Skip to content

Mania and mixed states

TL;DR — Acute mania is treated on a days-to-weeks timescale, usually with lithium, valproate or an antipsychotic; severe agitation, psychosis and immediate danger make speed, containment and tolerability as important as mean symptom change. In a 72-trial network meta-analysis, 13 medicines beat placebo for response, but only aripiprazole, olanzapine, quetiapine and risperidone also reduced all-cause discontinuation (Kishi 2022, PMID 34642461). Older comparative evidence ranked haloperidol, risperidone and olanzapine highly for efficacy, while also exposing meaningful movement, sedation and metabolic trade-offs (Cipriani 2011, PMID 21851976). Mixed presentations are not simply “mania plus depression”: prospective treatment trials are scarce, most estimates are post-hoc, and antidepressant exposure requires particular caution (Takeshima 2017, PMID 28004626; Xiao 2025, PMID 40808264). The evidence base is short-term, selectively enrolled and weak on coercion, functioning, patient-defined recovery and long-term consequences of acute drug choice.

What the acute evidence actually measures

Most antimanic trials last about three to four weeks and use change on the Young Mania Rating Scale (YMRS), response (usually at least 50% improvement), remission thresholds and discontinuation. The largest recent synthesis included 72 double-blind RCTs, 16,442 participants and 23 drugs; mean duration was 3.96 weeks (Kishi 2022, PMID 34642461). Thus, “effective” in this literature usually means reduced clinician-rated mania over several weeks, not restored judgment, housing, employment or relationships.

Mixed-state evidence has an additional classification problem. DSM-5 replaced the narrow categorical mixed episode with a dimensional mixed-features specifier, so studies may contain DSM-IV mixed episodes, mania with depressive symptoms, bipolar depression with hypomanic symptoms, or broader research definitions (Takeshima 2017, PMID 28004626; Shim 2018, PMID 30466209). These populations should not be pooled without naming the definition.

Monotherapy scoreboard

Intervention Placebo-controlled acute efficacy Acceptability / harms visible in abstracts Interpretation
Haloperidol SMD −0.56 (95% CI −0.69 to −0.43) in the 2011 network; YMRS WMD −5.85 (−7.69 to −4.00) as monotherapy Tremor RR 3.01 (1.55–5.84); less weight gain than olanzapine, RR 0.28 (0.12–0.67) High efficacy, movement-disorder cost (Cipriani 2011, PMID 21851976; Cipriani 2006, PMID 16856043)
Risperidone SMD −0.50 (−0.63 to −0.38); severe-mania RCT separated from placebo by week 1 EPS was the most frequent adverse-event category Fast, effective option; quantify EPS in individual selection (Cipriani 2011, PMID 21851976; Khanna 2005, PMID 16135859)
Olanzapine SMD −0.43 (−0.54 to −0.32); response 48.6% vs 24.2% placebo in a 3-week RCT Somnolence, dizziness, dry mouth and weight gain more frequent Strong acute signal with metabolic/sedative liability (Cipriani 2011, PMID 21851976; Tohen 1999, PMID 10327902)
Lithium SMD −0.37 (−0.63 to −0.11) More discontinuation than olanzapine, risperidone or quetiapine in the older network Proven antimanic efficacy, slower practical titration and monitoring burden (Cipriani 2011, PMID 21851976)
Quetiapine SMD −0.37 (−0.51 to −0.23) Fewer discontinuations than lithium, lamotrigine, placebo, topiramate or gabapentin Balances efficacy and trial acceptability; sedation remains clinically relevant (Cipriani 2011, PMID 21851976)
Aripiprazole SMD −0.37 (−0.51 to −0.23) Pooled acute review found small but significant EPS increase Effective; akathisia can undermine apparent “activation” or adherence (Cipriani 2011, PMID 21851976; Smith 2007, PMID 17845269)
Carbamazepine SMD −0.36 (−0.60 to −0.11) Withdrawal excess could not be excluded in older pooled evidence Effective but interaction/monitoring burden limits convenience (Cipriani 2011, PMID 21851976; Smith 2007, PMID 17845269)
Asenapine SMD −0.30 (−0.53 to −0.07) Sublingual administration; review reported generally mild–moderate AEs Effective, but comparative precision is weaker (Cipriani 2011, PMID 21851976; Kishi 2022, PMID 34642461)
Valproate SMD −0.20 (−0.37 to −0.04) Mood-stabilizer pooled response RR 2.01 (1.66–2.43) vs placebo Established option; acute mean effect smaller in the older network (Cipriani 2011, PMID 21851976; Smith 2007, PMID 17845269)
Ziprasidone SMD −0.20 (−0.37 to −0.03) Discontinuation due to inefficacy lower than placebo in newer network Effective, with drug-specific cardiac/administration issues outside these abstracts (Cipriani 2011, PMID 21851976; Kishi 2022, PMID 34642461)
Cariprazine Response 58.9% vs 44.1%; remission 51.9% vs 34.9%; separation by day 4 Akathisia, EPS, tremor, dyspepsia and vomiting; small mean metabolic change Direct phase III support, short duration (Sachs 2015, PMID 25532076)
Gabapentin, lamotrigine, topiramate Not significantly better than placebo in the 2011 network Topiramate had higher all-cause discontinuation in the 2022 network Negative acute-mania evidence should not be obscured by use in other phases (Cipriani 2011, PMID 21851976; Kishi 2022, PMID 34642461)

