Last week, an article looking at the cognitive and motor development of children with prenatal exposure to antidepressants was published in JAMA Psychiatry.  Since that time, we have received a lot of questions regarding the article and the clinical implications of its findings.  When I did a quick Google search, I found 36 articles reviewing this study published within the last week.  It is unfortunate, however, that none of the headlines seem to accurately reflect the results of the study and may be misleading and unnecessarily alarming.  

This was a prospective birth cohort study which analyzed data from medical register data in Finland between 1996 to 2010. The sample included 845,345 pregnant women and their singleton offspring; for all subjects included in the final analysis, data was available on the maternal use of antidepressants and depression-related psychiatric disorders during pregnancy.  

The researchers compared outcomes in three groups of children:

  • 15,596 children with SSRI exposure and a mother with a depression-related diagnosis during pregnancy (medicated)
  • 9,537 with no SSRI exposure and a mother diagnosed with depression-related psychiatric disorder during pregnancy (unmedicated)
  • 31,207 with no SSRI exposure and no depression-related psychiatric disorder during pregnancy (unexposed)

Children up to 14 years of age were included; speech/language, scholastic, and motor disorders were identified in 829, 187, and 285 children, respectively.  The researchers observed that the offspring of mothers who purchased SSRIs at least twice during pregnancy had a 1.37-fold increase in risk for speech/language disorders compared to the offspring in the unmedicated group. Compared to the unexposed group, there was an 1.63-fold increase in risk of speech/language disorders in the SSRI-exposed offspring.  There were no significant differences in risk for the other disorders in the SSRI-exposed group and the unmedicated group.

Based on these findings, it appears that if the mother suffers from depression (or a “depression-related disorder”) during pregnancy, the risk of having a child with a speech/language disorder is increased.  The risk is elevated whether or not you take an antidepressant.  One way to look at this result is to say that antidepressants confer some degree of risk, but also that the underlying psychiatric illness (or some other factors associated with the illness) contributes to this risk.  

But there are other issues we must consider as we interpret these data.  First of all, this is not a randomized controlled trial, and we cannot conclude that the women with depression who decide to remain on medication are clinically indistinguishable from those who elect to come off medication during pregnancy.  It is likely that the women who remain on medication have more severe depressive illness or comorbid psychiatric illness.  Other studies have demonstrated that more severe depressive illness may be associated with worse outcomes in the children.  In addition, it is likely that women with more severe illness will continue to have recurrent and more severe illness after pregnancy and that this depressive illness may contribute to developmental delays.

Finally, we wonder that while this finding is statistically significant, is it clinically significant? Even if we are assume that prenatal SSRI exposure increases risk for speech/language disorders, the increase in risk demonstrated here — 1.37-fold — is very small.  Furthermore, there was no observed association between SSRI exposure and risk for scholastic or motor disorder.

Ruta Nonacs, MD PhD

Brown AS, Gyllenberg D, Malm H, et al. Association of selective serotonin reuptake inhibitor exposure during pregnancy with speech, scholastic, and motor disorders in offspring [published online October 12, 2016]. JAMA Psychiatry. doi:10.1001/jamapsychiatry.2016.2594.

Nulman I, Koren G, Rovet J, et al.  Neurodevelopment of Children Following Prenatal Exposure to Venlafaxine, Selective Serotonin Reuptake Inhibitors, or Untreated Maternal Depression.  Am J Psychiatry 2012; 169: 1165–1174.


We had the privilege of submitting a commentary on the above article from Brown and colleagues.

