Version July 2025
INTRODUCTION —
Among females with an established pregnancy, surveys estimate that psychotropic drugs are taken by 21 to 33 percent [1,2]. These medications are often necessary to control psychiatric illnesses such as bipolar, depressive, and psychotic disorders that predate or emerge during pregnancy. However, pharmacotherapy may entail risks of structural malformations, pregnancy complications, neonatal toxicity and withdrawal, and adverse developmental effects.
This topic reviews the risks of antipsychotics, benzodiazepines, lithium, and neuromodulation during pregnancy. The antenatal risks of antidepressants and antiepileptics and general information about congenital anomalies are discussed separately, as is pharmacotherapy for antenatal unipolar depression and bipolar mood episodes:
●(See "Congenital anomalies: Epidemiology, types, and patterns" and "Congenital anomalies: Causes".)
●(See "Severe antenatal unipolar major depression: Choosing treatment".)
●(See "Bipolar disorder in pregnant women: Treatment of major depression".)
DEFINITION OF A TERATOGEN —
Teratogens are factors that can alter normal intrauterine development of fetal growth, anatomic structures, physical functioning, and postnatal development. This definition encompasses environmental exposures, maternal medical disorders, infectious agents, and genetic conditions. (See "Congenital anomalies: Causes", section on 'Teratogens'.)
Many discussions of teratogens usually center on drug exposures. In determining whether a drug is a teratogen, many authorities stipulate that the exposure causes a pattern of defects [3]. Thus, if exposure is associated with an increase in congenital anomalies greater than that expected in the general population, but the defects vary and there is no discernible pattern, the drug is generally not considered teratogenic.
GENERAL PRINCIPLES —
All psychotropic drugs presumably cross the placenta and are present in the amniotic fluid [4]. Embryonic and fetal exposure to maternal pharmacotherapy can cause [2,5-7]:
●Pregnancy loss (miscarriage or spontaneous abortion)
●Major and minor structural malformations
●Fetal growth restriction and low birth weight
●Preterm delivery
●Neonatal toxicity and withdrawal (neonatal adaptation syndrome)
●Postnatal developmental effects upon behavior, cognition, and emotional regulation
Medication effects upon the fetus vary according to gestational age [2,5,7]. As an example, the fetus is most vulnerable to major morphologic teratogenesis during organogenesis in the embryonic period of the first trimester, between the third and eighth week of gestation (weeks of gestation are counted from the first day of the last menstrual period) (figure 1 and figure 2). Organogenesis occurs 5 to 10 weeks from the first day of the last menstrual period, or three to eight weeks from conception (conception occurs approximately two weeks after the first day of the last menstrual period). By contrast, neonatal toxicity and withdrawal are the result of third-trimester exposure.
The estimated risk of major congenital malformations appears to vary among psychotropic medications; the rank order from greatest to least teratogenic risk is [8-13]:
●Antipsychotics
●Antidepressants
Within drug classes such as antidepressants or benzodiazepines, there are generally little data to determine whether the risk of adverse outcomes differs among the specific drugs within the class.
In discussing teratogenic effects, patients should be informed that the base rate for congenital defects in the general population is approximately 2 to 4 percent. (See "Congenital anomalies: Epidemiology, types, and patterns", section on 'Epidemiology'.)
The estimated risks of congenital defects from pharmacotherapy are typically based upon birth registry and observational studies [6]. Although these studies probably provide the best evidence of the risks, the accuracy of these estimates is uncertain due to ascertainment bias. In addition, information about teratogenic effects is usually presented in terms of monotherapy, whereas many studies are confounded due to concomitant exposure to multiple psychotropic and nonpsychotropic medications [14,15], and acutely ill pregnant patients often require medication combinations [16-19]. Further, it is often not possible to separate the effects of drugs from the psychiatric illness itself (confounding by indication), and studies often do not control for potential confounding factors, such as comorbid substance use disorder, maternal age, maternal body mass index, and prior pregnancy loss. Also, some congenital defects are not recognized at birth and are discovered only later.
Although a drug may increase the risk of a congenital anomaly, the absolute risk may be low [20]. As an example, the estimated risk of Ebstein anomaly (abnormalities of the tricuspid valve and right ventricle) in the general population is nearly 1 in 20,000 live births [21]. Following first-trimester exposure to lithium, the risk increases 20-fold to approximately 1 in 1000, which many authorities consider low [5,20,22].
Resources — Current, general information about the possible teratogenic effects of medications is available from several resources that are discussed separately. (See "Congenital anomalies: Causes", section on 'Resources'.)
In addition, the possible teratogenic effects of drugs and suggestions for managing them can be found for all drugs included in the UpToDate drug database; search on the drug name, choose the drug information topic for that drug, and click on the “Pregnancy Considerations” section of the topic outline. Clicking on the name of a drug cited within any UpToDate topic will also bring you to the drug information topic.
LIMITATIONS OF THE EVIDENCE —
The quality of evidence that is available about the risks of psychotropic drugs during pregnancy is discussed separately. (See "Antenatal use of antidepressants and the potential risk of teratogenicity and adverse pregnancy outcomes: Selective serotonin reuptake inhibitors", section on 'Limitations of the evidence'.)
ANTIPARKINSONIAN DRUGS USED FOR TREATING EXTRAPYRAMIDAL SYMPTOMS —
Among the antiparkinsonian drugs that are used to treat extrapyramidal symptoms secondary to antipsychotics, reviews suggest that the risk of teratogenicity with antihistamines (eg, diphenhydramine) appears to be low [23], and that organ malformation appears to be less likely with diphenhydramine than amantadine, benztropine, and trihexyphenidyl [24].
Additional information about the teratogenicity of diphenhydramine is discussed separately. (See "Nausea and vomiting of pregnancy: Treatment and outcome", section on 'Diphenhydramine'.)