The 2007 placebo-controlled meta-analysis found broadly similar pooled response magnitudes for antipsychotics, RR 1.74 (95% CI 1.54–1.96), and mood stabilizers, RR 2.01 (1.66–2.43), but these class averages conceal different adverse effects and trial populations (Smith 2007, PMID 17845269). The newer network supports response efficacy for aripiprazole, asenapine, carbamazepine, cariprazine, haloperidol, lithium, olanzapine, paliperidone, quetiapine, risperidone, tamoxifen, valproate and ziprasidone; only four agents also beat placebo on all-cause discontinuation (Kishi 2022, PMID 34642461).

Response is not remission

Acute mania papers often use several endpoints that answer different questions:

Endpoint Typical operational meaning What it misses
Mean YMRS change Average symptom-scale movement Averages responders and nonresponders; may reward sedation
Response Usually at least 50% YMRS reduction A person can respond while remaining substantially manic
Remission YMRS below a specified cutoff Does not require restored functioning or insight
All-cause discontinuation Stayed in the assigned arm Mixes efficacy, adverse effects, preference and logistics
Discontinuation for inefficacy Left because symptoms did not improve adequately Investigator attribution can vary

Cariprazine illustrates the distinction: 58.9% met response but 51.9% met the study's remission threshold, versus 44.1% and 34.9% on placebo (Sachs 2015, PMID 25532076). Olanzapine's pivotal trial reported response in 48.6% versus 24.2%, but the abstract did not establish functional recovery (Tohen 1999, PMID 10327902). A page or guideline that reports only “effective” without the endpoint hides this clinically important denominator.

Combination treatment

Combination treatment is most defensible when illness severity, psychosis, aggression, poor sleep or insufficient early response makes monotherapy inadequate. Placebo-controlled adjunctive evidence supports adding olanzapine, risperidone, quetiapine or haloperidol to lithium or valproate, and supports adjunctive valproate in selected designs (Sachs 2007, PMID 17688460).

Question Evidence Boundary
Does adding an antipsychotic improve acute control? Adjunctive haloperidol reduced YMRS by WMD −5.20 (95% CI −9.26 to −1.14) vs adjunctive placebo (Cipriani 2006, PMID 16856043) Short trials; sedation may contribute to early rating improvement
Which combinations have positive placebo-controlled evidence? Lithium/valproate plus olanzapine, risperidone, haloperidol or quetiapine (Sachs 2007, PMID 17688460) The review found no placebo-controlled studies for several commonly used combinations
Which adjuncts had negative or failed studies? Carbamazepine, gabapentin, lamotrigine, topiramate, oxcarbazepine and ziprasidone (Sachs 2007, PMID 17688460) “Failed” may include design or power failure, not proof of zero effect
Is combination always superior to clean monotherapy? Evidence did not conclusively answer this in patients naïve to both treatments (Sachs 2007, PMID 17688460) Run-in and inadequate-response enrichment restrict generalizability

Combination therapy raises total adverse-effect load and makes attribution harder. The clinical comparison should therefore be symptom control gained per additional burden, not simply whether two active drugs lower a scale more than one.

