Nausea and vomiting of pregnancy affects approximately 80% of pregnancies, beginning as early as 5 weeks of gestation and typically ending before the second trimester.1 Among affected women, 15% use prescription or over-the-counter medications for treatment.2 Although nausea and vomiting of pregnancy overlaps with embryologic development, there are few data on safety with respect to risks of structural malformations for many medications.3
With no U.S. Food and Drug Administration (FDA)–approved medications available until recently, pregnant women have been using treatments such as herbal products, supplements, over-the-counter medications, and off-label use of prescription antiemetics. Ondansetron (Zofran and generics) is a serotonin 5-HT3 receptor antagonist that is approved by the FDA only for the treatment of nausea and vomiting related to chemotherapy, radiation therapy, and surgery. Although not approved for nausea and vomiting of pregnancy, ondansetron has become the most commonly used prescription antiemetic for this condition in the United States.4,5
Although some studies found no association between ondansetron and risks of malformations overall,6–9 grouping malformations into a single outcome category is problematic because teratogens do not affect the development of all defects.10 Findings from studies focused on specific defects have identified increased risks of renal obstructive defects11 and cardiovascular defects,6,12 specifically septal defects,6 but these studies were based on small numbers and specific types of septal defects were not examined. The National Birth Defects Prevention Study, covering births from 1997 to 2004, examined neural tube defects, cleft palate, cleft lip with or without cleft palate, and hypospadias; it identified an increased risk for cleft palate.2 However, these studies were also limited by small numbers.
We used data from the National Birth Defects Prevention Study and another large case–control study of birth defects, the Birth Defects Study at Slone Epidemiology Center, to examine previously reported associations between ondansetron and birth defects and to identify possible associations with birth defects not previously described.
MATERIALS AND METHODS
The National Birth Defects Prevention Study and Birth Defects Study were multisite, case–control studies designed to investigate risk factors such as medication use for birth defects. Detailed information on the study designs13,14 is available elsewhere.
Briefly, from 1997 to 2011, the National Birth Defects Prevention Study identified more than 30 selected major birth defects among live births, stillbirths, and (in selected sites) elective terminations through surveillance programs in Arkansas, California, Georgia, Iowa, Massachusetts, New Jersey, New York, North Carolina, Texas, and Utah. Diagnoses were confirmed by clinical geneticists. Control neonates were a random sample of liveborn neonates without birth defects identified from birth certificates or birth hospitals in the same geographic regions and time periods as those in the case group. Mothers were contacted up to 24 months after delivery to complete a computer-assisted telephone interview.
Between 1976 and 2014, the Birth Defects Study identified neonates with any major birth defects among live births, stillbirths, and elective terminations. For the years of this analysis (1997–2014), participants in the case group were identified by review of discharge records or registry data at participating hospitals or birth defect registries in parts of Massachusetts, areas surrounding Philadelphia, Pennsylvania, San Diego, California (since 2001), parts of New York State (since 2004), and Nashville, Tennessee (since 2012). Control participants were liveborn neonates without malformations identified from study hospitals and birth certificates in the same catchment areas as case participants. Mothers were contacted within 6 months of delivery by a study nurse who conducted a computer-assisted telephone interview. Mothers were asked for permission to obtain their children's medical records, but all interviewed women were included.
Although both studies had centers in Massachusetts, California, and New York, eligibility criteria prevented participants from being included in both studies. We obtained institutional review board approval for all participating sites. In both data sets, the occurrence of nausea and vomiting of pregnancy was self-reported in the interview conducted up to 24 months and 6 months after delivery in the National Birth Defects Prevention Study and Birth Defects Study, respectively. Among women reporting nausea and vomiting of pregnancy, data were obtained on frequency, duration, and timing. We restricted this analysis to women reporting first-trimester nausea and vomiting of pregnancy.
“First trimester” was defined differently in the two studies. The National Birth Defects Prevention Study interview used the estimated date of conception as the interview reference date to collect information with the first three 30-day months (ie, 90 days) after the estimated date of conception constituting the first trimester. The Birth Defects Study used the last menstrual period date as the interview reference date to collect information with the 4 lunar months (112 days) after the last menstrual period constituting the first trimester. Thus, the Birth Defects Study first trimester included the 2 weeks before and 98 days after the estimated date of conception.