ANTIDEPRESSANTS —
Most of the evidence about the teratogenicity, pregnancy complications, and postnatal risks associated with antidepressants is based upon observational studies of pregnant patients with unipolar major depression. (See "Antenatal use of antidepressants and the potential risk of teratogenicity and adverse pregnancy outcomes: Selective serotonin reuptake inhibitors" and "Antenatal use of antidepressants and risks of teratogenicity and adverse pregnancy outcomes: Drugs other than selective serotonin reuptake inhibitors" and "Antenatal exposure to selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs): Neonatal outcomes".)
ANTIEPILEPTICS —
Most observational studies that have examined the association between antiepileptics and congenital malformations and postnatal risks have been conducted primarily in pregnant patients with epilepsy. (See "Risks associated with epilepsy during pregnancy and the postpartum period", section on 'Effects of ASMs on the fetus and child'.)
ANTIPSYCHOTICS
Overview — Antipsychotics are classified as either first- or second-generation. First-generation antipsychotics include chlorpromazine, haloperidol, perphenazine, prochlorperazine, and trifluoperazine. Second-generation antipsychotics include aripiprazole, olanzapine, quetiapine, risperidone, and ziprasidone. In the general population of patients receiving antipsychotics, second-generation antipsychotics are often better tolerated than first-generation antipsychotics and may be less likely to adversely affect fertility [25].
First-generation antipsychotics have often been used and studied during pregnancy. The reproductive safety risks of first-generation antipsychotics are generally regarded as low, based upon a literature that extends back to the 1960s [2,4,18,26]. However, the risks of prenatal exposure to first-generation antipsychotics have been examined primarily in patients with hyperemesis gravidarum who received low doses of antipsychotics as antiemetics and were presumed to not have a psychiatric disorder, and secondarily in patients with bipolar, depressive, or psychotic disorders [26,27].
Second-generation antipsychotics are also typically regarded as safe during pregnancy, and the general consensus is that their benefits for psychiatric disorders, such as psychotic disorders and bipolar disorders, far outweigh the pregnancy safety risks [28,29]. Second-generation antipsychotics are more widely used during pregnancy than first-generation drugs. In studies of administrative claims data for pregnant females (n>580,000 and >1,300,00), antenatal exposure to either second-generation antipsychotics or first-generation antipsychotics occurred in approximately 0.7 and 0.1 percent of the females [27,30]. Among second-generation antipsychotics, quetiapine is prescribed most often during pregnancy [27,31].
The subsections below discuss the teratogenicity, perinatal mortality, pregnancy complications, placental passage, labor and delivery outcomes, and postnatal effects of first- and second-generation antipsychotics.
General information about first- and second-generation antipsychotics, including adverse effects, is discussed elsewhere. (See "First-generation antipsychotic medications: Pharmacology, administration, and comparative side effects" and "Second-generation and other antipsychotic medications: Pharmacology, administration, and side effects".)
Teratogenicity — Most studies suggest that prenatal exposure to first- or second-generation antipsychotics does not increase the risk of major physical malformations above rates observed in the general population [1,3,4,15,26,27,32,33].
As an example, a study of an administrative health care claims database (n>1,200,000 live births) examined teratogenicity in offspring with first-trimester exposure to first-generation antipsychotics (n = 727) or second-generation antipsychotics (n>9000) [27]. The analyses controlled for more than 250 potential confounding baseline factors (eg, maternal age, indications for antipsychotics, and concomitant medications) that can affect the probability (propensity) of receiving an antipsychotic. The results included the following:
●Overall rate of congenital malformations – The overall rate of congenital malformations with first-generation antipsychotics was comparable in the exposed and unexposed children (relative risk 0.9, 95% CI 0.6-1.3), as was the rate with second-generation antipsychotics (relative risk 1.05, 95% CI 0.96-1.16).
●Cardiac malformations – The rate of cardiac malformations with either first- or second-generation antipsychotics was comparable in the exposed and unexposed children.
In addition, the risk of teratogenicity with either first- or second-generation antipsychotics appears similar. In a prospective observational study of pregnant women who received either first-generation (n = 213) or second-generation antipsychotics (n = 430) during the first trimester, the likelihood of major congenital malformations in the offspring was comparable for the two groups (odds ratio 0.8, 95% CI 0.4-1.8) [34].
Multiple studies of specific second-generation antipsychotics also suggest that first-trimester exposure is not associated with congenital malformation, including aripiprazole [27,35-37], clozapine [3], olanzapine [27,36,38], quetiapine [3,27,36,39], risperidone [36,40], and ziprasidone [3,27].
Although some studies have found that in utero exposure to antipsychotics is associated with congenital malformations [41], later studies appear to have used more rigorous statistical methods to better account for potential confounding factors [27,28].
Perinatal mortality — Antenatal use of first- or second-generation antipsychotics does not appear to increase perinatal mortality (pregnancy loss [spontaneous abortion or miscarriage] or stillbirth), based upon prospective observational studies [1,3,16,42,43]. As an example, a study of pregnant females who received first-generation antipsychotics (n = 284), second-generation antipsychotics (n = 561), or nonteratogenic drugs (n = 1122) found that pregnancy loss and stillbirth were each comparable across the three groups [34].
In addition, the reported rates of fetal death for specific second-generation drugs, including olanzapine, risperidone, and ziprasidone, generally do not exceed rates in the general population [3,38,40].
General information about perinatal mortality is discussed separately. (See "Perinatal mortality".)
Pregnancy complications
●Maternal weight gain – Using antipsychotics during pregnancy may lead to maternal weight gain. A national birth registry of pregnancies (n>900,000) found that a body mass index ≥26 (overweight or obesity) following delivery was twice as likely in patients treated with antipsychotics than untreated patients (odds ratio 2) [26].
