Selective serotonin reuptake inhibitors (SSRIs) are frequently used to treat depression during pregnancy and the postpartum period. These drugs are capable of crossing the placenta and being transferred to the newborn during lactation. This report reviews the available information regarding the effects of SSRIs on the fetus and newborn; including long-term neurodevelopmental outcomes.
The lifetime risk for depression in women varies from 10 to 30%, with peak prevalence during the childbearing years.1, 2 Studies of postpartum depression suggest a 10 to 15% incidence in all parturients, making it one of the most common complications of pregnancy.3 There has been a significant body of literature documenting the adverse effects of maternal depression on the newborn. Infants of depressed mothers have been found to have lower activity levels, decreased motor tone, less vocalization, fewer positive and more negative facial expressions during interactions, more protest and despair behaviors, increased irritability, and poor consolation.4 In addition, infants of chronically depressed mothers score inferiorly on the Bayley Mental and Motor Scales, achieve lower growth percentiles at 1 year of age,4 and have delayed linguistic development and behavioral difficulties.5 Depressed women have elevated levels norepinephrine and cortisol as markers of stress response, as do their infants in the immediate newborn period, suggesting an in utero effect.6 Women who are depressed during pregnancy have more obstetrical complications, such as premature delivery, preterm onset of labor, low birth weight (LBW) infants, and pre-eclampsia.7, 8, 9, 10, 11
SELECTIVE SEROTONIN REUPTAKE INHIBITORS
The class of medications known as selective serotonin reuptake inhibitors (SSRIs) is widely prescribed to treat a number of psychiatric disorders, including major depression, anxiety, and chronic pain. Their use in pregnancy continues to be controversial. As a class, the SSRIs selectively inhibit reuptake of serotonin (5-HT) into presynaptic nerve terminals resulting in an increase in the synaptic concentration of serotonin (Figure 1). SSRIs were developed to selectively inhibit the neuronal 5-hydroxytryptamine (5-HT) uptake pump without affecting other neuroreceptors, such as histamine, acetylcholine, and adrenergic receptors. This is their major advantage over the tricyclic antidepressants (TCA), and the reason for their increased tolerance and fewer adverse side effects.12
EFFECTS DURING PREGNANCY AND DELIVERY
Serotonin, as a neurotransmitter, appears very early in fetal development and has a broad role in the brain morphogenesis. Its exact role has yet to be well delineated, but it appears to autoregulate the development of serotonin neurons and aid in the development of other neurons, which produce neurotransmitters.13 There have been many studies in rats to evaluate the fetal effects of prenatal exposure to fluoxetine, the prototype SSRI. Both fluoxetine and its active metabolite, norfluoxetine, cross the placenta and enter fetal brain tissue.14 Studies to evaluate the neurochemical toxicity of fluoxetine on the developing fetal rat have demonstrated both age-specific and site-specific alterations in brain 5-HT content, 5-HT2A/2C receptor density, and 5-HT2A/2C receptor-mediated hormone secretion, as well as region-specific alterations in the density of 5-HT transporters.15 The limbic system appears particularly vulnerable to the effects of fluoxetine exposure in utero.16 There is also some evidence to suggest that 5-HT may play an indirect, but critical role in processing sensory stimuli,17 and the 5-HT transporter may play an integral part in molding human behavior and neurodevelopment.18 These findings lead to a hypothetical concern for adverse human fetal brain development and subsequent subtle neurodevelopmental abnormalities that have yet to be substantiated. To date, animal studies have not shown behavioral or learning abnormalities after prenatal exposure to SSRIs.15
A meta-analysis of several smaller studies has documented that the use of SSRIs during the first trimester of pregnancy does not increase the risk of congenital malformations above that seen in the general population.19 One study did note an increased incidence of three or more minor congenital malformations in infants exposed prenatally to SSRIs.20
EFFECTS ON THE HUMAN FETUS
