Skip to content

Bipolar disorder — overview

TL;DR — Bipolar disorder is defined by mania or hypomania and is usually dominated longitudinally by depression, but it cannot safely be treated as unipolar MDD (Nierenberg 2023, PMID 37815563). Lifetime bipolar-spectrum prevalence was 2.4% across World Mental Health surveys (Merikangas 2011, PMID 21383262). Lithium, anticonvulsants and atypical antipsychotics have phase-specific evidence; quetiapine, lurasidone, cariprazine and olanzapine–fluoxetine have randomized evidence in bipolar depression, while lithium remains a foundational maintenance treatment (Thase 2006, PMID 17110817; Loebel 2014, PMID 24170180; Earley 2019, PMID 30845817; Tohen 2003, PMID 14609883). Antidepressant benefit is contested and rapid cycling may worsen in a subgroup (Sachs 2007, PMID 17392295; El-Mallakh 2015, PMID 26142612). No biomarker currently establishes diagnosis or chooses treatment.

Definition and diagnostic anchors

  • Bipolar I disorder requires at least one manic episode: a distinct period of elevated, expansive or irritable mood plus increased energy/activity, severe enough for marked impairment, hospitalization or psychosis.
  • Bipolar II disorder requires hypomania plus major depression and no lifetime mania; hypomania is observable and functionally changed but not severely impairing or psychotic.
  • Mixed features can occur during mania, hypomania or depression. Substance/medication effects and medical causes must be excluded.
  • A depressive episode does not by itself identify polarity. History of activation, decreased need for sleep, episodic overactivity, family history and antidepressant-associated switching can raise suspicion, but none is a standalone test (Nierenberg 2023, PMID 37815563).
  • Unipolar MDD is covered in the depression condition; this condition owns bipolar depression because treatment benefits and harms differ.

Spectrum map

Presentation Longitudinal anchor What the label does not establish Research implication
Bipolar I At least one manic episode Whether future burden will be predominantly manic or depressive Trials and maintenance plans should report polarity-specific outcomes (Nierenberg 2023, PMID 37815563)
Bipolar II Hypomania plus major depression, without mania That illness is intrinsically mild; depression can dominate disability Hypomania must be sought longitudinally rather than inferred from depression severity (Nierenberg 2023, PMID 37815563)
Cyclothymic disorder Chronic subthreshold hypomanic and depressive symptoms Eventual diagnostic stability Repeated assessment is more informative than one cross-sectional interview (Cegla-Schvartzman 2019, PMID 30358663)
Mixed features Opposite-polarity symptoms within a mood episode A single mechanism or uniform drug response Trials need explicit mixed-feature definitions and switch outcomes (Keramatian 2023, PMID 38695002)
Subthreshold bipolar spectrum Bipolar features below categorical thresholds That treatment evidence from bipolar I generalizes Epidemiological prevalence depends strongly on the operational definition (Merikangas 2011, PMID 21383262)

Epidemiology and burden

World Mental Health Initiative data estimated lifetime prevalence of 0.6% for bipolar I, 0.4% for bipolar II and 1.4% for subthreshold bipolar disorder, totaling 2.4%; 12-month prevalence was 1.5% across the spectrum (Merikangas 2011, PMID 21383262).

GBD 2019 estimated 125.3 million DALYs across 12 mental disorders, with bipolar disorder a persistent disability contributor and little decline in age-standardized mental-disorder burden since 1990 (GBD 2019 Mental Disorders Collaborators 2022, PMID 35026139).

Burden domain Evidence signal
Disability Recurrent depression, mania and interepisode symptoms contribute to functional loss (Merikangas 2011, PMID 21383262)
Mortality Elevated all-cause, suicide, natural-cause and cardiovascular mortality (Biazus 2023, PMID 37491460)
Diagnostic instability Polarity and competing diagnoses may change longitudinally (Cegla-Schvartzman 2019, PMID 30358663)
Treatment complexity Acute mania, acute depression and maintenance require different evidence hierarchies (Keramatian 2023, PMID 38695002)

Mechanism sketch

Bipolar disorder is strongly familial and highly polygenic. Common and rare variation overlaps partly with schizophrenia and major depression, but current genetic findings are probabilistic rather than diagnostic (Gordovez 2020, PMID 31907381).

