Lithium Embryopathy

Malformation and perinatal risk of prenatal exposure to lithium salts, with Ebstein's anomaly as the signal malformation, the management of lithium in pregnancy, at delivery and during breastfeeding, and the differential diagnosis with genetically caused Ebstein's anomaly

Update of the literature and regulatory information: 29 October 2026.

0. Initial Reference Table

1. Introduction

1.1. Lithium and bipolar disorder in the reproductive years

Lithium salts were introduced into psychiatry in 1949 by the Australian psychiatrist John Cade; in Europe, their use followed the work of Mogens Schou in Denmark, who conducted a randomised placebo-controlled trial in patients with mania, and in the United States the medicine was only introduced in 1970 [1]. Lithium continues to be regarded as the reference maintenance treatment for bipolar disorder, because of its efficacy in preventing manic and depressive recurrences and suicidal behaviour, and it is the only mood stabiliser with documented efficacy in the perinatal period [25]. Bipolar disorder affects about 1% to 2% of the population, with a mean onset around 25 years of age, so that many women of reproductive age are taking lithium when they become pregnant [2, 3, 21]. Unplanned pregnancies are more frequent in women with bipolar disorder than in controls, which makes exposure in the 1st trimester a real and not merely theoretical clinical problem [25].

1.2. The register of «lithium babies» and the Ebstein's anomaly hypothesis

Concern about teratogenicity arose from the Register of Lithium Babies, started in Denmark and extended internationally in the early 1970s. Schou and colleagues published the first report in 1973, which suggested increased rates of cardiovascular malformations in the children of mothers treated in early pregnancy (indirect attribution, according to the Oxford Desk Reference) [27]. On the basis of only two cases, a risk of Ebstein's anomaly 400 times that of the general population and an overall risk of heart disease five times higher were then estimated; the final report of 1979 included 225 exposed children, of whom 18 (8%) had congenital heart disease and 6 (3%) Ebstein's anomaly [2]. Schou coined the expression «lithium babies» and, in 1976, published the 5-year follow-up of 60 children exposed in the 2nd and 3rd trimesters, with no differences in development compared with unexposed siblings [1, 27]. On the basis of these data, the regulatory agencies concluded that there was evidence of human fetal risk, but that the benefits might justify use in pregnancy [2].

1.3. The epidemiological reappraisal

The register had serious methodological limitations: it had no control group and enrolment was voluntary and retrospective, with probable over-representation of malformed cases [2]. In 1994, Cohen and colleagues reviewed the literature and found that none of the mothers in four case-control studies of Ebstein's anomaly (with 25, 34, 59 and 89 affected children) had taken lithium, and that two cohorts gave relative risks of 3.0 and 1.5 for all anomalies and of 7.7 and 1.2 for cardiac ones, with wide confidence intervals; they concluded that the risk from the 1st trimester was lower than initially suggested [32]. The multicentre prospective study of Jacobson and colleagues (1992), with 148 women exposed in the 1st trimester followed by four teratology information centres in the United States and Canada, found major malformations in 2.8% of the exposed and 2.4% of the controls, with one case of Ebstein's anomaly detected by fetal echocardiography [33]. The review by Yacobi and Ornoy (2008) counted 222 children with Ebstein's anomaly and 44 with tricuspid atresia in case-control studies, with no mother exposed to lithium, and concluded that the medicine did not appear to be a significant teratogen, while still recommending fetal echocardiography [34].

1.4. The era of the large cohorts

From 2014 onwards, studies with sufficient power to quantify the risk appeared. The prospective cohort of the Israeli teratology information service (Diav-Citrin, 2014) showed more cardiovascular anomalies in those exposed in the 1st trimester [35]. In 2017, Patorno and colleagues analysed 1,325,563 pregnancies from the United States Medicaid programme and demonstrated a modest, dose-dependent increase in the risk of cardiac malformations, including right ventricular outflow tract defects, «much lower than initially postulated» [2]. In 2018, the primary data meta-analysis of six cohorts by Munk-Olsen and colleagues confirmed the increase in major malformations with exposure in the 1st trimester [21]. The Swedish (2021) and Australian (2026) registry studies added associations with spontaneous preterm birth and with the newborn large for gestational age [3, 36].

1.5. Definition of the entities within scope

There is no dysmorphic lithium syndrome comparable to the embryopathies caused by valproate, hydantoin or retinoids; the term «lithium embryopathy» designates, in practice, the set of risks associated with intrauterine exposure: malformations (mainly cardiac) with exposure in the 1st trimester, obstetric and growth outcomes, neonatal toxicity with exposure at the end of pregnancy and exposure through milk. The signal malformation is Ebstein's anomaly, defined by Orphanet as the apical displacement of the functional annulus of the tricuspid valve due to incomplete delamination of the septal and inferior leaflets, which remain inserted within the right ventricle; the anterosuperior leaflet is often redundant, fenestrated or anchored to an abnormal subvalvular apparatus, and the atrioventricular junction and the «atrialised» portion of the right ventricle are dilated, with variable thinning of the wall [19]. Ebstein's anomaly also has genetic and chromosomal causes, which constitute the main differential diagnosis in an exposed child with this malformation [9, 14].

2. Molecular Genetics and Pathophysiology

2.1. Nature of the exposure

Lithium is a monovalent ion eliminated exclusively by the kidney and with a narrow therapeutic window: toxic concentrations (≥1.5 mEq/L) lie close to therapeutic ones (0.8 to 1.2 mEq/L in the US label; 0.5 to 1.0 mmol/L for prophylaxis according to the Portuguese summary of product characteristics, SmPC) [4, 37]. Lithium crosses the placenta completely: in ten deliveries with simultaneous measurement and 32 cases from the literature, the ratio between cord blood and maternal blood concentrations was 1.05 (standard deviation 0.13), uniform for maternal lithium concentrations between 0.2 and 2.6 mEq/L [38]. In a Dutch cohort of 29 neonates, maternal and neonatal lithium concentrations were strongly correlated (Pearson coefficient 0.703) [39]. A 2024 review states that the placenta protects the fetus from lithium entry, but this assertion contradicts direct measurements of maternal–fetal equilibrium and should not be followed [1, 38].

2.2. Pharmacokinetics in pregnancy and the postpartum period

The rise in cardiac output and glomerular filtration rate in pregnancy accelerates lithium elimination, which may lower lithium concentrations below the therapeutic range or below the pre-pregnancy level [22]. The serum lithium concentration falls gradually during the first half of pregnancy, with the nadir at around 18 weeks, and rises again in the third trimester, approaching the preconception value [25, 39]. Clearance increases by 30% to 50% over the course of pregnancy, according to a 2026 Swiss review [40]. At delivery, vascular volume falls rapidly and renal clearance may return to pre-pregnancy values, which underpins the US label recommendation to reduce or withdraw lithium 2 to 3 days before the expected date of delivery [4]. This premise has been challenged by a cohort of 100 pregnancies in which the lithium concentration/dose ratio did not vary between the week before and the week after delivery (mean concentrations of 0.73 and 0.70 mmol/L) [39]. Sodium restriction and diuretics should be avoided, because they reduce lithium elimination [4].

2.3. Proposed mechanisms of teratogenicity

The mechanisms have not been studied directly in humans; the hypotheses derive from animal and cellular models [1]. The first involves the phosphoinositide cycle: lithium inhibits inositol monophosphatase and inositol polyphosphate 1-phosphatase, and the maturational arrest induced by lithium in sea urchin embryos is reversed by myo-inositol, although myo-inositol does not protect rat embryos exposed to lithium chloride [1]. The second involves inhibition of glycogen synthase kinase 3β (GSK-3β) and the consequent interference with Wnt/β-catenin signalling, which alters morphogenesis in eukaryotic cells, ascidians, zebrafish and pig [1]. In rodents, doses producing lithium concentrations similar to human therapeutic ones cause fetal mortality, lower fetal weight, skeletal abnormalities, cleft palate (in the mouse only) and implantation changes [4]. Animal teratogenicity depends on species and dose [1].

2.4. Dose, timing and duration of exposure

Malformation risk is concentrated in cardiac organogenesis, in the first trimester; no increase in malformations has been found with exposure limited to the second and third trimesters [25, 39]. In the Medicaid cohort, the relative risk of cardiac malformation was 1.11 with a daily dose up to 600 mg, 1.60 between 601 and 900 mg and 3.22 above 900 mg, and all right ventricular outflow tract defects among exposed infants occurred with doses above 600 mg/day; lamotrigine showed no dose–response relationship [2]. A 2020 meta-analysis described, in a qualitative synthesis, neonates who were more reactive and without increased cardiac risk when the maternal lithium concentration was below 0.64 mEq/L and the dose below 600 mg/day [41]. Neonatal effects (hypotonia, respiratory depression, bradycardia), by contrast, depend on the lithium concentration at the time of delivery [4, 38], and the time-weighted mean lithium concentration in pregnancy was associated with shorter gestation in a Rotterdam cohort [42].

2.5. Effects on the thyroid, the kidney and growth

Lithium has thyroid (hypothyroidism and hyperthyroidism) and renal effects that are well recognised in adults and may occur in the exposed neonate [4, 27]. Lithium-induced nephrogenic diabetes insipidus has been proposed as the mechanism of the severe polyhydramnios described in a 42-year-old pregnant woman on long-term treatment [43]. Exposure is associated with greater fetal growth: at 18–22 weeks, exposed fetuses had larger head and abdominal circumferences and femur length than the reference population, and birth weight was on average 142 g higher [44]. Neonatal hypoglycaemia, observed in analyses restricted to women with psychiatric illness, is consistent with the excess of neonates large for gestational age and with the effects of lithium on blood glucose in adults [3].

2.6. Genetics of Ebstein anomaly in the differential diagnosis

Ebstein anomaly is genetically heterogeneous [9]. About 20% of patients have extracardiac anomalies in the context of chromosomal or Mendelian disorders, and the non-syndromic form may be sporadic or familial [14]. In an Italian series of 44 consecutive patients, 27% were syndromic (CHARGE syndrome in 2; VACTERL association, Noonan syndrome, Kabuki syndrome, Holt-Oram syndrome and Cornelia de Lange syndrome in 1 each) and the 1p36 and 8p23.1 deletions were the most frequent chromosomal imbalances; the 28 non-syndromic patients tested had no variants in GATA4 or NKX2-5 [9, 14].

The sarcomeric gene MYH7 (14q11.2), which encodes the cardiac β-myosin heavy chain, is the best established monogenic cause. In a cohort of 141 probands with Ebstein anomaly, heterozygous MYH7 variants were found in 8 (6%), almost all missense and one de novo; left ventricular non-compaction cardiomyopathy was present in 6 of the 8 probands with a variant according to the original abstract (7 confirmed and 1 uncertain according to OMIM) and in none of the 133 without a variant [6, 16]. Variants associated with non-compaction with or without Ebstein anomaly cluster in the region of the ATPase active site of the N-terminal myosin head [45]. Conversely, in the non-compaction cardiomyopathy series compiled in Clinical Cardiogenetics, Ebstein anomaly is one of the most frequent associated congenital heart defects, present in 11 of 130 patients (8%), in addition to 14 published isolated cases [45]. Documented examples are NM_000257.4(MYH7):c.847T>G p.(Tyr283Asp), likely pathogenic in ClinVar, in two families with Ebstein anomaly in the probands, and NM_000257.4(MYH7):c.5754C>G p.(Asn1918Lys), pathogenic in ClinVar, in a family with four affected members across three generations in which the proband had marked non-compaction and Ebstein anomaly diagnosed at the age of 3 years [16, 46, 47]. The variant NM_000257.4(MYH7):c.3658_3660del p.(Glu1220del), described in a family in which the father and four children had Ebstein anomaly, has an aggregate classification of uncertain significance in ClinVar [48, 49]. The Ebstein-with-non-compaction phenotype linked to MYH7 behaves as an autosomal dominant disorder with variable penetrance [11].