Among women with nausea and vomiting of pregnancy, both studies used standardized interviews to capture treatments with prescription and nonprescription medications, herbal products, and supplements. Both used the Slone Drug Dictionary (Kelley K, Kelley T, Kaufman D, Mitchell A. The Slone Drug Dictionary: a research driven pharmacoepidemiology tool [abstract]. Pharmacoepidemiol Drug Saf 2003;12:S168–9.) to code and classify all exposures. Women were classified into hierarchical first-trimester treatment categories: 1) ondansetron (with or without other prescription antiemetics); 2) other prescription antiemetic drugs or intravenous (IV) fluids; and 3) no treatment for nausea and vomiting of pregnancy with any medication (including prescription medications, IV fluids, over-the-counter, herbal products, and supplements) (“no treatment”); this last category served as the reference in all analyses. The other prescription antiemetic drugs or IV fluids group was created to address potential confounding by indication, because these women presumably had nausea and vomiting of pregnancy severe enough to prompt use of a prescription drug or IV fluids (other prescription antiemetic drugs or IV fluids are provided in Appendix 2, available online at http://links.lww.com/AOG/B102). Women reporting only over-the-counter medications, herbal products, or supplements for treatment were excluded from the etiologic analyses as were women who used ondansetron or other prescription antiemetic drugs or IV fluids only outside the first trimester.
In each data set, specific birth defects or defect groups were included in the analysis if there were a minimum of 100 cases in total and a minimum of four ondansetron-exposed cases. Several birth defects were not considered in one data set as a result of nonascertainment of the defect or insufficient sample size. Participants in the case group with chromosomal abnormalities and single-gene disorders were excluded.
To accurately reflect the prevalence of first-trimester ondansetron use among control mothers in each study population, this analysis included all control women reporting first-trimester nausea and vomiting of pregnancy. Two-year averages were calculated based on estimated date of delivery year (1997–2011) in the National Birth Defects Prevention Study and last menstrual period year (1997–2014) in the Birth Defects Study.
In the National Birth Defects Prevention Study , we included participants with estimated delivery dates from 1997 to 2011. However, for the four defects (neural tube defects, cleft lip with or without cleft palate, cleft palate, hypospadias) that were previously analyzed (1997–2004),2 we restricted the repeat analyses to participants with estimated delivery dates from 2005 to 2011. In the Birth Defects Study, analyses included all participants with last menstrual periods from 1997 to 2014. In both data sets, we excluded multiple gestations and terminations. As a result of differences in case ascertainment and control selection, data from each study were analyzed separately.
The distributions of maternal characteristics by exposure category (ondansetron, other prescription antiemetic drugs or IV fluids, no treatment) among participants in the control group were described for each data set. Using the no treatment category as the reference, we investigated associations between ondansetron and other prescription antiemetic drugs or IV fluids with the risk of birth defects. Other prescription antiemetic drugs or IV fluids were included as a secondary exposure group only when there were at least four exposed participants in the case group. First, we revisited birth defects previously studied in the National Birth Defects Prevention Study2: neural tube defects, cleft palate, cleft lip with or without cleft palate, and hypospadias. Second, we considered defects reported in other studies: septal defects6 and renal collecting system anomalies.11 Third, we sought to identify unreported associations between ondansetron and birth defects. When numbers permitted, specific defects instead of a collapsed category were considered (eg, atrial septal defects and ventricular septal defects instead of septal defects overall).
We used logistic regression models to calculate adjusted odds ratios (ORs) and 95% CIs. All adjusted models include the following covariates selected a priori: maternal age (younger than 25, 25–34, 35 years or older), maternal education (less than 12, 12, greater than 12 years), periconceptional folic acid use (yes, no), study year (1997–1999, 2000–2002, 2003–2005, 2006–2008, 2009–2011, and for the Birth Defects Study only, 2012–2014), and study site. Crude ORs are included in the supplementary materials (Appendix 3, available online at http://links.lww.com/AOG/B102).
Sensitivity analyses were performed to test the robustness of findings when elevated point estimates were observed and discrepant between the two data sets.
Of the 11,471 control mothers in the National Birth Defects Prevention Study and 10,701 control mothers in the Birth Defects Study, first-trimester nausea and vomiting of pregnancy was reported by 70% (n=8,034) and 62% (n=6,619), respectively (Appendix 4, available online at http://links.lww.com/AOG/B102). Among these mothers, first-trimester ondansetron use increased from less than 1% before 2000 to 10.4% in 2011 (National Birth Defects Prevention Study) and 13.4% in 2013–2014 (Birth Defects Study). The increase in use was similar in both studies (Fig. 1).