●Other complications – A study used administrative health care datasets to examine pregnancy complications in pregnant females who either received an antipsychotic during pregnancy or did not [31]. The two groups (n = 1021 in each group) were matched on potential confounding baseline factors (eg, psychiatric and general medical diagnoses and prescribed medications) that can affect the probability (propensity) of receiving an antipsychotic. Approximately 90 percent of the patients treated with an antipsychotic received a second-generation antipsychotic. The incidence of each of the following complications was comparable in those who received or did not receive an antipsychotic:
•Gestational diabetes – 7 and 6.1 percent
•Hypertensive disorders of pregnancy – 4.7 and 4.1 percent
•Venous thromboembolism – 1.2 and 1.3 percent
Placental passage — Fetal exposure to antipsychotics may vary due to differences in placental permeability to these drugs. A prospective observational study of 50 pregnant patients examined the placental passage of antipsychotics, defined as the ratio of umbilical cord serum drug concentration to maternal serum drug concentration [18]. (Umbilical cord and maternal plasma concentrations were drawn at delivery; drugs cross the placenta more readily late in pregnancy). The placental passage ratios were as follows:
●Olanzapine – 72 percent
●Haloperidol – 66 percent
●Risperidone – 49 percent
●Quetiapine – 24 percent
The ratio for quetiapine was lower than that for olanzapine and haloperidol.
Labor and delivery outcomes — Antenatal exposure to first-generation antipsychotics may be associated with preterm birth. In a prospective observational study of pregnant females exposed to either first-generation antipsychotics (n = 284) or nonteratogenic drugs (n = 1122), premature delivery occurred more often in the group with first-generation antipsychotic exposure (16 versus 9 percent) [34].
It is not clear if prenatal exposure to first-generation antipsychotics is associated with abnormal birth weight, due to discrepant results across studies. However, larger prospective studies have failed to find such an association [34,42], whereas smaller prospective observational studies each found that birth weights were less (approximately 220 grams) in babies exposed to typical antipsychotics compared with unexposed babies [33,44].
In utero exposure to second-generation antipsychotics is generally not associated with adverse labor and delivery outcomes. As an example, a study of administrative health care datasets examined labor and delivery outcomes in pregnant females who either received an antipsychotic during pregnancy or did not [31]. The two groups (n = 1021 in each group) were matched on potential confounding baseline factors (eg, psychiatric and general medical diagnoses and prescribed medications) that can affect the probability (propensity) of receiving an antipsychotic. Approximately 90 percent of the patients treated with an antipsychotic received a second-generation antipsychotic. The incidence of each of the following labor and delivery outcomes was comparable in patients who received or did not receive an antipsychotic:
●Cesarean delivery – 31.1 and 31.8 percent
●Preterm birth – 14.5 and 14.3 percent
●Small for gestational age – 6.1 and 5.1 percent
●Large for gestational age – 3.6 and 2.3 percent
Other studies have also found that antenatal exposure to second-generation antipsychotics is not associated with preterm birth [34], as well as birth weight [34,43].
Postnatal effects
Neonatal adaptation syndrome — Although it is not clear whether chronic administration of antipsychotics during the third trimester causes symptoms of the neonatal adaptation syndrome (neonatal toxicity and withdrawal), the best evidence suggests that other factors play a bigger role.
The US Food and Drug Administration issued a warning in 2011 that described extrapyramidal symptoms (eg, abnormal movements, restlessness, and tremor) and other symptoms of toxicity and withdrawal in neonates who were exposed in utero to antipsychotics [45]. The warning was based upon 69 case reports. Symptoms included [1,5,45]:
●Abnormal movements (dyskinesia)
●Abnormally increased or decreased muscle tone
●Agitation
●Crying
●Hyperactivity
●Hyperreflexia
●Irritability
●Motor restlessness
●Sedation
●Tremor
●Hypotension
●Tachycardia
●Difficulty breathing
●Difficulty feeding
●Gastrointestinal dysfunction (eg, functional bowel obstruction)
The symptoms typically subside within hours to days, but may persist for weeks to months after birth [3]. Specific treatment is usually not necessary, but more severely affected newborns may require longer hospital stays [45].
However, later and more rigorous evidence suggests that these symptoms are better attributed to other factors. A study used administrative health care datasets to examine postnatal outcomes in pregnant females who either received an antipsychotic during pregnancy or did not [31]. The two groups (n = 1021 in each group) were matched on potential confounding baseline factors (eg, psychiatric and general medical diagnoses and prescribed medications) that can affect the probability (propensity) of receiving an antipsychotic. Approximately 90 percent of the patients treated with an antipsychotic received a second-generation antipsychotic. The incidence of the neonatal adaptation syndrome in the two groups was comparable (13 and 11 percent). Whereas many of the case reports were confounded by other factors that could cause the syndrome, the later study controlled for such factors, including alcohol and other substance use and concomitant medications.
Neurobehavioral development — Intrauterine exposure to antipsychotics may adversely affect neuromotor functioning during infancy. In a prospective observational study that was conducted six months postpartum and controlled for maternal psychiatric status and other variables, neuromotor performance (eg, posture, muscle tone, and reflexes) was poorer in babies with prenatal antipsychotic exposure (n = 22) compared with babies with no prenatal psychotropic exposure (n = 85) [19]. However, information processing and learning were comparable for the two groups.
Adverse developmental effects due to in utero exposure to antipsychotics may not persist beyond infancy. Multiple reviews have concluded that prenatal exposure to first-generation antipsychotics does not appear to adversely affect behavioral, cognitive, or emotional development in children [2,46,47].
BENZODIAZEPINES —
Approximately 1 to 5 percent of pregnant females are prescribed benzodiazepines to manage problems such as agitation, anxiety, depression, and insomnia [14,48-50]. All benzodiazepines readily cross the placenta [4]; thus, drugs with short half-lives (eg, lorazepam) are generally preferred to limit fetal exposure, which is consistent with practice guidelines from the United Kingdom National Institute for Health and Care Excellence [51]. In many cases, benzodiazepines are used in conjunction with other drugs, such as antidepressants and opioids [14,49]. However, most studies of benzodiazepines during pregnancy do not distinguish the risk of adverse outcomes among specific benzodiazepines.
Later studies of prenatal benzodiazepines often include hypnotic benzodiazepine receptor agonists such as zaleplon, zolpidem, and zopiclone because both benzodiazepines and hypnotic benzodiazepine receptor agonists have the same mechanism of action; they modulate the gamma-aminobutyric acid receptor [14].