The gene TPM1, which encodes thin-filament α-tropomyosin, was the second to be associated with Ebstein anomaly with non-compaction: a de novo variant NM_001018005.2(TPM1):c.475G>A p.(Asp159Asn), normalised in VariantValidator, in a 2-year-old child who died during assessment for transplantation, and the variant p.Leu113Val cosegregating in an autosomal dominant family with five affected members, two of them with Ebstein anomaly [7, 12, 50]. The transcription factor NKX2-5 (5q35) causes autosomal dominant congenital heart disease in which atrial septal defect with progressive atrioventricular block predominates, but in which Ebstein anomaly and other tricuspid abnormalities have been described [8, 13]. Microarray studies in isolated Ebstein anomaly have found candidate copy number variants in 46% to 51% of cases and rare variants in cardiomyopathy genes in 23%, involving genes of the NOTCH (HEY1) and BMP pathways and of myocardial development (NODAL, PDLIM5, SIX1, ASF1A, FGF12), which points to a disturbance of the myocardial programme in the genesis of the malformation [51, 52].

3. Clinical Spectrum and Manifestations by System

3.1. Magnitude of the malformation risk

The risk estimate has changed radically between the 1970s register and current cohorts. Table 1 summarises the principal studies, with the population, design and effect measure of each.

In the Medicaid cohort, which is the reference study for cardiac risk, the increase corresponded to about one additional case per 100 live births exposed in the first trimester, with an interval compatible with up to two additional cases [2]. The relative risk of non-cardiac malformations was 1.22 (95% CI 0.81–1.84) and that of any malformation 1.37 (95% CI 1.01–1.87) [2, 21]. The 2024 CANMAT (Canadian Network for Mood and Anxiety Treatments) guideline summarises the risk as a relative risk of about 1.7 for cardiovascular malformations, dose-dependent, without a clear association with non-cardiac malformations [22]. The UKTIS leaflet translates this risk for families: about 1 in every 100 babies in the United Kingdom has a heart defect and some studies suggest about 2 in every 100 among those exposed to lithium, that is, 98 in every 100 are born with a normal heart [28].

3.2. Cardiovascular system

Ebstein anomaly is the malformation historically associated with lithium, but its absolute frequency among exposed infants is very low and most large cohorts lack the power to study it in isolation. In the Medicaid cohort, right ventricular outflow tract defects (the group that includes Ebstein anomaly) occurred in 0.60% of exposed and 0.18% of unexposed infants (adjusted relative risk 2.66; 95% CI 1.00–7.06), while the remaining cardiac defects had a relative risk of 1.46 (0.84–2.57); none of the defects in exposed infants was coded specifically as Ebstein anomaly, but most were compatible with defects that accompany it, such as pulmonary atresia and stenosis [2]. The authors chose this outcome group because a right ventricular outflow tract defect is more likely to be labelled as Ebstein anomaly in the infant of a mother treated with lithium, which would make the diagnosis of Ebstein anomaly subject to differential misclassification [2]. In the meta-analysis of six cohorts no case of Ebstein anomaly was observed [21]; in the Israeli study there was one exposed fetus with Ebstein anomaly and two retrospective cases excluded because contact with the service followed the ultrasound diagnosis [35]. In the Swedish cohort, two thirds of the cardiac malformations in exposed infants were septal defects [3]. The Oxford Desk Reference lists Ebstein anomaly, tricuspid atresia and atrial septal defect as the defects described with lithium, classifying the risk as “equivocal” [54].

Ebstein anomaly has a spectrum ranging from severe fetal and neonatal disease to virtually asymptomatic survival into adulthood [14]. The right ventricle is divided into an atrialised inlet portion, functionally integrated with the right atrium, and a functional trabecular and outlet ventricle; the atrialised portion usually occupies one third of the ventricle and may exceed half in extreme cases [55]. The anomaly predisposes to right bundle branch block, ventricular pre-excitation and sudden death, and atrial fibrillation occurs in about one third of patients [14]. In a Danish and Swedish national cohort of 794 patients born between 1930 and 2017, the lifetime cumulative incidence was about 70% for supraventricular arrhythmia and 19% for ventricular pre-excitation, and supraventricular arrhythmia multiplied the risk of ischaemic stroke by 22.6 [56].

3.3. Other malformations

There is no extracardiac dysmorphic pattern associated with lithium. The meta-analysis of six cohorts found an increase in major malformations overall with first-trimester exposure, without describing a pattern by organ [21], and the 2024 meta-analysis by Callovini and colleagues pooled the increase in “any anomaly” without separating systems [1]. CANMAT concludes that there is no clear association with non-cardiac malformations [22]. In an exposed child with multiple malformations or dysmorphism, the possibility of a genetic cause or of another exposure (valproate, carbamazepine, alcohol, maternal diabetes) should be considered before attribution to lithium (section 6).

3.4. Pregnancy outcomes

Spontaneous abortion was more frequent in exposed women in the Israeli study (adjusted odds ratio 1.94; 95% CI 1.08–3.48), as was elective termination of pregnancy (9.3% vs 2.0%) [35]. In the Dutch cohort of women with bipolar I disorder, miscarriage occurred in 20.8% of exposed and 10.9% of unexposed pregnancies (adjusted odds ratio 2.94; 95% CI 1.39–6.22), whereas Jacobson's study found no difference (9% vs 8%) [57]. Fornaro's meta-analysis estimated an odds ratio of 3.77 for miscarriage with first-trimester exposure, which ceased to be significant when the comparison was restricted to unexposed pregnancies [41]; CANMAT considers that miscarriage, pre-eclampsia and postpartum haemorrhage do not appear to be associated with lithium [22]. In the meta-analysis of six cohorts there was no association with pre-eclampsia, gestational diabetes, fetal distress, postpartum haemorrhage, caesarean section, preterm birth, low birth weight or small for gestational age neonates [21].

Spontaneous preterm birth is the obstetric outcome on which the evidence has diverged most. In the Swedish cohort, it occurred in 8.7% of exposed and 3.0% of unexposed women (adjusted relative risk 2.64; 95% CI 1.82–3.82), with a median of 35.2 weeks, and the association persisted when the comparison was made with unexposed women with psychiatric illness (adjusted relative risk 2.34) or with women who had stopped lithium before pregnancy (2.08) [3]. In the 2026 Australian cohort the association was reproduced (8.1% vs 2.4%; adjusted relative risk 2.18; 95% CI 1.45–3.30) and persisted in women with bipolar disorder or schizophrenia (1.88) [36]. Earlier meta-analyses did not distinguish spontaneous from iatrogenic preterm birth, which may explain the absence of an association [3]. The mean lithium concentration in pregnancy was negatively associated with the duration of gestation in a cohort of 101 pregnancies [42], and exposure was associated with a gestation 1.41 weeks shorter in a Dutch cohort [44]. The CANMAT guideline and the review by Bergink and colleagues consider that well-controlled data do not support a risk of preterm birth [22, 25], a position that predates the Australian study.

3.5. Fetal growth

Large for gestational age neonates were more frequent in the Swedish (9.0% vs 3.5%; adjusted relative risk 2.64) and Australian (13.7% vs 6.4%; adjusted relative risk 1.94) cohorts, without an increase in small for gestational age neonates or in pre-eclampsia [3, 36]. As early as 1992, birth weight was higher in exposed infants (3475 g vs 3383 g), despite more mothers who smoked [33]. In a cohort of 119 exposed pregnancies with advanced ultrasound at 18–22 weeks, the abdominal circumference was 5.54 mm larger than that of the reference population and birth weight 142 g higher [44].

3.6. The neonate: toxicity and functional effects

Exposure late in pregnancy may cause neonatal toxicity similar to that in adults, described as floppy baby syndrome, with hypotonia, respiratory distress syndrome, cyanosis, lethargy, feeding difficulties, depressed reflexes, neonatal depression, apnoea and bradycardia; the signs may last up to 14 days, until lithium is eliminated [4]. In a series of 10 neonates and 32 cases from the literature, neonatal lithium concentrations above 0.64 mEq/L were associated with lower Apgar scores, longer hospital stays and more central nervous system and neuromuscular complications [38]. In the Dutch cohort of 29 neonates with a mean lithium concentration of 0.61 mmol/L, 48.3% had some complication, mostly mild and transient: hyperbilirubinaemia (6), hypotonia (3), tremor, irritability, respiratory distress, bradycardia, transient thyroid change and one case of disseminated intravascular coagulation; 44.8% were admitted to intermediate or intensive care (median of 3 days), under a protocol that provides for observation of all exposed neonates for 5 days, with no association between neonatal lithium concentration and outcomes [39]. In the meta-analysis of six cohorts, admission to a special care unit in the first 28 days was 27.5% vs 14.3% (adjusted odds ratio 1.62; 95% CI 1.12–2.33), but the association disappeared when the reference group was restricted to women with bipolar disorder [21].

Neonatal hypoglycaemia was associated with lithium in the analyses restricted to women with psychiatric illness (adjusted relative risk 1.59) and in the comparison with women who had stopped lithium (2.72) [3]. Lithium affects neonatal thyroid function [27], and CANMAT accepts a slight increase in hypoglycaemia, thyroid changes, nephrogenic diabetes insipidus and hypotonia (level 4 evidence) [22]. Severe polyhydramnios attributed to nephrogenic diabetes insipidus is described in a case report [43].

3.7. Neurodevelopment

The available data are reassuring, but come from small samples. Schou's follow-up (60 exposed children vs 57 siblings, with developmental milestones by questionnaire) showed no differences, and three smaller studies (22 exposed assessed by telephone interview; 15 exposed with standardised tests compared with norms, with poorer performance on a block design subtest; 20 exposed at 4–5 years) likewise found no differences in intelligence quotient [27, 58]. In the largest prospective study, with 56 exposed and 43 unexposed children of mothers with bipolar disorder assessed between 6 and 14 years of age, there was no association with intelligence quotient or with the lithium concentration in pregnancy, and the results were comparable to those of the general population [58]. Magnetic resonance imaging of 30 exposed and 33 unexposed children, at 8–14 years, showed no structural differences after correction for multiple comparisons, with a non-significant trend towards lower subcortical grey matter volume [59]. Preclinical studies have suggested effects on motor activity, reflexes, spatial memory and brain weight [58].

3.8. Natural history and the effect of treatment

The risks described apply to treated pregnancies. The comparison relevant to the clinical decision is not between exposure and absence of disease, but between exposure and untreated bipolar disorder: bipolar disorder in itself is associated with preterm birth, gestational hypertension, microcephaly, low birth weight and malformations, through mechanisms that include smoking, cannabis use, obesity and poorer adherence to antenatal care [24]. In the European EUROCAT registry study, Ebstein anomaly was associated with maternal mental illness in general (adjusted odds ratio 2.64; 95% CI 1.33–5.21 compared with controls with other heart defects), and not with lithium or benzodiazepines in particular, which led the authors to suggest that changing or stopping medication may not be preventive [20]. Exposed women in the cohorts are older and smoke and have obesity more often, factors whose under-recording in registers may bias estimates in either direction [2, 3]. A 2024 review argues that small risks do not establish causality, but the first author declares employment by a lithium manufacturer [60].

4. Clinical Recognition and Warning Signs

4.1. Before conception

The moment of greatest impact is the consultation of a woman of reproductive age treated with lithium. The 2018 CANMAT/ISBD (International Society for Bipolar Disorders) guideline recommends preconception counselling at least 3 months before pregnancy, or immediately when pregnancy is already under way, and contraceptive counselling as part of the treatment plan [61]. The Portuguese SmPC recommends that women on lithium use adequate contraception [37], and the US label includes pregnancy status in the screening carried out before starting treatment [4]. Obtaining a preconception lithium concentration during euthymia provides a reference for management during pregnancy [25].

