Skip to content

Red flags and safety concerns

TL;DR — Bipolar-disorder emergencies are defined by danger and loss of safe functioning, not by a mood-score threshold: suicidal intent or behavior, severe mania or depression, psychosis, escalating agitation, inability to care for basic needs, intoxication/withdrawal, and suspected medication toxicity require urgent assessment (Yatham 2018, PMID 29536616; Izadi 2023, PMID 38695007). Suicide mortality is approximately 11.7 times the general-population rate in a 57-study meta-analysis, while all-cause mortality is doubled; safety therefore includes both crisis response and sustained physical-health care (Biazus 2023, PMID 37491460). Lithium toxicity may present with gastrointestinal, neurologic, renal or cardiac features and can worsen despite a “therapeutic” prescribed dose when dehydration, kidney injury or interacting medicines alter clearance (McKnight 2012, PMID 22265699; Sun 2018, PMID 29189921). Abrupt treatment changes can precipitate relapse or withdrawal effects; medication should not be stopped reflexively outside an emergency. This page is a research synthesis, not individualized emergency advice.

Immediate safety red flags

Red flag Why it matters Assessment focus
Current suicidal intent, plan, preparation, recent attempt or inability to maintain safety Suicide risk is markedly elevated in bipolar disorder (Plans 2019, PMID 30173059; Biazus 2023, PMID 37491460) Intent, access to lethal means, recent behavior, intoxication, supports and safe setting
Severe mania with rapidly decreasing sleep, escalating impulsivity or dangerous behavior Risk can rise through impaired judgment, psychosis, spending, driving, sexual behavior or conflict Speed of change, sleep, insight, psychosis, substances, safeguarding and capacity
Psychosis, severe disorganization or catatonic features May prevent self-care and signal severe mood episode or alternative medical cause Mood congruence, delirium, neurologic/medical causes, medicines and substances
Threats or violence, severe agitation, inability to de-escalate Requires a least-restrictive but rapid safety response (Zeller 2016, PMID 26973742) Environment, triggers, weapons, intoxication, delirium and immediate danger
Inability to eat, drink, sleep or perform basic self-care Can produce medical deterioration and indicates functional severity Hydration, nutrition, vital signs, physical illness and available supervision
New confusion, ataxia, coarse tremor, persistent vomiting/diarrhea or reduced consciousness while taking lithium Compatible with lithium poisoning; severity is clinical, not concentration alone (Decker 2015, PMID 25583292) Serum lithium, renal function, electrolytes, ECG, timing/formulation and interacting exposures
Fever, severe rigidity, autonomic instability or altered consciousness after dopamine-blocking treatment Potential life-threatening drug reaction Urgent medical evaluation, differential diagnosis and recent dose changes
New rash during lamotrigine titration, especially with mucosal/systemic symptoms Potential severe cutaneous reaction Immediate clinical assessment; do not self-rechallenge
Pregnancy or pregnancy planning while taking a high-risk medicine Treatment interruption and exposure both carry risks (ACOG 2023, PMID 37486661) Urgent specialist risk-benefit review, not abrupt unsupervised cessation

Suicide and self-harm

The largest recent mortality synthesis included 57 studies and 678,353 people with bipolar disorder. All-cause mortality was elevated (RR 2.02, 95% CI 1.89-2.16), as were suicide mortality (RR 11.69, 95% CI 9.22-14.81), unnatural causes (RR 7.29, 95% CI 6.41-8.28) and natural causes (RR 1.90, 95% CI 1.75-2.06) (Biazus 2023, PMID 37491460). An earlier meta-analysis estimated suicide SMR 14.44 (95% CI 12.43-16.78) amid high heterogeneity (Hayes 2015, PMID 25735195).

A systematic review of completed suicide reported rates approximately 20-30 times the general population across included studies, while emphasizing limited and heterogeneous evidence (Plans 2019, PMID 30173059). The ISBD Task Force meta-analysis associated attempts with younger onset, depressive first/current polarity, anxiety, substance use, alcohol use, illicit-substance use, cluster B/borderline personality disorder and first-degree family history; suicide deaths were associated with male sex and first-degree family history. These are correlates, not deterministic scores (Schaffer 2015, PMID 25329791).

