Antidepressant-Induced Psychosis

Product labels warn that they can cause mania, hallucinations, or psychosis; hospital series keep finding people admitted shortly after a new prescription; and a thin but consistent case literature describes frank delusions after a dose increase. The clinically useful question is not whether this ever happens. It is what is happening when it does.

Most of the time, “antidepressant-induced psychosis” is not a miniature version of schizophrenia created by a serotonin reuptake inhibitor. It is one of four overlapping problems: a manic or mixed switch with psychotic features in unrecognized or known bipolar disorder; a rare, more purely psychotic reaction—best documented with bupropion; an uncovered psychotic depression that was treated with an antidepressant alone; or a toxic state (serotonin excess, severe hyponatremia, akathisia) mistaken for a primary thought disorder. Sorting those apart changes what you stop, what you start, and what diagnosis the patient carries for the next decade.

How official diagnosis treats it

DSM-5-TR places these episodes under substance/medication-induced psychotic disorder when delusions or hallucinations begin during or soon after antidepressant exposure, the drug is capable of producing those symptoms, the picture is not solely a delirium, and a primary psychotic disorder does not better explain it. If the psychosis is clearly part of a manic syndrome, the more precise formulation is often treatment-emergent mania (or a mixed episode) with psychotic features—not a stand-alone induced psychosis. Both can be true in the same patient.

Every major antidepressant class appears in that warning language: SSRIs, SNRIs, tricyclics, MAOIs, bupropion, mirtazapine, and others. Frequency and mechanism are not the same across the class.

The common pathway: a mood switch, sometimes with psychosis

The oldest and best-supported story is polarity switch. Imipramine was linked to new mania within a few years of its introduction. Later work made the pattern quantitative. In a meta-analysis of more than 100,000 patients, mania or hypomania occurred in about 12.5% of people exposed to antidepressants versus 7.5% without them. Tricyclics were riskier than SSRIs; mood stabilizers offered less protection than clinicians hoped, probably because they are given to the people already most likely to switch.

Among people already diagnosed with bipolar disorder, rates are higher still. In a prospective cohort of 1,629 antidepressant-treated bipolar patients followed about five years, mood switching occurred in 27.6%—more often hypomania than full mania, and somewhat more often in bipolar II than bipolar I. Psychotic features in that cohort clustered in bipolar I. Predictors included a depression–mania–interval course, more episodes per year, family psychiatric history, older current age, and prior suicide attempts. A history of switching predicts the next switch more reliably than the raw number of past antidepressant trials.

In people carrying a unipolar depression diagnosis, new mania-like responses still occur. Pooled observational data put the risk around 8% over roughly two and a half years of antidepressant treatment, or about 3.4% per year—and several times higher in juveniles than in adults. Only a fraction of those switches immediately produce a revised bipolar diagnosis; diagnostic conversion from MDD to bipolar disorder within three months of starting an antidepressant is on the order of 1–2% in large health-system data, higher in specialty clinics than in primary care. That gap is not reassuring. It means many switches are coded as side effects or “activation” rather than as a change in illness class.

The International Society for Bipolar Disorders has long ranked switch liability as higher with tricyclics, MAOIs, and some SNRIs—especially venlafaxine—than with SSRIs and bupropion. A 2025 network meta-analysis of randomized trials in bipolar depression found no antidepressant statistically worse than placebo for switch, but venlafaxine again sat at the top of the risk estimates. Evidence quality was low, and most trial patients were already on a mood stabilizer. Monotherapy in undiagnosed bipolar depression is a different, riskier setting. One large Swedish analysis found roughly a threefold rise in manic switch when antidepressants were used without a mood stabilizer in bipolar disorder.

Psychosis, when it appears in this pathway, usually wears a manic costume: grandiosity, religious or erotic delusions, decreased need for sleep, pressured speech, then loss of insight. Calling that “SSRI-induced schizophrenia” misses the treatment target, which is mood stabilization, not a first-episode psychosis protocol copied wholesale from a cannabis case.

The rarer pathway: psychosis without a party

A smaller set of patients become paranoid or hallucinated without a clear manic syndrome. The drug with the most coherent mechanistic story is bupropion.

Bupropion inhibits reuptake of dopamine and norepinephrine. Premarketing trial aggregates reported hallucinations in about 2.8% of participants and delusions in about 1.4%—figures that bundled mild perceptual changes with more serious events and led early authors to advise against first-line use in people with a psychosis history. Systematic reviews of published cases conclude that bupropion can both ignite new psychotic symptoms and worsen existing ones in selected people. Higher dose, rapid titration, substance use (especially cocaine), head trauma or structural brain lesions, older age, liver impairment, and pharmacokinetic interactions—bupropion is a strong CYP2D6 inhibitor—show up repeatedly as risk markers. Concurrent antipsychotics at adequate doses appear protective. Immediate-release formulations dominate older case series.