Neurobiological models implicate distributed prefrontal–limbic circuits, circadian and sleep regulation, synaptic plasticity, calcium signaling, mitochondrial function and immune/metabolic pathways. None is sufficiently specific to define an individual case (Harrison 2018, PMID 29169634).

The episodic structure matters mechanistically: sleep loss, reward sensitivity and rhythm disruption can both reflect and amplify transitions. Longitudinal sampling is therefore more informative than a single blood draw or scan, motivating digital phenotyping; current studies cannot yet reliably discriminate unipolar from bipolar depression in routine care (Zhong 2025, PMID 40408762).

Treatment landscape

Phase-specific decision map

Clinical phase Evidence question Main outcome domain Recurring limitation
Acute mania How quickly does treatment reduce mania without intolerable sedation, neurological or metabolic harm? Symptom response, remission, discontinuation Short trials incompletely represent recovery and subsequent depression (Kishi 2022, PMID 34642461)
Acute bipolar depression Which intervention produces clinically meaningful depressive response without mania/hypomania? Response, remission, treatment-emergent switch Trial populations and endpoint timing differ across agents (Yildiz 2023, PMID 37595997)
Maintenance Which treatment prevents manic and depressive recurrence over years? Time to any, manic or depressive recurrence Enriched designs may favor the drug tolerated during stabilization (Kishi 2021, PMID 33177610)
Longitudinal recovery Which combination preserves function, relationships and physical health? Function, quality of life, mortality Symptom scales dominate; rare outcomes and multimorbidity require much larger cohorts (Biazus 2023, PMID 37491460)

Acute mania

Antipsychotics, lithium and valproate are central options, often combined when severity is high. Network meta-analysis confirms multiple effective agents but shows material tolerability differences, so speed, prior response, metabolic risk, reproductive context and psychosis shape choice (Kishi 2022, PMID 34642461). CANMAT/ISBD ranks treatments by efficacy, safety and evidence, with electroconvulsive therapy available for severe or refractory states (Keramatian 2023, PMID 38695002).

Bipolar depression

  • Quetiapine monotherapy improved bipolar I and II depression in BOLDER II (Thase 2006, PMID 17110817).
  • Olanzapine–fluoxetine outperformed placebo and olanzapine monotherapy in bipolar I depression, with metabolic tradeoffs (Tohen 2003, PMID 14609883).
  • Lurasidone monotherapy and cariprazine have positive randomized trials (Loebel 2014, PMID 24170180; Earley 2019, PMID 30845817).
  • Network meta-analysis compares these and other acute interventions but cannot eliminate differences in trial populations and endpoint timing (Yildiz 2023, PMID 37595997).

Antidepressant controversy. In STEP-BD, adding a standard antidepressant to a mood stabilizer did not increase durable recovery or switch rates versus placebo (Sachs 2007, PMID 17392295). A randomized STEP-BD analysis found antidepressant continuation worsened course in rapid-cycling patients (El-Mallakh 2015, PMID 26142612). This does not prove uniform harm, but it argues against reflexive antidepressant monotherapy and for polarity- and course-specific use.

Maintenance

Lithium prevents relapse and remains a benchmark, while lamotrigine is more depression-polarity oriented. An 18-month trial compared lamotrigine and lithium with placebo after mania/hypomania (Bowden 2003, PMID 12695317), and systematic reviews support lamotrigine maintenance efficacy with limitations in acute mania (Hashimoto 2021, PMID 34523118).

Network meta-analysis supports several mood stabilizers and antipsychotics for maintenance but highlights differing efficacy, metabolic, neurological and renal/thyroid profiles (Kishi 2021, PMID 33177610). Lithium's suicide-preventive effect is clinically influential, yet meta-analyses of randomized trials differ in certainty because suicide events are rare (Cipriani 2013, PMID 23814104; Nabi 2022, PMID 36111461).

Psychotherapy and monitoring

Adjunctive psychoeducation, family and cognitive-behavioral components improve relapse-related outcomes when added to pharmacotherapy, although effects differ by component and study design (Miklowitz 2021, PMID 33052390). Sleep/rhythm stabilization, relapse-signature recognition, substance-use treatment and physical-health monitoring are core longitudinal tasks.