Safety principle Evidence boundary
Ask directly and repeatedly Risk changes with episode, substance use, stress and access; one historical “low-risk” assessment does not persist
Use diagnosis-specific correlates without false precision Meta-analytic correlates are heterogeneous and do not establish causal ranking (Schaffer 2015, PMID 25329791)
Treat the mood episode and the suicide crisis Medication alone is not a complete safety plan (Izadi 2023, PMID 38695007)
Restrict access to lethal means collaboratively A practical prevention layer independent of diagnostic formulation
Include follow-up after discharge or attempt Transition periods require active contact and continuity

Lithium has an important but nuanced suicide evidence base. A 48-RCT meta-analysis (6,674 participants) found fewer suicides with lithium than placebo (OR 0.13, 95% CI 0.03-0.66) and fewer all-cause deaths (OR 0.38, 95% CI 0.15-0.95), but deliberate self-harm was not clearly reduced (OR 0.60, 95% CI 0.27-1.32) (Cipriani 2013, PMID 23814104). Conversely, a 519-participant VA trial after recent suicide-related events stopped for futility; lithium augmentation did not reduce the composite outcome (HR 1.10, 95% CI 0.77-1.55) (Katz 2022, PMID 34787653). Population and trial design matter.

Mania, mixed states, psychosis and agitation

Rapidly reduced need for sleep, escalating speech/goal-directed activity, irritability, grandiosity, disinhibition and impaired judgment are concerning when they depart from baseline and accelerate. Mixed presentations can combine activation with depressive distress and suicidality; they should not be assumed “less severe” because euphoria is absent (Grunze 2018, PMID 29098925).

Emergency agitation guidance emphasizes assessment of cause, verbal de-escalation and route-appropriate medication while minimizing unnecessary coercion (Zeller 2016, PMID 26973742; Garriga 2016, PMID 26912127). Delirium, intoxication, withdrawal, endocrine disease, infection and neurologic illness can mimic or amplify mania; first-episode or atypical late-onset presentations warrant an active medical differential.

Lithium: prevention, chronic toxicity and poisoning

Long-term safety

The lithium toxicity-profile review screened 5,988 abstracts and included 385 studies. Urinary concentrating ability was reduced by a weighted mean 158.43 mOsm/kg (95% CI 87.07-229.78); clinical hypothyroidism was more common than with placebo (OR 5.78, 95% CI 2.00-16.67); calcium increased by 0.09 mmol/L (95% CI 0.02-0.17). Renal replacement therapy occurred in 18 of 3,369 patients (0.5%) in the relevant data (McKnight 2012, PMID 22265699).

Toxicity triggers

Trigger / context Mechanism or concern
Vomiting, diarrhea, fever or inadequate fluid intake Volume depletion and reduced lithium clearance
Acute kidney injury or chronic kidney-function decline Reduced clearance
Interacting medicines affecting renal perfusion or sodium handling Rising serum concentration despite unchanged lithium dose
Dose error, formulation change or intentional overdose Acute or acute-on-chronic poisoning
Older age, polypharmacy and medical comorbidity Lower physiologic reserve and atypical presentation

In a review of 38 published toxicity cases in adults aged at least 65, mean age was 71.4 years, mean toxic serum concentration 2.55 mmol/L, polypharmacy was present in 63.2% and medical comorbidity in 76.3%; case-report evidence is strongly selected and cannot estimate incidence (Sun 2018, PMID 29189921).

Extracorporeal treatment

The EXTRIP systematic review covered 166 articles and 418 patients, but evidence quality was very low because reports were mostly cases. It recommended extracorporeal treatment for severe poisoning, including impaired kidney function with lithium >4.0 mEq/L or decreased consciousness, seizures or life-threatening dysrhythmias regardless of concentration; it suggested treatment at >5.0 mEq/L, significant confusion, or expected time to <1.0 mEq/L exceeding 36 hours. Hemodialysis was preferred (Decker 2015, PMID 25583292). The 2015 Cochrane review identified no randomized trial, and a live PubMed re-search through 2026-08-30 retrieved case reports, cohorts and reviews but no randomized comparison establishing hemodialysis benefit. The positively stated evidence gap is therefore that current thresholds rest on toxicokinetics, case-level evidence and expert consensus rather than randomized efficacy evidence (Lavonas 2015, PMID 26374731).

