Medication-Induced Psychosis

Medication-induced psychotic disorder is the DSM-5-TR name for hallucinations or delusions that begin during, or soon after, exposure to a drug capable of producing those symptoms—and that are severe enough to need care in their own right, not merely a footnote to ordinary intoxication, sedation, or delirium.

The category covers street drugs and prescriptions together. This article focuses on medicines people are given on purpose: corticosteroids, Parkinson drugs, ADHD stimulants, antimalarials, interferons, antibiotics, anticholinergics, and a long tail of other agents. The practical stakes are different from cannabis- or methamphetamine-induced psychosis. The “substance” is often treating a life-threatening illness. Stopping it is a negotiation, not an obvious first step.

What the diagnosis requires

DSM-5-TR criteria for substance/medication-induced psychotic disorder are straightforward and easy to misuse.

  • Delusions, hallucinations, or both

  • A clear temporal link to a medication (or to withdrawal from one) that is known to be capable of causing psychosis

  • The picture is not better explained by a primary psychotic disorder

  • Symptoms do not occur only as part of a delirium

  • The disturbance causes real distress or impairment

The diagnosis should be used instead of plain “intoxication” or “withdrawal” only when psychosis dominates the presentation. If symptoms started before the drug, persist about a month after the drug is gone, or have already occurred in drug-free periods, a primary psychotic disorder is more likely.

That one-month rule is a guide, not a law of nature. Some agents—amphetamines, cocaine, phencyclidine—can leave psychosis smoldering for weeks after the last dose. Steroid reactions can appear during a taper. Parkinson’s disease itself causes hallucinations; the medicine may only be the last straw. Age of onset, medical context, and a collateral history matter more than any single cutoff.

Two neighboring problems must be separated:

  • Delirium is a fluctuating disturbance of attention and awareness, often with visual hallucinations. Medication-induced psychosis can coexist with delirium, but if the whole syndrome is a toxic-metabolic encephalopathy, that is the diagnosis that should lead.

  • The underlying disease can cause psychosis on its own. In systemic lupus erythematosus, brain involvement explains more new psychosis than the steroids used to treat it. Attributing every psychotic break on a medical ward to prednisone is a common error.

Which medicines are implicated

No complete list exists. The WHO pharmacovigilance database has linked dozens of molecules to new delusions; the leading classes in that analysis were antidepressants, antiepileptics, dopaminergic agents, opioids, anti-infectives, benzodiazepines, anti-dementia drugs, and psychostimulants. Amantadine, mefloquine, and carbidopa/levodopa showed the strongest disproportionality signals. In most reports, delusion began within the first two months. For about a third of the implicated drugs, psychosis is not even mentioned in the product label.

The agents that matter most in ordinary practice fall into a shorter set of families.

Corticosteroids

Glucocorticoids are the classic iatrogenic cause. About 5–6% of people on systemic corticosteroids develop a severe psychiatric reaction (mania, major depression, delirium, or psychosis). Milder mood and sleep changes are far more common, on the order of one in four. “Steroid psychosis” is a sloppy umbrella: mania or hypomania is the most frequent severe presentation; isolated schizophrenia-like psychosis is less common than the name implies. Mixed manic–psychotic pictures are typical in published cases.

Dose is the clearest risk factor. In the Boston Collaborative Drug Surveillance Program, psychiatric reactions occurred in:

  • 1.3% of people on less than 40 mg of prednisone a day

  • 4.6% on 40–80 mg

  • 18.4% on more than 80 mg

Dose predicts frequency, not timing, severity, or which symptom cluster appears. Median onset is about 11–12 days; most cases declare themselves within two weeks, but reactions can start after a single large pulse or during a taper. Age over 50, female sex, prolonged high-dose therapy, and prior psychiatric illness show up as risk markers in reviews, though not every study agrees.

Dopaminergic drugs for Parkinson’s disease

Levodopa, dopamine agonists (pramipexole, ropinirole, rotigotine), and amantadine can produce hallucinations—often visual—and paranoid delusions. The problem is entangled with the disease: advancing Parkinson’s, cognitive impairment, and sleep disruption all raise risk even before the dose is increased. Treatment is a tightrope. Cutting the Parkinson drug can worsen motor function as fast as it clears the psychosis. Low-dose quetiapine, clozapine, and pimavanserin are the usual psychiatric options because they spare motor symptoms better than high-potency D2 blockers.