4.2. In pregnancy

Exposure should be identified at the first antenatal visit. A low lithium concentration in the first or second trimester (the nadir occurs at around 18 weeks) may present as irritability or relapse, as in the Canadian case in which the concentration fell from 0.9 to 0.4 mmol/L at 14 weeks [24, 25]. A lithium concentration that rises without a change in dose may herald pre-eclampsia, through reduced renal function, as in the case published in 2026 of a 37-year-old woman with bipolar I disorder [62]. Gastrointestinal symptoms, tremor, polydipsia and polyuria suggest toxicity and warrant immediate measurement of the lithium concentration [24]. Polyhydramnios should prompt consideration of nephrogenic diabetes insipidus [43]. On ultrasound, the high-risk signs of Ebstein anomaly are severe cardiomegaly, retrograde or bidirectional flow in the ductus arteriosus, pulmonary valve atresia, circular shunt, left ventricular dysfunction and hydrops [26].

4.3. In the neonate

The warning signs in the first two weeks are hypotonia, lethargy, feeding or respiratory difficulty, cyanosis, apnoea and bradycardia [4]. Emergency signs are severe cyanosis with cardiomegaly (Ebstein anomaly with circular shunt or functional pulmonary atresia), cardiogenic shock and apnoea [4, 26]. Paroxysmal tachycardia suggests pre-excitation associated with Ebstein anomaly [14]. Jaundice, altered thyroid function and polyuria with dehydration point to hepatic, thyroid or renal effects [22, 24, 39].

4.4. In children and adults

Ebstein anomaly may be diagnosed late: 2% of live births with Ebstein anomaly in the European registers were diagnosed only after the first year of life [20], and the case described in section 10 was diagnosed at the age of 19 years. In a child or adult with Ebstein anomaly, the Oxford Desk Reference recommends asking about fetal exposure to lithium [54], and in Ebstein anomaly with non-compaction cardiomyopathy, dysmorphism, intellectual disability or a family history a genetic cause should be sought [6, 9].

4.5. Newborn screening

The sources consulted do not describe a specific newborn screening test for lithium exposure; surveillance of the exposed neonate is clinical, with targeted investigations (sections 7 and 11) [22, 39].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. Absence of formal criteria

There are no formal diagnostic criteria for lithium embryopathy, nor biological markers that prove a causal relationship between exposure and a specific malformation. The diagnosis is one of association: documented exposure during cardiac organogenesis and a compatible malformation, once alternative diagnoses have been considered. Given that the absolute risk of cardiac malformation in exposed infants is 2% to 3%, most exposed children have no malformations and, among those who do, a proportion would be explained by the background risk of about 1% [2, 28].

5.2. Documenting the exposure

The history should record the salt and formulation (carbonate or citrate, immediate or modified release), the daily dose and the start and stop dates in relation to the date of the last menstrual period, the available lithium concentrations and concomitant medication (other psychotropic drugs, antiepileptics, diuretics, non-steroidal anti-inflammatory drugs) [2, 4]. The exposure relevant to malformations corresponds to the first 90 days after the last menstrual period, the definition used in the Medicaid cohort [2]. To convert doses, 400 mg of lithium carbonate contains 10.8 mmol of lithium, and citrate doses are multiplied by 0.395 to obtain the carbonate equivalent (lithium citrate 564 mg = 6 mmol) [37, 39]. Alcohol, tobacco, other substances, maternal diabetes and obesity should also be recorded, as independent cardiac risk factors [2, 54].

5.3. Prenatal diagnosis

The US label recommends considering fetal echocardiography between 16 and 20 weeks after first-trimester exposure [4]; the Israeli study and Jacobson's study recommended fetal echocardiography and level II ultrasound [33, 35]; the 2025 clinical review and the 2026 Canadian guide recommend level II ultrasound and fetal echocardiography, the latter at 16–18 weeks [24, 25]; and the 2018 CANMAT/ISBD guideline recommends fetal ultrasound when lithium is used in the first trimester [61]. UKTIS takes a different position: the routine detailed ultrasound at around 20 weeks is considered sufficient, without additional monitoring of fetal development after early exposure [28]. The Oxford Desk Reference mentions fetal echocardiography at 14, 19 and 22 weeks for the prenatal diagnosis of complex heart defects, with the limitation that it does not reliably detect transposition of the great arteries [54]. In the European registers, 54% of cases of Ebstein anomaly or associated anomalies were diagnosed prenatally [20]. The American fetal cardiology societies differ on routine fetal echocardiography. The ASE (American Society of Echocardiography, 2023) guideline assigns lithium exposure an absolute risk of congenital heart disease of 1% to 2% and considers fetal echocardiography “not indicated” when the cardiac screening of the second-trimester anomaly scan is normal; the same table records that AIUM (American Institute of Ultrasound in Medicine, 2020) considers it “possibly indicated” and that the AHA (American Heart Association, 2014) scientific statement assigned it class IIb (“may be considered”) [23]. The ASE rationale is the referral threshold: fetal echocardiography is justified when the probability of heart disease exceeds about 1%, that is, about three times the population risk (birth prevalence of 0.3% to 1.2% in the USA), and, with a prior probability of up to 2% (three-plane screening) or 3% (five-plane axial screening), a normal screen lowers the residual probability to the background level; the ASE notes that the decision to refer despite a normal screen should take into account the sensitivity of local screening [23]. The optimal window for transabdominal fetal echocardiography is 18 to 22 weeks; the examination may be performed from 12–14 weeks in high-risk fetuses, but a normal early examination should be repeated in the second trimester [23]. The BAP (British Association for Psychopharmacology, 2017) guideline recommends informing the obstetrician of any first-trimester exposure and requesting particular attention to screening for fetal cardiovascular anomalies [63], and the MotherToBaby fact sheet (2025) indicates prenatal ultrasound as a means of screening for Ebstein anomaly [64]. The ACOG (American College of Obstetricians and Gynecologists) guideline (Clinical Practice Guideline no. 5, June 2023, endorsed by the Society for Maternal-Fetal Medicine) recommends that a pregnant woman exposed to lithium in the first trimester have a detailed second-trimester ultrasound, with particular attention to cardiac anatomy (strong recommendation, moderate-quality evidence), and accepts fetal echocardiography in centres where it is available, with reference to the 2020 AIUM practice parameter; it bases the recommendation on the odds ratios of 1.81 (95% CI 1.35–2.41) for congenital anomalies overall and 1.86 (95% CI 1.16–2.96) for cardiac anomalies from Fornaro's meta-analysis, against the 400-fold increase in Ebstein anomaly described in the early literature, and stresses that the absolute risk and the absolute risk increase are low [41, 65].

When a fetal heart defect is present, the ASE recommends serial assessments at intervals of 2 to 8 weeks, depending on the lesion, until 34–36 weeks; in Ebstein anomaly, which it includes among the lesions at risk of fetal heart failure, it recommends complete functional assessment every 2 to 4 weeks, Doppler interrogation of the tricuspid regurgitation jet as a measure of right ventricular function and consideration of fetal magnetic resonance imaging to assess the lungs [23]. When an exposed fetus has Ebstein anomaly, assessment should include the high-risk markers of the AATS (American Association for Thoracic Surgery, 2024) consensus: severe cardiomegaly, retrograde or bidirectional flow in the ductus arteriosus, pulmonary valve atresia, circular shunt, left ventricular dysfunction and hydrops [26]. The presence of extracardiac malformations, growth restriction or other ultrasound markers warrants offering genetic diagnosis by amniocentesis or chorionic villus sampling with chromosomal microarray, because about 20% of cases of Ebstein anomaly are associated with chromosomal or Mendelian disorders and the 1p36 and 8p23.1 deletions are the most frequent [9, 14]. Prenatal diagnosis of a malformation is also the basis for counselling on continuation of the pregnancy, which the NICE (National Institute for Health and Care Excellence) guideline requires to be offered after exposure to psychotropic drugs with a known teratogenic risk [66].

5.4. Assessment of the exposed neonate

Assessment includes a complete physical examination with attention to hypotonia, respiratory function, feeding and cardiac rhythm, and echocardiography when adequate fetal echocardiography was not performed or when there is a murmur, cyanosis or arrhythmia [4, 39]. The neonatal lithium concentration (from the cord or by venepuncture in the first 24 hours, methods with equivalent results) correlates with the maternal concentration and is useful when there are signs of toxicity [39]. Samples should be collected in tubes without lithium heparin, because samples collected in such tubes have given falsely therapeutic or toxic values in asymptomatic neonates [5]. Thyroid function (thyroid-stimulating hormone, TSH, and thyroxine), renal function, blood glucose and bilirubin are guided by the clinical findings [3, 22, 39].

5.5. Genetic testing in Ebstein anomaly

Lithium exposure does not remove the need for genetic investigation of a child with Ebstein anomaly when there is an independent indication. Chromosomal microarray detects the 1p36 and 8p23.1 deletions and the copy number variants associated with isolated Ebstein anomaly; in microarray studies of isolated Ebstein anomaly, candidate copy number variants were found in about half of cases, but most lack functional validation and recurrence and should not be interpreted as causal without classification criteria [9, 51, 52]. When Ebstein anomaly and non-compaction cardiomyopathy coexist, the yield of MYH7 is high (a variant in 6 to 7 of 8 probands with non-compaction vs 0 of 133 without it) and testing of MYH7 and TPM1 is recommended [6, 12]. The three genes considered “known” in Ebstein anomaly in studies of isolated cases are NKX2-5, GATA4 and MYH7, joined by TPM1 in the forms with non-compaction [12, 51]. Classification should follow the criteria of the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP), and the report should state the gene, the reference transcript with version (MANE, Matched Annotation from NCBI and EMBL-EBI: NM_000257.4 for MYH7, NM_001018005.2 for TPM1), the HGVS (Human Genome Variation Society) nomenclature, the zygosity and the class [47, 50]. The same variant may have different classifications depending on the phenotype and the date: the MYH7 deletion p.(Glu1220del), which cosegregated with Ebstein anomaly in one family, has an aggregate classification of uncertain significance [48, 49]. Variants of uncertain significance require segregation study and echocardiography of first-degree relatives, because non-compaction may be asymptomatic in carriers [6, 16]. A negative genetic result does not prove causality of lithium, because most cases of isolated Ebstein anomaly have no genetic cause identifiable with current tests [9, 51].

5.6. Practical algorithm

1. Woman of reproductive age on lithium: contraception, planning and a reference lithium concentration during euthymia [25, 61]. 2. Confirmed pregnancy with first-trimester exposure: do not stop abruptly; joint assessment by psychiatry and obstetrics; baseline lithium concentration [22, 66]. 3. Anomaly scan and fetal echocardiography between 16 and 20 weeks, according to the US label, or echocardiography at 18–22 weeks when the second-trimester cardiac screen is not normal or is of limited quality, according to the ASE [4, 23]. 4. If a malformation is present: high-risk assessment of Ebstein anomaly, offer of genetic diagnosis when there are extracardiac anomalies, counselling and planning of delivery in a centre with paediatric cardiology [9, 26]. 5. Monitoring of lithium concentrations until delivery and a plan for labour (section 7). 6. Neonate: clinical observation, echocardiography if indicated, lithium concentration and laboratory tests if signs are present [4, 39]. 7. Ebstein anomaly in the neonate or later: microarray and gene panel if there is non-compaction, dysmorphism, extracardiac malformations or a family history [6, 9].

6. Differential Diagnosis

The essential differential lies between Ebstein anomaly attributed to lithium and Ebstein anomaly of genetic cause. The presence of left ventricular non-compaction cardiomyopathy is the most useful sign: in studies of Ebstein anomaly, non-compaction accompanies almost only the cases with a MYH7 variant, and Ebstein anomaly is the most prevalent congenital heart defect in patients with non-compaction (8% of those with associated heart disease in pooled studies) [6, 45]. A family history of cardiomyopathy, sudden death, pacemaker implantation or congenital heart disease, and the presence of atrioventricular block, should prompt testing of MYH7, TPM1 and NKX2-5 [11, 13]. Recurrence in siblings of an isolated non-syndromic Ebstein anomaly is about 1%, and the Oxford Desk Reference recommends, in that setting, asking about lithium exposure [54].

