Tardive Dyskinesia: Risk, Screening, and What the Treatment Trials Show

Creator: Ryan S. Sultan, MD
Published: 2026-11-12

By Ryan S. Sultan, MD
Psychiatrist · Columbia University Irving Medical Center
Published November 12, 2026 · 6-minute read

Tardive dyskinesia is a syndrome of involuntary movements that can develop in people treated with antipsychotic drugs, and the risperidone label describes it as potentially irreversible.¹ In a meta-analysis of 41 studies with 11,493 patients, the mean prevalence was 25.3%: 30.0% with current first-generation antipsychotic treatment and 20.7% with current second-generation treatment.²

The FDA approved two treatments in 2017, valbenazine (Ingrezza) and deutetrabenazine (Austedo).^{15, 17} New involuntary movements are a reason to contact the prescriber, who examines them and decides on any medication change.

What tardive dyskinesia looks like

A Cochrane review describes involuntary movements of the face, mouth, tongue, and jaw.²³ Clinicians rate them with the Abnormal Involuntary Movement Scale (AIMS), a 12-item scale in which items 1 to 7 score the severity of movements across body regions.⁸

The risperidone label states that the risk rises with treatment duration and cumulative dose.¹

How common it is with older and newer antipsychotics

Prevalence was 7.2% in 4 studies of people who had never taken a first-generation drug, and 23.4% in second-generation cohorts with likely prior first-generation exposure.²

In a meta-analysis of 57 head-to-head randomized trials, the annualized incidence of new cases was 6.5% across first-generation arms and 2.6% across second-generation arms. In exploratory comparisons, olanzapine and aripiprazole (Abilify) had an advantage over other non-clozapine second-generation drugs, and the abstract gives no separate rate for aripiprazole.³

In a 1998 study of 261 patients aged 55 or older starting antipsychotic treatment, cumulative rates were 25%, 34%, and 53% after 1, 2, and 3 years.⁴ In a 2011 study of 207 older patients, the cumulative rate with risperidone was 5.3% after 1 year and 7.2% after 2 years.⁵ A 2018 review also lists female sex, diabetes, smoking, and early parkinsonian side effects as risk factors.⁶

How often screening with the AIMS is recommended

An American Psychiatric Association practice tool, based on the association's schizophrenia guideline, calls for a clinical assessment for abnormal movements at each visit. It calls for a structured instrument such as the AIMS at least every 6 months in patients at high risk, at least every 12 months in other patients, and whenever new or worsening movements are detected.⁷

What the valbenazine and deutetrabenazine trials found

Placebo-controlled trials of the two approved treatments
Trial	Length and size	Fall in AIMS score	Endpoint or dose that fell short
KINECT 3 (valbenazine)	6 weeks, 225 participants	3.2 points at 80 mg/day and 1.9 at 40 mg/day, compared with 0.1 on placebo	None reported in the abstract
ARM-TD (deutetrabenazine)	12 weeks, 117 patients	3.0 points, compared with 1.6 on placebo (p = 0.019)	Clinician-rated treatment success, 48.2% vs 40.4%, was not significant
AIM-TD (deutetrabenazine)	12 weeks, 298 randomized	3.3 points at 36 mg/day and 3.2 at 24 mg/day, compared with 1.4 on placebo	12 mg/day (2.1 points) did not differ from placebo (p = 0.217)

Sources: trial abstracts.^{10, 11, 12} Doses are the trial arms. The prescriber sets the dose for a given patient.

A 2018 meta-analysis found no head-to-head comparison among these drugs and less consistent results on clinician and patient global ratings than on AIMS scores.¹³

The APA tool recommends a reversible VMAT2 inhibitor, the class both drugs belong to, when moderate to severe or disabling tardive dyskinesia persists. The 2023 VA/DoD guideline suggests a trial of one and rates that suggestion "Weak for."^{7, 14}

The FDA approved valbenazine for tardive dyskinesia on April 11, 2017.¹⁵ Deutetrabenazine was approved on April 3, 2017, for chorea associated with Huntington's disease, and on August 30, 2017, for tardive dyskinesia.^{16, 17}

What the labels warn about

Both labels carry a boxed warning, and each one names patients with Huntington's disease. The Ingrezza label states that VMAT2 inhibitors "can increase the risk of depression and suicidal thoughts and behavior in patients with Huntington’s disease."⁸ The Austedo boxed warning uses the same wording for Austedo XR and Austedo, adding the term "suicidality."⁹

In pooled placebo-controlled trials of valbenazine, somnolence occurred in 10.9% of treated patients and 4.2% on placebo, and parkinson-like events in 3% and under 1%.⁸ In deutetrabenazine's tardive dyskinesia trials, akathisia, agitation, or restlessness occurred in 2% and 1%.⁹ Both labels say the drug should be avoided in patients with congenital long QT syndrome.^{8, 9}

Movements returned toward baseline after valbenazine was stopped

In the one-year extension of KINECT 3, 124 of 198 participants completed treatment. Improvement was sustained on valbenazine, and scores returned toward baseline after 4 weeks of washout.¹⁸

Is tardive dyskinesia permanent?