Placental passage of antidepressants has also been documented in humans.21, 22 One study compared maternal serum concentrations of SSRIs to infant umbilical vein serum concentrations at birth. Antidepressant and/or metabolite concentrations were detectable in the umbilical vein blood in 87% of the samples collected.22 Fluoxetine and citalopram had the highest ratio of umbilical vein-to-maternal serum concentration, indicating greater transfer from mother to fetus. Sertraline and paroxetine had the lowest ratio. In two separate publications, Heikkinen et al.21, 23 evaluated the pharmacokinetics of two commonly used SSRIs, fluoxetine and citalopram, during pregnancy and lactation. During pregnancy, maternal trough levels of fluoxetine and its metabolite, norfluoxetine, were found to be decreased by as much as 50% compared to the prepregnant state. Trough levels remained low in the first few days after delivery but gradually increased and were 55% higher at 2 months postpartum than during pregnancy. The decreased trough plasma concentrations during pregnancy are likely secondary to the normal physiologic changes of pregnancy, such as volume expansion, increased hepatic blood flow and decreased plasma protein binding. Infant plasma levels of fluoxetine and norfluoxetine at birth were 65 and 72% of mothers’ serum levels, respectively. At 2 weeks of life, the majority of infants no longer had detectable fluoxetine levels, but norfluoxetine continued to be measurable, although significantly decreased from birth. At 2 months of age, fluoxetine was not detectable in any infant. Norfluoxetine levels continued to be detectable, but again, significantly lower than at birth and 2 weeks of life. The estimated half lives of fluoxetine and norfluoxetine in the newborn are 5 and 14 days, respectively. All exposed infants in this study were breastfed, and mothers continued taking fluoxetine in the postpartum period. No adverse pregnancy outcomes were reported, except for a lower 15 minute Apgar score in fluoxetine-exposed infants. No adverse events were reported in the neonatal period, and growth and development were normal at 1 year. The pharmacokinetics of citalopram and its metabolites during pregnancy and lactation are similar to fluoxetine and norfluoxetine in the mother and infant. Infant serum concentrations of citalopram at birth are 64% of the maternal serum level and decline to almost zero by 2 months of age.
EFFECTS ON THE NEWBORN
Neonatal complications associated with exposure to maternal SSRIs have been widely reported, including prolonged hospitalization or increased Neonatal Intensive Care Unit (NICU)/Special Care Nursery (SCN) admissions, LBW, increased incidence of prematurity, and decreased Apgar scores. In addition, infants with third trimester exposure to SSRIs have been described as having “poor neonatal adaptation”(PNA), a conglomeration of diagnoses that include respiratory distress, hypoglycemia, jitteriness, lethargy, hypotonia, weak or absent cry, feeding difficulties, neonatal convulsions, and hyperbilirubinemia. Currently, there is a debate as to whether the signs in newborns are a result of acute cessation of exposure (i.e., withdrawal) to the SSRIs at birth, or a result of overstimulation of the serotonergic system secondary to prenatal exposure.24 SSRI discontinuation syndromes have been described in the adult population with symptoms that commonly include dizziness, parasthesias, electric shock-like sensations, lethargy, headache, tremor, sweating, anorexia, insomnia, nausea, vomiting, diarrhea, irritability and anxiety, or agitation.25 The syndromes have primarily been associated with the SSRIs with the shortest half lives, such as paroxetine. Fluoxetine, which has the longest half life, has the lowest risk of discontinuation syndrome in the adult population. In adults, serotonergic overstimulation has symptoms similar to serotonin discontinuation syndromes.26 Although there is placental transfer of SSRIs in utero, infant plasma concentrations at birth have not correlated with neonatal signs. Oberlander et al.27 evaluated serum drug concentrations in 46 infants with late in utero exposure to SSRIs and found no correlation between infant drug levels in symptomatic infants with poor neonatal adaptation versus nonsymptomatic infants.27
Despite the frequency of side effects reported in the literature, cohort studies designed to evaluate neonatal outcomes have produced contradictory results. Low or decreased birth weight after in utero exposure to SSRIs was noted in three studies,28, 29, 30 but no difference was found in several other studies.31, 32, 33, 34, 35, 36, 37 Similarly, several reports have documented a tendency for infants exposed to SSRIs to have decreased gestation and/or a higher incidence of premature birth,20, 28, 29, 30, 38, 39, 40 while several others have found no difference when compared to unexposed infants.31, 33, 34, 35, 36, 37 Several studies have noted prolonged hospitalization or increased NICU/SCN admissions among infants exposed to maternal SSRIs.20, 33, 41, 42, 43 In contrast, a recent investigation by Suri et al.31 found no increase in NICU admission among their exposed population. Finally, since numerous case reports have documented aspects of “poor neonatal adaptation” in infants with third trimester exposure to maternal SSRIs, several cohort studies have attempted to look at these diagnoses as an outcome measure after in utero exposure. The majority have noted an increase incidence of transitional difficulties in the neonatal period with no one diagnosis predominating.20, 27, 37, 42, 44
