What is Postpartum Psychosis: understanding the headlines

Postpartum (puerperal) psychosis is a rare psychiatric emergency arising in the immediate weeks after childbirth, most often as an affective episode, usually mania or a mixed state, accompanied by psychotic features. “Puerperal mania” is the historical term for the manic presentation, which remains the most characteristic phenotype. This review covers epidemiology, pathophysiology, phenomenology, diagnostic classification, differential diagnosis, acute and prophylactic management, and breastfeeding safety in adults. Because the condition sits at the intersection of obstetrics, psychiatry, and neurology, secondary organic causes receive dedicated attention. The evidence base is notably thin, fewer than 30 publications have focused specifically on treatment, and there are no randomized controlled trials, so much of the management framework rests on well-conducted observational cohorts and expert consensus (Bergink et al., 2016). See the visual infographic summary of postpartum psychosis

Epidemiology and timing

Postpartum psychosis affects approximately 1 to 2 per 1,000 deliveries, with DSM-5 citing 1 in 500 to 1 in 1,000 and a JAMA Psychiatry report citing 1.1 per 1,000 births (Jeste et al., 2022; VanderKruik et al., 2017; Wisner et al., 2024). When restricted to first-lifetime-onset cases with no prior psychiatric history, population-based register data give a lower incidence of 0.25 to 0.6 per 1,000 births (Bergink et al., 2016; Valdimarsdóttir et al., 2009). The defining epidemiologic feature is the sharp temporal clustering after delivery: in a large Swedish cohort, the incidence of psychosis hospitalization was 1.2 per 1,000 within 90 days postpartum but dropped to 0.49 per 1,000 person-years thereafter, confirming a dramatic peak in the immediate puerperium (Valdimarsdóttir et al., 2009).

Onset is abrupt and early. The median time from delivery to symptom onset was 10 days in the largest phenomenology cohort to date (n=248), with most cases presenting within the first 2 weeks (Cohen et al., 2025). Manic and psychotic episodes present earlier than depressive ones (Di Florio et al., 2013). The condition is markedly more common in primiparous women, and more than half of affected women have no prior psychiatric history (Di Florio et al., 2021; Valdimarsdóttir et al., 2009).

Pathophysiology

The etiology is multifactorial, integrating endocrine, immune, circadian, genetic, and stress-system contributions in vulnerable women (Bergink et al., 2016; Meltzer-Brody et al., 2018). The abrupt postpartum withdrawal of estrogen and progesterone, which rise 100 to 1,000-fold during pregnancy, is hypothesized to destabilize affect in susceptible women, with animal models showing that estrogen withdrawal alters oxytocin-system neuroplasticity (Carlson & Kulkarni, 2026; Essali et al., 2013; Hedges et al., 2021). A landmark study of 63 women with first-onset postpartum psychosis demonstrated immune-neuroendocrine dysregulation: a failure of the normal postpartum T-cell elevation, with monocyte activation instead and a decreased glucocorticoid receptor α/β ratio (Bergink et al., 2013). Circadian disruption acts as a proximal trigger, and loss of sleep during labor and delivery is strongly associated with onset in bipolar women (Perry et al., 2024; Tsokkou et al., 2024). Heritability is high and relatively specific: recent Swedish register and whole-genome data estimate family-based heritability at 55% and the relative recurrence risk in full siblings at 10.69 (95% CI 6.60-16.26), with linkage signals on chromosome 16p13 in bipolar families with puerperal psychosis (Jones et al., 2007; Jung et al., 2026; Kępińska et al., 2025).

Clinical presentation and phenomenology

Postpartum psychosis is almost always a mood disorder with psychotic features, not a primary schizophrenia-spectrum illness (Bergink et al., 2016; Essali et al., 2013). In the Massachusetts General Hospital Postpartum Psychosis Project (n=248), 71.8% met criteria for bipolar I disorder with psychotic features (Cohen et al., 2025). Symptom frequencies in that cohort were odd beliefs or delusions in 87.6%, persecutory delusions in 75.2%, ideas of reference in 55.8%, visual hallucinations in 52.3%, and auditory hallucinations in 48.1% (Cohen et al., 2025). For background on how these experiences are described clinically, see our overview of what psychosis is.

