However, it soon became clear that EPS can be mistaken for or worsen psychotic symptoms, are sometimes irreversible or lethal, necessitate additional burdensome side effects from antiparkinsonian agents, can be disfiguring and stigmatizing, and have been shown to influence compliance, relapse and rehospitalization. ^2-4 As a result, EPS dominated concerns about tolerability of antipsychotic drugs and fueled efforts for new drug research and development.

In 1988, Kane et al. reported that clozapine had broader efficacy in schizophrenia with negligible EPS, stimulating the search for other second generation antipsychotics (SGAs) with less toxicity.⁵ Industry-sponsored clinical trials suggested that newer SGAs were superior to first generation antipsychotics (FGAs) in reducing psychotic symptoms and causing fewer EPS. ^6-13 Cumulative and convincing evidence confirming reduced liability for EPS with SGAs contributed to their market dominance and fostered the concept of “atypicality” in their mechanism of action. ^14-21

Subsequent studies raised questions about the advantages of SGAs seen in earlier trials. Although haloperidol was a very reasonable choice as a comparator in industry-sponsored trials because of its widespread use in clinical practice, new evidence suggested that the advantages of SGAs in reducing EPS were diminished when lower doses or lower-potency FGAs are used, or if prophylactic antiparkinsonian drugs are administered.^{19, 22-27}

In view of these conflicting findings, the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) Schizophrenia Trial offered an opportunity to address the liability for movement disorders between first and second generation antipsychotics.^28-30 The strengths of the CATIE study include its large sample size, diverse representation of clinical settings, independence from industry sponsorship, and a well-controlled, double-blind, head-to-head treatment comparison. The rationale, design, methods and statistical analysis of the CATIE trial have been described previously.²⁸ Briefly, CATIE was designed to address the overall effectiveness between four SGAs (olanzapine, risperidone, quetiapine and ziprasidone) and a mid-potency FGA (perphenazine) including the influence of specific EPS on tolerability and effectiveness.

The differential diagnosis of dystonia is extensive including primary genetic disorders and secondary forms including neurodegenerative disorders, structural abnormalities of the brain, and metabolic and toxic etiologies.^34-37 Drug-induced dystonia is distinguished by recent antipsychotic treatment, negative family history, focal and non-progressive course, and absence of associated neurologic signs. The co-occurrence of dystonia or other EPS with behavioral disturbances narrows the differential to disorders associated with both features, e.g., Huntington’s and Wilson’s disease, neuroacanthocytosis, etc. Although most often associated with the initiation of antipsychotics, dystonia may also be seen if dosages are increased, a second antipsychotic is added, for the few days each time after long-acting injectable antipsychotics are administered, ³⁴ if another drug, e.g., paroxetine, is added that inhibits antipsychotic metabolism, or following the discontinuation of antiparkinsonian agents. Dystonia also occurs in a tardive form, which may respond to anticholinergic treatment, but first appears or worsens rather than resolves when antipsychotic drugs are discontinued. The relationship between the acute and tardive forms is unknown. Apart from antipsychotic drugs, dopamine antagonists used as anti-emetics or sedatives (prochlorperazine, metoclopramide), anticonvulsants, antimalarials, stimulants and serotonergic agents have been implicated. Dystonias are encountered with levodopa during the course of treatment of Parkinson’s disease. Toxins (manganese, carbon monoxide, carbon disulfide) have also been implicated. Finally, dystonia was erroneously ascribed to psychogenic factors for much of the twentieth century, and patients with drug-induced dystonia still may be dismissed as hysterical if clinicians are unaware that it is often intermittent and fluctuates with stress and relaxation. However, some patients may feign symptoms to avoid taking antipsychotic drugs or to obtain anticholinergic drugs for euphoric effects.³³

Dystonia is the earliest EPS to be seen after antipsychotic drug administration. Usually observed within a few hours of a single dose especially following parenteral administration, dystonias may appear after a delay of several hours to a few days. ³² In 95% of cases, dystonia appears within the first five days of treatment after a drug is started or increased in dosage. ^{32, 33} Dystonic reactions last a few seconds or several hours and are often recurrent even after a single drug dose. After drug discontinuation, dystonia resolves within 24-48 hours. ³⁴