Safety and special populations

  • Lithium requires serum-level, renal, thyroid and interaction monitoring; toxicity can emerge with dehydration or interacting medicines.
  • Antipsychotics require metabolic and movement-disorder surveillance; valproate has major reproductive safety constraints.
  • Pregnancy and postpartum decisions require balancing fetal/neonatal exposure against high relapse risk; abrupt discontinuation can be dangerous.
  • Suicide risk and cardiovascular/metabolic disease contribute materially to the mortality gap (Biazus 2023, PMID 37491460).

Research frontier

  • Prospective identification of bipolarity before antidepressant exposure or recurrent episodes.
  • Treatment-response prediction for lithium and polarity-specific maintenance.
  • Biomarker replication: metabolomic and late-life reviews find heterogeneous candidate signals but no clinically validated test (Chen 2024, PMID 38218254; Chancel 2024, PMID 38470559).
  • Digital measures of sleep, activity, speech and communication that predict relapse without unacceptable privacy costs (Zhong 2025, PMID 40408762).
  • Reducing excess natural-cause mortality as well as suicide (Biazus 2023, PMID 37491460).

Symptom control is not functional recovery

The burden map must distinguish symptom remission from cognitive, occupational and social recovery. An early systematic review found 65 health-related-quality-of-life studies, 14 work-impairment studies and 28 utilization/cost studies; inpatient care was the largest cost component, while bipolar-disorder quality of life was similar to unipolar depression and no better than other chronic illnesses (Dean 2004, PMID 15006007). The age of that evidence base is itself a limitation: it establishes the breadth of burden, not a current global cost estimate.

Cognitive remediation illustrates the outcome gap. Across seven independent randomized trials (n=586), it produced small improvements in working memory (Hedges g=0.32, 95% CI 0.11–0.52), planning (g=0.30, 0.03–0.56) and verbal learning (g=0.40, 0.15–0.66), but no significant functional benefit at treatment completion or follow-up (Samamé 2023, PMID 37485698). A separate longitudinal meta-analysis of 22 reports found no average cognitive decline over mean 1.5-year or 5.5-year follow-up in established bipolar disorder, arguing against inevitable progressive cognitive deterioration even though stable deficits remain clinically important (Bora 2017, PMID 28585513). The unresolved contradiction is therefore not “impairment versus no impairment,” but substantial impairment without consistent longitudinal decline or reliable functional response to cognitive training.

Prediction and early recognition remain research tools

Only three bipolar risk-calculator development studies met a 2020 systematic review: one predicted conversion from major depression, one new-onset bipolar-spectrum disorder in high-risk offspring, and one conversion in bipolar disorder not otherwise specified. Discrimination was described as good, but independent replication and clinical-impact validation were missing (Silva Ribeiro 2020, PMID 32781733). Pediatric misdiagnosis evidence is also narrow: a 15-study review found ADHD, schizophrenia and major depressive disorder were the most frequent alternative labels, while calling for quantitative work on false-negative consequences (Afzal 2024, PMID 38109022). These findings support longitudinal assessment; they do not justify population screening or automated diagnosis.

Open questions

  • Which depressed patients will later manifest mania or hypomania, and can prediction improve without overdiagnosis? (Nierenberg 2023, PMID 37815563; Cegla-Schvartzman 2019, PMID 30358663)
  • Which subgroups benefit from antidepressants without cycle acceleration? (Sachs 2007, PMID 17392295; El-Mallakh 2015, PMID 26142612)
  • Is lithium's antisuicide effect larger than randomized-event data can reliably estimate? (Cipriani 2013, PMID 23814104; Nabi 2022, PMID 36111461)
  • Can omics or digital phenotyping yield a replicable treatment selector? (Chen 2024, PMID 38218254; Zhong 2025, PMID 40408762)
  • Which intervention most reduces excess cardiovascular and natural-cause mortality? (Biazus 2023, PMID 37491460)