After exclusions, there were 6,751 National Birth Defects Prevention Study and 5,873 Birth Defects Study participants in the control groups (Appendix 4, http://links.lww.com/AOG/B102) and 14,667 National Birth Defects Prevention Study participants in the case group and 8,533 Birth Defects Study participants in the case groups. The overall prevalences of ondansetron use among participants in the control group with first-trimester nausea and vomiting of pregnancy were 3.7% (National Birth Defects Prevention Study) and 6.4% (Birth Defects Study); the prevalences of use of other prescription antiemetic drugs or IV fluids were 5.9% (National Birth Defects Prevention Study) and 2.4% (Birth Defects Study). Table 1 presents the distributions of maternal characteristics according to exposure group. In both studies, mothers reporting ondansetron use were more likely to be white, non-Hispanic (81.0% National Birth Defects Prevention Study and 70.7% Birth Defects Study) compared with mothers using other prescription antiemetic drugs or IV fluids (70.5% and 66.2%, respectively) or no treatment (58.3% and 68.3%, respectively). The same pattern was observed for women with more than 12 years of education. In the National Birth Defects Prevention Study, ondansetron users were more likely to be older than those in both the other prescription antiemetics and no treatment groups, whereas in the Birth Defects Study, participants in the ondansetron and other prescription antiemetics groups were younger than those in the no treatment group. Differences in use of ondansetron also varied by folic acid use and study site.
The analyses of defects previously examined in the 1997–2004 National Birth Defects Prevention Study data are presented in Table 2; spina bifida and anencephaly are described separately instead of the neural tube defect category used previously. In National Birth Defects Prevention Study data from 2005 to 2011, adjusted ORs for ondansetron and both anencephaly and spina bifida were null. In Birth Defects Study data, the association between ondansetron and spina bifida was 1.4 with a 95% CI that included the null (95% CI 0.8–2.5). There was no association between ondansetron and hypospadias or cleft lip with or without cleft palate in either study. However, for cleft palate, the association with ondansetron was elevated in the National Birth Defects Prevention Study but not the Birth Defects Study. Adjusted ORs were 1.6 (95% CI 1.1–2.3) in the National Birth Defects Prevention Study and 0.5 (95% CI 0.3–1.0) in the Birth Defects Study, whereas adjusted ORs were similar for other prescription antiemetic drugs or IV fluids in the two studies (1.2 and 1.5, respectively).
To explore the discrepant findings for cleft palate in the two studies, we conducted several sensitivity analyses (Appendix 5, available online at http://links.lww.com/AOG/B102). First, because there is no consensus regarding when palate closure occurs, we repeated the cleft palate analyses with exposure windows that were shorter than those used in the both main analyses. Second, we excluded participants with Pierre Robin sequence. Third, we restricted analyses to isolated participants in the case group. Finally, we separated first-trimester ondansetron users into exposure to categories of only ondansetron and ondansetron plus other prescription antiemetic drugs or IV fluids. None of these sensitivity analyses explained the differences observed in the main analysis.
For both groups of septal defects explored, ventricular septal defect and atrial septal defect secundum, there was no increased risk associated with ondansetron. The adjusted OR for the association between ondansetron and renal collecting system defects in the Birth Defects Study was also null (Table 3).
Among noncardiac defects not previously described (Table 4) that were analyzed in both studies, there were no increased risks associated with ondansetron exposure compared with no treatment. Among defects analyzed in only one study, we observed in the National Birth Defects Prevention Study an increased risk of diaphragmatic hernia for ondansetron (adjusted OR 1.5, 95% CI 1.0–2.4), but not for other prescription antiemetic drugs or IV fluids (adjusted OR 1.0, 95% CI 0.7–1.5). In the Birth Defects Study, an increased risk for renal agenesis–dysgenesis was observed (adjusted OR 1.8, 95% CI 1.1–3.0); there were insufficient exposed cases among other prescription antiemetic drugs or IV fluids for analysis. In sensitivity analyses exploring this association further, we limited cases to those whose diagnosis was confirmed by the medical record and separately considered isolated cases (Appendix 6, available online at http://links.lww.com/AOG/B102). The adjusted OR for confirmed cases was 1.9 (95% CI 1.2–3.1) and for isolated cases was 2.6 (95% CI 1.3–4.9). Among cardiac defects, we observed adjusted ORs of 1.5 (95% CI 0.9–2.5) and 1.2 (95% CI 0.6–2.4) for ondansetron exposure and hypoplastic left heart syndrome in the National Birth Defects Prevention Study and Birth Defects Study, respectively.