Other medication safety

Exposure Red flag Prevention / monitoring logic
Antipsychotics Hyperglycemia, severe rigidity/fever, movement disorder, marked sedation, orthostasis Baseline and follow-up metabolic monitoring; urgent assessment for severe systemic reaction
Valproate/divalproex Pregnancy exposure, hepatic dysfunction, bleeding/bruising, pancreatitis symptoms Reproductive-risk controls and hepatic/hematologic monitoring (ACOG 2023, PMID 37486661)
Lamotrigine Rash during titration, mucosal lesions, systemic symptoms Slow titration and urgent assessment of rash
Antidepressants Emerging reduced sleep need, activation, agitation, mixed symptoms or mania Avoid monotherapy in bipolar I; monitor after starts and dose changes (Goldberg 2003, PMID 14636364)
Sedatives / polypharmacy Falls, cognitive impairment, respiratory depression, driving risk Review cumulative burden and substances
Any maintenance medicine Abrupt cessation Relapse and withdrawal risk; planned taper with monitoring when clinically possible (Kishi 2021, PMID 33177610)

Substance use and compounding risk

Substance use can worsen diagnostic uncertainty, adherence, impulsivity and acute risk. Cannabis use disorder has a statistically significant but small association with suicide attempts in bipolar disorder in meta-analysis; association does not establish causation and residual confounding is likely (Bartoli 2019, PMID 31121199). A systematic review also examined cigarette smoking as a suicide-risk correlate, reinforcing the need to assess tobacco and other substances rather than treating them as peripheral lifestyle variables (García-Jiménez 2023, PMID 37275988).

Physical-health red flags

Safety extends beyond psychiatric crisis. In the 2023 meta-analysis, cardiovascular mortality RR was 1.76 (95% CI 1.53-2.01), respiratory mortality RR 3.18 (95% CI 2.55-3.96), and cerebrovascular mortality RR 1.57 (95% CI 1.34-1.84) (Biazus 2023, PMID 37491460). A Finnish cohort of 53,273 people followed for 428,426 person-years recorded 5,988 deaths; 26% were cardiovascular. Relative excess was greatest for cardiomegaly (SMR 4.51, 95% CI 3.58-5.43), venous thromboembolism (3.03, 2.26-3.81) and cardiomyopathy (2.46, 1.95-2.97) (Paljärvi 2024, PMID 38826056).

Annual physical-health checks are a minimum framework, not a ceiling. New chest pain, focal neurologic symptoms, severe dyspnea, syncope, hyperglycemic symptoms or rapid systemic deterioration require condition-appropriate urgent care, not attribution to anxiety or bipolar disorder.

Digital monitoring and safety

Passive monitoring may detect sleep/activity change, but false reassurance and false alarms are both hazards. Reviews identify privacy, data security, legal and patient-safety concerns, while randomized evidence has not shown that self-monitoring alone reduces depressive or manic symptom severity (Faurholt-Jepsen 2018, PMID 29510813). Any alert system needs a named recipient, operating hours, response threshold, failure mode and explicit statement that it is not continuous emergency monitoring.

Family and carer roles

Carers may observe reduced sleep, escalating behavior or loss of insight before services do, but responsibility should not be offloaded onto them. Qualitative interviews identify carer-health effects, unpredictability, silencing and need for tailored support (Speirs 2023, PMID 36656805). Prospective plans should specify what information may be shared, what carers can report, and when immediate safety overrides ordinary confidentiality boundaries under local law.

Safety handoff checklist

Element Minimum content
Current state Polarity, mixed symptoms, psychosis, agitation, sleep and self-care
Suicide risk Recent thoughts/behavior, intent, means access and protective supports
Medicines Exact names, doses, last doses, recent changes and adherence
Toxicity Symptoms, serum concentrations when relevant, renal/hepatic function and interactions
Substances Alcohol, cannabis, stimulants, sedatives and withdrawal risk
Physical health Vital signs, pregnancy context, metabolic/cardiovascular issues and acute medical differential
Plan Observation setting, named responsible clinician/team, reassessment time and escalation threshold

Incomplete handoffs are especially hazardous when behavioral change is attributed to psychiatric illness before delirium, intoxication, withdrawal or toxicity has been assessed.

The safety frame must extend beyond the immediate crisis. Reviews synthesize the high suicide burden and its dynamic clinical correlates, while large-scale cardiovascular evidence shows that severe mental illness carries excess cardiovascular incidence and mortality requiring active prevention rather than deferral to later care (Miller 2020, PMID 31955273; Correll 2017, PMID 28498599).