References

  1. Nierenberg AA, et al. Diagnosis and treatment of bipolar disorder: a review. JAMA. 2023. PMID 37815563
  2. Merikangas KR, et al. Prevalence and correlates of bipolar spectrum disorder in the World Mental Health Survey Initiative. Arch Gen Psychiatry. 2011. PMID 21383262
  3. GBD 2019 Mental Disorders Collaborators. Global burden of 12 mental disorders, 1990–2019. Lancet Psychiatry. 2022. PMID 35026139
  4. Cegla-Schvartzman FB, et al. Diagnostic stability in bipolar disorder. Harv Rev Psychiatry. 2019. PMID 30358663
  5. Gordovez FJA, et al. The genetics of bipolar disorder. Mol Psychiatry. 2020. PMID 31907381
  6. Harrison PJ, et al. The emerging neurobiology of bipolar disorder. Trends Neurosci. 2018. PMID 29169634
  7. Keramatian K, et al. CANMAT and ISBD bipolar guidelines: 2023 evidence update. Focus. 2023. PMID 38695002
  8. Kishi T, et al. Pharmacological treatment for bipolar mania: network meta-analysis. Mol Psychiatry. 2022. PMID 34642461
  9. Thase ME, et al. Quetiapine monotherapy in bipolar I and II depression: BOLDER II. J Clin Psychopharmacol. 2006. PMID 17110817
  10. Tohen M, et al. Olanzapine and olanzapine–fluoxetine in bipolar I depression. Arch Gen Psychiatry. 2003. PMID 14609883
  11. Loebel A, et al. Lurasidone monotherapy in bipolar I depression. Am J Psychiatry. 2014. PMID 24170180
  12. Earley W, et al. Cariprazine for bipolar depression. Am J Psychiatry. 2019. PMID 30845817
  13. Yildiz A, et al. Acute bipolar depression pharmacotherapy network meta-analysis. Lancet Psychiatry. 2023. PMID 37595997
  14. Sachs GS, et al. Adjunctive antidepressant treatment for bipolar depression. N Engl J Med. 2007. PMID 17392295
  15. El-Mallakh RS, et al. Antidepressants worsen rapid-cycling course in bipolar depression. J Affect Disord. 2015. PMID 26142612
  16. Bowden CL, et al. Lamotrigine and lithium maintenance after mania/hypomania. Arch Gen Psychiatry. 2003. PMID 12695317
  17. Hashimoto Y, et al. Lamotrigine in maintenance treatment of bipolar disorder. Cochrane Database Syst Rev. 2021. PMID 34523118
  18. Kishi T, et al. Maintenance pharmacotherapy network meta-analysis. Mol Psychiatry. 2021. PMID 33177610
  19. Cipriani A, et al. Lithium in prevention of suicide in mood disorders. BMJ. 2013. PMID 23814104
  20. Nabi Z, et al. Lithium and suicide: randomized-trial meta-analysis. Epidemiol Psychiatr Sci. 2022. PMID 36111461
  21. Miklowitz DJ, et al. Adjunctive psychotherapy for bipolar disorder. JAMA Psychiatry. 2021. PMID 33052390
  22. Biazus TB, et al. All-cause and cause-specific mortality in bipolar disorder. Mol Psychiatry. 2023. PMID 37491460
  23. Chen J, et al. Metabolomic biomarkers of bipolar disorder. J Affect Disord. 2024. PMID 38218254
  24. Chancel R, et al. Biomarkers of bipolar disorder in late life. Curr Psychiatry Rep. 2024. PMID 38470559
  25. Zhong R, et al. Digital phenotyping to discriminate unipolar and bipolar disorder. J Med Internet Res. 2025. PMID 40408762
  26. Dean BB, et al. A systematic review evaluating health-related quality of life, work impairment, and healthcare costs and utilization in bipolar disorder. Curr Med Res Opin. 2004;20:139–154. PMID 15006007
  27. Samamé C, et al. Efficacy of cognitive remediation in bipolar disorder: systematic review and meta-analysis of randomized controlled trials. Psychol Med. 2023;53:5361–5373. PMID 37485698
  28. Silva Ribeiro J, et al. Risk calculators in bipolar disorder: a systematic review. Brain Sci. 2020;10:525. PMID 32781733
  29. Afzal T, et al. A systematic review of misdiagnosis of pediatric bipolar disorder: assessments and recommendations. Res Child Adolesc Psychopathol. 2024;52:659–670. PMID 38109022
  30. Bora E, Özerdem A. Meta-analysis of longitudinal studies of cognition in bipolar disorder: comparison with healthy controls and schizophrenia. Psychol Med. 2017;47:2753–2766. PMID 28585513