Dystonia is less frequent than parkinsonism or akathisia, occurring in 2% to 5% of patients generally. ^32-34 However, in young men receiving high potency antipsychotics parenterally, the frequency approaches 90% in some series. ³²

Patient risk factors for dystonia include age, male sex, race, previous dystonic reactions, family history of dystonia, cocaine use, mood disorders, hypocalcemia, hypoparathyroidism, hyperthyroidism and dehydration.^{33, 34, 38} Children and young adults are highly vulnerable, whereas drug-induced dystonia is rare over 45 years of age. ³³

Drug dosage and potency have been associated with risk of dystonia.^{33, 34, 39} Moderate to high doses of antipsychotics are associated with dystonia, whereas low or very high doses are less often involved. ³⁸ Antipsychotics with weak dopamine antagonism and prominent anticholinergic effects diminish the risk of dystonia, while the newer SGAs appear to have reduced liability as well. ^{14, 15} In a sample of published industry-sponsored trials, haloperidol had up to four times greater risk of causing dystonia than SGAs, ^{6, 8, 40} with clozapine unlikely to cause it at all.

There were only six cases of acute dystonia in the CATIE trial (6/1460 or 0.4%) that were not present at baseline, four of which resulted in treatment discontinuation. Of these six cases, none were receiving olanzapine, one was receiving perphenazine, one was receiving quetiapine, one was receiving risperidone, and three were receiving ziprasidone. ³⁰ The CATIE data suggest that in older patients with chronic schizophrenia, use of a mid-potency FGA at modest doses presents no greater risk for dystonia than SGAs.

Moreover, Satterthwaite et al. performed a meta-analysis of studies comparing intramuscular haloperidol versus three injectable SGAs, and found that while the SGAs were significantly less likely to induce dystonia, this advantage was abolished when an anticholinergic agent was administered with haloperidol. ³¹ However, the use of prophylactic anticholinergic medications to prevent dystonia had been controversial prior to the introduction of the SGAs due to atropinic side effects. But in young or other high risk patients receiving parenteral, high potency antipsychotics, or in paranoid or other patients ambivalent about treatment, the benefits of preventing a dystonic reaction far outweigh potential risks. ⁴¹

Dystonia is responsive within 10 to 20 minutes to anticholinergic or antihistaminic agents administered intramuscularly and repeated if necessary. Intravenous administration is necessary only in an emergency, e.g., for laryngeal stridor. Benzodiazepines have been used in some cases. If response is not achieved, a search for underlying disorders, e.g., hypocalcemia, should be conducted or tardive dystonia considered.⁴² After dystonia is suppressed, oral anticholinergics are continued for 24 to 48 hours if the antipsychotic is discontinued, or for at least several days if antipsychotic treatment is continued with gradual tapering to prevent recurrence. However, at-risk patients who tolerate anticholinergic drugs may need continued prophylaxis.

It is important to differentiate drug-induced parkinsonism from negative or withdrawal symptoms of schizophrenia and psychomotor retardation associated with depression. It can be difficult to distinguish drug-induced from idiopathic Parkinson’s disease. Parkinson’s disease is usually asymmetric in symptom presentation, has a slow progressive course, may have associated dysautonomia, should precede treatment and not resolve after antipsychotic drugs are discontinued, and may show nigrostriatal degeneration on dopamine transporter scans and sympathetic dysregulation on ¹²³IMIBG cardiac scintigraphy. ⁴⁴ Interestingly, olfactory deficits have recently been reported in drug-induced parkinsonism.⁴⁵ The differential diagnosis also includes essential tremor and other causes of parkinsonism, including vascular parkinsonism which also tends to be asymmetric, and parkinsonism caused by other drugs including anti-emetics, tetrabenazine, and reserpine.