Most of the aforementioned studies were of small sample size, making definitive interpretation difficult. In order to address this in a meaningful way, we performed a meta-analysis of prospective cohort trials with outcomes that included the incidence of prematurity, LBW, SCN/NICU admission, and the diagnosis of “poor neonatal adaptation.”
SELECTION OF STUDIES
Studies were included if they described late pregnancy exposure (at least third trimester exposure) to any SSRI, and prospective reports of the outcomes of pregnancy. The diagnosis of “poor neonatal adaptation” appears more closely associated with use of SSRIs through the third trimester. This is the rational for excluding studies reporting only first trimester exposure pregnancy outcomes.
In all, 19 studies and one abstract were identified after an initial search yielded 194 articles (using MEDLINE) with MeSH headings: “Serotonin Uptake Inhibitors” or “Antidepressive Agents” and “pregnancy outcome” and “Serotonin Uptake Inhibitors/adverse events” or “Antidepressive Agents/adverse events” limited to newborn (birth to 1 month). In addition, a search was made of ISI Web of Knowledge, EMBASE, and CINAHL using the above search terms and “SSRI” and “pregnancy” as keywords. No further studies were identified in these databases. One additional abstract was found under publications on the University of California-San Diego (UCSD) HealthCare and California Teratogen Information Service and Clinical Research Program Website.45 No additional studies were identified from citations referenced in pertinent studies.
Nine of the original 20 articles were excluded for the following reasons: retrospective study design,29, 42 combination prospective and retrospective study design,43 outcomes following first trimester exposure only,30, 46, 47 outcomes including first trimester and late exposure without differentiation,28 and first trimester exposure and study drug not a true SSRI.48 Nulman's 1997 study was excluded after realizing that the Nulman 2002 study protocol included 18 late exposed women common to the 1997 study. Of the 11 remaining studies, two were excluded because of insufficient information for data analysis even after attempting to contact the authors.20, 45
When raw data or absolute number of infants with prematurity and/or LBW were unavailable from the published study or the author, an estimate was obtained assuming a normal distribution of data and using the given standard deviations and means to calculate the area under the curve and the proportion of patients with these diagnoses. This calculated estimate does introduce a margin of error, especially since it is unlikely that gestational ages were normally distributed. Results are likely to be skewed to the left. However, the same calculations and assumptions were also applied to the control infant data in these studies. Data were analyzed using SAS 9.1 software where a nonlinear mixed effect model was used to account for the homogeneity within each study. The nine studies included in this meta-analysis and their outcomes are listed in Tables 1 and 2.20, 27, 31, 32, 33, 34, 40, 41, 49
Table 3 shows the results of the meta-analysis and lists the odds ratios (OR) and 95% confidence intervals (CI) for neonatal factors significantly impacted by in utero exposure to SSRIs. This suggests that neonates exposed to SSRIs in utero are more likely to be LBW and to be admitted to a SCN/NICU at birth. This is consistent with results recently published by Kallen.28 This study utilized the large Swedish Medical Database. A total of 558 women were identified who used SSRIs during pregnancy. Infant outcomes were prospectively assessed and compared to the entire population. His study was excluded from this meta-analysis as the population studied included infants exposed to SSRIs exclusively in the first trimester as well as infants exposed in the third trimester. Kallen feels that this population is representative of late SSRI exposure, but exact numbers could not be provided (personal communication). Kallen also examined the incidence of prematurity, LBW, and the incidence of respiratory distress, jaundice, hypoglycemia, and convulsions, diagnoses often included in the category of “poor neonatal adaptation.” The study concluded that infants exposed to SSRIs in utero were more likely to be born prematurely with an OR of 2.06 (95% CI 1.58, 2.69), and were more likely to be LBW with an OR of 1.98 (95% CI 1.42, 2.76). When Kallen's data are incorporated into the above meta-analysis, the OR for prematurity becomes 2.03 (95% CI 1.61, 2.60), with a highly significant p-value of 0.0001. The OR for LBW becomes 2.37 (95% CI 1.65, 3.41) with a significant p-value of 0.0006. Additionally, the existing trend towards the increased diagnosis of PNA becomes significant with a p-value of 0.003 and OR of 1.99 (95% CI 1.43, 2.77). Hence, Kallen's results are uniformly consistent with the findings of the present meta-analysis and inclusion would serve to strengthen all results.