The second distinction is from syndromic Ebstein anomaly, which accounts for about 20% to 27% of cases. The 1p36 deletion, the most frequent terminal deletion (about 1/5000 births), combines heart disease (including Ebstein anomaly and non-compaction) with hypotonia, cognitive impairment, seizures, deafness and recognisable facial dysmorphism [18, 67]. The 1p36 deletions described in patients with Ebstein anomaly are distal and share loss of PRDM16, and a new candidate locus at 1p36.2 has been proposed [68]; the role of PRDM16 in the cardiomyopathy of 1p36 deletion has been challenged [18]. The 8p23.1 deletion removes GATA4, a candidate for explaining Ebstein anomaly in those patients [9]. A child exposed to lithium with extracardiac malformations or developmental delay should be investigated like any other child with syndromic heart disease.

The third distinction is from the other embryopathies and maternal diseases, above all because lithium is often an alternative or an addition to other mood stabilisers. Valproate carries a malformation risk of about 18% and a characteristic malformation and neurodevelopmental pattern, and carbamazepine is not recommended in the first trimester [24]. Maternal diabetes and alcohol are frequent causes of congenital heart disease and should be excluded by the history [54]. Finally, maternal mental illness itself is associated with Ebstein anomaly in the European register, which may explain part of the association observed with lithium [20].

7. Treatments and Risk Management Strategies

7.1. Principles

There is no treatment of lithium embryopathy as such. Management has three components: deciding, before and during pregnancy, between continuing, reducing, switching or withdrawing lithium, weighing fetal risk against the risk of maternal relapse; monitoring the lithium concentration throughout pregnancy, delivery and the postpartum period to avoid subtherapeutic and toxic levels; and treating complications in the child, above all heart disease and neonatal toxicity. Guidelines differ on important points (the position of lithium among the options, management at delivery, breastfeeding), and these divergences are presented throughout this section.

7.2. The risk of not treating

The decision rests on the magnitude of the relapse risk. In a meta-analysis cited by the 2025 review, postpartum relapse was 66% (95% CI 57%–75%) in women without medication and 23% (95% CI 14%–37%) in those on medication, mostly lithium; in a retrospective cohort, relapse was 23.8% with lithium and 40.1% without medication (odds ratio 0.47; 95% CI 0.27–0.81) [25]. The meta-analysis by Sharma and colleagues, published in October 2026, estimated a postpartum recurrence proportion of 0.21 (95% CI 0.15–0.29) with prophylactic lithium and 0.41 (95% CI 0.33–0.50) without lithium, across eight studies with 554 women, with the caveat of confounding by indication [69]. Fornaro's meta-analysis estimated an odds ratio of 0.16 for postpartum relapse with lithium (number needed to treat 3), based on only 48 women [41]. During pregnancy, in a tertiary centre, 85% of the women who stopped the mood stabiliser and 37% of those who continued it had an episode, with a median time to recurrence of 2 weeks after abrupt discontinuation and 22 weeks after gradual tapering [61]. The 2026 Canadian guide estimates a relapse risk of 50% in the 2 weeks following abrupt discontinuation [24]. The first 6 months postpartum carry the highest lifetime risk, about 35% overall and greatest in the first month, with a risk of psychosis, mania, depression, mixed episodes and, in extreme cases, suicide and infanticide; 28% of mothers with bipolar disorder had their first episode in the peripartum period [25].

7.3. Planning before conception

The 2018 CANMAT/ISBD guideline recommends preconception counselling at least 3 months before pregnancy, a shared decision between continuing, withdrawing or switching medication, a preference for psychosocial measures in the first trimester where possible and, if medication is required, monotherapy at the lowest effective dose [61]. The review by Bergink and colleagues recommends simplifying the regimen and avoiding polypharmacy, considering a gradual dose reduction only when the patient has been stable on a lower dose or has never tried one, and not reducing when a previous reduction caused relapse; it also recommends a written perinatal relapse prevention plan with six points (medication in pregnancy, medication after delivery, mode of delivery, infant feeding, sleep strategies and early signs of relapse) [25]. The 2024 meta-analysis by Callovini and colleagues suggests that euthymic women planning pregnancy may consider withdrawing lithium, but that, once pregnancy has begun, there is no reason to withdraw it [1]. The Canadian guide accepts reducing and withdrawing it in the first trimester when the patient has been stable for at least a year, with mild previous episodes and a rapid response to treatment [24].

The NICE CG192 guideline (2014, updated in 2020) is the most restrictive: do not offer lithium to women planning pregnancy or who are pregnant, unless antipsychotics have been ineffective; when it is offered, explain the risk of cardiac malformations “of uncertain magnitude”, the risk of high levels in breast milk and the need for more frequent monitoring [66]. The 2024 CANMAT guideline places lithium as second line in acute mania and in bipolar I depression during pregnancy (“highly effective, including in relapse prevention, but with a high risk of multiple adverse outcomes and complex management”), with quetiapine and aripiprazole (mania) and quetiapine and lamotrigine (depression) as first line [22]. The 2017 BAP guideline considers antipsychotics the mainstay of perinatal treatment of bipolar disorder, because they have a better risk–benefit profile than valproate and carbamazepine and a larger reproductive evidence base than lamotrigine and lithium; for lithium, it concludes that the data do not allow an effect on the overall malformation rate or on cardiovascular anomalies to be excluded, but that this risk, if it exists, is probably small, and that there are insufficient data on miscarriage, birth weight and duration of gestation [63]. Bergink's review considers lithium the gold standard in the peripartum period and lithium's teratogenicity far lower than that of carbamazepine and valproate [25]. Valproate is contraindicated in pregnancy and should be avoided in anyone who may become pregnant [24].

ACOG guideline no. 5 (2023) takes the opposite position to NICE. It recommends not withdrawing mood stabilisers during pregnancy, with the exception of valproate, because of the risk of recurrence or worsening of symptoms (strong recommendation, moderate-quality evidence), and not withdrawing or withholding psychiatric medication solely because of pregnancy or breastfeeding (strong recommendation, low-quality evidence); it recommends not using valproate as first line and avoiding it in pregnancy, because alternatives are generally available (strong recommendation, moderate-quality evidence) [65]. In a woman who is stable at the time of conception, the guideline considers it preferable to continue the same regimen rather than switch drug, because switching requires cross-tapering, exposes the fetus to more medicines and may increase recurrence; in treatment-naive patients, lithium is often used as first line in bipolar I disorder with mania, because of its efficacy, with the proviso that it should be initiated in consultation with psychiatry [65]. To support continuation, it cites the AHRQ (Agency for Healthcare Research and Quality) systematic review by Viswanathan and colleagues, in which, across two cohorts, withdrawal of the mood stabiliser in pregnancy was associated with more recurrences (adjusted hazard ratio 2.2; 95% CI 1.2–4.2) and a shorter time to recurrence (2 weeks versus 28 weeks; adjusted hazard ratio 12.1; 95% CI 1.6–91) [65]. It considers that current estimates of malformation risk make lithium a reasonable option, particularly in women with a history of mania or with a previous response to lithium, and that treatment of bipolar disorder in pregnancy should be led by psychiatry in collaboration with obstetrics, preferably starting from a preconception consultation with shared decision-making [65]. The guideline also recalls that untreated bipolar disorder increases the risk of postpartum psychosis by 25% to 50% (about 100 times that of the general population), multiplies the odds of postpartum hospitalisation sevenfold and is associated with fetal growth restriction, preterm birth and adverse neurodevelopmental outcomes [65].

7.4. Ongoing pregnancy with first-trimester exposure

NICE recommends that, in a pregnant woman taking lithium who is well, gradual withdrawal over 4 weeks be considered, explaining that withdrawal may not remove the malformation risk (cardiac organogenesis may already have occurred) and that there is a risk of relapse, particularly postpartum; if the woman is not well or is at high risk of relapse, consider gradual switching to an antipsychotic, withdrawal with restarting in the second trimester (if she does not plan to breastfeed and responded better to lithium) or continuation of lithium when an antipsychotic is unlikely to be effective [66]. NICE also recommends, after first-trimester exposure to a psychotropic drug with teratogenic risk, confirming the pregnancy as early as possible, offering screening for fetal anomalies and counselling on continuation of the pregnancy [66]. Bergink's review recommends, in an unplanned pregnancy, reducing the dose at the time of discovery rather than withdrawing it, given the risk of relapse after withdrawal [25]. The Portuguese SmPC states that lithium should not be administered in pregnancy, particularly in the first trimester, “unless it is considered essential” [37]. BAP recommends that, if the plan is to withdraw lithium, the reduction be gradual over 4 weeks, with the possibility of restarting it in the second or third trimester [63]. In any scenario, abrupt withdrawal should be avoided, and the UKTIS leaflet advises women not to stop or reduce lithium without medical supervision [28, 61].

7.5. Dosing and monitoring of lithium during pregnancy

Lithium is here the drug for the maternal disease and not a treatment directed at the embryopathy; doses are those for bipolar disorder, adjusted by the lithium concentration. Table 2 compares the monitoring recommendations and targets of the guidelines with the labels.

The usual maintenance dose in adults is 900 to 1200 mg/day of lithium carbonate in the Lithobid® label (USA), for lithium concentrations of 0.6 to 1.2 mEq/L [71], the same dose and concentration range indicated in the Brazilian Carbolitium® label for maintenance [70]. The MotherToBaby fact sheet notes that the dose may have to be increased during pregnancy to prevent relapse, and that withdrawal, particularly if rapid, may cause symptoms to return; after withdrawal, elimination of most of the lithium takes on average up to 9 days in a healthy non-pregnant adult, and longer in patients treated for longer [64]. The Priadel® SmPC does not set a dose for adults in general: it requires individualisation by lithium concentration and establishment of the minimum effective dose, and treats a change between formulations as the start of a new course of treatment, because of differences in bioavailability; in elderly patients or those weighing less than 50 kg it indicates 600 to 1200 mg/day for prophylactic concentrations of 0.4 to 0.8 mmol/L [37]. The 2025 review recommends considering a dose reduction in the first trimester, given the dose-dependent malformation risk (above 900 mg/day in the Medicaid cohort) [2, 25], and Fornaro's meta-analysis recommends prescribing within the lowest therapeutic range, particularly in the first trimester and in the days preceding delivery [41]. Over the second trimester, the increase in clearance often requires the dose to be raised to maintain the pre-pregnancy level: in the Canadian case, the dose went from 600 to 750 mg every 12 hours when the lithium concentration fell to 0.4 mmol/L at 14 weeks [24]. A lithium concentration that rises without a change in dose should raise the suspicion of pre-eclampsia, and CANMAT recommends withdrawing lithium in pre-eclampsia, considering another mood stabiliser if withdrawal is prolonged [22, 62]. Diuretics, non-steroidal anti-inflammatory drugs, renin–angiotensin system antagonists and metronidazole should be avoided, and the lithium concentration should be repeated after any change of dose, of concomitant medication, of intense physical activity or of intercurrent illness [4]. Folic acid should be given at the standard antenatal dose of 0.4 mg/day, according to the Canadian guide, whereas the 2018 CANMAT/ISBD guideline recommended 5 mg/day in women with bipolar disorder on mood stabilisers [24, 61]. ACOG recommends 4 mg/day before conception and in early pregnancy only for anticonvulsants, and notes that some experts also propose 4 mg/day with lithium to reduce congenital heart defects, a proposal based on a review of folic acid in the prevention of heart defects and not on a study in pregnant women exposed to lithium [65]. The ACOG guideline further draws attention to the obstetric situations that alter glomerular filtration or plasma volume (pre-eclampsia, hyperemesis, acute haemorrhage) and to non-steroidal anti-inflammatory drugs postpartum, which require closer monitoring of the lithium concentration, and describes toxicity from 1.5 mEq/L with nausea, vomiting, lethargy, tremor and fatigue, progressing at higher levels to confusion, agitation, seizures, hypothermia and loss of consciousness [65].

