Skip to content

Pregnancy and reproductive health in bipolar disorder

TL;DR — Pregnancy is not reliably protective against bipolar recurrence, and the postpartum period carries an approximately 35–39% pooled relapse risk. Stopping maintenance treatment, especially abruptly, markedly increases recurrence; medication-free pregnancy was associated with 66% postpartum relapse versus 23% with prophylaxis. Fetal and neonatal risks differ substantially by drug: valproate has the clearest teratogenic and neurodevelopmental harms; lithium has a smaller, dose-related cardiac-malformation signal than historical estimates suggested; lamotrigine is comparatively reassuring but not risk-free. Bipolar disorder itself is associated with adverse obstetric outcomes, so treated-versus-general-population comparisons confound illness and medication effects. Decisions should be planned before conception when possible and cover relapse prevention, dosing, delivery, sleep, feeding and rapid postpartum response.

Reproductive planning is part of longitudinal care

Bipolar disorder commonly begins during reproductive years, making pregnancy planning a routine component of maintenance care rather than an exceptional consultation. The core comparison is not “medication risk versus no risk,” but fetal/neonatal exposure risk versus maternal recurrence and the consequences of untreated illness.

A written plan should specify maintenance treatment, monitoring during physiological change, delivery dosing, immediate postpartum prophylaxis, sleep protection, feeding preferences, early warning signs and escalation routes (Bergink 2025, PMID 40593436).

Decision point Evidence-informed question
Before conception Is the current regimen the lowest-risk effective option, especially regarding valproate?
Early pregnancy What is the individual recurrence history after discontinuation, and what fetal screening is appropriate?
Late pregnancy How will renal clearance, dose and serum monitoring be handled?
Delivery Will dosing remain stable or be briefly interrupted, and why?
Postpartum What prophylaxis begins immediately, and who protects sleep and observes early signs?
Feeding What infant exposure and monitoring follow from breastfeeding with the selected drug?

Recurrence during pregnancy

In a prospective cohort of 89 euthymic pregnant women with bipolar disorder, 71% had at least one recurrence during pregnancy (Viguera 2007, PMID 18056236).

Women who discontinued mood stabilizers had twice the recurrence risk, more than fourfold shorter median time to recurrence and five times more weeks ill than women continuing treatment. Abrupt or rapid discontinuation shortened recurrence latency 11-fold compared with gradual discontinuation (Viguera 2007, PMID 18056236).

Viguera cohort finding Estimate
Any recurrence during pregnancy 71%
Recurrences that were depressive or mixed 74%
Recurrences occurring in first trimester 47%
Relative recurrence risk after discontinuation ~2-fold
Weeks ill after discontinuation 5-fold greater
Recurrence latency after abrupt versus gradual withdrawal 11-fold shorter

This was observational rather than randomized; women who continued and discontinued treatment may have differed. The size and consistency of temporal effects nevertheless make abrupt discontinuation a clinically important exposure.

In a smaller cohort of 26 women, recurrence was 30% with continued lamotrigine versus 100% after stopping all mood stabilizers. Time to 25% recurrence was 28 versus 2 weeks, HR 12.1 (95% CI 1.6–91.7) (Newport 2008, PMID 18402631). The wide interval reflects the small sample.

Postpartum relapse

A meta-analysis of 37 articles covering 5,700 deliveries in 4,023 patients estimated overall postpartum relapse at 35% (95% CI 29–41) (Wesseloo 2016, PMID 26514657).

Group Postpartum outcome Estimate
Bipolar disorder and/or postpartum psychosis Any relapse 35% (95% CI 29–41)
Bipolar disorder Severe postpartum episode 17% (13–21)
Prior postpartum psychosis Severe postpartum episode 29% (20–41)
Bipolar, medication-free in pregnancy Relapse 66% (57–75)
Bipolar, prophylaxis in pregnancy Relapse 23% (14–37)

A newer bipolar I/schizoaffective-bipolar meta-analysis included 12 studies, 3,595 deliveries and 2,183 women. Overall postpartum relapse was 39% (95% CI 29–49; I²=96.31%); among relapses, manic or mixed episodes comprised 38% (28–50) (Sharma 2024, PMID 38258551).