In analyses of data from these two studies, we observed that off-label use of ondansetron for the treatment of nausea and vomiting of pregnancy increased rapidly over the study period, with an estimated 13% of these women using ondansetron by 2013–2014. Etiologic analyses for the majority of specific birth defects that met inclusion criteria identified no increased risk for first-trimester use of ondansetron when compared with no treatment.
As was observed in 1997–2004,2 more recent National Birth Defects Prevention Study data (2005–2011) provided support for no association for neural tube defects (specifically, spina bifida and anencephaly), and the Birth Defects Study analysis identified a modestly increased risk for spina bifida with a 95% CI that included the null. Both data sets also supported earlier findings of no increased risk for cleft lip with or without cleft palate and hypospadias. The findings for cleft palate were less consistent. Anderka et al,2 using earlier National Birth Defects Prevention Study data, report an adjusted OR of 2.4 (95% CI 1.2–4.8) for cleft palate based on 11 exposed cases. In the more recent National Birth Defects Prevention Study data, we observed an attenuated, yet still elevated, adjusted OR of 1.6 (95% CI 1.1– 2.4) based on 40 exposed cases. By contrast, the adjusted OR in Birth Defects Study data was 0.5 (95% CI 0.3–1.0). We conducted a number of sensitivity analyses, but none explained the discrepant findings for cleft palate.
For defects with increased risks identified in other prior studies, we did not confirm the finding for renal obstructive defects11 in the Birth Defects Study data. For septal defects,6 we took advantage of both data sets classifying specific types of septal defects (atrial septal defects and ventricular septal defects) and also found no associations.
We also analyzed data on approximately 40 additional defects. The adjusted ORs for the majority of these defects did not support increased risks; for those that did, confidence bounds included the null, except for renal agenesis–dysgenesis in the Birth Defects Study. Diagnostic misclassification did not account for that finding, and the adjusted OR was modestly higher for isolated cases; however, numbers were insufficient to compare this finding with exposure to other prescription antiemetic drugs or IV fluids in the Birth Defects Study. The National Birth Defects Prevention Study had too few cases to analyze. This association warrants attention in other studies.
We addressed concern about confounding by indication by restricting the analysis to women reporting nausea and vomiting of pregnancy, because it may be associated with a minimal reduction in the risk of birth defects.2 Furthermore, we created a group of women who used prescription antiemetics other than ondansetron. Thus, when elevated adjusted ORs were identified for ondansetron, we were able to control for severity, at least in part, by comparing those estimates with those observed for women who took prescription antiemetics other than ondansetron. Women relying on over-the-counter treatments were excluded because their severity of nausea was likely not comparable.
Data from the two studies were analyzed separately because of differences in the conduct of each. Although the definition of first trimester differed, both included the etiologic-relevant window for most structural birth defects. Another difference is the longer time to interview in the National Birth Defects Prevention Study, which may increase the likelihood of inaccurate recall, resulting in greater exposure misclassification and possible recall bias if the misclassification differed among case and control mothers. Finally, differences existed in case inclusion and classification. We excluded terminations as a result of inconsistencies in the collection of these cases, which may have resulted in underestimates of associations if ondansetron use was more common among women who terminated their pregnancies.
Although the use of two large case–control studies provided robust estimates for some defects, partly as a result of the increase in ondansetron exposure, the number of exposed cases for many specific birth defects was small, leading to imprecise estimates. Nonetheless, estimates for the large majority of defects did not support increases in risks. Two exceptions were cleft palate in the National Birth Defects Prevention Study and renal agenesis–dysgenesis in the Birth Defects Study. The cleft palate finding was not replicated in the Birth Defects Study analysis, and renal agenesis–dysgenesis has not previously been considered. For both outcomes, the possibility of chance cannot be excluded given the large number of defects studied.
Strengths include the use of two data sets with sufficient sample sizes to investigate many specific major birth defects. Maternal interviews provided information on nausea and vomiting of pregnancy and medication exposure, and the latter reflected reported use and gestational timing rather than prescriptions issued or filled but not taken.
