. Paradoxically, we know more about ondansetron (due to the large sample sizes) than most other drugs, including other antiemetics that are now considered first line treatment for NVP. Women need to be given information about risks and benefits and be allowed to make an informed choice, rather than a blanket ban on the use of effective treatment based on small risks. The statement issued by the EMA PRAC has substantial ramifications for patients. Hospitals in the UK have already re-written guidelines forbidding ondansetron use, despite our own Medicines and Healthcare product Regulatory Authority (MHRA) issuing a more balanced statement urging clinicians to consider the risks and benefits of ondansetron when considering its use in early pregnancy. ⁷ By contrast, the MHRA are to be commended on seeking a wide range of professional and patient opinion including patient groups in their approach to the same signal.

By adopting a paternalistic, ‘black-and-white’ approach to drug safety in pregnancy, the EMA have failed pregnant women. It is now down to informed and interested clinicians and patient groups to correctly counsel women and to try to undo the damage caused by negative publicity. This episode raises important questions about how and when to issue drug safety warnings in pregnancy, the process through which recommendations are made and the implications to clinical practice and to pregnant women themselves in getting the message wrong.