Although it appears that maternal SSRI use does have an effect on the incidence of prematurity, LBW, SCN/NICU admissions, and the diagnosis of PNA, heterogeneity exists between the exposed populations of the included studies. Many of the studies included women taking other psychotropic medications, such as benzodiazepine, in the SSRI-exposed group.20, 33, 34, 49 Withdrawal syndromes have also been associated with benzodiazepine, and have included such signs as hypotonia, respiratory distress, hypothermia, and poor feeding, similar to signs included in the diagnosis of “poor neonatal adaptation” with SSRIs.27 Several studies also included women who smoked cigarettes and used alcohol.20, 34, 49 Cigarette smoking is known to adversely affect infant growth in utero, and alcohol abuse has been associated with prematurity and infants that are small for gestational age.50 In addition, many of the studies included women using SSRIs for indications other than depression (e.g., anxiety).41, 49 As noted previously, maternal depression during pregnancy has been associated with premature delivery and LBW.7, 9, 10 Finally, all but one study compared SSRI-exposed women to healthy control populations with no psychiatric diagnosis. The study by Suri utilized two control groups: one with no history of psychiatric illnesses, and one with a diagnosis of major depressive disorder (MDD), but no pharmacotherapy.31 The study found no difference between the three groups in gestational age, adjusted birth weight, or admissions to the NICU. Ideally, comparisons should be made between women diagnosed with MDD and receiving SSRI therapy and women diagnosed with MDD and receiving no pharmacotherapy.
EFFECTS DURING LACTATION
Infant exposure to maternal SSRI therapy during breastfeeding is another concern. Postpartum depression occurs in 10 to 15% of mothers, and half of affected women are treated with pharmacotherapy.51 Weissman et al.51 recently reported a pooled analysis of 57 published and unpublished studies looking at antidepressant levels in lactating women, breast milk, and nursing infants. They found that breastfeeding infants exposed to paroxetine or sertraline seemed unlikely to develop detectable or elevated plasma drug levels (defined as greater than 10% of maternal serum levels). Infants exposed to fluoxetine appeared at highest risk of developing elevated serum drug levels, especially following prenatal exposure, or if levels were high in the breast milk. Limited data suggested that elevated infant levels of citalopram may also occur with lactation. Maternal serum levels of citalopram were highly correlated with infant levels, but no correlation was found for fluoxetine or paroxetine. Drug levels in maternal breast milk were significantly correlated with infant plasma levels of citalopram, fluoxetine, and paroxetine. Another recent publication concluded that infants exposed to SSRIs during lactation had low or undetectable serum drug levels.52 Citalopram was the most prescribed antidepressant and, in contrast to Weissman's analysis, drug levels were undetectable or just above the detectable limit in exposed infants. Sertraline and paroxetine were not detected in any drug-exposed infants. Infants exposed to venlafaxine, a combination of SSRI and norepinephrine uptake inhibitor, had high serum levels of venlafaxine and its metabolite, estimated as 10% of maternal serum concentration. However, there were only three infant–mother pairs in this group. The one infant exposed to fluoxetine also had high serum drug levels, consistent with Weissman's analysis. There was no increase in adverse symptoms in drug-exposed infants compared to a nonexposed group.52
Several studies have sought to establish the peak drug concentration in breast milk after maternal dosing as a strategy to minimize infant exposure. Peak breast milk concentrations for sertraline and fluoxetine (and its metabolite norfluoxetine) occur 8 to 9 hours after maternal dosing,53, 54 and breast milk can be “pumped and dumped” at this time interval in order to minimize infant exposure. It should also be noted that drug concentrations tend to be higher in hindmilk because of its lipophilic nature.