7.6. Labour and delivery

Management at delivery is the point of greatest divergence between sources (Table 3).

The rationale for withdrawal comes mainly from the study by Newport and colleagues (2005): withdrawal 24 to 48 hours before delivery lowered the maternal lithium concentration by 0.28 mEq/L, and neonatal concentrations above 0.64 mEq/L were associated with more complications [38]. The contrary data come from the Dutch cohort, in which concentrations were within the therapeutic window and no toxic levels were observed; the authors attribute the discrepancy to a high-exposure group made up mainly of case reports with neonatal concentrations above 0.7 and at times above 1.2 mmol/L, subject to publication bias [39]. CANMAT attempts to reconcile the two positions with the threshold of 0.7 mmol/L [22]. In all sources, delivery should take place in a hospital with monitoring of the lithium concentration and of hydration, and the obstetric team should be aware of the psychiatric plan [25, 66].

7.7. Postpartum period

The 2025 review recommends a target lithium concentration of 0.8 to 1.0 mEq/L in the first month after delivery, with measurement at least weekly in that month and monthly for the following 3 months if stable; if lithium was withdrawn before or during pregnancy, it recommends restarting it immediately after delivery with the same target, and, in patients never treated with lithium, considering starting it on the day of delivery [25]. The Dutch cohort recommends a lithium concentration on the second postpartum day and thereafter weekly or fortnightly, with a target of 0.8 to 1.0 mmol/L in the first month [39]. CANMAT recommends resuming the pre-pregnancy dose immediately after delivery (or 150 to 300 mg less than the pregnancy dose, if lithium was started during pregnancy), because of volume contraction, and measuring the concentration 5 to 7 days later [22]. NICE recommends offering an antipsychotic as prophylaxis when a woman with bipolar disorder has withdrawn lithium or intends to breastfeed [66]. ACOG recommends that the postpartum plan, including the anticipated dose adjustments, be defined before delivery, and that, with the rapid return of glomerular filtration to baseline, the dose be reduced carefully but without delay to the pre-pregnancy regimen, to avoid toxicity; it does not generally recommend reducing the dose of psychotropic drugs in the third trimester, because such a reduction does not improve neonatal outcomes and increases the risk of maternal deterioration [65]. In a woman who withdrew or did not start treatment in pregnancy, ACOG recommends starting or restarting it postpartum, particularly after previous postpartum psychosis, whose recurrence without pharmacotherapy is 66%; in the prevention of postpartum psychosis, starting lithium at a high dose immediately after delivery is the intervention with the strongest evidence, and forgoing night-time breastfeeding to protect sleep may help in the initial phase [65]. Postpartum psychosis is a psychiatric emergency: ACOG indicates, while psychiatric assessment is awaited, a sedating antipsychotic (olanzapine or haloperidol) combined with a short-acting benzodiazepine (lorazepam), with continuous observation and, generally, admission, and recalls that with prompt treatment complete remission may be achieved by 2 months postpartum [65]. Sleep protection (guaranteed consecutive hours of sleep, with the partner taking the night feeds) is a central non-pharmacological measure, because the loss of at least one full night's sleep near delivery increases the risk of postpartum psychosis [24, 25]. Screening with the Edinburgh Postnatal Depression Scale and enquiry about manic symptoms should form part of antenatal and postnatal visits [24, 61].

7.8. The exposed neonate

Treatment of neonatal toxicity is supportive: monitor and support the neonate until lithium is excreted and the signs resolve, which may take up to 14 days [4]. The protocol of the Rotterdam centre provides for clinical observation of all exposed neonates for 5 days [39], and the Portuguese SmPC recommends careful clinical observation and, if necessary, measurement of the lithium concentration [37]. Preterm, dehydrated or infected neonates should be hydrated and assessed for toxicity, because immature renal function and dehydration reduce elimination [5, 24]. Hypoglycaemia, hyperbilirubinaemia and thyroid changes are treated according to the usual neonatal protocols [3, 39]. For severe intoxication in adults, haemodialysis is the treatment of choice and activated charcoal does not adsorb lithium, with a possible rebound rise in the concentration [4]; the sources consulted do not describe the use of haemodialysis in the neonate.

7.9. Breastfeeding

The sources differ. The Portuguese SmPC classifies breastfeeding as a contraindication to lithium and requires a choice between withdrawing the drug or stopping breastfeeding [37]; NICE recommends encouraging breastfeeding except in women treated with carbamazepine, clozapine or lithium, supporting each woman's choice [66]; the US label does not recommend breastfeeding, but accepts it with close monitoring of the infant and recommends considering regular measurement of the infant's lithium concentration and thyroid function [4]. The 2024 CANMAT guideline considers the decision an individual one, weighing the risk of relapse, the infant's health and the woman's preference, advises against exposure through milk in preterm infants, in infants with renal malformations or with conditions causing dehydration, and recommends, in the face of changes in behaviour or tone, measurement of the infant's lithium concentration, electrocardiography and thyroid and renal function tests; postpartum, it places lithium as second line because of concerns about lactation [22]. The LactMed database (revised in June 2026) considers that most sources see no absolute contraindication in healthy term neonates, particularly after 2 months of age and on monotherapy, in mothers able to monitor the infant [5]. A 2026 review considers lithium a reasonable option throughout the peripartum period, including lactation [72]. The 2017 BAP guideline is the most restrictive: lithium, valproate and carbamazepine should not be prescribed to breastfeeding women with affective disorders, on the basis of relative infant doses of up to 42% and infant lithium concentrations of up to 50% of the maternal value, although it acknowledges reports of successful breastfeeding with close monitoring of the infant's concentrations [63]. The MotherToBaby fact sheet (2025) accepts that the benefit of lithium and of breastfeeding may outweigh the risks, given that continuing lithium after delivery reduces relapse, and recommends monitoring the infant for agitation, hypotonia and feeding difficulties, with measurement of the lithium concentration and thyroid and renal function particularly in very young or preterm infants [64]. The Brazilian Carbolitium® label does not recommend use during breastfeeding [70]. ACOG considers the recommendations on lithium in lactation controversial and, if breastfeeding is contemplated, requires coordination between obstetrician, paediatrician and psychiatrist, because of the neonate's renal immaturity and risk of relative dehydration; it cites lithium concentrations in breastfed infants of between 10% and 60% of maternal values, above the upper limit of 50% given by other sources [5, 65]. In general terms, the same guideline does not discourage breastfeeding in women with mental illness, advises against switching postpartum a drug on which the woman was stable in pregnancy, because fetal exposure exceeds exposure through milk, and takes a relative infant dose below 10% as compatible with breastfeeding, to be weighed against prematurity and the infant's age [65].

The pharmacokinetic data explain these positions. Lithium in milk corresponds to about 40% to 45% of the maternal concentration (0.12 to 0.7 mEq/L), the relative infant dose was estimated at 12.2% (0% to 30%) in a recent series and at about 26% in data predating 1990, and the lithium concentration in breastfed infants varies between 10% and 50% of the maternal value [4, 5]. The Canadian guide indicates a relative infant dose of up to 30%, above the threshold of 10% usually considered acceptable [24]. In a study of 24 women in whom lithium was withdrawn during labour, the infant's lithium concentration took 6 to 8 days to fall below 0.2 mEq/L with formula feeding, 7 to 8 days with mixed feeding and 53 to 60 days with exclusive breastfeeding, without accumulation [5]. The cases of toxicity described occurred in situations of compromised elimination: a 5-day-old neonate with cyanosis, lethargy and T-wave inversion, whose mother had taken chlortalidone before delivery, and an infant who became intoxicated after a common cold with probable dehydration; two infants had a slight rise in TSH [5]. The proposed infant monitoring schedules vary: lithium concentration, creatinine, urea and TSH every 4 to 12 weeks; maternal and infant concentrations on days 2 and 10 (and also 30 and 60 with exclusive breastfeeding); a single assessment at 10 days, repeated only if the infant's concentration is ≥0.3 mEq/L or signs are present; or paediatric surveillance with selective laboratory tests [5]. Reducing the proportion of breast milk lowers the infant's lithium concentration [5]. Exclusive breastfeeding also carries the cost of interrupting maternal sleep, a risk factor for postpartum relapse [25].

7.10. Treatment of Ebstein anomaly

Lithium-associated Ebstein anomaly is treated like any other. The 2024 AATS consensus for neonates and infants defines as high risk severe cardiomegaly, prematurity below 32 weeks, intrauterine growth restriction, pulmonary valve atresia, circular shunt, left ventricular dysfunction and cardiogenic shock [26]. Unstable neonates with a circular shunt should undergo emergency interruption of the shunt; those in refractory cardiogenic shock may be palliated with the Starnes procedure, followed by biventricular repair or univentricular palliation; stable neonates without risk factors may be observed until spontaneous closure of the ductus arteriosus, or undergo an attempt at pharmacological ductal closure when there is significant pulmonary regurgitation or functional pulmonary atresia with normal right ventricular systolic pressure; stable neonates with inadequate antegrade pulmonary flow may receive a ductal stent or a systemic-to-pulmonary shunt [26]. Palliative care may be reasonable when prematurity, genetic syndromes, major comorbidities, ventricular dysfunction or sepsis coexist [26]. In the older child and in adults, surgical repair of the right heart, including the tricuspid valve, is the definitive treatment, and fewer than 5% of patients without surgery survive beyond 50 years, according to the review cited in the case in section 10 [55]. Arrhythmias (pre-excitation, supraventricular tachycardia, atrial fibrillation) are frequent throughout life and require electrophysiological assessment [14, 56].

When the cause is genetic, treatment acquires additional components. In the forms linked to MYH7 or TPM1, the cardiomyopathy follows the cardiomyopathy guidelines, with assessment of relatives [6, 10]. In NKX2-5-linked disease, atrioventricular block is progressive and postnatal, the risk does not end with repair of the septal defect, electrocardiographic surveillance is lifelong and pacing is the treatment of high-grade block; consideration of an implantable cardioverter-defibrillator has been proposed in carriers from families with sudden death [13].

7.11. Regulatory status and designations

7.12. Studies supporting the recommendations

There are no randomised clinical trials of medicines in pregnancy, and the CANMAT guideline accepts that the perinatal safety evidence is observational (level 3 or 4) [22]. Table 1 (section 3) summarises the studies of malformation risk; Table 4 summarises the studies of obstetric, neonatal and neurodevelopmental outcomes.

These estimates come from observational studies, with common limitations: dispensing of a medicine does not prove that it was taken, confounding by disease severity, smoking, alcohol and obesity is difficult to eliminate, studies of live births do not capture terminations of pregnancy for malformation (which may underestimate the risk), and the increased surveillance of exposed neonates may overestimate mild outcomes [2, 3, 21]. Comparisons between studies with different reference groups (general population, women with mood disorder, women who stopped lithium, women on lamotrigine) are not directly comparable. Negative results form part of the body of evidence: the meta-analysis of six cohorts found no significant increase in cardiac malformations, no case of Ebstein anomaly and no association with obstetric outcomes [21], and an umbrella review of 21 meta-analyses found no associations with convincing or highly suggestive evidence for any psychotropic drug in pregnancy [74].

