Another Study on Antidepressants and Birth Defects: Does it Change Our Decision-Making?
by Ruta Nonacs, MD PhD, MGH Center for Women's Mental Health
Aug 26, 2020

Prozac (fluoxetine) was the first SSRI or selective serotonin reuptake inhibitor. Since its release in 1988, we have been gathering data on the reproductive safety of fluoxetine and other SSRIs. Over the last 32 years, we have collected an enormous amount of data on the use of SSRIs during pregnancy - far more information than we have on other medications commonly used during pregnancy. The data has been reassuring, and based on this veritable mountain of data, we can conclude that SSRIs do not carry significant teratogenic risk. In addition, a smaller but also reassuring body of data indicates that prenatal exposure to the SNRIs venlafaxine and duloxetine are not associated with an increased risk of malformations. While we feel comfortable using SSRIs and SNRIs during pregnancy in light of the evidence base we currently have, articles on the reproductive safety of antidepressants continue to be published in high impact journals. For example, this month JAMA Psychiatry published a study from Anderson and colleagues looking at risk for congenital malformations associated with prenatal antidepressant exposure in a case control study, the National Birth Defects Prevention Study.

In our opinion, the findings of this study are of interest but are unlikely to influence our decision making, as this study is but one study looking at the outcomes of children exposed to antidepressants in utero. Looking at the entire body of studies addressing the reproductive safety of antidepressants, there is little evidence to suggest that SSRI and SNRI antidepressants are teratogens.

Nonetheless, this is an opportunity to discuss some of the limitations and problems inherent in studying the reproductive safety of antidepressants and other medications.

Findings of the Study
The National Birth Defects Prevention Study (October 1997-December 2011) is a population-based, multicenter case-control study consisting of 30,630 infants ("cases") with selected birth defects identified from surveillance systems. Controls (n= 11,478) were randomly sampled live-born infants without any major birth defects. Mothers provided information regarding exposures in an interview after the expected delivery date.

Early pregnancy antidepressant use was reported by 1562 case mothers (5.1%) and 467 control mothers (4.1%). Exposure to several different SSRI antidepressants was associated with increased risk of several different types of cardiovascular malformations. However, these associations were attenuated and no longer significant after controlling for potential confounding factors. This is consistent with other studies examining SSRI use and the risk of cardiovascular malformations, which have also concluded that associations between antidepressants and heart defects may be attributable to the underlying condition or other associated factors.

The authors noted other associations between antidepressant exposure and non-cardiovascular malformations. For example, citalopram was associated with diaphragmatic hernia (aOR, 5.11; 95% CI, 1.29-20.24). Venlafaxine exposure was associated with multiple defects which persisted after partially accounting for underlying conditions, including certain neural tube defects (aOR, 9.14; 95% CI, 1.91-43.83). While these findings are arguably the most novel aspects of this study, it is perplexing why the data are not included in the article but are provided instead in a less accessible supplemental section. It is important to note that no other studies have documented an association between these antidepressants and these particular malformations. Furthermore, the numbers of cases included in this analysis are very small and contribute to considerable statistical variability.

Understanding the Limitations of this Study
In an excellent accompanying article, Drs, Katherine Wisner, Tim Oberlander, and Krista Huybrechts, discuss many of the limitations inherent in the Anderson study. These limitations apply to many of the studies looking at the reproductive safety of medications in the absence of randomized trials.

Study Design: Case control studies like the one used here can be very useful when looking for risk factors affecting the incidence of rare outcomes or when there is a long latency period between exposure and the manifestation of a particular outcome. However, there are many disadvantages. Case control studies are susceptible to selection bias. With studies of birth defects, selection bias can occur when prenatal diagnosis of certain malformations increases the probability of pregnancy termination and when termination differs by exposure status. In addition, they are not reliable for accurately calculating the incidence of a particular incidence and often overestimate risk.

Recall Bias: In this study, mothers reported medication usage using a computer-assisted telephone interview 6 weeks to 24 months after delivery. In women interviewed retrospectively, how accurate are reports regarding medication start and stop dates, duration, and frequency of medication use? Do they accurately report the frequency of alcohol and tobacco use? What about other pertinent exposures? Furthermore, it is likely that the recall of women who gave birth to an infant with a congenital malformation will differ from the recall of women who gave birth to a healthy infant. Specifically, women who have an infant with a malformation may report more accurately or even over-report exposures, whereas women who deliver a healthy infant may not recall pertinent exposures; this pattern of under- and over-reporting would result in an overestimation of risk associated with exposure.

Role of Confounding Factors: In these non-randomized studies, women who use antidepressants typically differ from women who do not take medication. While antidepressant exposure is the primary variable being measured, there are potentially other factors which may be more common among women who take antidepressants and may influence outcomes. The researchers were able to control for some of these potential confounders, including body mass index, use of tobacco and alcohol, but other factors including socioeconomic status and adherence to prenatal care were not included. Most importantly, they were not able to control for one of the more important confounding factors: exposure to underlying psychiatric illness.

Putting this Study into a Clinical Context
Rather than focusing on the findings of a single study, we must put this study into a larger context. Looking at the vast body of data we have accumulated over the last two decades, including several large meta-analyses combining studies with thousands of exposures to SSRIs, we have not observed an increase in risk of congenital malformations in children exposed to these antidepressants. Our data include smaller numbers of exposures to the SNRIs, venlafaxine and duloxetine; however, a pooled analysis of 3186 venlafaxine-exposed and 668 duloxetine-exposed infants demonstrated no increase in teratogenic risk.


  1. Anderson KN, Lind JN, Simeone RM, Bobo WV, Mitchell AA, Riehle-Colarusso T, Polen KN, Reefhuis J. Maternal Use of Specific Antidepressant Medications During Early Pregnancy and the Risk of Selected Birth Defects. JAMA Psychiatry. 2020 Aug 5:e202453.
  2. Huybrechts KF, Palmsten K, Avorn J, Cohen LS, Holmes LB, Franklin JM, Mogun H, Levin R, Kowal M, Setoguchi S, Hernández-Díaz S. Antidepressant use in pregnancy and the risk of cardiac defects. N Engl J Med. 2014 Jun 19;370(25):2397-407.
  3. Wisner KL, Oberlander TF, Huybrechts KF. The Association Between Antidepressant Exposure and Birth Defects-Are We There Yet? JAMA Psychiatry. 2020 Aug 5.