Adverse effects with breastfeeding exposure have been mainly documented in case reports. Weissman tabulated six case reports that reported adverse neonatal effects and infant drug levels after SSRI exposure through lactation. The most commonly reported infant signs were uneasy sleep or irritability and poor feeding or sucking. Five of the cases demonstrated signs after fluoxetine exposure, and half of the infants were exposed prenatally. The long-term effects on newborns after chronic low-dose antidepressant exposure through lactation has not been well studied and information is limited. Heikkinen et al.23, 55 found no concerning adverse events in the immediate neonatal period and no difference in body weight or neurodevelopment at 1 year in infants exposed to fluoxetine and citalopram prenatally and through lactation compared to nonexposed control infants. Chambers et al.56 found a small growth deficit in fluoxetine-exposed infants at 6 months of age.
There does not appear to be a clear way to estimate neonatal drug exposure from lactation by monitoring maternal or breast milk drug levels. Monitoring infant drug levels is an option but involves repeated blood sampling and discomfort. In addition, the clinical significance of low infant drug levels remains unclear. Of the available SSRIs, paroxetine and sertraline seem to be the most favorable choices during lactation as they appear to minimize infant exposure as shown by undetectable infant drug levels.51 It is clear that better controlled studies of short-term and long-term infant outcomes are needed. The benefit breast milk offers to infants has been well established. However, the impact of an untreated maternal mood disorder on mother–infant attachment and infant development must also be considered. Decisions about lactation involve careful risk–benefit analysis and ideally should be assessed on an individual basis by the patient, her family and the infant's pediatrician when possible.
NEURODEVELOPMENT IN THE NEONATAL PERIOD
Few studies have attempted to evaluate subtle neurodevelopmental findings in the neonatal period after SSRI exposore in utero. Oberlander et al.57 found that infants exposed to SSRIs prenatally have an attenuated response to acute pain during heelsticks compared to infants not exposed to antidepressants. This attenuated response may result from increased serotonin (5-HT) and GABA agonists in the fetal human brain as a consequence of maternal SSRI use. 5-HT and GABA are known to be active during early fetal neurologic growth and play a role in pain modulation.13, 57 A recently published case report also described an infant with an attenuated pain response and an abnormal EEG suggestive of encephalopathy, who had been exposed to paroxetine in utero.58 The infant subsequently had a normal pain response and normal EEG at 2 weeks of life without intervention. Another recent prospective controlled study evaluated serotonergic signs at birth, 2 weeks, and 2 months of age, and cord blood 5-HT levels in infants exposed to SSRIs prenatally compared to infants born to healthy mothers with no prenatal medication use.49 This study found a four-fold increase in serotonergic signs, including tremor, restlessness, and rigidity in infants exposed to SSRIs compared to controls. In addition, cord blood levels of serotonin at birth were significantly reduced when compared to controls. Serotonergic signs in both groups were similar at 2 weeks and 2 months of age. One potential problem with this study is the scale used to evaluate the serotonergic signs in the infants was originally developed to evaluate adults with serotonergic overstimulation. Zeskind and Stephens32 published a report comparing the neonatal neurobehavior of infants exposed to SSRIs prenatally with infants of healthy mothers and no antidepressant exposure. All infants were evaluated for behavioral state using the Neonatal Behavioral Assessment Scale, amount and quality of REM sleep, startle and tremulousness, motor activity, and heart rate variability for 1 hour between feedings at 14 and 39 hours of life. Infants exposed to SSRIs