7.13. Developments in the past 12 months

Between November 2025 and October 2026 the following studies were published, all observational or preclinical; no trials, regulatory submissions or label changes were identified in the sources consulted. The Australian study by Roddy Mitchell and colleagues (BJOG, July 2026; published result) confirmed the increase in spontaneous preterm birth, in large for gestational age neonates and in cardiac malformations [36]. The meta-analysis by Sharma and colleagues (October 2026; published result) reinforced the efficacy of lithium in preventing postpartum recurrence [69]. The United Kingdom primary care study by Chauhan and colleagues (April 2026) showed that prescription of lithium in pregnancy is rare (1.89 per 10,000 pregnancies between 1995 and 2018; 1.03 in 2018) and that, among the 227 pregnancies with lithium before conception, only 15.4% continued treatment throughout pregnancy [75]. A 2026 case report described an unexpected rise in the lithium concentration as the first sign of pre-eclampsia [62], and the CMAJ (Canadian Medical Association Journal) practice guide of July 2026 summarised management for primary care [24]. A Swiss review of September 2026 and an American one of May 2026 argued for an individualised approach [40, 72]. The network meta-analysis by Wang and colleagues (2025) received a correction published in 2026 [53]. Outside therapeutic exposure, a Californian ecological study of October 2026 linked high lithium concentrations in drinking water to a higher prevalence of Wolff-Parkinson-White syndrome in childhood (odds ratio 1.71; 95% CI 1.44–2.03 above the 75th percentile), a hypothesis to be confirmed in individual-level studies [76]. At the preclinical level, lithium administered to lactating rats altered the pups' thyroxine, an effect reversed by iodine supplementation [77], and environmental lithium exposure was associated with trophoblast cuproptosis and miscarriage in mice and in a case–control study of unexplained miscarriage [78]. The October 2025 clinical update of ACOG guideline no. 5 deals with zuranolone and brexanolone in postpartum depression and does not change the 2023 recommendations on lithium.

7.14. Agents to avoid, the perioperative period and transition

In the setting of pregnancy on lithium, sodium restriction, diuretics (chlortalidone before delivery was the origin of one case of neonatal toxicity), non-steroidal anti-inflammatory drugs, renin–angiotensin system antagonists and metronidazole should be avoided, and dehydration (vomiting of pregnancy, prolonged labour, perioperative fasting) should be corrected with monitoring of the lithium concentration [4, 5, 66]. The Portuguese SmPC contraindicates lithium in severe renal impairment, in heart failure, in Addison's disease and in Brugada syndrome or a family history of it [37]. Valproate is contraindicated in pregnancy, and carbamazepine is not recommended in the first trimester [24, 66]. For a planned caesarean section, CANMAT accepts withdrawing lithium 24 hours beforehand when the concentration is ≥0.7 mmol/L [22]. In the adolescent with bipolar disorder on lithium, the US label establishes efficacy and safety from the age of 7 years, and contraception and planning should be discussed during transition to adult care [4, 61]. In the child with Ebstein anomaly, transition to an adult congenital heart disease clinic is necessary, and a woman with Ebstein anomaly considering pregnancy should be assessed in that clinic and offered fetal echocardiography [54].

8. Reproductive Options and Family Counselling

8.1. Recurrence risk

Lithium embryopathy is not hereditary. The risk in a subsequent pregnancy depends on renewed exposure, on the dose and on the timing of exposure, and not on a genetic mechanism; in a pregnancy without lithium in the first trimester, the malformation risk returns to the background risk, with the caveat that maternal mental illness was itself associated with Ebstein anomaly in the European registers [2, 20]. Paternal exposure to lithium around conception has no evidence of harm to the fetus [28]. In rats, lithium caused changes in spermatogenesis, with possible reduction of fertility [37].

8.2. Counselling when the child has Ebstein anomaly

If Ebstein anomaly is isolated and non-syndromic and no genetic cause is identified, the empirical recurrence risk in siblings is about 1% [54]. If a pathogenic variant in MYH7, TPM1 or NKX2-5 is identified, inheritance is autosomal dominant and the risk to the offspring of a carrier is 50%, with variable penetrance and expressivity (carriers with asymptomatic non-compaction, cardiomyopathy, septal defects, bicuspid aortic valve or coarctation); the variant may be de novo [6, 7, 11, 12]. In that case, cascade testing and echocardiography of first-degree relatives are indicated, and in NKX2-5-linked disease electrocardiography as well [6, 13]. In the 1p36 and 8p23.1 deletions, usually de novo, study of the parents defines the recurrence risk. The discovery of a genetic cause in a child exposed to lithium alters counselling and frees the mother from an incorrect causal attribution.

8.3. Prenatal and preimplantation diagnosis

For lithium exposure, prenatal diagnosis is by ultrasound (anomaly scan and fetal echocardiography, section 5.3). For the genetic forms of Ebstein anomaly with a known familial variant, invasive prenatal diagnosis and preimplantation genetic diagnosis are possible, with the limitation that the variable penetrance of MYH7 and TPM1 makes phenotype prediction imprecise [11]. A woman with Ebstein anomaly should be assessed in an adult congenital heart disease clinic before pregnancy, to evaluate the impact of pregnancy on her health; the Oxford Desk Reference reports a miscarriage risk of 20% when the mother has congenital heart disease, compared with 10% when it is the father [54].

8.4. Counselling the woman treated with lithium

Counselling should convey absolute numbers: the probability of a cardiac malformation rises from about 1% to about 2% to 2.4%, and that of a right ventricular outflow tract defect from 0.18% to 0.60%; the number needed to harm for one additional cardiac anomaly was estimated at 71 [2, 28, 41]. The CANMAT guideline expresses the relative risk of about 1.7 for cardiovascular malformations, including Ebstein anomaly, as a rise from 1 to 5 to 2 to 8 cases per 200,000 births; these absolute values are far lower than the absolute risks of cardiac malformation observed in the cohorts (1.15% in unexposed and 2.41% in exposed infants in the Medicaid cohort), so the cohort figures should be used in counselling [2, 22]. Counselling should also address the obstetric and neonatal risks (spontaneous preterm birth, large for gestational age neonate, transient neonatal toxicity) and the risk of relapse without treatment, particularly postpartum [3, 25]. The decision is shared and should be recorded in a written plan, known to the obstetric and paediatric teams [25]. Women who have already become pregnant on lithium should know that stopping after confirmation of the pregnancy may not remove the malformation risk, because cardiac organogenesis occurs early [66]. Elective termination of pregnancy was more frequent in exposed women in the Israeli study (9.3% vs 2.0%) [35], which shows the weight that risk perception carries in decisions and the importance of counselling based on current figures.

9. Special Considerations: Population Diversity and Epidemiology

9.1. Epidemiology of Ebstein anomaly

Ebstein anomaly is rare and the published figures measure different things. Orphanet gives a point prevalence in Europe of 1.25/100,000 and a European birth prevalence of 3.5/100,000 (EUROCAT, 2010) [19]. The EUROCAT national figures vary widely, from 0.6/100,000 in Spain to 27.2/100,000 in Switzerland, passing through 3.5 in the United Kingdom, 4.1 in Italy, 5.7 in the Netherlands, 7.8 in France, 9.8 in Norway, 11.6 in Germany and 18.2 in Denmark, between 2005 and 2010; the United States has 6.0/100,000 and Taiwan 4.7/100,000 [19]. These differences reflect the inclusion or otherwise of fetal deaths and terminations of pregnancy and the registration methods, and should not be read as real population differences. In the European study of 15 registers from 12 countries (5.6 million births, 1982–2011), the total prevalence rose from 0.29 to 0.48 per 10,000, a rise attributed to better and earlier diagnosis; 81% of cases were live births and 54% were diagnosed prenatally [20]. The traditional estimate of 1/20,000 births appears in the Oxford Desk Reference and in a 2024 genetics review (0.5 to 1 per 20,000), and Ebstein anomaly accounts for about 0.5% of congenital heart defects [27, 79]. In the Medicaid cohort, the prevalence observed in unexposed infants was about 7/100,000 [2]. The sources consulted do not include prevalence data for Ebstein anomaly in Portugal or Brazil.

9.2. Epidemiology of exposure

First-trimester lithium exposure occurred in 0.05% of US Medicaid pregnancies (2000–2010), 0.05% of Swedish deliveries (2007–2014) and 0.03% of deliveries in Victoria, Australia (2009–2020) [2, 3, 36]. In United Kingdom primary care, prescription during pregnancy was 1.89 per 10,000 pregnancies between 1995 and 2018, falling to 1.03 in 2018, and discontinuation was frequent, particularly in the second and third trimesters [75]. Women who stopped lithium were younger and more often had obesity, smoked and had a diagnosis of bipolar disorder [75]. In the Medicaid cohort, 73.9% of exposed women were white and 65.8% had a diagnosis of bipolar disorder; in the Swedish cohort, 93.3% had bipolar disorder and 90.1% had been born in Sweden [2, 3]. The sources consulted do not describe exposure data for Portugal or Brazil.

9.3. Special populations

Bipolar disorder is more severe in women with peripartum episodes, and unplanned pregnancies are more frequent in this population, through impulsivity, limited access to contraception or incorrect use [25]. Adolescents treated with lithium require early contraceptive counselling [61]. Women with renal disease, heart failure or Brugada syndrome have contraindications to lithium in the Portuguese SmPC [37]. The woman with Ebstein anomaly who becomes pregnant is a distinct special population, whose management falls to adult congenital heart disease cardiology [54].

9.4. Reference networks and associations

In Europe, ERN GUARD-Heart brings together the centres for rare heart diseases, including complex congenital heart disease, and ERN ITHACA the centres for rare congenital malformations and intellectual disability [29, 30]. ENTIS brings together the European teratology information services, and UKTIS publishes leaflets for families (bumps) [27, 28, 31]. In the United States and Canada, the MotherToBaby service provides information on exposures in pregnancy. Relevant patient associations include Bipolar UK, Action on Postpartum Psychosis and, for heart disease, the Children's Heart Federation [54, 80, 81].

10. Illustrative Clinical Case

Gupta and colleagues published in 2024 the case of a 19-year-old man of Indian origin and low socioeconomic status, who presented with acute chest pain of 2 days' duration with a sensation of retrosternal beating, exertional dyspnoea of New York Heart Association class II and fatigue for one month [55]. The family history revealed that his mother, who had since died, had taken lithium carbonate 300 mg twice daily (600 mg/day) for bipolar disorder during the first trimester of pregnancy, stopping it at the end of the second trimester [55]. On examination he had a heart rate of 184/min, regular, blood pressure of 130/96 mmHg, peripheral oxygen saturation of 82% on room air, precordial asymmetry, a systolic thrill at the lower left sternal border, a systolic murmur and lower limb oedema [55]. Echocardiography showed a markedly dilated right atrium (33.5 cm²; figure 1, https://pmc.ncbi.nlm.nih.gov/articles/PMC12139639/figure/heartviews.heartviews_72_24-F1/), a moderately dysfunctional right ventricle (14.9 cm²), apical displacement of the septal leaflet of the tricuspid valve, a “sail-like” and unstable anterior leaflet (figure 3, https://pmc.ncbi.nlm.nih.gov/articles/PMC12139639/figure/heartviews.heartviews_72_24-F3/), severe tricuspid regurgitation, a left ventricular ejection fraction of 60%, no atrial septal defect or patent foramen ovale, and a Great Ormond Street echocardiographic (GOSE) index of 1.5 [55]. He was treated with esmolol for the tachyarrhythmia and furosemide for the oedema, declined surgery and was referred to electrophysiology; after 2 weeks he had neither improved nor deteriorated [55].

The authors attributed the Ebstein anomaly to lithium and stressed that the dose of 600 mg/day was below the threshold of 900 mg/day associated with greater risk in the Medicaid cohort [2, 55]. The case illustrates three points: Ebstein anomaly may be diagnosed only in adulthood, a history of lithium exposure should be sought in a patient with Ebstein anomaly, and a lower daily dose does not eliminate the risk [54, 55]. It also illustrates the limitations of causal attribution in an individual case: no genetic study was performed, no assessment for non-compaction cardiomyopathy or pre-excitation is described, the exposure is based on the family history and the measurement of leaflet displacement is presented in a unit of area (mm²), which prevents comparison with the usual thresholds. According to sections 5 and 6, a patient with this profile should be assessed for left ventricular non-compaction and, if present, have MYH7 and TPM1 testing [6, 12]. Refusal of surgery and exclusively pharmacological treatment do not represent the recommended management of symptomatic Ebstein anomaly with severe tricuspid regurgitation and hypoxaemia.