Teratogenicity — Due to conflicting results across studies, it is not known if exposure to either benzodiazepines or hypnotic benzodiazepine receptor agonists during pregnancy is associated with an increased risk of congenital malformations. However, the best data, which are derived from systematic reviews of prospective observational studies, suggest that benzodiazepines are not associated with an increased risk of congenital anomalies [49,52-57]. As an example, a systematic review identified nine observational studies and conducted the following meta-analyses [51]:
●A meta-analysis of five studies (n>3000 exposed infants and >127,000 unexposed infants) found that the risk of major congenital malformations was nearly identical for the two groups.
●A meta-analysis of five studies (n>6000 exposed infants and >1,000,000 unexposed infants) found that the risk of cardiac malformations was similar for the two groups (however, heterogeneity across studies was large).
●A meta-analysis of two studies (n>3000 exposed infants and >893,000 unexposed infants) found that benzodiazepines were not associated with an increased risk of orofacial cleft.
Conversely, some retrospective studies suggest that benzodiazepines or hypnotic benzodiazepine receptor agonists may be associated with major congenital malformations [20,57-59]. However, the results should be viewed cautiously. To the extent that benzodiazepines/hypnotics are associated with teratogenic effects, many authorities consider the absolute risk small [5,20,57]. In addition, retrospective case-control studies may be subject to recall bias [52]. Another issue is that the association between in utero exposure to benzodiazepines/hypnotics and congenital anomalies may be spurious because of confounding maternal characteristics that increase the risk of congenital anomalies. The pregnant females who use these drugs may be older and more likely to smoke and use other drugs that are regarded as teratogens (eg, antiepileptics), compared with pregnant patients who do not use benzodiazepines/hypnotics [60].
Combination treatment — Benzodiazepines are often used concurrently with antidepressants, and the combination appears to be associated with a small increased risk of congenital anomalies. A meta-analysis of three prospective studies included infants exposed in utero during the first trimester to benzodiazepines plus antidepressants and infants not exposed to this combination (sample sizes not reported) [56]. The rate of all major congenital malformations was greater in the group exposed to benzodiazepines plus antidepressants (odds ratio 1.4, 95% CI 1.1-1.8). However, the clinical significance is probably small, because the result suggests that the absolute risk is very small [61].
Pregnancy loss — Antenatal benzodiazepines are associated with a nearly twofold increased risk of pregnancy loss (also referred to as miscarriage or spontaneous abortion) [51]:
●A meta-analysis of five prospective observational studies included pregnant females who were either treated with benzodiazepines (n>3000) or were not (n>1,000,000); benzodiazepines were associated with an increased risk of pregnancy loss (odds ratio 1.9, 95% CI 1.4-2.4) [62].
●Two subsequent studies, using relatively large administrative health care claims datasets, compared females with pregnancies that either ended in pregnancy loss (cases) or did not (controls) [63,64]. After adjusting for potential confounding factors, the analyses found that exposure to benzodiazepines in early pregnancy was nearly two times greater in the cases than the controls (odds ratio 1.9, 95% CI 1.6-2.1; odds ratio 1.7, 95% CI 1.5-1.9). The studies also found that short- and long-acting benzodiazepines were both associated with an elevated risk of pregnancy loss, as were each of the commonly prescribed drugs, including alprazolam and lorazepam.
General information about pregnancy loss is discussed separately. (See "Pregnancy loss (miscarriage): Terminology, risk factors, and etiology".)
Hypertensive disorders of pregnancy — Multiple prospective studies suggest that benzodiazepines are not associated with hypertensive disorders during pregnancy [48,65].
Labor and delivery outcomes
●Cesarean delivery – Antenatal use of benzodiazepines may be associated with an increased risk for cesarean delivery. A prospective study followed pregnant females who were either treated with benzodiazepines (n = 67) or were not (n>2500) [48]. After adjusting for potential confounding, the analyses found that cesarean delivery occurred more often in females treated with benzodiazepines (odds ratio 2.5, CI 1.4-4.4).
●Preterm birth – It is not clear if antenatal benzodiazepines are associated with preterm birth (eg, <37 weeks gestational age), due to conflicting results across studies:
•A national registry study identified pregnant females with preterm deliveries (n>42,000), including females treated with benzodiazepines or hypnotic benzodiazepine receptor agonists (n = 211) [59]. Preterm birth occurred more often with exposure early in the pregnancy compared with no exposure (odds ratio 1.5, 95% CI 1.3-1.8), and also with later exposure (odds ratio 2.6, 95% CI 1.9-3.4).
•Conversely, in a prospective study of pregnant females who were either treated with benzodiazepines (n = 67) or were not (n>2500), the analyses found that after adjusting for potential confounding factors, preterm birth was comparable in the two groups (odds ratio 1.98, CI 0.97-4.04) [48].
An overview of preterm delivery is discussed separately. (See "Preterm labor: Clinical findings, diagnostic evaluation, and initial treatment".)
●Birth weight – It is not clear if antenatal benzodiazepines are associated with low birth weight (eg, <2500 g) due to conflicting results across studies. Evidence that suggests benzodiazepines are not associated with low birth weight includes a meta-analysis of three prospective studies, which included pregnant females who were treated with benzodiazepines (n = 478) or were not (n = 559) [51]. The risk of low birth weight was nearly identical in the two groups.
However, other studies suggest that antenatal benzodiazepines are associated with low birth weight. A prospective study followed pregnant females who were either treated with benzodiazepines (n = 67) or were not (n>2500) [48]. After adjusting for potential confounding factors, the analyses found that low birth weight occurred more frequently in neonates exposed to benzodiazepines (odds ratio 3.4, CI 1.6-7.3).
In addition, a national registry study identified low-birth-weight infants (n>27,000), including those exposed in utero to benzodiazepines or hypnotic benzodiazepine receptor agonists (n = 131) [59]. Low birth weight occurred more often with early exposure compared with no exposure (odds ratio 1.3, 95% CI 1.1-1.6) and was also associated with exposure later in the pregnancy (odds ratio 1.9, 95% CI 1.3-2.8).