Meta-regression estimated relapse at 58.1% in studies with no medication use and 25.9% in studies with 100% medication use (P=0.0359), but study-level meta-regression cannot prove an individual treatment effect (Sharma 2024, PMID 38258551).

Postpartum depressive onset can be the first recognized bipolar presentation. In the FACE-BD cohort, 93 of 759 women whose bipolar disorder began with depression had postpartum onset (12.2%) (Tebeka 2021, PMID 33347983).

Lithium: congenital risk

The largest Medicaid cohort included 1,325,563 pregnancies, of which 663 had first-trimester lithium exposure. Cardiac malformations occurred in 2.41% of lithium-exposed infants, 1.15% unexposed and 1.39% lamotrigine-exposed (Patorno 2017, PMID 28591541).

Lithium result Estimate
Adjusted cardiac-malformation RR 1.65 (95% CI 1.02–2.68)
≤600 mg/day RR 1.11 (0.46–2.64)
601–900 mg/day RR 1.60 (0.67–3.80)
>900 mg/day RR 3.22 (1.47–7.02)
Right-ventricular outflow defect prevalence 0.60% exposed vs 0.18% unexposed
Right-ventricular outflow adjusted RR 2.66 (1.00–7.06)

The dose gradient supports a real association but the absolute excess is much smaller than early registry estimates (Patorno 2017, PMID 28591541).

A 29-study systematic review/meta-analysis estimated any congenital anomaly prevalence 4.1%, OR 1.81 (95% CI 1.35–2.41), NNH 33 (22–77); cardiac anomaly prevalence was 1.2%, OR 1.86 (1.16–2.96), NNH 71 (48–167) (Fornaro 2020, PMID 31623458).

First-trimester spontaneous abortion was estimated at 8.1%, OR 3.77 (95% CI 1.15–12.39), but the association lost significance in some lithium-exposed versus unexposed comparisons (Fornaro 2020, PMID 31623458). This inconsistency should remain visible.

Lithium around delivery

Physiological renal changes alter lithium handling during pregnancy and after birth. The evidence does not establish one universal predelivery strategy.

In 66 mother–infant pairs managed with brief interruption, the mean cord:maternal lithium ratio was 1.10 (SD 0.17), confirming near-complete placental transfer. Fifty-six percent of neonates had transient acute complications; hypotonia occurred in 15, and neonatal lithium was 0.178 mEq/L higher with hypotonia (Imaz 2024, PMID 39187196).

In another observational study, 233 maternal levels showed no relation between delivery timing and lithium level/dose ratio (r=−0.03, P=0.63). Among 29 neonates, maternal and neonatal levels strongly correlated but neonatal level was not associated with measured neonatal outcomes (Molenaar 2021, PMID 32526071).

Predelivery question Evidence tension
Stop/reduce briefly? May reduce peak neonatal exposure but risks subtherapeutic maternal levels
Continue stable dose? Avoids destabilization; neonatal exposure still requires preparation
Universal protocol? Not supported by small observational cohorts

The practical implication is individualized dosing with serum monitoring, hydration/renal assessment and coordinated obstetric, psychiatric and neonatal care, rather than automatic cessation.

Valproate

Valproate has the least favorable reproductive profile among established mood stabilizers. Risks include congenital malformations and later neurodevelopmental outcomes; safer effective alternatives should be considered before conception.

In a Nordic cohort of 4,494,926 children, among offspring of women with epilepsy, age-eight autism prevalence was 1.5% and intellectual disability 0.8% without antiseizure-medication exposure. With valproate, autism was 2.7% and intellectual disability 2.4%; adjusted HRs were 2.4 (95% CI 1.7–3.3) and 2.5 (1.7–3.7) (Bjørk 2022, PMID 35639399).