Agitation, psychosis and safety

Severe mania can be life-threatening and can impair capacity, judgment, sleep and behavioral control; the risperidone placebo-controlled trial specifically enrolled inpatients with mean baseline YMRS 37.2 and found separation from placebo by week 1 (Khanna 2005, PMID 16135859). Trial populations nevertheless under-represent the most medically unstable, intoxicated, delirious or immediately violent presentations.

The acute evidence does not reduce safety care to pharmacology. A defensible research-informed pathway distinguishes:

Domain Immediate research/clinical question Why the drug trial does not answer it
Medical causes Is the syndrome secondary to substances, medication, delirium, endocrine or neurologic disease? Efficacy trials generally enroll diagnostically established bipolar disorder
Suicide/violence Is there current intent, access to means, severe impulsivity or inability to maintain safety? Rating-scale response is not a validated substitute for imminent-risk assessment
Capacity Can the person understand, retain, weigh and communicate treatment choices? Capacity is jurisdiction- and decision-specific and rarely an RCT endpoint
Nutrition/sleep Is prolonged wakefulness, dehydration or inability to eat driving medical risk? Acute antimanic trials seldom report these as primary outcomes
Pregnancy Does reproductive risk change the relative acceptability of valproate, lithium or antipsychotics? General adult mania trials do not establish perinatal safety

These are evidence boundaries, not reasons to withhold urgent care. The red-flags and safety, comorbidity and differential diagnosis, and pregnancy pages hold the condition-wide safety context.

Mixed presentations

Mixed presentations predict complexity, but the evidence is thinner than the treatment confidence often implies. A review found only two trials prospectively designed around mixed mania/hypomania; most evidence came from post-hoc subgroups of acute manic or mixed-episode RCTs (Takeshima 2017, PMID 28004626). Another systematic review found just 11 eligible reports for major depressive episodes with mixed features, with much of the antipsychotic evidence post-hoc (Shim 2018, PMID 30466209).

Evidence slice Finding Confidence constraint
Mixed mania/hypomania review Strongest acute signals for aripiprazole, asenapine, carbamazepine, olanzapine and ziprasidone; quetiapine and valproate also supported Mostly post-hoc subgroup evidence (Takeshima 2017, PMID 28004626)
SGA meta-analysis in bipolar depression with mixed features Manic-symptom SMD −0.74 (95% CI −1.20 to −0.28); MADRS SMD −1.08 (−1.35 to −0.81) Publication/measurement bias and heterogeneous definitions (Fornaro 2016, PMID 26891297)
Updated mixed-depression review Lurasidone, olanzapine, cariprazine, lumateperone, quetiapine and ziprasidone associated with depressive improvement Only lumateperone used mixed features as a prespecified primary outcome across relevant diagnoses (Xiao 2025, PMID 40808264)
Switching No significant treatment–placebo difference in hypomanic symptom intensification in the small modern mixed-feature evidence set Studies were 6–8 weeks; long-term cycle destabilization remains uncertain (Xiao 2025, PMID 40808264)
Lithium/valproate Minimal direct evidence for depressive episodes with mixed features Absence of purpose-designed evidence is not evidence of ineffectiveness (Xiao 2025, PMID 40808264)

Mixed symptoms should be counted and followed rather than treated as a binary label. The antidepressant question becomes more hazardous when activation, irritability, agitation, reduced sleep or racing thoughts coexist with depression; STEP-BD observational evidence found higher follow-up mania severity without faster recovery when antidepressants were used in bipolar depression with concurrent manic symptoms (Goldberg 2007, PMID 17728419).

Rapid cycling and episode burden

Rapid cycling is commonly defined as at least four recurrences in a year, but studies vary in whether they require this pattern in the past year or any year. A meta-review estimated one-year prevalence at 22.3% (95% CI 14.4–32.9) and lifetime prevalence at 35.5% (27.6–44.3), with high heterogeneity; associations with suicide attempts and unsatisfactory mood-stabilizer response had the strongest evidence (Miola 2023, PMID 37429185).

Rapid cycling is therefore both a prognostic marker and a source of treatment-effect heterogeneity. It should not be inferred from ordinary mood lability, nor should one negative average trial result be assumed to apply identically to high-cycle-burden patients.