Delayed recognition is a safety exposure

In a Hungarian specialist-care cohort of 8,935 people, mean time from first specialist contact to bipolar diagnosis was 6.46 years; only 11.85% were diagnosed without delay. Prior schizophrenia, nonpsychotic unipolar depression and personality-disorder diagnoses were associated with longer delay (Lublóy 2020, PMID 32075625). Administrative pathways cannot identify every clinically appropriate diagnostic revision, but they quantify prolonged exposure to uncertain treatment.

A risk-factor review reported antidepressant monotherapy, tricyclic use and more prior depressive episodes as the most consistent clinical signals for antidepressant-associated mania, with weak genetic predictors; the frequently quoted ~14% switch rate is background context in that review, not a figure it pooled (Melhuish Beaupre 2020, PMID 32134853). Activation after treatment still requires differential assessment for akathisia, substance effects, sleep loss and illness-course emergence.

Calming, coercion and route-specific hazards

Emergency consensus distinguishes calming from inducing sleep: de-escalation and continued interaction preserve diagnostic assessment and participation, while rapid tranquilization is reserved for danger after less restrictive measures fail (Gonzalez 2013, PMID 23323879). Inhaled loxapine offers rapid non-injection delivery, but it is contraindicated in asthma, COPD or other bronchospasm-associated lung disease: in dedicated safety trials symptomatic bronchospasm occurred in 53.8% of loxapine-treated asthma patients (n=52) and 19.2% of COPD patients (n=53) versus 11.5% and 11.1% on placebo, and the label warns of respiratory distress and arrest. The same expert review argues for extending inhaled loxapine to unsupervised community self-administration on the basis of preliminary results from the first ten participants of a phase 4 self-administration study (NCT02525991); that study has since completed (actual completion 2019-12-31, n=323, status verified Nov 2020) with no results posted on ClinicalTrials.gov as of 2026-09-03, so unsupervised community use remains a proposal rather than demonstrated safety (Pacciardi 2019, PMID 30721526).

Prediction performance can be misleadingly precise

In an eight-week ecological-momentary-assessment pilot of 35 interepisode adults, only eight nonzero suicidal-ideation events occurred in six people. Cross-validated affect trajectories yielded 88% sensitivity and 95% specificity one week ahead (Thompson 2014, PMID 24903771). With so few events and no external cohort, these impressive figures are hypothesis-generating and must not replace direct suicide-risk assessment or emergency response.

Open questions

  • Which dynamic combinations of sleep, mixed symptoms, substance use and recent behavior improve near-term suicide prediction without unacceptable false positives (Schaffer 2015, PMID 25329791)?
  • How should lithium's protective observational/meta-analytic signal be reconciled with the negative VA augmentation trial (Cipriani 2013, PMID 23814104; Katz 2022, PMID 34787653)?
  • Can personalized digital warning systems reduce severe episodes or hospitalization when paired with a tested response protocol (Faurholt-Jepsen 2018, PMID 29510813)?
  • What monitoring schedule best prevents lithium toxicity in older adults and during intercurrent illness (Sun 2018, PMID 29189921)?
  • Which interventions reduce the cardiovascular mortality gap rather than only improving intermediate metabolic markers (Paljärvi 2024, PMID 38826056)?
  • How can crisis planning include carers without increasing surveillance, conflict or loss of autonomy (Speirs 2023, PMID 36656805)?