Three classic presentations are recognized (Brockington, 2004; Essali et al., 2013):

• Manic (puerperal mania): euphoria or irritability, overactivity, markedly decreased need for sleep, pressured speech, flight of ideas, disinhibition, and grandiose or religious delusions.
• Depressive psychotic: severe depression with mood-congruent delusions, hallucinations, mutism, or stupor.
• Acute polymorphic (cycloid) psychosis: rapidly shifting, “kaleidoscopic” symptoms varying in type and intensity within a single day, with perplexity and confusion.

Importantly, a clinical cohort of 130 women found the three most prevalent symptoms were irritability, abnormal thought content, and anxiety (about 75% each), with irritability far exceeding elevated mood, suggesting that dysphoric rather than euphoric mania predominates (Kamperman et al., 2017). Latent profile analysis of this cohort identified three subgroups.

Distinctive atypical features, including perplexity, confusion, rapid fluctuation of mental state, catatonia, and extremes of fear or ecstasy, help differentiate postpartum psychosis from non-puerperal psychosis (Essali et al., 2013). These fluctuations can mimic delirium, but true fluctuating impairment of consciousness is actually uncommon (roughly 10 to 20%) and concentrated in the atypical subgroup (Kamperman et al., 2017). A rare severe variant, postpartum delirious mania, combines delirium, mania, and psychosis and may be refractory to pharmacotherapy while responding to ECT (Song et al., 2026). Catatonia occurs in roughly 20% of women (Nahar et al., 2017).

Diagnostic classification

Postpartum psychosis is not a distinct diagnostic entity in either DSM-5 or ICD-11. It is coded under an existing mood or psychotic disorder with a “with peripartum onset” specifier (Bergink et al., 2026; Di Florio et al., 2021; Michalczyk et al., 2023), which in DSM-5 applies to onset during pregnancy or within 4 weeks postpartum, while the ICD uses a 6-week window (Di Florio et al., 2021; Jeste et al., 2022). The specifier can be applied to manic, hypomanic, or depressive episodes in bipolar I and II disorder and to brief psychotic disorder (Jeste et al., 2022). Postpartum psychosis was a recognized diagnosis in DSM-I and DSM-II but was removed in DSM-III in 1980, and a 2020 proposal for reinstatement was not accepted (Spinelli, 2021).

The specifier has been criticized for conflating prepartum and postpartum onset and for a 4-week window that excludes many clinically relevant later-onset episodes, with some experts advocating extension to 6 months (Sharma & Mazmanian, 2014). A 2026 expert consensus panel in Biological Psychiatry recommends classifying postpartum psychosis as a distinct category within the bipolar disorders chapter, citing prominent affective symptoms, excellent response to lithium and ECT, the observation that first-onset postpartum psychosis is the first onset of bipolar disorder in roughly half of cases, and a distinct but overlapping genetic architecture (Bergink et al., 2026).

Relationship to bipolar disorder

The link to bipolar disorder is strong but not unidimensional. Bipolar I disorder is the single strongest risk factor. A Dutch cohort of 436 women with bipolar I (762 live births) found a postpartum episode risk of 30.1% versus 5.2% during pregnancy (Gilden et al., 2021). A prospective study found that 23% of women with bipolar I or schizoaffective-bipolar disorder experienced mania or psychosis within 6 weeks postpartum, versus only 4% with bipolar II (Perry et al., 2021). Longitudinally, after a first episode of postpartum psychosis, roughly 20 to 50% have “isolated” postpartum psychosis with episodes confined to the postpartum period, while the remainder develop non-perinatal episodes, usually within the bipolar spectrum (Bergink et al., 2016). A Danish study found that women with psychiatric onset within 30 days postpartum had approximately triple the risk of eventually receiving a bipolar diagnosis, about 14% over 15 years (Munk-Olsen et al., 2012).

Genetic data refine this picture: polygenic risk score analysis shows that first-onset postpartum psychosis overlaps with bipolar disorder and schizophrenia in genetic vulnerability but has significantly lower genetic loading for major depression, closer to controls, arguing that it is not simply a bipolar subtype but a partially distinct entity (Di Florio et al., 2021).

Differential diagnosis

Because onset is acute and organic mimics carry different treatments, the workup must exclude secondary causes before attributing symptoms to a primary affective psychosis (Bergink et al., 2016). Our companion piece on the medical workup of first-episode psychosis details the laboratory and imaging sequence.