Because the indication was epilepsy, dose, seizures and maternal characteristics may differ from bipolar populations. The direction aligns with a UK/Swedish cohort of 3,182,773 children in which valproate exposure was associated with autism, intellectual disability and ADHD (Madley-Dowd 2024, PMID 39548057).

Regulation changes prescribing but do not eliminate exposure. In 1,018 French women aged 16–50 with bipolar disorder, valproate use fell from 32.6% before May 2015 restrictions to 17.3% afterward; 26.9% of the overall cohort used it at a mean 968 mg/day (Samalin 2020, PMID 32745833).

Valproate also affects reproductive endocrine health outside pregnancy: PCOS OR 6.74 (95% CI 1.66–27.32), menstrual disorder OR 1.81 (1.02–3.23), and hyperandrogenism OR 2.02 (1.11–3.65) versus non-valproate treatment (Zhang 2016, PMID 27160812).

Lamotrigine and other antiseizure medicines

Lamotrigine is often selected for depression-predominant illness because observational pregnancy evidence is comparatively reassuring and continuation may reduce recurrence. “Comparatively reassuring” does not mean proven risk-free.

The Medicaid cardiac-malformation cohort observed 1.39% cardiac malformations among 1,945 lamotrigine-exposed infants, compared with 1.15% unexposed (Patorno 2017, PMID 28591541).

The Nordic neurodevelopment cohort found no consistent increase after lamotrigine monotherapy, unlike valproate and topiramate (Bjørk 2022, PMID 35639399). The UK/Swedish analysis similarly found little evidence for neurodevelopmental diagnoses after lamotrigine (Madley-Dowd 2024, PMID 39548057).

Topiramate exposure in the Nordic epilepsy cohort was associated with autism HR 2.8 (95% CI 1.4–5.7) and intellectual disability HR 3.5 (1.4–8.6) (Bjørk 2022, PMID 35639399). These data argue against treating all anticonvulsants as one reproductive-risk class.

Antipsychotics

Antipsychotic reproductive evidence is complicated by drug heterogeneity, metabolic effects and confounding by illness severity. A recent review found no statistically significant malformation association for antipsychotics as a group or atypical antipsychotics as a subgroup, while noting a possible risperidone cardiac signal (Cantilino 2025, PMID 40679799).

A Hong Kong cohort found bipolar disorder associated with gestational diabetes at adjusted OR 1.75 (95% CI 1.15–2.70); treated bipolar disorder had OR 2.09 (1.21–3.70), but no significant association emerged for most other pregnancy/neonatal outcomes (Chan 2024, PMID 38914040).

Illness versus treatment confounding

Swedish registry data compared 320 treated women, 554 untreated women with bipolar disorder and 331,263 controls. Induction/planned caesarean occurred in 37.5% treated, 30.9% untreated and 20.7% controls; preterm-birth risk was increased by about 50% in both bipolar groups (Bodén 2012, PMID 23137820).

Untreated bipolar disorder was associated with microcephaly 3.9% versus 2.3% (OR 1.68, 95% CI 1.07–2.62) and neonatal hypoglycemia 4.3% versus 2.5% (OR 1.51, 1.04–2.43) (Bodén 2012, PMID 23137820).

A Medicaid study of 1,472,672 pregnancies found unadjusted placental/preterm RRs of 1.15–1.56 for mood stabilizers, but adjusted estimates were 0.89–1.16. Continuation after 20 weeks did not increase outcomes versus discontinuation (Cohen 2019, PMID 31237992).

Together, these studies show why a poor outcome after exposure is not automatically medication-caused.

Lactation

The lactation literature is much smaller than pregnancy registries and long-term infant neurodevelopment is particularly uncertain.

A 56-study systematic review considered lithium a possible monitored option; carbamazepine and valproate relatively compatible; lamotrigine usable at low dose in selected cases; quetiapine and olanzapine preferred among antipsychotics; and clozapine/amisulpride contraindicated by the review authors (Pacchiarotti 2016, PMID 27568278).