Reading efficacy against harms

Drug property Acute advantage Acute or downstream cost
Strong dopamine blockade Rapid reduction in mania/psychosis EPS, akathisia, subjective dysphoria; haloperidol tremor RR 3.01 (Cipriani 2006, PMID 16856043)
Sedation Sleep restoration and behavioral containment Falls, impaired participation, difficulty separating sedation from core recovery
Appetite/weight effect Usually no acute therapeutic advantage Olanzapine weight gain appeared within a 3-week RCT (Tohen 1999, PMID 10327902)
Longer-term polarity coverage May simplify transition into maintenance Acute ranking does not establish prevention efficacy; see maintenance
Simple administration Supports adherence in disorganized states Sublingual, food-dependent or titration requirements can complicate real use

The best acute drug is not automatically the best maintenance drug. Once safety and meaningful stabilization are achieved, continuation should be re-justified using polarity-specific relapse evidence and patient-important harms.

Evidence limitations

  • Average trial duration is under four weeks, while weight, metabolic, renal, endocrine and functional consequences unfold over months to years (Kishi 2022, PMID 34642461).
  • Network rankings assume enough comparability across trials for indirect comparisons; differences in baseline severity, rescue medication and outcome handling can violate that assumption (Cipriani 2011, PMID 21851976).
  • Mixed-feature evidence is dominated by post-hoc analyses and inconsistent definitions (Takeshima 2017, PMID 28004626; Shim 2018, PMID 30466209).
  • Acceptability is usually all-cause discontinuation, a composite that cannot tell whether a participant stopped for inefficacy, adverse effects, rapid improvement or external reasons.
  • Functional recovery, therapeutic alliance, coercive-care exposure and patient-valued outcomes are sparsely measured.

What the evidence changes in younger patients

Youth evidence is not a miniature copy of adult evidence. A network meta-analysis of 18 randomized trials (n=2,844; mean age 11.74 years; mean duration 5.4 weeks) found all six tested second-generation antipsychotics reduced manic symptoms versus placebo. Risperidone had the largest estimate (SMD −1.18, 95% CI −1.45 to −0.92), while pooled antipsychotics outperformed mood stabilizers (SMD −0.61, −0.82 to −0.40); sedation, weight gain and metabolic effects moved in the opposite direction (Vita 2025, PMID 39128561). Lithium response barely crossed the null (RR 1.35, 1.00–1.83; very-low confidence), so this analysis should not be read as proof of lithium inefficacy in youth.

Valproate and mixed-state uncertainty

An overview of 26 systematic reviews found valproate superior to placebo for acute-mania response (RR 1.42, 95% CI 1.19–1.71 in one high-quality synthesis; OR 2.05, 1.32–3.20 in another), broadly similar to lithium, similar to quetiapine and less effective than risperidone (Mari 2024, PMID 39500601). The apparent precision comes from overlapping reviews of 31 trials rather than 31 independent new trials.

Mixed-state treatment remains much less certain. A systematic review found limited randomized evidence for olanzapine, aripiprazole, asenapine, valproate and carbamazepine, while noting that nearly all mixed-state trials were industry funded and DSM-5 lowered the symptom threshold relative to DSM-IV mixed episodes (Betzler 2017, PMID 28417647). Older broad treatment synthesis supports antipsychotics, lithium, valproate and carbamazepine in mania but also shows that combination efficacy must be traded against greater adverse-effect burden (Fountoulakis 2008, PMID 18752718).

Treatment-resistant mania is not a standardized entity. In a 17-study review (n=928), evidence for resistant mania was too sparse for a stable pooled hierarchy; rapid versus standard clozapine titration was described as promising, whereas most analyzable data concerned resistant depression (Fornaro 2020, PMID 32750614). Failure of one adequate antimanic strategy therefore does not define a validated biological subtype.

Open questions

  • Can acute-treatment trials use time to restored sleep, judgment and functioning alongside YMRS response, without allowing sedation to masquerade as recovery (Kishi 2022, PMID 34642461)?
  • Which prospectively defined mixed-feature phenotypes respond to which agent, rather than inheriting estimates from post-hoc subgroups (Xiao 2025, PMID 40808264)?
  • Does early combination treatment improve durable outcomes enough to justify greater adverse-effect burden in treatment-naïve severe mania (Sachs 2007, PMID 17688460)?
  • Why do rapid-cycling presentations show poorer treatment response, and can treatment-by-cycle-burden interactions be estimated prospectively (Miola 2023, PMID 37429185)?
  • Can acute drug choice be optimized jointly for immediate control and the subsequent predominant relapse polarity (Cipriani 2011, PMID 21851976; Miura 2014, PMID 26360999)?