Suicide and infanticide risk

Postpartum psychosis is a psychiatric emergency because of the elevated risk of maternal suicide and, more rarely, infanticide. Women admitted psychiatrically in the first postpartum year had a standardized mortality ratio for suicide of 1,719, rising to 7,216 within the first year after diagnosis, roughly a 70-fold increase over the general population (Appleby et al., 1998). Suicidal ideation was present in 38% of inpatients with postpartum psychosis, with 18% attempting during the current episode (Babu et al., 2008). Untreated postpartum psychosis carries an infanticide rate of approximately 4%, most often driven by command hallucinations to kill the infant or delusions that the infant is possessed, and the filicide rate is highest in depressive psychoses at 4.5% (Alford et al., 2025; Brockington, 2017b; Jeste et al., 2022). These risks make urgent inpatient evaluation mandatory.

Acute management

Inpatient admission is universally recommended for safety and diagnostic workup. Mother-baby units are considered the gold standard where available, yielding higher service satisfaction, though a quasi-experimental study found no significant difference in 12-month readmission versus non-mother-baby-unit care (aOR 0.95, 95% CI 0.86-1.04) (Bergink et al., 2016; Howard et al., 2022; Osborne, 2018).

The most influential framework is the Bergink sequential algorithm, tested in 64 consecutively admitted women with first-onset postpartum psychosis (Bergink et al., 2015b):

1. Benzodiazepine (lorazepam) for acute agitation, insomnia, and anxiety.
2. Antipsychotic monotherapy if benzodiazepines are insufficient.
3. Lithium added to the antipsychotic if remission is not achieved.
4. ECT for refractory cases.

In that study, 98.4% achieved complete remission within the first three steps and none required ECT. At 9 months, sustained remission was 79.7%, and patients maintained on lithium relapsed significantly less than those on antipsychotic monotherapy (Bergink et al., 2015b). The Nature Reviews Disease Primers algorithm similarly positions lithium as the primary agent for mania and psychosis, with adjunctive antipsychotics or benzodiazepines, and ECT available at any step.

Lithium is the cornerstone, with “excellent” treatment response cited by expert consensus (Bergink et al., 2015b; Bergink et al., 2026). Postpartum renal changes cause lithium levels to rise about 9% relative to late pregnancy, so close serum monitoring is essential (Fornaro et al., 2020). Antipsychotics are a mainstay for psychosis, mania, and agitation. Both first- and second-generation agents are used, with olanzapine the most frequently reported, though no head-to-head efficacy data exist (Teodorescu et al., 2020). See our overview of antipsychotic pharmacotherapy for general prescribing considerations. Benzodiazepines, with lorazepam preferred, serve as initial stabilization and for catatonia, where the APA recommends not exceeding 4 mg/day in pregnancy and considering ECT for non-responders (Bergink et al., 2015b; Wilson et al., 2025).

Valproate and carbamazepine should be avoided in women of childbearing potential given teratogenicity, and antidepressants are generally avoided because of the bipolar-spectrum nature of the illness and the risk of inducing mania (Bergink et al., 2025; Sharma et al., 2017).

ECT is highly effective and may be particularly so postpartum. A population-based Swedish study found that 87.0% of postpartum patients responded to ECT versus 73.5% of non-postpartum patients (p=0.001) (Rundgren et al., 2018). A 2026 systematic review reported symptom improvement in all cases with mostly complete remission after 5 to 11 sessions and only transient side effects (Herrera-Barrón et al., 2026). ECT is indicated when pharmacotherapy is insufficient, when rapid response is required for safety such as suicidality or infanticide risk, or for catatonia unresponsive to lorazepam. No adverse effects have been noted in breastfed infants of mothers receiving ECT (Babu et al., 2013; Essali et al., 2013; Herrera-Barrón et al., 2026).

Prognosis and recurrence

The acute prognosis is generally favorable: with structured treatment, about 98% remit and 74% report good global functioning at 9 months (Bergink et al., 2015b; Burgerhout et al., 2017). Long-term outcome is bifurcated. A meta-analysis (N=645, 11 to 26 year follow-up) found that 43.5% had isolated postpartum psychosis with no non-puerperal episodes, while about 56% experienced episodes outside the perinatal period (Gilden et al., 2020).

Recurrence risk after a subsequent pregnancy is substantial but not inevitable. For isolated postpartum psychosis, relapse risk is 31% (95% CI 22-42). For bipolar disorder with a prior puerperal psychotic episode, a well-characterized sample found 57% (95% CI 44-69) recurrence (Bergink et al., 2016; Robertson et al., 2005). Meta-analysis shows that women with bipolar disorder who are medication-free in pregnancy relapse postpartum at 66% (95% CI 57-75) versus 23% (95% CI 14-37) on prophylaxis (Wesseloo et al., 2016).