Postnatal effects
●Neonatal adaption syndrome – Chronic administration of benzodiazepines proximal to delivery is associated with the neonatal adaption syndrome (toxicity and withdrawal), including [2,5,52,59]:
•Apgar scores that are relatively low
•Apnea
•Hypothermia
•Hyperreflexia
•Hypertonia or hypotonia
•Irritability
•Lethargy
•Restlessness
•Tremor
•Diarrhea
•Poor feeding
•Vomiting
These signs and symptoms are widely reported and have been observed secondary to several benzodiazepines, including alprazolam, chlordiazepoxide, clonazepam, diazepam, and lorazepam [4,58]; some experts assert that all benzodiazepines are associated with neonatal toxicity and withdrawal [66]. In addition, toxicity and withdrawal may occur more often in preterm infants than term infants [67], necessitate neonatal ventilatory support [48], and persist for up to three months or longer [4,66].
●Neurobehavioral development – Multiple prospective studies suggest that intrauterine exposure to benzodiazepines or hypnotic benzodiazepine receptor agonists is not associated with adverse effects upon neurobehavioral development at age three, five, and six years, including impaired fine and gross motor skills, impaired cognitive functioning and communication skills, aggressive behavior, and anxiety and symptoms of attention deficit hyperactivity disorder [14,68,69]. Other studies suggest an association with negligible adverse effects at age 1.5 and 3 years that resolve by age 5 years [70,71].
LITHIUM —
Lithium appears to completely equilibrate across the placenta. An observational study of 27 infant-mother pairs found that the ratio of lithium serum concentrations in umbilical cord blood to maternal blood was 1.1, across a wide range of maternal concentrations [72].
Maternal lithium levels may drop as pregnancy progresses, and doses may need to be increased. (See "Bipolar disorder in women: Preconception and prenatal maintenance pharmacotherapy", section on 'Refractory patients'.)
Teratogenicity — Observational studies suggest that fetal lithium exposure increases the probability of major teratogenic effects twofold, but the absolute risk is considered small:
●In a meta-analysis of four prospective and retrospective studies that compared exposed infants (n = 365) and unexposed infants (n>974,000), the risk of congenital malformations was two times greater in exposed infants (odds ratio 2, 95% CI 1-4) [51]. However, the increased absolute risk of congenital abnormalities was 7 per 1000, which was considered small.
●A subsequent meta-analysis of a different set of four studies included pregnant females who were either treated with lithium during the first trimester (n = 941) or were not (n>21,000) [73]. The risk of any major congenital anomaly was greater in the lithium-exposed females (odds ratio 1.8, 95% CI 1.2-3.0). However, the absolute risk was deemed small.
The finding that intrauterine exposure to lithium is associated with a small increased risk of major teratogenic effects should be viewed cautiously because some of the studies did not control for confounding factors, such as alcohol use disorder, obesity, smoking, and substance use disorder, which are also associated with congenital malformations and are relatively common in patients with bipolar disorder [74].
Cardiac — Antenatal lithium exposure is generally thought to be associated with teratogenic cardiac effects [2,5]. Cardiac malformations that have been observed include Ebstein anomaly (abnormalities of the tricuspid valve and right ventricle), right ventricular outflow tract obstruction defects, coarctation of the aorta, and mitral atresia [5,75-77]. However, the absolute risk may be small:
●A meta-analysis of four observational studies included pregnant females with mood disorders who were either treated with lithium during the first trimester (n = 1436) or were not (n>23,000) [73]. The risk of any major cardiac anomaly was greater in the lithium-exposed females (odds ratio 1.8, 95% CI 1.1-2.8). However, the absolute risk was deemed small.
●A retrospective study of an administrative claims database examined cardiac defects in infants with first-trimester exposure to either lithium (n = 663) or lamotrigine (n = 1945). After adjusting for the probability (propensity) to receive lithium or lamotrigine based upon potential confounding factors, the analyses found that cardiac defects occurred in more infants exposed to lithium than lamotrigine (2.4 versus 1.4 percent; relative risk 2.3, 95% CI 1.2-4.3) [78]. In addition, right ventricular outflow tract obstruction defects were more common in lithium-exposed infants than infants who were not exposed to either lithium or lamotrigine (0.6 versus 0.2 percent; relative risk 2.7, 95% CI 1.0-7.1).
Other results from the same study suggested that the association between lithium exposure and cardiac defects may perhaps be dose-dependent, such that cardiac malformations occurred in more infants with exposure to a maternal dose >900 mg/day, compared with unexposed infants.
Ebstein anomaly may be the most common fetal cardiac defect associated with prenatal lithium exposure. The estimated risk in the general population is 1 in 20,000 live births [79-81]; following first-trimester exposure to lithium, the risk may increase 20-fold to approximately 1 in 1000 [82], an absolute risk that many authorities consider low [2,5,20,22,83].
However, the evidence that lithium causes Ebstein anomaly is uncertain [32]. As an example, a meta-analysis of six observational studies (n = 264 pregnant females) found that antenatal exposure to lithium did not differ statistically between cases of Ebstein anomaly and controls without the anomaly [84]. Other studies of pregnant females treated with lithium found that none of their infants developed Ebstein anomaly [78,85].
General information about Ebstein anomaly is discussed separately. (See "Ebstein anomaly: Clinical manifestations and diagnosis".)
Perinatal mortality — It is not clear if prenatal exposure to lithium increases perinatal mortality, due to conflicting results across prospective observational studies:
●One study found that the frequency of pregnancy loss (miscarriages) and stillbirths were each similar in patients treated with lithium (n = 138) and controls (n = 148) [86].
●However, a second study found that pregnancy loss occurred in more patients exposed to lithium during the first trimester (n = 183) than females exposed to nonteratogenic substances (n = 748; 16 versus 6 percent) [87].
