What does the evidence say about stopping antipsychotics in dementia?
The best-known withdrawal research is the Dementia Antipsychotic Withdrawal Trial (DART-AD), led by Clive Ballard in the UK. People with Alzheimer disease living in care homes who were already taking an antipsychotic were randomized either to continue it or to switch to placebo for 12 months.
- Behavior and function (PLoS Medicine, 2008): for most participants, withdrawal had no overall harmful effect on function or cognition. People with more severe neuropsychiatric symptoms at the start (NPI score 15 or above) did somewhat better on continued treatment.
- Survival (Lancet Neurology, 2009): long-term follow-up showed lower survival in the group that continued antipsychotics. Survival at 24 months was 46% versus 71% on placebo, and at 36 months 30% versus 59%. The trial was small (128 people started treatment), so the size of the effect is uncertain, but the direction matched other studies.
A 2018 Cochrane review of 10 withdrawal trials (632 participants) concluded, on low-quality evidence, that antipsychotics can often be discontinued in people with dementia who have taken them for at least three months, with little or no effect on behavioral symptoms overall. Two exceptions matter: people whose psychosis, agitation or aggression had clearly responded to the drug, and people with more severe symptoms at baseline, were more likely to relapse. That is why reduction should be gradual, individualized and monitored rather than a blanket stop.
Ballard's later WHELD programme found that a structured antipsychotic review in UK care homes cut antipsychotic use by about half, but behavioral symptoms tended to worsen when the review was not paired with more social interaction. Reducing drugs works best alongside the non-drug approaches that replace them.
Which antipsychotics are commonly used in nursing homes?
| Drug | Type | Notes for dementia care |
|---|---|---|
| Quetiapine | Atypical | Often used at low doses for sleep or agitation; sedation and falls are common concerns |
| Risperidone | Atypical | Most studied for aggression in dementia; stroke and parkinsonism risks |
| Olanzapine | Atypical | Anticholinergic and metabolic effects; sedation |
| Aripiprazole | Atypical | Less sedating; akathisia (restlessness) can mimic agitation |
| Haloperidol | Conventional | Highest estimated mortality risk in a large US veterans study; strong movement side effects |
For drug-by-drug information see the atypical antipsychotic class overview. A 2015 JAMA Psychiatry study estimated one extra death over about six months for every 26 people with dementia treated with haloperidol, 27 with risperidone, 40 with olanzapine and 50 with quetiapine, compared with similar non-users.
How does the CMS long-stay antipsychotic measure work?
CMS publishes the percentage of long-stay residents (101 or more cumulative days in the facility) who received an antipsychotic, and the measure feeds the Five-Star quality rating on Care Compare.
| Feature | MDS-only measure (to Jan 2026) | Hybrid measure (from Jan 2026) |
|---|---|---|
| Data | MDS 3.0 only | MDS plus Medicare and Medicaid claims and Medicare Advantage encounter data |
| What counts | Antipsychotic received in the 7-day MDS look-back | Also antipsychotic prescriptions filled during the stay that the MDS window missed |
| Diagnosis exclusions | Schizophrenia, Huntington disease, Tourette syndrome coded on the MDS | Same three conditions, but the MDS diagnosis must be supported by claims |
| National rate | 14.64% (CMS figure cited in 2025) | About 16.98% (CMS projection) |
CMS explained the change in memo QSO-25-20-NH after a 2021 HHS Office of Inspector General report found the MDS under-counted antipsychotic use. Star rating cut points for the new measure are set by deciles of facility performance. More detail on how measures feed ratings is on our Five-Star ratings page.
What were the CMS schizophrenia coding audits?
Because schizophrenia removes a resident from the antipsychotic measure, an inaccurate schizophrenia diagnosis can hide antipsychotic use. In memo QSO-23-05-NH (January 18, 2023), CMS announced off-site audits of MDS schizophrenia coding. Its pilot audits found missing psychiatric evaluations and behaviors that were related to dementia rather than schizophrenia. For facilities found to have inaccurate coding, CMS said:
- The overall and long-stay quality measure ratings drop to one star for six months, which lowers the overall star rating by one star.
- The short-stay quality measure rating is suppressed for six months.
- The long-stay antipsychotic measure is suppressed for 12 months.
- Facilities that admit errors after notice but before the audit may receive a lesser action, such as suppression rather than a downgrade.
Schizophrenia rarely begins in late life. A defensible diagnosis needs a documented psychiatric evaluation, history, symptoms over time and the exclusion of dementia, delirium and other causes. CMS has not published national pass or fail counts for the audits.
What do federal rules require for antipsychotic prescribing?
Under 42 CFR 483.45, a resident's drug regimen must be free from unnecessary drugs, meaning any drug used in excessive dose (including duplicates), for excessive duration, without adequate monitoring, without adequate indication, or despite adverse effects. For psychotropics specifically:
- A psychotropic may be started only to treat a specific condition diagnosed and documented in the record.
- Residents on psychotropics receive gradual dose reductions (GDR) and behavioral interventions, unless clinically contraindicated, in an effort to discontinue them.
- PRN (as needed) psychotropic orders are limited to 14 days unless the prescriber documents a rationale and duration.
- PRN antipsychotic orders are limited to 14 days and cannot be renewed without the prescriber evaluating the resident.