Prophylaxis and breastfeeding

Lithium is the best-supported prophylactic. For isolated postpartum psychosis, lithium started immediately postpartum, which avoids fetal exposure, is highly effective: no relapses occurred among 20 women who accepted prophylaxis versus 44% among those who declined (Bergink et al., 2012). For bipolar disorder, continuous prophylaxis through pregnancy and postpartum is recommended (Wesseloo et al., 2016). Sleep protection is an emerging non-pharmacologic strategy, since the loss of at least one complete night of sleep around delivery conferred roughly 5-fold odds of postpartum psychosis (OR 5.19, 95% CI 1.45-18.54) (Perry et al., 2024). A written perinatal relapse-prevention plan should specify maintenance medication, delivery mode, immediate postpartum pharmacoprophylaxis, feeding plan, sleep and circadian protection, and early-warning protocols (Bergink et al., 2025). Our WRAP planning guide describes how such plans are built with families.

On breastfeeding, lithium carries an FDA “not recommended” label, but contemporary data are increasingly reassuring: infant serum levels are typically low, about 0.16 mEq/L in one series and at or below 0.30 mEq/L in 80% of a systematic review, with declining exposure over months and no developmental delay reported, making it a reasonable option with monitoring of infant lithium, renal, and thyroid function (Betcher et al., 2026; Imaz et al., 2019; Imaz et al., 2021; Viguera et al., 2007). Among antipsychotics, olanzapine and quetiapine have the most favorable lactation profiles, risperidone and aripiprazole are compatible under supervision, and clozapine and amisulpride are contraindicated because of very high milk-to-plasma ratios (Pacchiarotti et al., 2016; Schoretsanitis et al., 2020; Teodorescu et al., 2020; Uguz, 2016). Valproate is contraindicated in women of childbearing potential, with an approximately 9.3% major malformation rate, 30 to 40% neurodevelopmental delay, and roughly 5-fold autism risk, and it showed no preventive benefit postpartum in a controlled trial (Degremont et al., 2022; Essali et al., 2013; Hernandez-Diaz et al., 2025).

Evidence gaps and emerging data

The most striking gap is the absence of any randomized controlled trials for either acute treatment or prevention, and the Cochrane review on prevention found no eligible RCTs (Bergink et al., 2016; Essali et al., 2013). The Bergink algorithm, while influential, was tested in a single center in first-onset cases only, limiting generalizability to recurrent episodes (Bergink et al., 2015b). No comparative efficacy data distinguish specific antipsychotics (Teodorescu et al., 2020). Whether sleep disruption is a trigger for or a symptom of the illness remains unresolved (Carr et al., 2023). Long-term neurodevelopmental outcomes of infant lithium exposure through breast milk are understudied (Imaz et al., 2019). Emerging work on immune dysregulation such as regulatory T-cell deficiency and CCN proteins, rare genetic variants such as HMGCR, and the estrogen-withdrawal and kisspeptin-CCN3 hypothesis may eventually yield biomarkers or targeted therapies, and the DSM and ICD reclassification debate continues to gain expert momentum (Bergink et al., 2026; Davies, 2026; Dazzan et al., 2018; Jung et al., 2026).

Summary

Postpartum psychosis is a rare (1 to 2 per 1,000) but severe emergency of the early puerperium, presenting abruptly at a median of about 10 days as a bipolar-spectrum affective episode, most characteristically puerperal mania, with psychosis, perplexity, and a fluctuating course (Cohen et al., 2025; VanderKruik et al., 2017). Bipolar I disorder, prior postpartum psychosis, primiparity, family history, and perinatal sleep loss are the dominant risk factors (Bauer et al., 2018; Perry et al., 2021; Perry et al., 2024). Organic mimics, especially autoimmune anti-NMDA receptor and thyroid encephalopathies, must be excluded (Bergink et al., 2011; Bergink et al., 2015a). Management is urgent inpatient care with a sequential regimen of benzodiazepine, antipsychotic, and lithium, the cornerstone for both acute treatment and prophylaxis, with ECT highly effective for refractory, catatonic, or high-risk cases (Bergink et al., 2015b; Rundgren et al., 2018). Prognosis is often good, but recurrence risk with subsequent pregnancies is high at 31 to 57%, making an immediate-postpartum lithium prophylaxis plan central to future reproductive planning (Bergink et al., 2012; Robertson et al., 2005; Wesseloo et al., 2016).

References