General information about perinatal mortality is discussed separately. (See "Perinatal mortality".)
Pregnancy complications
●Pregnancy loss – Exposure to lithium during the first trimester does not appear to be associated with pregnancy loss (also referred to as miscarriage or spontaneous abortion). A meta-analysis of two prospective observational studies included pregnant females with mood disorders who were either treated with lithium during the first trimester (n = 321) or were not treated with lithium (n = 220); the frequency of spontaneous abortion in the two groups was comparable (odds ratio 2.5, 95% CI 0.6-10.8) [73].
General information about pregnancy loss is discussed separately. (See "Pregnancy loss (miscarriage): Terminology, risk factors, and etiology".)
●Hypertensive disorders of pregnancy – Antenatal use of lithium is not associated with an increased risk for hypertensive disorders of pregnancy [65]. As an example, a meta-analysis of five observational studies included pregnant females with mood disorders who were either treated with lithium at any time during pregnancy (n = 612) or were not (n>21,000); the prevalence of pre-eclampsia in both groups was 2 percent [85].
●Diabetes – Antenatal use of lithium is not associated with an increased risk for diabetes during pregnancy. In a meta-analysis of five observational studies that included pregnant females with mood disorders who were either treated with lithium at any time during pregnancy (n = 489) or were not (n>7000), the prevalence of diabetes was comparable in females taking lithium and controls (6 and 5 percent). [85].
●Fetal distress – Antenatal use of lithium is not associated with an increased risk for fetal distress. As an example, a meta-analysis of six observational studies included pregnant females with mood disorders who were either treated with lithium at any time during pregnancy (n = 727) or were not (n>21,000) [85]. The prevalence of fetal distress was comparable in the lithium-exposed group and controls (14 and 13 percent).
●Postpartum hemorrhage – Antenatal use of lithium is not associated with an increased risk for postpartum hemorrhage. In a meta-analysis of five observational studies that included pregnant females with mood disorders who were either treated with lithium at any time during pregnancy (n = 489) or were not (n>7000), the prevalence of postpartum hemorrhage in both groups was 7 percent [85].
Labor and delivery outcomes
●Cesarean delivery – Antenatal use of lithium is not associated with an increased risk for cesarean delivery. As an example, a meta-analysis of six observational studies included pregnant females with mood disorders who were either treated with lithium at any time during pregnancy (n = 727) or were not (n>21,000) [85]. The prevalence of cesarean delivery was comparable in the lithium-exposed group and controls (27 and 26 percent).
●Preterm birth – Use of lithium during pregnancy is not associated with preterm delivery. A meta-analysis of six observational studies included pregnant females with mood disorders who were either treated with lithium at any time during pregnancy (n>1000) or were not (n>21,000); the frequency of preterm birth in the two groups was comparable [73].
An overview of preterm delivery is discussed separately. (See "Preterm labor: Clinical findings, diagnostic evaluation, and initial treatment".)
●Birth weight – Antenatal exposure to lithium is not associated with low birth weight. A meta-analysis of three observational studies included pregnant females with mood disorders who were either treated with lithium at any time during pregnancy (n = 980) or were not (n>21,000); the frequency of low birth weight in the two groups was comparable [73].
Postnatal effects — Use of lithium during the second and third trimester can result in neonatal complications, including [4,6,7,88-91]:
●Cardiomegaly
●Gastrointestinal bleeding
●Goiter and hypothyroidism
●Hepatomegaly
●Hypoglycemia
●Arginine vasopressin resistance (previously called nephrogenic diabetes insipidus)
●Polyhydramnios
●Premature labor
●Shock
Lithium toxicity can also occur in newborns with late pregnancy exposure; symptoms include [4,6,7,88-91]:
●Low Apgar scores
●Apnea, shallow respirations, and cyanosis
●Bradycardia or tachycardia
●Cardiac arrhythmias and abnormal electrocardiogram
●Feeding difficulties
●Lethargy or coma
●Muscle flaccidity and hypotonia
●Poor suck, grasp, and Moro reflexes
●Seizures
●Twitching
Neonatal lithium toxicity and complications are more common in newborns with higher serum lithium concentrations. In a study of 24 infants exposed in utero to lithium, lower Apgar scores, higher rates of central nervous system and neuromuscular complications, and longer hospital stays were observed in infants born with umbilical cord serum lithium concentrations >0.6 mEq/L (0.6 mmol/L), compared with infants with lower concentrations [72]. Lithium toxicity generally resolves in one to two weeks.
Antenatal use of lithium is associated with an increased risk for rehospitalization of the infant within one month of birth. As an example, a meta-analysis of six observational studies included pregnant women with mood disorders who were either treated with lithium at any time during pregnancy (n = 718) or were not treated with lithium (n>21,000) [85]. Neonatal readmission within 28 days of birth occurred in twice as many infants who were exposed in utero to lithium, compared with controls (28 versus 14 percent).
Available studies suggest that prenatal lithium exposure does not adversely affect neurobehavioral development [92]:
●A prospective study found that major developmental milestones (eg, smiling, lifting head, sitting, crawling, standing, talking, and walking) were achieved at comparable ages for lithium-exposed children (n = 22) and unexposed controls (n = 148) [86].
●Retrospective studies found that behavioral measures of development were comparable for children exposed to lithium prenatally (n = 60) and their unexposed siblings (n = 57) [93] and that intelligence quotient was comparable in children who were exposed to lithium prenatally or were not [94].
NEUROMODULATION —
Patients with unipolar major depression or bipolar disorder who decline or do not respond to standard treatment with pharmacotherapy and psychotherapy are candidates for neuromodulation procedures [95]. Noninvasive neurostimulation therapies include electroconvulsive therapy (ECT), repetitive transcranial magnetic stimulation (TMS), and transcranial direct current stimulation (tDCS).
General information about neuromodulation procedures is discussed separately. (See "Unipolar depression in adults: Overview of neuromodulation procedures".)