Our Commentary Published in JAMA Psychiatry:


The last 20 years has seen an extensive examination of relevant research questions regarding the reproductive safety of selective serotonin reuptake inhibitors (SSRIs). These have included studies describing the risk of developing teratogenesis (overall and specific malformations),1 transient neonatal adaptation syndromes, and problems such as persistent pulmonary hypertension of then ewborn (PPHN).2 As an example of the evolution of interest in perinatal psychiatry, more than several hundred articles have been published during the last decade on the risk associated with first trimester exposure to SSRIs during pregnancy; the vast majority of these reports do not support a significant teratogenic

Few would argue that one of the critical missing pieces of information needed to weigh relative risks of psychotropic drug use during pregnancy is information regarding the effect of fetal SSRI exposure and untreated depression on long-term neurodevelopmental outcomes. To date, studies regarding long term effects of fetal exposure to SSRIs have been limited to relatively small cohort studies. Comorbid substance use, such as alcohol or tobacco during pregnancy, is a critical variable in the model vis-á-vis neurodevelopmental outcomes, yet accurate data on fetal exposure to these agents are frequently absent. In addition, the use of structured and consistent child development assessments in studies has been lacking.3

One of the major advances of reports regarding the reproductive safety of antidepressants over the last decade has been an appreciation of the need to account for other relevant exposures during pregnancy, such as the effect of untreated maternal psychiatric illnesses on obstetrical, neonatal, and long term behavioral outcomes.4 This is particularly relevant given the multiple articles demonstrating adverse effects of untreated depression on a host of obstetrical and neonatal outcomes and the effect of maternal psychiatric illness on child development. Perhaps one of the most critical unanswered questions in maternal and child health is the extent to which untreated depression during pregnancy affects fetal brain development and whether or how untreated psychiatric illness during pregnancy modulates risk for neurodevelopmental dysregulation or risk for the development of psychopathology in children over the long-term.5

In this issue of JAMA Psychiatry, Brown et al6 examine the critical question of whether SSRI exposure during pregnancy is associated with disorders of speech, language, scholastic achievement, and neuromotor function in children observed through early adolescence.With a very substantial sample of 845,345 pregnant women from a Finnish database, the authors describe a greater risk of speech/language disorders in the SSRI-exposed group (i.e., offspring exposed to SSRIs during pregnancy) as well as in the unmedicated group (ie, offspring with no SSRI exposure but with mothers diagnosed as having a depression-related psychiatric disorder during pregnancy based on diagnosis coding in a health service register) compared with nonexposed offspring.6 Comparing offspring exposed to SSRIs with those exposed to maternal depression yielded no differences in any of the neurodevelopmental outcomes studied. However, a difference was seen between infants exposed to SSRIs and infants not exposed to SSRIs (but whose mothers were diagnosed as having a depression-related psychiatric disorder) when a greater number of SSRI prescriptions were purchased during pregnancy.

Given the obvious methodologic limitations that are associated with analysis of “big data” from administrative databases, which are not specifically designed to provide answers to the questions for which we decide to probe, what are the clinical implications of the data presented? While these studies yield large sample sizes,we are forced to deal with some fuzziness of data with regard to diagnosis, precise duration of exposures, and severity of illness as well as whether medication is even taken. Should treatment of depression during pregnancy shift based on the findings noted in Table 2.6 where absolute risks vs relative risks are provided? The frequency of speech/language problems following referral to specialized health care services (without specific criteria for such referral) are relatively small; disorders occurred in 1.6% of patients from the SSRI-exposed group, 1.9% from the unmedicated group, and 1.0% from the nonexposed group.6 Are the data presented a signal of concern requiring further study or just background noise? Is it also possible that the finding of increased risk for speech/language disorders among those who purchased more SSRIs during pregnancy reflects a group of women who have more severe depression during pregnancy compared with those who had a diagnosis of depression at some point in the year before pregnancy or across pregnancy but who did not take antidepressants?

Residual confounding is a major methodologic quagmire for studies7 such as the one presented by Brownet al,6 and the authors note clearly that numerous unmeasured variables may affect the neurodevelopmental trajectory of children, particularly when assessments of that trajectory occur across an age continuum from birth to 14 years. Critical unanswered questions in the current study are: How many children aged 4 to 5 years, for example,noted to have speech or developmental disorders still are noted to have these disorders at 14 years? And is there any hypothesis to explain why a neurodevelopmental signal would be found along the domain of speech/language disorder vs the other domains of scholastic performance or motor function when there was no finding across the different groups examined? Is there something specific about serotonergic neuromodulation during developmentally sensitive periods that would confer risk for dysregulation across one domain of neurobehavioral function and not another. Despite the authors’ very careful look at a cutting-edge area of scientific inquiry, questions linger about the clinical implications of the findings.