Surveyor guidance says GDR compliance may be shown by attempting a reduction in two separate quarters, at least one month apart, in the first year after admission on (or initiation of) a psychotropic, then annually, unless contraindicated with documented clinical reasons. "Resident is stable" alone is not a contraindication. In revisions effective April 28, 2025, CMS moved psychotropic guidance from F758 into F605 (chemical restraints) and added an explicit expectation of informed consent (F552) before starting or increasing a psychotropic.
How should a gradual dose reduction be run?
- 1Confirm the targetWrite down the specific symptom the drug was meant to treat and how often it happens now, using a behavior log for at least a week.
- 2Agree a plan with the familyExplain why a reduction is being tried, what to watch for, and how quickly the team will respond if symptoms return.
- 3Reduce in modest stepsThe prescriber lowers the dose in small increments over weeks, not days, to limit withdrawal effects and allow monitoring. (Dosing decisions belong to the prescriber.)
- 4Strengthen non-drug supportsAdd or intensify the personalized approaches that address the trigger, and make sure every shift knows them.
- 5Re-measure and decideCompare behavior logs before and after. If symptoms return, document it; that record supports a clinical contraindication to further reduction for now.
For care-planning templates and huddle scripts, see the professional toolkit. For the wider medication safety picture, see medication management in nursing homes.
Frequently asked questions
Is it safe to stop antipsychotics in someone with dementia?
For many people it is. Trials, including DART-AD and those pooled in a 2018 Cochrane review, found that stopping long-term antipsychotics usually made little difference to behavior overall. People with severe symptoms that had clearly responded to the drug are more likely to relapse. Reductions should be gradual, planned with the prescriber and family, and monitored with behavior records.
What is a gradual dose reduction in a nursing home?
A gradual dose reduction (GDR) is a planned, stepwise lowering of a psychotropic dose to find the lowest effective dose or stop the drug. Federal guidance expects attempts in two separate quarters in the first year a resident is on the drug, then annually, unless the prescriber documents why a reduction would likely cause harm.
Which diagnoses are excluded from the antipsychotic quality measure?
CMS excludes residents with schizophrenia, Huntington disease or Tourette syndrome from the long-stay antipsychotic measure. Since January 2026, an exclusion diagnosis coded on the MDS must also be supported by Medicare or Medicaid claims or encounter data. Hospice residents are also excluded under the updated measure. Bipolar disorder is not an exclusion.
What happens if a nursing home miscodes schizophrenia?
Under CMS memo QSO-23-05-NH, facilities whose MDS schizophrenia coding is found inaccurate in an audit have their overall and long-stay quality measure ratings lowered to one star for six months, which drops the overall star rating by one star. The long-stay antipsychotic measure is suppressed for 12 months. Residents may also have received unnecessary antipsychotics.
How long can a PRN antipsychotic order last in a nursing home?
Federal rules limit as-needed (PRN) antipsychotic orders to 14 days. The order cannot be renewed unless the attending physician or prescribing practitioner evaluates the resident for the appropriateness of the medication. Other PRN psychotropics are also limited to 14 days unless the prescriber documents a rationale and a specific duration.
Why did the antipsychotic rate on Care Compare go up in 2026?
In January 2026 CMS switched to a hybrid measure that adds Medicare and Medicaid claims and Medicare Advantage encounter data to MDS data. It captures prescriptions the MDS look-back missed and requires claims support for exclusion diagnoses. CMS projected the national rate would rise from about 14.6% to about 17% for that reason alone.
Get help choosing care or coverage
Talk with a licensed agent about Medicare, Medigap, Medicare Advantage and ways to pay for a nursing home stay. Or call 1-800-MEDIGAP (1-800-633-4427).
Sources
- Ballard C et al. DART-AD long-term follow-up. Lancet Neurology 2009 (PubMed 19138567)
- Ballard C et al. DART-AD: withdrawal of neuroleptics in Alzheimer disease. PLoS Medicine 2008
- Van Leeuwen E et al. Withdrawal vs continuation of long-term antipsychotics in dementia. Cochrane Review 2018
- 42 CFR 483.45: Pharmacy services, unnecessary and psychotropic drugs
- CMS QSO-23-05-NH: Schizophrenia coding audits and quality measure ratings (Jan 2023)
- CMS QSO-25-20-NH (revised Sept 2025): Updated long-stay antipsychotic measure
Medical disclaimer. This page is general education, not medical, legal or financial advice, and it does not replace the judgment of a qualified clinician who knows the resident. Do not start, stop or change any medication without talking to the prescriber. In an emergency, call 911.
Independent resource. NHQualityCampaign.org is an independent educational resource. It is not affiliated with, endorsed by or operated by the Centers for Medicare & Medicaid Services (CMS), the U.S. Department of Health and Human Services (HHS), any government agency, or the original Advancing Excellence in America’s Nursing Homes campaign or its sponsors. Historical campaign information is summarized with attribution for educational purposes.
Related reading
Dementia careGood dementia care in a nursing home means knowing the person, treating causes of distress, and using…
Non-drug approachesThe best non-drug dementia care finds the cause of distress and changes care around it, using methods like…
National PartnershipThe National Partnership to Improve Dementia Care is a CMS initiative, launched in 2012, to cut unnecessary…
Medication managementSafe nursing home medication management means every drug has a clear reason, gets a monthly pharmacist…
Long-Stay MeasuresNine long-stay quality measures feed the CMS Five-Star QM rating: seven from the MDS and two from Medicare…
Behavioral symptomsBehavioral symptoms of dementia (responsive behaviors, or BPSD) are usually reactions to pain, illness…