ADHD stimulants

Prescribed amphetamines and methylphenidate can induce hallucinations or a full psychotic or bipolar episode. A 2025 meta-analysis in JAMA Psychiatry estimated that among people with ADHD treated with stimulants, about 2.8% developed psychotic symptoms, 2.3% a psychotic disorder, and 3.7% bipolar disorder, with substantial heterogeneity across studies. Amphetamines carried a higher odds of psychosis than methylphenidate (odds ratio about 1.6). High dose, rapid titration, sleep loss, cannabis or other stimulant use, and a personal or family history of psychosis raise risk. The mechanism is the same one that makes illicit methamphetamine psychotogenic: a surge of synaptic dopamine in mesolimbic pathways.

These percentages describe treated ADHD populations over follow-up, not the risk of a single prescription. They are large enough that patients and families should be told what to watch for. They are not large enough to treat stimulants as interchangeable with methamphetamine.

Antimalarials

Mefloquine is notorious. Severe neuropsychiatric reactions—anxiety, paranoia, psychosis, and suicidal thinking—occur in well under 1% at treatment doses and far less often in prophylaxis, but when they occur they can be delayed, vivid, and slow to fade. Chloroquine has a similar, rarer signal. A personal or family history of psychiatric illness is a reason to choose a different agent. Symptoms can begin after the first dose.

Interferons and other immunomodulators

Interferon-alfa carries a boxed warning for life-threatening neuropsychiatric effects. Depression is the dominant problem; mania and psychosis are less common but documented, especially in people with prior psychiatric disease. Interleukin-2 and some chemotherapy agents (including procarbazine and calcineurin inhibitors such as cyclosporine) appear in the same iatrogenic list.

Antibiotics

Psychosis is an uncommon but repeatedly reported adverse effect of several anti-infectives. Fluoroquinolones (ciprofloxacin, levofloxacin) and macrolides show the strongest signals in adverse-event analyses; penicillins, cephalosporins, trimethoprim–sulfamethoxazole, metronidazole, and doxycycline also appear. Prevalence of psychosis among reported psychiatric adverse events varies by agent; it is not a common reason to avoid a needed antibiotic, but it is a reason to stop one when a patient becomes newly paranoid on day three of a course and the workup for infection of the central nervous system is negative. Isoniazid and some antiretrovirals have case-level associations as well.

Anticholinergics and over-the-counter drugs

Diphenhydramine, other first-generation antihistamines, oxybutynin, benztropine, and similar agents can produce an anticholinergic delirium that includes hallucinations—especially in older adults, after overdose, or in combination. Dextromethorphan in high doses, pseudoephedrine, and phenylephrine belong on the same household list. These cases are easy to miss if the history stops at “prescriptions.”

Everything else that shows up in textbooks

DSM-5-TR and consultation-liaison reviews also list anesthetics and analgesics, anticonvulsants, antihypertensives, digoxin, muscle relaxants, NSAIDs, disulfiram, and anabolic-androgenic steroids. Anabolic steroid psychosis is idiosyncratic rather than neatly dose-related. Benzodiazepine withdrawal, more than benzodiazepine treatment, can produce a psychotic picture. Polypharmacy in later life shifts the epidemiology: the offending agent is more often a Parkinson drug or a cardiovascular medicine than a recreational stimulant.

How a pill can do this

There is no single pathway.

Dopamine-boosting drugs—stimulants, levodopa, some antidepressants in susceptible people—fit the classic mesolimbic-dopamine model of positive psychotic symptoms. Anticholinergic agents scramble cortical signaling and attention. Corticosteroids alter hippocampal and amygdala function, change monoamine availability, and can increase dopamine synthesis; the exact sequence is still inferred rather than proven. Fluoroquinolones and some other antibiotics interact with GABA and NMDA systems. Mefloquine has direct CNS toxicity. Interferons push a cytokine milieu that is itself associated with depression and, less often, mania or psychosis.

Two clinical implications follow. First, not every medication-induced psychosis will respond equally to a D2-blocking antipsychotic; steroid mania sometimes settles with a dose cut and a mood stabilizer. Second, pharmacokinetic interactions can manufacture a “new” psychosis by raising levels of a drug the patient was previously stable on—ciprofloxacin inhibiting CYP1A2 and lifting clozapine or olanzapine is a textbook example.

Who is most vulnerable

Risk stacks rather than residing in one variable:

  • High dose and rapid escalation (steroids, stimulants)

  • Older age and cognitive impairment (anticholinergics, dopaminergic drugs, polypharmacy)

  • Personal or family history of psychosis, bipolar disorder, or a prior reaction to the same class

  • Sleep deprivation

  • Concurrent cannabis, alcohol, or illicit stimulants

  • Female sex in several steroid and medication-induced-psychosis reviews

  • The medical illness itself (lupus, advanced Parkinson’s, CNS infection)

Absence of these factors does not make an episode impossible. Presence of several of them should change the consent conversation before a high-dose steroid burst or an amphetamine start.