Brown et al6 have identified a very important research question, and systematic research, preliminary findings, and scientific replication may lead to greater delineation of the risk-benefit decision for the use of antidepressants during pregnancy. Data suggest a prevalence of SSRI use of 5% to 10%during pregnancy,and regardless of the current reproductive safety data for SSRIs, even women with highly recurrent or long-term depression invariably prefer to avoid fetal exposure to these medications if at all possible. But the current report does not answer whether exposure to SSRIs or untreated depression during pregnancy are in equipoise with respect to neurodevelopmental toxicity or if, over the long-term, one confers greater risk. Given the extent to which depression during pregnancy predicts risk for postpartum depression with its attendant morbidity, and in light of the robust data describing the adverse effects of maternal psychiatric morbidity on long-term child development, clinicians will need to broaden the conceptual framework used to evaluate relative risk of SSRI use during pregnancy as they navigate this clinical arena with patients making individual decisions to match patient wishes.

Lee S. Cohen, MD; Ruta Nonacs, MD, PhD


Author Affiliations: Ammon-Pinizzotto Center for Women’s Mental Health, Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston.

Corresponding Author: Lee S. Cohen, MD, Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, 185 Cambridge St, Simches Research Blg, Boston, MA 02114 (

Published Online: October 12, 2016. doi:10.1001/jamapsychiatry.2016.2705

Conflict of Interest Disclosures: Dr Cohen has received research support from Cephalon, Takeda/Lundbeck Pharmaceuticals, GlaxoSmithKline, and JayMac Pharmaceuticals and support for the National Pregnancy Registry for Atypical Antipsychotics from Alkermes Biopharmaceuticals, AstraZeneca Pharmaceuticals, Bristol-Myers Squibb/Otsuka Pharmaceuticals, Forest/Actavis Pharmaceuticals, Ortho-McNeil Janssen Pharmaceuticals, and Sunovion Pharmaceuticals. No other disclosures were reported.

1. Huybrechts KF, Palmsten K, Avorn J, et al. Antidepressant use in pregnancy and the risk of cardiac defects.  N Engl J Med. 2014;370(25):2397-2407.
2. Huybrechts KF, Bateman BT, Palmsten K, et al. Antidepressant use late in pregnancy and risk of persistent pulmonary hypertension of the newborn. JAMA. 2015;313(21):2142-2151.
3. Suri R, Lin AS, Cohen LS, Altshuler LL. Acute and long-term behavioral outcome of infants and children exposed in utero to either maternal depression or antidepressants: a review of the literature. J Clin Psychiatry. 2014;75(10):e1142-e1152.
4. Salisbury AL, Wisner KL, Pearlstein T, Battle CL, Stroud L, Lester BM. Newborn neurobehavioral patterns are differentially related to prenatal maternal major depressive disorder and serotonin reuptake inhibitor treatment. Depress Anxiety.  2011;28(11):1008-1019.
5. Rifkin-Graboi A, Bai J, Chen H, et al. Prenatal maternal depression associates with microstructure of right amygdala in neonates at birth. Biol Psychiatry. 2013;74(11):837-844.
6. Brown AS, Gyllenberg D, Malm H, et al. Association of selective serotonin reuptake inhibitor exposure during pregnancy with speech, scholastic, and motor disorders in offspring [published online October 12, 2016]. JAMA Psychiatry.  doi:10.1001/jamapsychiatry.2016.2594
7. Palmsten K, Hernández-Díaz S. Can nonrandomized studies on the safety of antidepressants during pregnancy convincingly beat confounding, chance, and prior beliefs? Epidemiology. 2012;23(5):686-688.