11. Prognosis and Long-Term Follow-Up

11.1. Prognosis

For most exposed children, the prognosis is that of the general population: 97% to 98% have no cardiac malformation, neonatal effects are generally mild and transient, and studies of neurodevelopment and brain structure up to 14 years of age have shown no differences [2, 28, 39, 58, 59]. The prognosis of the child with Ebstein anomaly depends on the anatomy: in the Nordic cohort of 794 patients, those with complex anatomy had the greatest burden of morbidity, followed by those with an atrial septal defect and those with isolated Ebstein anomaly; supraventricular arrhythmia affected about 70% over a lifetime, pre-excitation 19%, heart failure 18% in the 10 years following the onset of a supraventricular arrhythmia, and ischaemic heart disease at 60 years was 9% to 20%, compared with 7% in controls [56]. Interventions (tricuspid valve surgery, antiarrhythmic treatment, closure of the atrial septal defect) accumulate over a lifetime [56]. In fetuses and neonates with high-risk markers, mortality is high and some cases warrant palliative care [26]. The GOSE index of 1.5 in the case in section 10 corresponds, according to the authors, to a mortality of 8% [55]. For the woman, the prognosis depends on control of the bipolar disorder: postpartum recurrence is about 20% to 23% with lithium and 40% to 66% without treatment [25, 69].

11.2. Surveillance table

The sources consulted do not define specific intervals for the cardiological follow-up of lithium-associated Ebstein anomaly; intervals should follow the congenital heart disease practices applicable to each patient's anatomy.

12. Summary and Final Considerations

Lithium is the mood stabiliser with the best evidence for preventing perinatal relapse of bipolar disorder, and it is used by women of reproductive age with a disease whose postpartum relapse risk is among the highest in medicine. The association with Ebstein anomaly, inferred in the 1970s from a voluntary register with over-representation of malformed cases, has been reassessed: current cohorts show an increase of about one additional case of cardiac malformation per 100 births exposed in the first trimester, with a dose–response relationship above 900 mg/day, a greater relative increase for right ventricular outflow tract defects and an increase in major malformations overall. The Swedish and Australian cohorts add spontaneous preterm birth and large for gestational age neonates, and exposure late in pregnancy may cause transient neonatal toxicity. The neurodevelopmental data are reassuring but come from small samples.

Management rests on planning pregnancy during euthymia, the lowest effective dose in the first trimester, frequent monitoring of the lithium concentration, which falls until 18 weeks and rises thereafter, and fetal echocardiography between 16 and 20 weeks. The guidelines differ on the position of lithium relative to antipsychotics (restrictive in NICE, second line in CANMAT, continuation recommended by ACOG), on withdrawal before delivery (from the US label, which recommends it 2 to 3 days beforehand, to the Dutch studies, which do not recommend it when the concentration is within the therapeutic window, with CANMAT proposing a threshold of 0.7 mmol/L) and on breastfeeding (contraindicated in the Portuguese SmPC, discouraged by NICE, individualised by CANMAT and LactMed). Postpartum, the target of 0.8 to 1.0 mEq/L and sleep protection are the central measures for preventing relapse.

In an exposed child with Ebstein anomaly, the exposure does not close the diagnosis. Left ventricular non-compaction cardiomyopathy, dysmorphism, extracardiac malformations, atrioventricular block or a family history point to MYH7, TPM1, NKX2-5 or to the 1p36 and 8p23.1 deletions, which change the recurrence risk and the surveillance of relatives. Ebstein anomaly, whatever its cause, requires lifelong cardiological follow-up, because of the high burden of arrhythmias and heart failure.