Lithium is detectable in breast milk and infant serum. A three-case series reported healthy early development with monitoring but cannot establish safety (Gehrmann 2021, PMID 34207460).

Feeding decisions also affect maternal sleep, an important relapse variable. Any lactation plan should include who protects consolidated sleep, how infant feeding is shared, and what infant renal/thyroid or drug-level monitoring is available.

Real-world prescribing gaps

In a 16-year Hong Kong cohort of 302 pregnant women with bipolar disorder, 59.6% filled at least one psychotropic prescription during pregnancy and 23.3% received polypharmacy (Kan 2022, PMID 35151952).

Among women exposed to valproate in the year before pregnancy, 16% remained exposed in the first trimester. Prenatal medication use was associated with pre-pregnancy treatment (OR 16.14, 95% CI 8.79–29.65) and psychiatric admission (OR 4.12, 1.66–10.24) (Kan 2022, PMID 35151952).

Guideline grading and lactation uncertainty

ACOG separates perinatal screening/diagnosis from treatment and grades recommendations with a modified GRADE evidence-to-decision framework; it includes bipolar disorder, suicidality and postpartum psychosis, and uses ungraded good-practice points where evidence is inadequate (ACOG 2023, PMID 37486660). This makes certainty visible rather than treating all recommendations as equivalently evidence based.

Breastfeeding and lithium: two reviews, the same uncertainty

A review of 13 case reports/series (39 mother–infant dyads) found mean maternal serum lithium 0.73 mEq/L, milk 0.84 mEq/L and infant serum 0.23 mEq/L. Twenty-six infants (80%) had concentrations no higher than 0.30 mEq/L without reported adverse effects; eight (20%) had transient toxicity or thyroid events, all after prenatal lithium exposure as monotherapy or polytherapy (Imaz 2019, PMID 31551795). Only 15% of included reports met at least half the quality checklist items.

A parallel review found only 12 eligible articles, all case reports, and concluded that adverse outcomes could not be separated cleanly from infant health, gestational exposure and concomitant medication (Newmark 2019, PMID 31180257). One review's conditional monitoring recommendations and another's emphasis on individualized decisions reflect the same sparse evidence rather than opposing safety findings.

An older mood-stabilizer synthesis ranked valproate worst for major malformations and neurodevelopment, identified lamotrigine as comparatively reassuring, and emphasized that carbamazepine still carries documented teratogenic risk (Costoloni 2014, PMID 25639010). Its qualitative design and older registry window mean contemporary registry estimates should take priority where available.

Open questions

  • What lithium dosing strategy around delivery best balances maternal relapse and neonatal adaptation in adequately powered prospective studies (Imaz 2024, PMID 39187196; Molenaar 2021, PMID 32526071)?
  • What are long-term neurodevelopmental outcomes after lithium and antipsychotic exposure with indication adequately controlled (Fornaro 2020, PMID 31623458)?
  • Which sleep-protection interventions prevent postpartum mania independently of medication (Sharma 2024, PMID 38258551)?
  • How can valproate exposure persist after strong regulatory restriction, and which system controls reduce it further (Samalin 2020, PMID 32745833)?
  • Which lactation regimens have acceptable infant exposure and preserve maternal sleep across the first postpartum months (Pacchiarotti 2016, PMID 27568278)?