References

  1. Yatham LN, et al. CANMAT/ISBD 2018 guidelines for the management of patients with bipolar disorder. Bipolar Disord. 2018;20:97-170. PMID 29536616
  2. Izadi N, et al. Suicide Assessment and Prevention in Bipolar Disorder. Focus. 2023;21:380-388. PMID 38695007
  3. Biazus TB, et al. All-cause and cause-specific mortality among people with bipolar disorder. Mol Psychiatry. 2023;28:2508-2524. PMID 37491460
  4. Hayes JF, et al. A systematic review and meta-analysis of premature mortality in bipolar affective disorder. Acta Psychiatr Scand. 2015;131:417-425. PMID 25735195
  5. Plans L, et al. Association between completed suicide and bipolar disorder: A systematic review. J Affect Disord. 2019;242:111-122. PMID 30173059
  6. Schaffer A, et al. ISBD Task Force on Suicide: meta-analyses and meta-regression. Bipolar Disord. 2015;17:1-16. PMID 25329791
  7. Miller JN, et al. Bipolar Disorder and Suicide: a Review. Curr Psychiatry Rep. 2020;22:6. PMID 31955273
  8. Cipriani A, et al. Lithium in the prevention of suicide in mood disorders. BMJ. 2013;346:f3646. PMID 23814104
  9. Katz IR, et al. Lithium Treatment in the Prevention of Repeat Suicide-Related Outcomes in Veterans. JAMA Psychiatry. 2022;79:24-32. PMID 34787653
  10. Grunze H, et al. WFSBP Guidelines: treatment of mixed states in bipolar disorder. World J Biol Psychiatry. 2018;19:2-58. PMID 29098925
  11. Zeller SL, et al. Managing Agitation Associated with Schizophrenia and Bipolar Disorder in the Emergency Setting. West J Emerg Med. 2016;17:165-172. PMID 26973742
  12. Garriga M, et al. Assessment and management of agitation in psychiatry: Expert consensus. World J Biol Psychiatry. 2016;17:86-128. PMID 26912127
  13. McKnight RF, et al. Lithium toxicity profile: a systematic review and meta-analysis. Lancet. 2012;379:721-728. PMID 22265699
  14. Sun M, et al. Lithium Toxicity in Older Adults: a Systematic Review of Case Reports. Clin Drug Investig. 2018;38:201-209. PMID 29189921
  15. Decker BS, et al. Extracorporeal Treatment for Lithium Poisoning: EXTRIP recommendations. Clin J Am Soc Nephrol. 2015;10:875-887. PMID 25583292
  16. Lavonas EJ, Buchanan J. Hemodialysis for lithium poisoning. Cochrane Database Syst Rev. 2015;2015:CD007951. PMID 26374731
  17. American College of Obstetricians and Gynecologists. Treatment and Management of Mental Health Conditions During Pregnancy and Postpartum. Obstet Gynecol. 2023;141:1262-1288. PMID 37486661
  18. Goldberg JF, Truman CJ. Antidepressant-induced mania: an overview of current controversies. Bipolar Disord. 2003;5:407-420. PMID 14636364
  19. Kishi T, et al. Mood stabilizers and/or antipsychotics for bipolar maintenance: network meta-analysis. Mol Psychiatry. 2021;26:4146-4157. PMID 33177610
  20. Bartoli F, et al. Cannabis use disorder and suicide attempts in bipolar disorder: A meta-analysis. Neurosci Biobehav Rev. 2019;103:14-20. PMID 31121199
  21. García-Jiménez J, et al. Cigarette smoking and risk of suicide in bipolar disorder: a systematic review. Front Psychiatry. 2023;14:1179733. PMID 37275988
  22. Paljärvi T, et al. Cardiovascular mortality in bipolar disorder: Population-based cohort study. Acta Psychiatr Scand. 2024;150:56-64. PMID 38826056
  23. Correll CU, et al. Cardiovascular disease in severe mental illness: large-scale meta-analysis. World Psychiatry. 2017;16:163-180. PMID 28498599
  24. Faurholt-Jepsen M. Electronic monitoring in bipolar disorder. Dan Med J. 2018;65:B5460. PMID 29510813
  25. Speirs B, et al. The lived experience of caring for someone with bipolar disorder. PLoS One. 2023;18:e0280059. PMID 36656805
  26. Lublóy Á, et al. Exploring factors of diagnostic delay for patients with bipolar disorder: a population-based cohort study. BMC Psychiatry. 2020;20:75. PMID 32075625
  27. Melhuish Beaupre LM, et al. Antidepressant-associated mania in bipolar disorder: a review and meta-analysis of potential clinical and genetic risk factors. J Clin Psychopharmacol. 2020;40:180–185. PMID 32134853
  28. Pacciardi B, et al. Inhaled loxapine for the management of acute agitation in bipolar disorder and schizophrenia. Drugs R D. 2019;19:15–25. PMID 30721526
  29. Gonzalez D, et al. Consensus statement on intramuscular aripiprazole for rapid control of agitation in bipolar mania and schizophrenia. Curr Med Res Opin. 2013;29:241–250. PMID 23323879
  30. Thompson WK, et al. Prediction of study-emergent suicidal ideation in bipolar disorder using ecological momentary assessment. Bipolar Disord. 2014;16:669–677. PMID 24903771