13. References

1. Callovini T, Montanari S, Bardi F, Barbonetti S, Rossi S, Caso R, et al. Obstetric Outcomes in Women on Lithium: A Systematic Review and Meta-Analysis. J Clin Med. 2024;13(16):4872. doi:10.3390/jcm13164872
2. Patorno E, Huybrechts KF, Bateman BT, Cohen JM, Desai RJ, Mogun H, et al. Lithium Use in Pregnancy and the Risk of Cardiac Malformations. N Engl J Med. 2017;376(23):2245-2254. doi:10.1056/NEJMoa1612222
3. Hastie R, Tong S, Hiscock R, Lindquist A, Lindström L, Wikström AK, et al. Maternal lithium use and the risk of adverse pregnancy and neonatal outcomes: a Swedish population-based cohort study. BMC Med. 2021;19(1):291. doi:10.1186/s12916-021-02170-7
4. U.S. Food and Drug Administration. Lithium carbonate tablets/capsules/oral solution — prescribing information (NDA 017812; SPL set_id 05312159-3d8a-45e0-99cb-b83a7b200cd3, effective version 2026-07-06). DailyMed. https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=05312159-3d8a-45e0-99cb-b83a7b200cd3
5. Drugs and Lactation Database (LactMed®). Lithium. Bethesda (MD): National Institute of Child Health and Human Development; last revised 15 Jun 2026. https://www.ncbi.nlm.nih.gov/books/NBK501153/
6. Postma AV, van Engelen K, van de Meerakker J, Rahman T, Probst S, Baars MJ, et al. Mutations in the sarcomere gene MYH7 in Ebstein anomaly. Circ Cardiovasc Genet. 2011;4(1):43-50. doi:10.1161/CIRCGENETICS.110.957985
7. Kelle AM, Bentley SJ, Rohena LO, Cabalka AK, Olson TM. Ebstein anomaly, left ventricular non-compaction, and early onset heart failure associated with a de novo α-tropomyosin gene mutation. Am J Med Genet A. 2016;170(8):2186-90. doi:10.1002/ajmg.a.37745
8. OMIM. *600584 NK2 homeobox 5; NKX2-5. https://omim.org/entry/600584
9. Digilio MC, Bernardini L, Lepri F, Giuffrida MG, Guida V, Baban A, et al. Ebstein anomaly: Genetic heterogeneity and association with microdeletions 1p36 and 8p23.1. Am J Med Genet A. 2011;155A(9):2196-202. doi:10.1002/ajmg.a.34131
10. GeneGuidance. MYH7 — clinical gene curation (version of 2026-09-26).
11. Vermeer AM, van Engelen K, Postma AV, Baars MJ, Christiaans I, De Haij S, et al. Ebstein anomaly associated with left ventricular noncompaction: an autosomal dominant condition that can be caused by mutations in MYH7. Am J Med Genet C Semin Med Genet. 2013;163C(3):178-84. doi:10.1002/ajmg.c.31365
12. Nijak A, Alaerts M, Kuiperi C, Corveleyn A, Suys B, Paelinck B, et al. Left ventricular non-compaction with Ebstein anomaly attributed to a TPM1 mutation. Eur J Med Genet. 2018;61(1):8-10. doi:10.1016/j.ejmg.2017.10.003
13. GeneGuidance. NKX2-5 — clinical gene curation (version of 2026-09-26).
14. OMIM. #224700 Ebstein anomaly. Baltimore (MD): Johns Hopkins University. https://omim.org/entry/224700
15. OMIM. #613426 Left ventricular noncompaction 5; LVNC5. https://omim.org/entry/613426
16. OMIM. *160760 Myosin, heavy chain 7, cardiac muscle, beta; MYH7. https://omim.org/entry/160760
17. OMIM. #108900 Atrial septal defect 7 with or without atrioventricular conduction defects; ASD7. https://omim.org/entry/108900
18. OMIM. #607872 Chromosome 1p36 deletion syndrome, distal. https://omim.org/entry/607872
19. Orphanet. Ebstein malformation of the tricuspid valve (ORPHA:1880). https://www.orpha.net/en/disease/detail/1880
20. Boyle B, Garne E, Loane M, Addor MC, Arriola L, Cavero-Carbonell C, et al. The changing epidemiology of Ebstein's anomaly and its relationship with maternal mental health conditions: a European registry-based study. Cardiol Young. 2017;27(4):677-685. doi:10.1017/S1047951116001025
21. Munk-Olsen T, Liu X, Viktorin A, Brown HK, Di Florio A, D'Onofrio BM, et al. Maternal and infant outcomes associated with lithium use in pregnancy: an international collaborative meta-analysis of six cohort studies. Lancet Psychiatry. 2018;5(8):644-652. doi:10.1016/S2215-0366(18)30180-9
22. Vigod SN, Frey BN, Clark CT, Grigoriadis S, Barker LC, Brown HK, et al. Canadian Network for Mood and Anxiety Treatments 2024 Clinical Practice Guideline for the Management of Perinatal Mood, Anxiety, and Related Disorders: Guide de pratique 2024 du Canadian Network for Mood and Anxiety Treatments pour le traitement des troubles de l'humeur, des troubles anxieux et des troubles connexes périnatals. Can J Psychiatry. 2025;70(6):429-489. doi:10.1177/07067437241303031
23. Moon-Grady AJ, Donofrio MT, Gelehrter S, Hornberger L, Kreeger J, Lee W, et al. Guidelines and Recommendations for Performance of the Fetal Echocardiogram: An Update from the American Society of Echocardiography. J Am Soc Echocardiogr. 2023;36(7):679-723. doi:10.1016/j.echo.2023.04.014
24. Gertsman S, Bengi D, Clark CT, Frey BN. Pregnancy in a young woman with bipolar disorder. CMAJ. 2026;198(27):E1069-E1072. doi:10.1503/cmaj.251879
25. Bergink V, Suleiman M, Hennen MA, Robakis T. Management of Bipolar Disorder in Pregnancy and Postpartum: A Clinicians' Guide. CNS Drugs. 2025;39(8):763-777. doi:10.1007/s40263-025-01202-7
26. Konstantinov IE, Chai P, Bacha E, Caldarone CA, Da Silva JP, Da Fonseca Da Silva L, et al. The American Association for Thoracic Surgery (AATS) 2024 expert consensus document: Management of neonates and infants with Ebstein anomaly. J Thorac Cardiovasc Surg. 2024;168(2):311-324. doi:10.1016/j.jtcvs.2024.04.018
27. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 718–721.
28. UK Teratology Information Service. Use of lithium in pregnancy (bumps — best use of medicines in pregnancy). Version 3, July 2022. https://www.medicinesinpregnancy.org/leaflets-a-z/lithium/
29. European Reference Network for rare, low prevalence and complex diseases of the heart (ERN GUARD-Heart). https://guardheart.ern-net.eu/ (accessed on 29 October 2026).
30. European Reference Network on Rare Congenital Malformations and Rare Intellectual Disability (ERN ITHACA). https://ern-ithaca.eu/ (accessed on 29 October 2026).
31. European Network of Teratology Information Services (ENTIS). https://www.entis-org.eu/ (accessed on 29 October 2026).
32. Cohen LS, Friedman JM, Jefferson JW, Johnson EM, Weiner ML. A reevaluation of risk of in utero exposure to lithium. JAMA. 1994;271(2):146-50.
33. Jacobson SJ, Jones K, Johnson K, Ceolin L, Kaur P, Sahn D, et al. Prospective multicentre study of pregnancy outcome after lithium exposure during first trimester. Lancet. 1992;339(8792):530-3. doi:10.1016/0140-6736(92)90346-5
34. Yacobi S, Ornoy A. Is lithium a real teratogen? What can we conclude from the prospective versus retrospective studies? A review. Isr J Psychiatry Relat Sci. 2008;45(2):95-106.
35. Diav-Citrin O, Shechtman S, Tahover E, Finkel-Pekarsky V, Arnon J, Kennedy D, et al. Pregnancy outcome following in utero exposure to lithium: a prospective, comparative, observational study. Am J Psychiatry. 2014;171(7):785-94. doi:10.1176/appi.ajp.2014.12111402
36. Roddy Mitchell A, Lindquist A, Hiscock RJ, Forsythe A, Gordon H, Homer CSE, et al. Lithium Use in Pregnancy and the Risk of Spontaneous Preterm Birth: An Australian Statewide Retrospective Cohort Study. BJOG. 2026. Online ahead of print. doi:10.1111/1471-0528.70301
37. INFARMED. Resumo das Características do Medicamento: Priadel 400 mg comprimidos de libertação modificada (lithium carbonate). Marketing authorisation holder: Essential Pharma Limited. SmPC approved on 15-11-2022. Infomed. https://extranet.infarmed.pt/INFOMED-fo/
38. Newport DJ, Viguera AC, Beach AJ, Ritchie JC, Cohen LS, Stowe ZN. Lithium placental passage and obstetrical outcome: implications for clinical management during late pregnancy. Am J Psychiatry. 2005;162(11):2162-70. doi:10.1176/appi.ajp.162.11.2162
39. Molenaar NM, Poels EMP, Robakis T, Wesseloo R, Bergink V. Management of lithium dosing around delivery: An observational study. Bipolar Disord. 2021;23(1):49-54. doi:10.1111/bdi.12955
40. Papapanagou Chroni M, Ydraios G, Nasserdine R, Tsartsalis S, Richard-Lepouriel H. [Sex-specific lithium management and perinatal specificities]. Rev Med Suisse. 2026;22(975):1489-1492. doi:10.53738/REVMED.2026.22.975.49124
41. Fornaro M, Maritan E, Ferranti R, Zaninotto L, Miola A, Anastasia A, et al. Lithium Exposure During Pregnancy and the Postpartum Period: A Systematic Review and Meta-Analysis of Safety and Efficacy Outcomes. Am J Psychiatry. 2020;177(1):76-92. doi:10.1176/appi.ajp.2019.19030228
42. Schrijver L, Kamperman AM, Bijma H, van Kamp IL, Wesseloo R, Hoogendijk WJG, et al. Dose response relationship between lithium serum levels during pregnancy and birth outcomes. Acta Psychiatr Scand. 2024;149(4):323-331. doi:10.1111/acps.13663
43. Stoppa A, Roda E, Garcia-Esteve L, Roca-Lecumberri A. A case of pregnancy with severe polyhydramnios related to long-term use of lithium. Bipolar Disord. 2024;26(4):405-408. doi:10.1111/bdi.13427
44. Poels EM, Sterrenburg K, Wierdsma AI, Wesseloo R, Beerthuizen A, van Dijke L, et al. Lithium exposure during pregnancy increases fetal growth. J Psychopharmacol. 2021;35(2):178-183. doi:10.1177/0269881120940914
45. Hoedemaekers YM, Germans T. Left ventricular noncompaction. In: Baars HF, van der Smagt JJ, Doevendans PAFM, van Tintelen JP, editors. Clinical Cardiogenetics. 3rd ed. Cham: Springer; 2020. p. 115–132. doi:10.1007/978-3-030-45457-9_7
46. ClinVar. VCV000066081 — NM_000257.4(MYH7):c.847T>G (p.Tyr283Asp). https://www.ncbi.nlm.nih.gov/clinvar/variation/66081/
47. ClinVar. VCV001175039 — NM_000257.4(MYH7):c.5754C>G (p.Asn1918Lys). https://www.ncbi.nlm.nih.gov/clinvar/variation/1175039/
48. Bettinelli AL, Mulder TJ, Funke BH, Lafferty KA, Longo SA, Niyazov DM. Familial ebstein anomaly, left ventricular hypertrabeculation, and ventricular septal defect associated with a MYH7 mutation. Am J Med Genet A. 2013;161A(12):3187-90. doi:10.1002/ajmg.a.36182
49. ClinVar. VCV000042968 — NM_000257.4(MYH7):c.3658_3660del (p.Glu1220del). https://www.ncbi.nlm.nih.gov/clinvar/variation/42968/
50. VariantValidator (v4.0.1; vvdb_2025_3). NM_001018005.2:c.475G>A. https://variantvalidator.org
51. Sicko RJ, Browne ML, Rigler SL, Druschel CM, Liu G, Fan R, et al. Genetic Variants in Isolated Ebstein Anomaly Implicated in Myocardial Development Pathways. PLoS One. 2016;11(10):e0165174. doi:10.1371/journal.pone.0165174
52. Giannakou A, Sicko RJ, Zhang W, Romitti P, Browne ML, Caggana M, et al. Copy number variants in Ebstein anomaly. PLoS One. 2017;12(12):e0188168. doi:10.1371/journal.pone.0188168
53. Wang E, Liu Y, Wang Y, Han X, Zhou Y, Zhang L, et al. Comparative Safety of Antipsychotic Medications and Mood Stabilizers During Pregnancy: A Systematic Review and Network Meta-analysis of Congenital Malformations and Prenatal Outcomes. CNS Drugs. 2025;39(1):1-22. doi:10.1007/s40263-024-01131-x
54. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 112–115.
55. Gupta V, Mishra P, Mirdha P, Patil A. Lithium-induced Ebstein's Anomaly. Heart Views. 2024;25(4):267-269. doi:10.4103/heartviews.heartviews_72_24
56. Eckerström F, Hjortdal VE, Eriksson P, Dellborg M, Mandalenakis Z. Morbidity Burden in Patients With Ebstein Anomaly: The Natural History. J Am Heart Assoc. 2024;13(21):e037127. doi:10.1161/JAHA.124.037127
57. Poels EMP, Kamperman AM, Vreeker A, Gilden J, Boks MP, Kahn RS, et al. Lithium Use during Pregnancy and the Risk of Miscarriage. J Clin Med. 2020;9(6):1819. doi:10.3390/jcm9061819
58. Poels EMP, Schrijver L, White TJH, Roza SJ, Zarchev MG, Bijma H, et al. The effect of prenatal lithium exposure on the neuropsychological development of the child. Bipolar Disord. 2022;24(3):310-319. doi:10.1111/bdi.13133
59. Poels EMP, Kamperman AM, Bijma HH, Honig A, van Kamp IL, Kushner SA, et al. Brain development after intrauterine exposure to lithium: A magnetic resonance imaging study in school-age children. Bipolar Disord. 2023;25(3):181-190. doi:10.1111/bdi.13297
60. Smith CJ, Payne VM. Epidemiology studies on effects of lithium salts in pregnancy are confounded by the inability to control for other potentially teratogenic factors. Hum Exp Toxicol. 2024;43:9603271241236346. doi:10.1177/09603271241236346
61. Yatham LN, Kennedy SH, Parikh SV, Schaffer A, Bond DJ, Frey BN, et al. Canadian Network for Mood and Anxiety Treatments (CANMAT) and International Society for Bipolar Disorders (ISBD) 2018 guidelines for the management of patients with bipolar disorder. Bipolar Disord. 2018;20(2):97-170. doi:10.1111/bdi.12609
62. Koire AM, Silver D, Clark CT, Bergink V, Teslyar P. Perinatal Bipolar Disorder Management: Lithium in the Setting of Obstetric Complications. Harv Rev Psychiatry. 2026. Online ahead of print. doi:10.1097/HRP.0000000000000481
63. McAllister-Williams RH, Baldwin DS, Cantwell R, Easter A, Gilvarry E, Glover V, et al. British Association for Psychopharmacology consensus guidance on the use of psychotropic medication preconception, in pregnancy and postpartum 2017. J Psychopharmacol. 2017;31(5):519-552. doi:10.1177/0269881117699361
64. MotherToBaby | Fact Sheets. Lithium. Brentwood (TN): Organization of Teratology Information Specialists (OTIS); March 2025. https://www.ncbi.nlm.nih.gov/books/NBK582797/
65. American College of Obstetricians and Gynecologists Committee on Clinical Practice Guidelines–Obstetrics. Treatment and Management of Mental Health Conditions During Pregnancy and Postpartum: ACOG Clinical Practice Guideline No. 5. Obstet Gynecol. 2023;141(6):1262-1288. doi:10.1097/AOG.0000000000005202
66. National Institute for Health and Care Excellence. Antenatal and postnatal mental health: clinical management and service guidance. Clinical guideline CG192. London: NICE; 2014 (updated 11 Feb 2020). https://www.nice.org.uk/guidance/cg192
67. Jones KL, Jones MC, del Campo M. Smith's Recognizable Patterns of Human Malformation. 8th ed. Philadelphia: Elsevier; 2022. p. 76–79.
68. Miranda-Fernández MC, Ramírez-Oyaga S, Restrepo CM, Huertas-Quiñones VM, Barrera-Castañeda M, Quero R, et al. Identification of a New Candidate Locus for Ebstein Anomaly in 1p36.2. Mol Syndromol. 2018;9(3):164-169. doi:10.1159/000488820
69. Sharma V, Dietrich EL, Cha B, Gonzalez-Torres C, Mazmanian D. Lithium in the Prevention of Postpartum Recurrence of Mood Disorders: A Systematic Review and Meta-Analysis. J Clin Psychiatry. 2026;87(4):26r16567. doi:10.4088/JCP.26r16567
70. Eurofarma Laboratórios. Carbolitium® (lithium carbonate) 300 mg coated tablet — patient information leaflet (Carbolitium_com rev_VP_V09; approved by ANVISA on 25/10/2024). https://static-webv8.jet.com.br/drogaosuper/bulas/147950.pdf
71. U.S. Food and Drug Administration. LITHOBID (lithium carbonate) extended-release tablets — prescribing information (ANI Pharmaceuticals; SPL set_id f7f5b69a-c2a1-4586-a189-1475d41387c0, effective 2026-08-24). https://dailymed.nlm.nih.gov/dailymed/lookup.cfm?setid=f7f5b69a-c2a1-4586-a189-1475d41387c0
72. Betcher HK, Vydysh A, Ellington BD, Clark CT. Management of Lithium in Pregnancy and Lactation: An Individualized Approach. Curr Psychiatry Rep. 2026;28(1):32. doi:10.1007/s11920-026-01674-4
73. Agência Nacional de Vigilância Sanitária. Open data — registered medicines (DADOS_ABERTOS_MEDICAMENTOS.csv). https://dados.anvisa.gov.br/dados/DADOS_ABERTOS_MEDICAMENTOS.csv
74. Fabiano N, Wong S, Gupta A, Tran J, Bhambra N, Min KK, et al. Safety of psychotropic medications in pregnancy: an umbrella review. Mol Psychiatry. 2025;30(1):327-335. doi:10.1038/s41380-024-02697-0
75. Chauhan D, Madley-Dowd P, Hayes JF, Rast JE, Ahlqvist VH, Martin FZ, et al. Lithium prescribing in the perinatal period: UK primary care cohort study. Br J Psychiatry. 2026:1-7. Online ahead of print. doi:10.1192/bjp.2026.10632
76. Teraoka JT, Agarwal A, Tang JJ, Dewland TA, Marcus GM. Elevated lithium levels in drinking water are associated with higher prevalence of Wolff-Parkinson-White diagnoses. J Clin Invest. 2026. Online ahead of print. doi:10.1172/JCI205020
77. Ahmed I, Khan MS, Cheung P, Magsi H, Ali Z, Zhang Y, et al. Minimizing neonatal hypothyroidism induced by lithium exposure through breast milk. J Trace Elem Med Biol. 2025;89:127653. doi:10.1016/j.jtemb.2025.127653
78. Xing S, Lin Y, Sun Y, Yue X, Fan Q, Nie J, et al. Lithium Exposure Causes Trophoblast Cuproptosis by Upregulating FOXO1/STEAP4 Axis in Unexplained Miscarriage. Adv Sci (Weinh). 2025;12(38):e02139. doi:10.1002/advs.202502139
79. Sedaghat-Hamedani F, Andelfinger GU, Meder B. Human Genetics of Ebstein Anomaly. Adv Exp Med Biol. 2024;1441:909-914. doi:10.1007/978-3-031-44087-8_57
80. Bipolar UK. https://www.bipolaruk.org/ (accessed on 29 October 2026).
81. Action on Postpartum Psychosis (APP). https://www.app-network.org/ (accessed on 29 October 2026).