Although dopamine receptor blockade occurs within a few hours after administration of antipsychotic drugs, the onset of parkinsonism may be delayed from days to weeks, with 50% to 75% of cases occurring within one month and 90% within three months. ³² Parkinsonism may also occur after doses are increased, a second antipsychotic is added, anticholinergic drugs are discontinued, or another drug is added that reduces central dopamine activity or increases plasma levels of the antipsychotic. In most cases, symptoms are reversible in days or weeks, but occasionally, especially in the elderly, or if long-acting injectable antipsychotics are used, symptoms may last for months. In about 15% of cases, parkinsonism may persist, raising the possibility of underlying Parkinson’s disease. ⁴⁴

The incidence of drug-induced parkinsonism is variable depending upon risk of the population studied, sensitivity of diagnosis and potency of the drugs used, but has been estimated to occur in at least 10 to 15% of patients treated in routine practice with FGAs. ³² It is the second most common cause of parkinsonism after Parkinson’s disease. The risk of drug-induced parkinsonism has been associated with increasing age, female gender, dementia, HIV infection, and pre-existing extrapyramidal disease or family history of Parkinson’s disease.^{32, 44, 46}

Although parkinsonism has correlated with increased dosages, potency and reduced anticholinergic properties of antipsychotics,⁴⁷ dose response relationships have not always been clear in view of differences in individual susceptibility. In industry-sponsored trials, haloperidol has been associated with two to four times the risk of parkinsonism compared with SGAs (22% to 38% versus 4% to 14%). ^{6, 8, 14, 15, 40, 48, 49} But judging by severe motor worsening experienced by patients with Parkinson’s disease after receiving antipsychotics, even SGAs may cause significant parkinsonism in susceptible individuals except for clozapine and quetiapine. ^{14, 15} Patients with Lewy body dementia may experience a potentially lethal “neuroleptic sensitivity syndrome” when exposed to FGAs or SGAs.

In contrast to previous studies using haloperidol as the representative FGA, there were no significant differences in the CATIE trial between perphenazine and SGAs in the proportion of patients exhibiting parkinsonism defined as a mean score of one or more on the Simpson-Angus Extrapyramidal Signs Scale (SAS) (Table 1). ^{28, 50} However, significantly more patients discontinued perphenazine (8%) than SGAs (2% to 4%) as a result of all EPS combined, and perphenazine (10%) had a significantly higher rate of concomitant anticholinergic drug use relative to SGAs (3% to 9%). Quetiapine itself was associated with a higher rate of anticholinergic side effects (31% versus 20 to 25%), and conversely, had the lowest rate of concomitant anticholinergic drugs.

In a more sensitive analysis of the CATIE data, examination of the proportion of patients showing no evidence of parkinsonism at baseline who met at least one of three criteria for parkinsonism (scored at least one moderate symptom or two mild symptoms on the SAS scale, treatment discontinuation for parkinsonism, treatment with antiparkinsonian medication) revealed again no significant differences between treatment groups (Table 2). ^{29, 30} Rates of parkinsonism were 37 to 44% among the four SGAs and 37% for the FGA perphenazine.

Prophylaxis of parkinsonism with anticholinergic drugs is less compelling than for dystonia and introduces significant risk of atropinic toxicity. Given the delayed onset, close monitoring for parkinsonian symptoms with prompt consideration of lowering dosages or switching to lower risk antipsychotics takes precedence, albeit with attendant risk of relapse. If a given antipsychotic is effective and cannot be changed, and if parkinsonism persists, treatment may include anticholinergic drugs or amantadine. However, there is limited controlled evidence for the use of these agents. ⁴² Specific dopaminergic therapy is ineffective, due to ongoing drug-induced blockade of dopamine receptors, and raises the risk of worsening psychotic symptoms. Once patients have been maintained on adjunctive antiparkinsonian therapy for three to six months, cautious tapering may be attempted.³² However, several studies have shown 62% to 96% of patients may experience worsening parkinsonism following antiparkinsonian drug discontinuation. ⁵¹