How clinicians sort it out

The workup is the new-onset psychosis workup plus a forensic look at the medicine cabinet.

A timeline is the most useful tool: when the drug started or changed, when sleep broke, when the first odd belief appeared. Collateral history from family often beats the patient’s account. Bring in over-the-counter products, eye drops, “supplements,” and drugs borrowed from a spouse.

Urine toxicology catches only the panel it is built to catch. It will not show levodopa, prednisone, mefloquine, or interferon. A negative screen does not clear the diagnosis.

Medical mimics still have to be excluded: infection (including encephalitis), metabolic derangement, thyroid storm, hypoxia, stroke, autoimmune encephalitis, and the disease being treated. Cognitive testing helps separate delirium from a purely psychotic state. In an older postoperative patient, toxic-metabolic delirium is the leading hypothesis until proven otherwise. In a young person on a new amphetamine prescription who has never been psychotic before, the stimulant is a leading hypothesis—but prodromal schizophrenia remains on the list.

Treatment

There are almost no randomized trials devoted to medication-induced psychosis as a class. Management is built from case series, pharmacovigilance, and first principles.

1. Reduce or stop the culprit when the underlying disease allows. For steroids, that usually means tapering to under 40 mg of prednisone-equivalent per day, or off entirely if the indication permits. Abrupt stops risk adrenal insufficiency and, sometimes, a rebound psychiatric crash. For stimulants, hold the drug. For Parkinson medicines, peel back the most psychotogenic agent first (often a dopamine agonist or anticholinergic) while protecting mobility. For mefloquine or interferon, switch classes if treatment must continue.

2. Secure the environment. A quiet room, sitters if needed, and attention to suicide risk. Medication-induced states can include command hallucinations and impulsive violence; they are not “less serious” because a pill caused them.

3. Treat the syndrome.

  • Dopamine-driven psychoses (stimulants, many Parkinson cases) usually need an antipsychotic.

  • Steroid-induced mania and psychosis often respond to olanzapine, risperidone, quetiapine, or haloperidol; lithium or valproate is sometimes added when mood elevation dominates and steroids cannot be stopped.

  • Hallucinogen-like presentations may settle with observation and a benzodiazepine.

  • Anticholinergic toxicity may need physostigmine in selected, monitored settings rather than more dopamine blockade.

Second-generation antipsychotics at modest doses are the usual first choice when a drug is required. Duration should be short if the episode clears after the offending medicine is gone, and longer if symptoms persist, risk of conversion to a primary disorder is high, or the culprit drug must stay on board.

4. Do not call it finished at discharge. Follow-up is there to answer two questions: Did this unmask bipolar disorder or a schizophrenia-spectrum illness? And does the patient still need the medicine that caused the break? Substance-induced psychoses as a group convert to schizophrenia or bipolar disorder in a substantial minority of people—highest after cannabis and stimulants, much lower after alcohol. Prescribed-drug cases are less well quantified, but a first psychotic episode in a young adult on an amphetamine deserves the same longitudinal caution as any other first episode.

Outlook

Most medication-induced episodes shrink once the drug is reduced and, if needed, an antipsychotic is started. Steroid cases can clear in days or take several weeks; some linger up to two months. Stimulant-induced symptoms often fade with abstinence but can persist. A reaction to one steroid burst predicts caution with the next, not a lifetime diagnosis by itself.

The errors that cause harm are predictable. One is continuing a high-dose culprit while adding layer upon layer of sedatives. Another is stopping a necessary immunosuppressant so abruptly that the original disease flares. A third is filing the episode under “drug-induced,” discharging the patient, and never checking whether psychosis returns off the medicine.

What patients and families can do

Before a high-risk prescription—especially high-dose steroids, a new amphetamine, mefloquine, or interferon—ask what psychiatric warning signs look like and whom to call at 2 a.m. During treatment, treat sudden insomnia, racing thoughts, new suspiciousness, or seeing things that others do not as a reason to contact the prescriber the same day, not as a personality change to manage privately.

Bring every bottle, including sleep aids and cold remedies, to the appointment. Do not stop a steroid, Parkinson drug, or transplant medicine without a plan for the illness it is treating.

Hallucinations, fixed false beliefs, severe agitation, or talk of suicide after a medication change are an emergency.