References

  1. Kishi T, et al. Pharmacological treatment for bipolar mania: a systematic review and network meta-analysis of double-blind randomized controlled trials. Molecular Psychiatry. 2022. PMID 34642461
  2. Cipriani A, et al. Comparative efficacy and acceptability of antimanic drugs in acute mania: a multiple-treatments meta-analysis. Lancet. 2011. PMID 21851976
  3. Smith LA, et al. Pharmacological interventions for acute bipolar mania: a systematic review of randomized placebo-controlled trials. Bipolar Disorders. 2007. PMID 17845269
  4. Cipriani A, et al. Haloperidol alone or in combination for acute mania. Cochrane Database of Systematic Reviews. 2006. PMID 16856043
  5. Tohen M, et al. Olanzapine versus placebo in the treatment of acute mania. American Journal of Psychiatry. 1999. PMID 10327902
  6. Khanna S, et al. Risperidone in the treatment of acute mania: double-blind, placebo-controlled study. British Journal of Psychiatry. 2005. PMID 16135859
  7. Sachs GS, et al. Cariprazine in the treatment of acute mania in bipolar I disorder: a double-blind, placebo-controlled, phase III trial. Journal of Affective Disorders. 2015. PMID 25532076
  8. Sachs GS, et al. Adjunctive treatment of acute mania: a clinical overview. Acta Psychiatrica Scandinavica Supplementum. 2007. PMID 17688460
  9. Takeshima M, et al. Treating mixed mania/hypomania: a review and synthesis of the evidence. CNS Spectrums. 2017. PMID 28004626
  10. Fornaro M, et al. Atypical antipsychotics in the treatment of acute bipolar depression with mixed features: a systematic review and exploratory meta-analysis. International Journal of Molecular Sciences. 2016. PMID 26891297
  11. Shim IH, et al. Pharmacological treatment of major depressive episodes with mixed features: a systematic review. Clinical Psychopharmacology and Neuroscience. 2018. PMID 30466209
  12. Xiao N, et al. The efficacy of pharmacological interventions in the treatment of major depressive disorder and bipolar depression with mixed features: a systematic review. Bipolar Disorders. 2025. PMID 40808264
  13. Miola A, et al. Prevalence and outcomes of rapid cycling bipolar disorder: mixed method systematic meta-review. Journal of Psychiatric Research. 2023. PMID 37429185
  14. Goldberg JF, et al. Adjunctive antidepressant use and symptomatic recovery among bipolar depressed patients with concomitant manic symptoms: findings from STEP-BD. American Journal of Psychiatry. 2007. PMID 17728419
  15. Miura T, et al. Comparative efficacy and tolerability of pharmacological treatments in the maintenance treatment of bipolar disorder: a systematic review and network meta-analysis. Lancet Psychiatry. 2014. PMID 26360999
  16. Vita G, et al. Efficacy and safety of antipsychotics versus antiepileptics or lithium for acute mania in children and adolescents: a systematic review and network meta-analysis. J Am Acad Child Adolesc Psychiatry. 2025;64:143–157. PMID 39128561
  17. Mari J, et al. The efficacy of valproate in acute mania, bipolar depression and maintenance therapy for bipolar disorder: an overview of systematic reviews with meta-analyses. BMJ Open. 2024;14:e087999. PMID 39500601
  18. Fornaro M, et al. The concept and management of acute episodes of treatment-resistant bipolar disorder: a systematic review and exploratory meta-analysis. J Affect Disord. 2020;276:970–983. PMID 32750614
  19. Fountoulakis KN, et al. Treatment of bipolar disorder: a systematic review of available data and clinical perspectives. Int J Neuropsychopharmacol. 2008;11:999–1029. PMID 18752718
  20. Betzler F, et al. Mixed states in bipolar disorder—changes in DSM-5 and current treatment recommendations. Int J Psychiatry Clin Pract. 2017;21:244–258. PMID 28417647