The risperidone label states that tardive dyskinesia "may remit, partially or completely, if antipsychotic treatment is discontinued."¹ Published figures vary widely:

Among 49 outpatients followed after discontinuation, complete and persistent reversibility was rare (2%), and many showed noticeable improvement within the first year, sometimes interrupted by psychological relapse.¹⁹
In a retrospective cohort of 108 clinic patients who withdrew the causative drug, most not psychotic, 13% had reversibility.²⁰
An editorial put remission rates in long-term studies at 0 to 73%, with most below 25%.²¹
Among 125 institutionalized patients who stayed on antipsychotics for 3 years, 28.6% of those affected recovered and 30% improved.²²

A Cochrane review of lowering the antipsychotic dose found 2 trials with 17 people, rated the evidence very low quality, and called the effect uncertain.²³ In one small trial in a 2025 network meta-analysis, antipsychotic washout worsened tardive dyskinesia.²⁴ Any change to the antipsychotic is a decision for the prescriber.

Switching to clozapine as a treatment

A meta-analysis of 16 studies with 1,060 patients found that a switch to clozapine monotherapy was followed by what the authors called a slight reduction in tardive dyskinesia (d = -0.40). The reduction was greater (d = -2.56) in the 4 studies, with 48 patients, that measured severity as a primary outcome.²⁵ A 2019 systematic review concluded that prospective randomized trials are necessary to substantiate the recommendation.²⁶ I've written separately about when to start clozapine.

The 2013 neurology guideline predates both drugs

The American Academy of Neurology's 2013 guideline rated clonazepam and ginkgo biloba Level B and tetrabenazine Level C. It found the data insufficient to support or refute withdrawing the causative drug or switching from a typical to an atypical one (Level U).²⁷ A 2018 update rated deutetrabenazine and valbenazine Level A and left the Level U rating in place.²⁸

Limitations of the evidence

This evidence has limitations. First, the placebo-controlled trials lasted 6 and 12 weeks, and the one-year valbenazine extension had no placebo group. Second, the reversibility figures come from small, older, or retrospective cohorts. Nevertheless, the screening interval is defined and both approved drugs lowered AIMS scores against placebo, so new movements warrant a prompt call to the prescriber and a structured examination.

Frequently asked questions

Is tardive dyskinesia permanent?

The risperidone label calls it potentially irreversible and also states it may remit, partially or completely, if antipsychotic treatment is discontinued. Published remission figures range from 2% to 13% in two cohorts, and an editorial put long-term rates at 0 to 73%, with most below 25%. Any change to the antipsychotic is a decision for the prescriber.^{1, 19, 20, 21}

Ingrezza vs Austedo: which works better for tardive dyskinesia?

The trials don't answer that. Each drug was tested against placebo. Valbenazine 80 mg/day lowered AIMS scores by 3.2 points compared with 0.1 on placebo over 6 weeks, and deutetrabenazine lowered them by 3.0 points compared with 1.6 on placebo over 12 weeks in ARM-TD. A 2018 meta-analysis found no head-to-head comparison among these drugs. Doses are set by the prescriber.^{13, 10, 11}

How common is tardive dyskinesia with newer antipsychotics?

In a meta-analysis of 41 studies, prevalence was 20.7% with current second-generation treatment and 30.0% with first-generation treatment. Across 57 randomized trials, the annualized incidence was 2.6% in second-generation arms and 6.5% in first-generation arms.^{2, 3}

Can Abilify (aripiprazole) cause tardive dyskinesia?

Tardive dyskinesia occurs with second-generation antipsychotics, at an annualized incidence of 2.6% across trial arms. In exploratory comparisons, olanzapine and aripiprazole had an advantage over other non-clozapine second-generation drugs, and the abstract gives no separate rate for aripiprazole. New involuntary movements are a reason to contact the prescriber.³

How often should the AIMS test be done?

An American Psychiatric Association practice tool calls for a structured instrument such as the AIMS at least every 6 months in patients at high risk of tardive dyskinesia, at least every 12 months in other patients, and whenever new or worsening movements are detected.⁷

In a crisis: call or text 988, or go to the nearest emergency department. A first episode of psychosis belongs in a hospital-based or coordinated specialty care program.

I start and manage clozapine in an outpatient practice in Chelsea, Manhattan. Details are on the schizophrenia care in NYC page. Clozapine should not be stopped, skipped, or restarted without the prescriber. This page is general information and does not replace an evaluation.