The typical published case is not subtle: persecutory delusions, auditory or visual hallucinations, sometimes catatonia, often within days of reaching 300 mg. Many patients have no prior psychotic episode. Stopping the drug and adding a short course of an antipsychotic often clears the picture within a week. In people with schizophrenia who are already on antipsychotics, added bupropion for depression or smoking cessation has a much thinner psychosis signal, which is why some clinicians still use it there with monitoring.

SSRIs and SNRIs produce far more case reports of isolated psychosis than they produce a measurable incidence rate. Sertraline, fluoxetine, paroxetine, and others have been linked to new auditory hallucinations or delusions that receded when the drug was stopped. Some of those patients were adolescents; some had psychotic depression that was incompletely recognized; some had no manic signs at all. These reports prove possibility. They do not prove that SSRIs are a common independent cause of schizophrenia-spectrum illness. Pharmacovigilance still lists antidepressants among the leading drug classes associated with new delusions in the WHO database, which is a signal to take a timeline seriously, not a license to treat every SSRI as amphetamine.

Two look-alikes that are not “induced schizophrenia”

Psychotic depression treated with an antidepressant alone. Major depression with delusions or hallucinations needs an antipsychotic, ECT, or both. An SSRI by itself can leave the mood slightly improved and the delusion untouched—or, in older case notes, appear to sharpen the delusion. Family history of bipolar disorder is over-represented in psychotic depression; a “new” psychosis after an antidepressant in that group may be the bipolar diathesis announcing itself.

Toxic and metabolic mimics. Severe SSRI or SNRI poisoning can include hallucinations as part of serotonin toxicity, alongside clonus, hyperreflexia, fever, and agitation. In older adults, antidepressant-associated hyponatremia—highest with SNRIs and SSRIs—can produce confusion that is charted as psychosis until the sodium is checked. Akathisia can look like agitation and, rarely, precede impulsive violence; it is a movement and inner-restlessness problem, not a thought disorder, but it is a reason to stop or change the drug immediately.

Who is at highest risk

Risk is not evenly spread.

  • Known or suspected bipolar disorder, mixed features during the depression, or a first-degree relative with bipolar illness

  • Prior antidepressant-associated switch

  • Younger age at illness onset; juvenile patients as a group

  • Psychotic features already present in the depression

  • Rapid cycling or a highly recurrent course

  • Bupropion at higher doses, after cocaine use, after brain injury, or combined with CYP2D6 substrates

  • Tricyclic, MAOI, or venlafaxine exposure more than SSRI or (in bipolar depression) carefully dosed bupropion

  • Sleep loss after the drug is started—often the first visible crack

A higher polygenic score for antidepressant response in major depression has even been associated with higher odds of treatment-emergent mania in bipolar I, a reminder that “this drug works strongly on this brain” is not always a compliment.

What to do when it happens

The acute sequence is more important than the brand name of the antidepressant.

  1. Stop the antidepressant. Taper is reasonable if the dose is high and serotonin-discontinuation symptoms are likely; it is not reasonable if the person is floridly manic or delusional.

  2. Treat the syndrome in front of you. Psychotic mania needs a mood stabilizer plus an antipsychotic, or an antipsychotic with mood-stabilizing properties (olanzapine, quetiapine, risperidone, lithium, valproate). Isolated bupropion psychosis often remits with discontinuation plus a short antipsychotic course.

  3. Do not restart the same drug as a test. A positive rechallenge is academically tidy and clinically reckless.

  4. Revise the diagnosis in writing. An MDD label that survives a psychotic manic switch is a planning failure. Even a clean, brief bupropion psychosis deserves follow-up measured in months, not days.

  5. Check the boring labs. Sodium, liver enzymes, and a toxicology screen prevent treating hyponatremia or a stimulant binge as a new axis-I disease.

If bipolar depression still requires an antidepressant later, ISBD-style caution applies: pair it with a mood stabilizer, prefer agents with lower switch signals, use the lowest effective dose, and watch sleep as if it were a vital sign.

What this is not

It is not a reason to withhold antidepressants from ordinary unipolar depression. Switch and de novo psychosis are uncommon in that group; untreated major depression is common and lethal. It is not proof that a given patient “was bipolar all along,” though that hypothesis should be first, not last. It is not interchangeable with cannabis-induced psychotic disorder, which carries a far higher published conversion rate to schizophrenia. And it is not a personality change or a moral failing. A brain that was stable on Tuesday and delusional on Friday after a dose increase is having a neuropsychiatric adverse event.

Patients and families should be told, before the first pill, what a bad turn looks like: sudden insomnia, racing thoughts, new suspiciousness, hearing voices, grandiose or bizarre beliefs, or a sense of being wound so tight that sitting still is impossible. Those symptoms warrant same-day contact with the prescriber and, if reality testing is gone, emergency care.