Antipsychotic-Induced Tardive Dyskinesia

Update on Epidemiology and Management

Christian G. Widschwendter; Alex Hofer


Curr Opin Psychiatry. 2019;32(3):179-184. 

In This Article


Estimates of prevalence rates of tardive dyskinesia vary widely, but it is thought that up to 30% of patients treated with antipsychotic drugs may be affected[10] with elderly patients being at higher risk.[11] The annual incidence rate is estimated at approximately 5% with a cumulative 5-year incidence of 25%.[10] Varying information on epidemiology may be attributed to several challenging factors in assessing tardive dyskinesia accurately. First, a patient may not be aware of mild tardive dyskinesia.[12] In addition, less sensitive thresholds for detecting tardive dyskinesia in several clinical assessments are quite likely to detect only moderately to severely affected patients. Second, this motor side-effect is known to fluctuate over time and may therefore not be present at the time of assessment.[13] Third, tardive dyskinesia symptoms can be masked by the hypokinetic effect of antipsychotic medication and substantial dose changes of antipsychotic drugs may elicit withdrawal dyskinesia.[1,14] Fourth, prevalence studies may be an inappropriate approach to reflect antipsychotic-associated tardive dyskinesia risks, as the causal antipsychotic may not be the one prescribed at the time of tardive dyskinesia assessment. Older, high-risk FGA are still in use and thousands of patients suffer from tardive dyskinesia caused by past treatment with these agents.[15,16] Moreover, not all randomized controlled trials (RCT) fulfill a length of at least 3 months and are therefore not suited to identify the occurrence of tardive syndromes.

In a recent meta-analysis, a prevalence rate of approximately 21% for tardive dyskinesia in patients receiving NGA was reported compared with 30% in those treated with FGA.[17] The lowest prevalence rate (approximately 7%) was observed in patients treated with NGA who had never been treated with FGA.[17] Although the prevalence of tardive dyskinesia differed significantly geographically (United States > Europe > Asia) in this meta-analysis, higher age, which has been considered as an established risk factor for developing tardive dyskinesia,[18,19] did not affect the discrimination of NGA and FGA.

Another large meta-analysis including 57 RCT found an annualized tardive dyskinesia incidence rate of 6.5% for FGA vs. 2.6% for NGA.[20] These results are close to the findings of earlier large incidence studies that demonstrated a significant risk reduction when using NGA vs. FGA.[21,22] Notably, although it had previously been suggested that the individual risk for tardive dyskinesia and clinically relevant extrapyramidal symptoms (EPS) increases with higher FGA doses,[23] in this meta-analysis the advantage of NGA was not affected by high doses of FGA. Similarly, both older age[18,19] and female sex,[18,24,25] which had previously been suggested to represent risk factors for developing tardive dyskinesia, were not found to be relevant in this meta-analysis.[20]

Regarding the risk of specific NGA agents to induce tardive dyskinesia, the following rank of lowest to highest risk of tardive dyskinesia has been suggested by Kenney et al.:[3] clozapine < quetiapine < aripiprazole < olanzapine = ziprasidone < risperidone. In contrast, the meta-analysis found annualized incidence rates of 1.7% for aripiprazole, 2.4% for risperidone, 2.5% for quetiapine, 2.9% for olanzapine, and 3.5% for ziprasidone.[20] Owing to methodological issues, tardive dyskinesia rates for individual NGA in this meta-analysis are not directly comparable and therefore should be interpreted with caution. However, the advantage of nonclozapine NGA over FGA was most prominent with olanzapine and aripiprazole and smallest with quetiapine. Significantly lower annualized rate ratios with aripiprazole and olanzapine relative to all other nonclozapine NGA have been found. Moreover, annualized rate ratios with olanzapine were also significantly lower relative to risperidone and quetiapine. These findings of olanzapine and clozapine showing the lowest risk for tardive dyskinesia are in line with an earlier clinical tardive dyskinesia incidence study.[21] In contrast, the superiority of NGA had neither been found in the "Clinical Antipsychotic Trials of Intervention Effectiveness" (CATIE) study,[26] nor in the "Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia Study" (CUtLASS-1).[27] However, when interpreting these data, one has to consider methodological limitations of these two studies. CATIE did not include patients with a history of tardive dyskinesia into the FGA arms, but into the NGA arms. Similarly, sulpiride was the antipsychotic agent used most frequently (58%) among FGA in CUtLASS-1. Owing to its rapid dissociation half-time at the dopamine receptor and consistent with its exceptionally low EPS risk,[27] this antipsychotic drug has previously been categorized as atypical[28,29] despite acting primarily as a dopamine D2 antagonist. Moreover, these two studies were designed to focus on antipsychotic efficacy and to a lesser extent to specific questions of EPS.