had increased tremulousness, fewer changes in behavioral state, fewer different behavioral states, greater amounts of uninterrupted REM sleep, more startles, more generalized motor activity related to the startles, and greater autonomic dysregulation (as determined by decreased quality of heart rate variability). After correcting for gestational age, the number of startles, motor activity, and rhythms in heart rate variability were no longer significant when compared to controls. Finally, Zeskind et al.59 reported that infants exposed in utero to SSRIs continue to have a significantly higher fundamental frequency in crying at 2 months, an indicator of the adverse effects of prenatal drug exposure on infant neurobehavioral development, compared to healthy, non-SSRI-exposed infants.
LONG-TERM NEURODEVELOPMENTAL OUTCOMES
Long-term neurodevelopmental outcomes of infants exposed to SSRIs during pregnancy have not yet been well studied, and results have been conflicting. Recently, in a prospective cohort trial, Nulman et al.34, 36 evaluated IQ, language development, and temperament in children exposed to TCAs or SSRIs in the first trimester only, and also throughout pregnancy, compared to infants born to mothers who were not depressed and were not exposed to any known teratogenic medications during pregnancy. Children were evaluated between 16 and 86 months of life with the Bayley Scales of Infant Development or the McCarthy Scales of Children's Abilities and the Reynell Developmental Language Scales. No differences in IQ, language development, or temperament were found in children who were exposed during the first trimester only or throughout pregnancy. In contrast, in another prospective cohort trial, Casper et al.43 compared mental and motor development using the Bayley Scales of Infant Development in infants 6 to 40 months of age exposed prenatally to SSRIs, compared to infants born to depressed mothers who were not receiving antidepressants. Children exposed to SSRIs were found to have lower Apgar scores and scored lower on the Bayley Psychomotor Development Index and the motor quality factor of the Bayley Behavioral Rating Scale, suggesting that SSRIs may have subtle effects on motor development and motor control. The broad time frame of infant assessment in all of these studies does not allow for adequate interpretation of the data in a developmental context.
It is tempting to speculate that prenatal exposure to SSRIs has some effect on newborn neurobehavioral outcomes. However, most of the studies cited above were of small sample size, used various neurobehavioral assessment tools that are not validated in infants, and compared SSRI-exposed infants to control infants born to healthy mothers with no history of depression. More importantly, the long-term neurodevelopmental outcomes of these infants remain unclear. The study by Casper et al.43 suggests that there may be subtle motor deficits in children with prenatal exposure to SSRIs. This continues to raise the question of whether pharmacologically treating depression during pregnancy is better or worse for mothers and infants than untreated maternal depression.
SSRIs are frequently prescribed to treat depression during pregnancy and in the postpartum period. Although the condition is very widespread, the available information suggests that there may be untoward subtle effects on the fetus and newborn. These may include serotonergic overstimulation and withdrawal syndromes, as well as long-term effects on neurobehavior and performance. The present systematic review and meta-analysis should aid the clinician in assigning probabilities to some infant outcomes. Until more complete long-term safety studies are forthcoming, careful risk-benefit analysis needs to be applied when considering their use in pregnancy or during lactation.
This research was supported by the National Institute of Mental Health, MH065062 to DMV, CRN and NK; and MH068489 to DMV and CRN. We thank Robert Schumacher, MD for his helpful comments during the preparation of this manuscript.
About this article
psychopraxis. neuropraxis (2017)