The current data covered a time in which there was no FDA-approved treatment of nausea and vomiting of pregnancy in the United States. In 2013, the FDA approved Diclegis, a rebranded formulation of Bendectin, for the treatment of nausea and vomiting of pregnancy. The National Birth Defects Prevention Study concluded before the introduction of Diclegis and the Birth Defects Study concluded shortly after it, capturing only 25 reports of Diclegis use among approximately 3,000 women. It is unclear how Diclegis availability will affect patterns of ondansetron use and other prescription antiemetics. A recent systematic review documented the effectiveness of ondansetron for severe nausea and vomiting of pregnancy,15 suggesting that ondansetron may remain an attractive treatment option. However, understanding the risks of specific birth defects associated with first-trimester ondansetron use remains challenging and ondansetron use for first-line treatment of nausea and vomiting of pregnancy is inconsistent with current guidelines. Our study contributes to the accruing body of literature regarding the safety of this medication, although additional studies are needed to confirm current and previous findings.
1. Chan RL, Olshan AF, Savitz DA, Herring AH, Daniels JL, Peterson HB, et al. Maternal influences on nausea and vomiting in early pregnancy. Matern Child Health J 2011;15:122–7.
2. Anderka M, Mitchell AA, Louik C, Werler MM, Hernandez-Diaz S, Rasmussen SA, et al. Medications used to treat nausea and vomiting of pregnancy and the risk of selected birth defects. Birth Defects Res A Clin Mol Teratol 2012;94:22–30.
3. Thorpe PG, Gilboa SM, Hernandez-Diaz S, Lind J, Cragan JD, Biggs G, et al. Medications in the first trimester of pregnancy: most common exposures and critical gaps in understanding fetal risk. Pharmacoepidemiol Drug Saf 2013;22:1013–8.
4. Mitchell AA, Gilboa SM, Werler MM, Kelley KE, Louik C, Hernández-Diaz S, et al. Medication use during pregnancy, with particular focus on prescription drugs: 1976–2008. Am J Obstet Gynecol 2011;205:51.e1–8.
5. Taylor LG, Bird ST, Sahin L, Tassinari MS, Greene P, Reichman ME, et al. Antiemetic use among pregnant women in the United States: the escalating use of ondansetron. Pharmacoepidemiol Drug Saf 2017;26:592–6.
6. Danielsson B, Wikner BN, Källén B. Use of ondansetron during pregnancy and congenital malformations in the infant. Reprod Toxicol 2014;50:134–7.
7. Einarson A, Maltepe C, Navioz Y, Kennedy D, Tan MP, Koren G. The safety of ondansetron for nausea and vomiting of pregnancy: a prospective comparative study. BJOG 2004;111:940–3.
8. Fejzo MS, MacGibbon KW, Mullin PM. Ondansetron in pregnancy and risk of adverse fetal outcomes in the United States. Reprod Toxicol 2016;62:87–91.
9. Pasternak B, Svanström H, Hviid A. Ondansetron in pregnancy and risk of adverse fetal outcomes. N Engl J Med 2013;368:814–23.
10. Mitchell AA. Special considerations in studies of drug-induced birth defects. In: Strom BL, editor. Pharmacoepidemiology. 2nd ed. Chichester (United Kingdom): John Wiley and Sons, 1994. p. 595–608.
11. Colvin L, Gill AW, Slack-Smith L, Stanley FJ, Bower C. Off-label use of ondansetron in pregnancy in Western Australia. Biomed Res Int 2013;2013:909860.
12. Andersen JT, Jimenez-Solem E, Andersen NL. Ondansetron use in early pregnancy and the risk of congenital malformation—a register based nationwide control study. Montreal (Canada): International Society of Pharmacoepidemiology; 2013.
13. Reefhuis J, Gilboa SM, Anderka M, Browne ML, Feldkamp ML, Hobbs CA, et al. The National Birth Defects Prevention Study: a review of the methods. Birth Defects Res A Clin Mol Teratol 2015;103:656–69.
14. Louik C, Lin AE, Werler MM, Hernández-Díaz S, Mitchell AA. First-trimester use of selective serotonin-reuptake inhibitors and the risk of birth defects. N Engl J Med 2007;356:2675–83.
15. O'Donnell A, McParlin C, Robson SC, Beyer F, Moloney E, Bryant A, et al. Treatments for hyperemesis gravidarum and nausea and vomiting in pregnancy: a systematic review and economic assessment. Health Technol Assess 2016;20:1–268.