Acute drug-induced akathisia must be distinguished from tardive akathisia, neurodegenerative conditions, and drug intoxication and withdrawal states. Restlessness can be observed with other medications including serotonin uptake inhibitors, calcium channel blockers, anti-emetics, anti-vertigo agents or sedatives. Akathisia resembles restless legs syndrome (Ekbom’s syndrome) which occurs during relaxation, rest or sleep mostly in the evening or night. Misdiagnosis of restless legs syndrome can lead to inappropriate prescription of dopamine agonists which could worsen psychosis leading to increased antipsychotic use further compounding akathisia. ⁵⁴ Finally, distinguishing akathisia from agitation and anxiety is challenging but crucial in determining whether to continue antipsychotic treatment.

Akathisia may begin within several days after treatment but usually increases with duration of treatment, occurring in up to 50% of cases within one month and 90% of cases within three months. ³² Akathisia should resolve after drug discontinuation, but could temporarily worsen or persist in withdrawal or tardive forms.

Estimates of the incidence of akathisia vary between 21% to 75% across studies of FGAs, with an estimated prevalence on average of at least 20% to 35% of patients in routine practice, depending upon the susceptibility of the sample population, sensitivity of diagnosis, and the potency of drug treatment. ⁵³

Patient risk factors for akathisia have not been well established, but increasing age, female sex, negative symptoms, cognitive dysfunction, iron deficiency, prior akathisia, concomitant parkinsonism, and mood disorders may entail greater risk. ^{54, 55} In most but not all clinical trials, SGAs have resulted in a significantly lower incidence of akathisia compared with FGAs. ⁵³ In a representative sample of trials, akathisia developed at an incidence rate of about two to seven times higher with haloperidol (15% to 40%) compared with SGAs (0% to 12%). ^{6, 8, 14, 15, 40, 48, 49}

In the CATIE study, there were no significant differences among treatment groups in the incidence of akathisia defined by scores of three more on the global assessment of the Barnes Akathisia Rating Scale (BAS) (Table 1). ^{28, 56} On subsequent analysis, examination of the proportion of patients showing no evidence of akathisia at baseline who met at least one of three criteria for akathisia (scored at least mild symptoms on the BAS global item, akathisia given as the reason for starting any medication, or discontinued antipsychotic medication due to akathisia) revealed no significant differences between treatment groups (Table 2). ^{29, 30} Rates of akathisia ranged from 26% to 34% for the SGAs and 35% for perphenazine.

There are no data on prophylaxis and given its subacute onset, close observation for early signs is the best preventive measure. Once developed, akathisia should prompt reassessment of antipsychotic therapy, with reduction in dosage, discontinuation, or switching to a less potent dopamine antagonist, all of which incur the risk of psychotic exacerbation or relapse. Evidence of treatments derives mostly from small, short-term clinical trials without active comparative groups. ^{55, 57} Beta-adrenergic blockers that are lipophilic and target beta₂-receptors have been effective in some studies though limited by hypotension, bradycardia and medical contraindications. Anticholinergics have been used, but are felt to be most effective in the presence of concomitant parkinsonism. Benzodiazepines have been useful due to their anxiolytic and sedative properties. Amantadine may be effective in some cases. Recently, 5-HT_2A receptor antagonists have attracted interest, with mirtazepine showing equal efficacy in treating akathisia compared to propranolol with better tolerability.⁵⁵

Catatonic symptoms that have been associated with antipsychotics include akinesia, stupor, and mutism (akinetic mutism), and less often catalepsy and waxy flexibility.^{62, 63} The more complex and qualitative catatonic behaviors typical of schizophrenia (stereotypies, echophenomena, verbigeration, automatic obedience) are rarely seen in drug-induced cases. In an unknown percentage of patients, antipsychotics could transform pre-existing catatonia into a more malignant form, consistent with NMS. ^{64, 65}

The differential diagnosis of catatonia includes a broad range of neurodegenerative, developmental, metabolic, toxic, infectious and structural conditions affecting brain function.⁶⁶ Catatonia overlaps and may be obscured by drug-induced parkinsonism. Catatonia also occurs in schizophrenia and mood disorders leading to the conundrum of the “catatonic dilemma”,⁶⁷ in which it may be difficult to distinguish catatonia due to the psychiatric disorder requiring antipsychotic drugs from the treatment itself. This is best resolved by discontinuing antipsychotics leading to resolution of the iatrogenic cases.