Electroconvulsive therapy — ECT has been more widely studied in pregnant women than any other type of neuromodulation, and is generally regarded as safe during all three trimesters by the American Psychiatric Association, American College of Obstetricians and Gynecologists, Canadian Network for Mood and Anxiety Treatments, Royal Australian and New Zealand College of Psychiatrists, and several other experts [29,95-99]. Nevertheless, obstetrical care should be readily available to manage patients at high risk for complications such as vaginal bleeding and preterm birth, manage fetal emergencies, and urgently deliver the baby [95,100].
Additional information about ECT is discussed separately. (See "Overview of electroconvulsive therapy (ECT) for adults" and "Unipolar major depression in adults: Indications for and efficacy of acute electroconvulsive therapy (ECT)" and "Medical evaluation for electroconvulsive therapy" and "Technique for performing electroconvulsive therapy (ECT) in adults", section on 'Pregnancy'.)
Perinatal mortality — ECT appears to be associated with a low risk of perinatal mortality:
●A review of 339 case reports of pregnant females treated with ECT found only one fetal death that was attributed to ECT (stillbirth due to maternal status epilepticus secondary to ECT) [96]. In another case, pregnancy loss (miscarriage) occurred 24 hours after an ECT treatment; however, it was thought that ECT was probably not the cause of death.
●Although a subsequent review of pregnant females (n = 169) treated with ECT found that fetal or neonatal death occurred in 7 percent, the investigators did not distinguish mortality attributed to ECT from mortality due to other causes [101].
●A third study found that among 27 pregnant females who received ECT, stillbirth occurred in one (4 percent) [102]. However, the cause of fetal death was not determined because the family refused an autopsy.
Teratogenicity — Neither the number nor pattern of major congenital malformations in children exposed in utero to ECT implicates it as a causal factor in organ dysgenesis [5,103]. The rate of malformations after exposure to ECT is approximately 2 to 4 percent of livebirths, which is consistent with the rate in the general population. (See "Congenital anomalies: Epidemiology, types, and patterns", section on 'Epidemiology'.)
Evidence regarding antenatal ECT and teratogenesis includes a review of 300 case reports of ECT, among which there were five cases (2 percent) of major congenital anomalies (hypertelorism, talipes equinovarus [clubfoot], optic atrophy, anencephaly, and pulmonary cysts) [104]. The review concluded that these malformations were not the result of ECT.
A subsequent review of 169 pregnant females treated with ECT found that fetal malformations occurred in 4 percent (n = 7), but the investigators did not determine whether the congenital anomalies were due to ECT [101]. The seven cases consisted of hyaline membrane disease, small left cerebellum and cortical infarct, transposition of the great vessels, intrauterine growth retardation, glaucoma and cleft palate, anencephaly, and lung cysts.
A third study found that among 27 neonates exposed in utero to ECT, hip dysplasia occurred in occurred in one (4 percent) [102]. Orthopedic consultation deemed it unlikely that the defect was due to ECT.
The risk of teratogenic effects and neonatal toxicity posed by ECT anesthetic drugs appears to be low [89,96,100,104,105]:
●Glycopyrrolate – The anticholinergic glycopyrrolate does not readily cross the placenta, and there do not appear to be any reports of teratogenesis with the injected solution.
●Methohexital or propofol – Although the general anesthetics methohexital and propofol cross the placenta, there do not appear to be any reports of teratogenesis with either drug.
●Succinylcholine – The muscle relaxant succinylcholine generally does not appear to significantly affect the fetus; typically, little or no succinylcholine crosses the placenta.
In addition, it is unlikely that these drugs cause congenital malformations and toxicity because of the relatively infrequent and brief exposure that occurs during a course of ECT.
Adverse effects — The most common adverse effect of ECT in pregnant females is premature labor; across different reviews of case reports, the incidence ranges from 4 to 11 percent [96,101]. A subsequent retrospective study of 27 pregnant females who received ECT found that three (11 percent) had transient uterine contractions during ECT, but no obstetric intervention was required and there were no preterm births [102].
In addition, case reports have described other maternal complications thought to be related to ECT, including abdominal pain, placental abruption, transient maternal hypertension, and vaginal bleeding [29,106].
In the fetus, the most common adverse effects are transient bradycardia and other cardiac arrhythmias, with an incidence of 3 to 9 percent [96,101]. There is little or no evidence that fetal exposure to ECT adversely affects intrauterine growth [5].
Postnatal effects — Studies of children who were exposed in utero to ECT and followed for up to six years after birth suggest that ECT is not associated with adverse postnatal effects [106,107]. A review of 300 case reports of ECT during pregnancy concluded that there was no evidence of adverse effects upon postnatal neurobehavioral development [104]. Similarly, a subsequent review found little or no evidence that fetal exposure to ECT causes neonatal toxicity or adverse developmental effects [5].
Transcranial magnetic stimulation — Repetitive TMS has been studied in fewer pregnant females than ECT [95]. Nevertheless, both high-frequency, left-sided TMS and low-frequency, right-sided TMS are generally regarded as safe because the fetus is not exposed to the magnetic field [95,108]. The only caveat is that TMS rarely causes seizures, which are associated with preterm labor. (See 'Adverse effects' above.)
Evidence regarding the safety of TMS includes a small randomized trial that compared right-sided active TMS with sham TMS in 22 pregnant females with unipolar major depression who completed the study protocol of 20 sessions [109]. There were no cases of perinatal mortality or major congenital malformations in the study, and the mean gestational age, birth length, and birth weight were normal in the two groups. However, late preterm birth occurred in 3 of the 11 (27 percent) mothers treated with active TMS, compared with none of the females treated with sham TMS. In other observational studies that included a total of 57 patients with antenatal depression, TMS was not associated with adverse pregnancy, neonatal, or postnatal effects [108].
Additional information about TMS is discussed separately. (See "Unipolar depression in adults: Indications, efficacy, and safety of transcranial magnetic stimulation (TMS)" and "Unipolar major depression: Administering transcranial magnetic stimulation (TMS)".)