References

  1. Bergink V, et al. Management of Bipolar Disorder in Pregnancy and Postpartum: A Clinicians' Guide. CNS Drugs. 2025. PMID 40593436.
  2. Viguera AC, et al. Risk of recurrence in women with bipolar disorder during pregnancy: prospective study of mood stabilizer discontinuation. American Journal of Psychiatry. 2007. PMID 18056236.
  3. Newport DJ, et al. Lamotrigine in bipolar disorder: efficacy during pregnancy. Bipolar Disorders. 2008. PMID 18402631.
  4. Wesseloo R, et al. Risk of Postpartum Relapse in Bipolar Disorder and Postpartum Psychosis. American Journal of Psychiatry. 2016. PMID 26514657.
  5. Sharma V, et al. Occurrence of postpartum manic or mixed episodes in women with bipolar I disorder. Bipolar Disorders. 2024. PMID 38258551.
  6. Tebeka S, et al. Clinical characteristics of bipolar disorders with postpartum depressive onset. Progress in Neuro-Psychopharmacology & Biological Psychiatry. 2021. PMID 33347983.
  7. Patorno E, et al. Lithium Use in Pregnancy and the Risk of Cardiac Malformations. New England Journal of Medicine. 2017. PMID 28591541.
  8. Fornaro M, et al. Lithium Exposure During Pregnancy and the Postpartum Period. American Journal of Psychiatry. 2020. PMID 31623458.
  9. Imaz ML, et al. Peripartum lithium management: Early maternal and neonatal outcomes. Journal of Affective Disorders. 2024. PMID 39187196.
  10. Molenaar NM, et al. Management of lithium dosing around delivery: An observational study. Bipolar Disorders. 2021. PMID 32526071.
  11. Bjørk MH, et al. Association of Prenatal Exposure to Antiseizure Medication With Risk of Autism and Intellectual Disability. JAMA Neurology. 2022. PMID 35639399.
  12. Madley-Dowd P, et al. Antiseizure medication use during pregnancy and children's neurodevelopmental outcomes. Nature Communications. 2024. PMID 39548057.
  13. Samalin L, et al. Evolution and characteristics of the use of valproate in women of childbearing age with bipolar disorder. Journal of Affective Disorders. 2020. PMID 32745833.
  14. Zhang L, et al. Reproductive and metabolic abnormalities in women taking valproate for bipolar disorder. European Journal of Obstetrics & Gynecology and Reproductive Biology. 2016. PMID 27160812.
  15. Cantilino A, et al. Mood Stabilizers, Antipsychotics, and Electroconvulsive Therapy in Patients With Bipolar Disorder During Pregnancy and Postpartum. Journal of Psychiatric Practice. 2025. PMID 40679799.
  16. Chan JKN, et al. Risk of adverse pregnancy, delivery and neonatal outcomes associated with bipolar disorder and prenatal use of mood stabilizers. Psychiatry Research. 2024. PMID 38914040.
  17. Bodén R, et al. Risks of adverse pregnancy and birth outcomes in women treated or not treated with mood stabilisers for bipolar disorder. BMJ. 2012. PMID 23137820.
  18. Cohen JM, et al. Anticonvulsant Mood Stabilizer and Lithium Use and Risk of Adverse Pregnancy Outcomes. Journal of Clinical Psychiatry. 2019. PMID 31237992.
  19. Pacchiarotti I, et al. Mood stabilizers and antipsychotics during breastfeeding: Focus on bipolar disorder. European Neuropsychopharmacology. 2016. PMID 27568278.
  20. Gehrmann A, et al. Lithium Medication in Pregnancy and Breastfeeding—A Case Series. Medicina. 2021. PMID 34207460.
  21. Kan ACO, et al. Psychotropic drug utilization patterns in pregnant women with bipolar disorder. European Neuropsychopharmacology. 2022. PMID 35151952.
  22. Imaz ML, et al. Clinical lactation studies of lithium: a systematic review. Front Pharmacol. 2019;10:1005. PMID 31551795
  23. Newmark RL, et al. Risk-benefit assessment of infant exposure to lithium through breast milk: a systematic review. Int Rev Psychiatry. 2019;31:295–304. PMID 31180257
  24. Costoloni G, et al. Mood stabilisers and pregnancy outcomes—a review. Psychiatr Pol. 2014;48:865–887. PMID 25639010
  25. American College of Obstetricians and Gynecologists. Screening and diagnosis of mental health conditions during pregnancy and postpartum: Clinical Practice Guideline No. 4. Obstet Gynecol. 2023;141:1232–1261. PMID 37486660