Transcranial direct current stimulation — tDCS appears to be safe for antenatal depression, given the relatively benign nature of the treatment and safety profile in the general population of depressed patients. (See "Unipolar depression in adults: Overview of neuromodulation procedures", section on 'Transcranial direct current stimulation'.)
In addition, a small randomized trial that compared active tDCS with sham treatment in 20 patients with antenatal depression found no evidence of adverse maternal or fetal consequences [110]. None of the study patients experienced a serious pregnancy complication and fetal monitoring revealed no abnormalities. In addition, the mean gestational age and birth weight were comparable for the two groups, as were postnatal developmental outcomes.
Other — Other neuromodulation interventions include invasive therapies that involve a surgical procedure and thus require anesthesia [95]:
●Vagus nerve stimulation – One case report describes a female with antenatal depression who received vagus nerve stimulation for three years prior to pregnancy and continued the treatment throughout pregnancy without any adverse perinatal events. Additional information about vagus nerve stimulation is discussed separately. (See "Unipolar depression in adults: Treatment with surgical approaches", section on 'Vagus nerve stimulation'.)
●Deep brain stimulation – Deep brain stimulation electrodes were implanted in three females with dystonia who subsequently had healthy pregnancies. Additional information about deep brain stimulation is discussed separately. (See "Unipolar depression in adults: Treatment with surgical approaches", section on 'Deep brain stimulation'.)
SOCIETY GUIDELINE LINKS —
Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Bipolar disorder".)
EDUCATIONAL MATERIAL FOR PATIENTS —
We encourage clinicians to educate their patients about the benefits and risks of pharmacotherapy and neuromodulation for psychiatric disorders. One useful resource for pregnant patients who are receiving pharmacotherapy is the JAMA Patient Page entitled “Safety of Medications Used During Pregnancy,” which can be downloaded and printed for patients [111].
In addition, patients can be directed to a website maintained by the Centers for Disease Control and Prevention, which includes multiple subpages such as “Facts about Medicine and Pregnancy” and “Guidelines and Recommendations.” Another useful website for patients (and clinicians) is MotherToBaby.
SUMMARY
●Definition of a teratogen – Teratogens are factors that can alter normal intrauterine development of fetal growth, anatomic structures, physical functioning, and postnatal development. This definition encompasses environmental exposures (eg, drugs), maternal medical disorders, infectious agents, and genetic conditions. (See 'Definition of a teratogen' above and "Congenital anomalies: Causes", section on 'Teratogens'.)
●Varying teratogenic risk of psychotropic drugs – The estimated risk of major congenital malformations appears to vary among psychotropic medications; the rank order from greatest to least teratogenic risk is (see 'General principles' above):
•Antipsychotics
•Antidepressants
●Antiparkinsonian drugs – Among the antiparkinsonian drugs that are used to treat extrapyramidal symptoms secondary to antipsychotics, reviews suggest that organ malformation appears to be less likely with diphenhydramine than amantadine, benztropine, or trihexyphenidyl. (See 'Antiparkinsonian drugs used for treating extrapyramidal symptoms' above.)
●Antidepressants – Much of the evidence about the teratogenicity, pregnancy complications, and postnatal risks associated with antidepressants is based upon observational studies of pregnant patients with unipolar major depression. (See "Antenatal use of antidepressants and the potential risk of teratogenicity and adverse pregnancy outcomes: Selective serotonin reuptake inhibitors" and "Antenatal use of antidepressants and risks of teratogenicity and adverse pregnancy outcomes: Drugs other than selective serotonin reuptake inhibitors" and "Antenatal exposure to selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs): Neonatal outcomes".)
●Antiepileptic drugs – Most observational studies that have examined the association between antiepileptics and congenital malformations and postnatal risks have been conducted primarily in pregnant patients with epilepsy. (See "Risks associated with epilepsy during pregnancy and the postpartum period", section on 'Effects of ASMs on the fetus and child'.)
●Antipsychotics – Most studies have found that exposure during pregnancy to first- and second-generation antipsychotics does not appear to increase the risk of major physical malformations above rates observed in the general population. Antenatal exposure to first-generation antipsychotics may be associated with preterm birth, whereas second-generation antipsychotics do not seem to carry the same risk. Although it is not clear whether chronic administration of antipsychotics during the third trimester causes symptoms of the neonatal adaptation syndrome (neonatal toxicity and withdrawal), the best evidence suggests that other factors play a bigger role. (See 'Antipsychotics' above.)
●Benzodiazepines – The best evidence suggests that exposure to benzodiazepines or to hypnotic benzodiazepine receptor agonists during pregnancy is not associated with an increased risk of congenital malformations or hypertensive disorders during pregnancy. However, antenatal exposure to benzodiazepines appears to be associated with a nearly twofold increased risk of pregnancy loss. In addition, chronic administration of benzodiazepines proximal to delivery is associated with the neonatal adaption syndrome (toxicity and withdrawal). (See 'Benzodiazepines' above.)
●Lithium – Although observational studies have found that fetal lithium exposure is associated with teratogenic effects, the increased absolute risk of congenital abnormalities (7 per 1000) is considered small. Antenatal lithium exposure is generally thought to be associated with teratogenic effects that primarily involve the heart, and the association may perhaps be dose dependent. Ebstein anomaly may be the most common fetal cardiac defect associated with prenatal lithium exposure. Second- and third-trimester lithium exposure can lead to neonatal complications and lithium toxicity in the newborn. (See 'Lithium' above.)
●Neuromodulation
•Electroconvulsive therapy (ECT) – ECT appears to be associated with a low risk of perinatal mortality. However, the number and pattern of congenital malformations in children exposed in utero to ECT suggests that it is not a causal factor in organ dysgenesis. The most common adverse effect of ECT in pregnant females is premature labor. (See 'Electroconvulsive therapy' above.)
•Transcranial magnetic stimulation (TMS) – TMS is generally regarded as safe because the fetus is not exposed to the magnetic field. The only caveat is that TMS rarely causes seizures, which are associated with preterm labor. (See 'Transcranial magnetic stimulation' above.)