Acute catatonia develops within hours to days after drug exposure and is expected to resolve in a similar period of time after drug discontinuation. In some cases, e.g., after resolution of NMS, mixed catatonic/parkinsonian features may persist for weeks or even months if not treated with electroconvulsive therapy (ECT).⁶⁸ The onset of catatonia has been reported to occur with antipsychotics after discontinuation of benzodiazepines or antiparkinsonian agents. The incidence of drug-induced catatonia is unknown. Stubner et al reported only five cases of “catatonic neuroleptic syndrome” among 86,439 patients receiving antipsychotics in a retrospective drug surveillance program.⁶⁹

Patient-related risk factors for drug-induced catatonia include past episodes and pre-existing catatonic symptoms observed in schizophrenia or mood disorders. Catatonia is mostly observed in association with high potency drugs.⁶³ The relative liability between FGAs and SGAs has not been studied. However, SGAs are not without risk considering published case reports documenting both occurrence or worsening of catatonia with the newer drugs.¹⁴ However, SGAs have also been proposed as treatments for catatonia.⁷⁰ In animal models, SGAs show an advantage over FGAs in reducing the risk of catalepsy,¹⁴ raising the possibility that these newer agents would be less likely to cause catatonia, and may be particularly useful in patients with pre-existing risk of catatonia. In fact, use of SGAs in trials that included patients with catatonic schizophrenia did not result in adverse effects in this subgroup and was beneficial in two of the studies. ¹⁴

There are no data on whether prophylaxis with benzodiazepines or other agents would prevent catatonia in vulnerable patients. A more conservative approach would be to avoid using antipsychotics and specifically treat pre-existing catatonia with benzodiazepines or ECT in patients presenting with the syndrome. For patients at risk based on past history, clinicians should consider using other agents for the underlying psychiatric disorder, e.g., lithium for mania. However, patients who require antipsychotic treatment merit careful observation and monitoring for incipient symptoms. Treatment of drug-induced catatonia has not been rigorously studied, but must include prompt discontinuation of the offending agent as patients with immobility and non-responsiveness are at high risk for medical complications, or may progress to NMS. Specific treatment includes benzodiazepines, but recent evidence has reinforced the utility of amantadine or memantine.^{63, 71} Use of direct dopaminergic agonists have not been studied. Patients who fail to respond to these measures may benefit from ECT, which offers the advantage of resolving both catatonia and the underlying psychosis or mood disorder.

The differential diagnosis of NMS includes other disorders with elevated temperatures and encephalopathy, such as malignant catatonia due to psychosis, ⁶⁵ central nervous system infections, ⁸² benign extrapyramidal side effects, agitated delirium, heatstroke, ⁸³ serotonin syndrome, and withdrawal from dopamine agonists, sedatives, or alcohol. In the peri-operative setting, NMS may be confused with malignant hyperthermia of anesthesia.⁸⁴ NMS may result from treatment with dopamine antagonists used in medical settings. Although no laboratory test is diagnostic for NMS, thorough assessment and neuroimaging studies are often necessary to exclude other serious medical conditions.

NMS may develop explosively within hours but usually evolves over days; about two-thirds of cases occur during the first 1–2 weeks after drug initiation. ⁷⁹ Once dopamine-blocking drugs are withheld, two-thirds of NMS cases resolve within 1–2 weeks, with an average duration of 7–10 days. ⁷⁷ Patients may experience prolonged symptoms if injectable long-acting drugs are implicated. Occasional patients develop a residual catatonic/parkinsonian state after acute metabolic symptoms subside that can last for weeks to months unless ECT is administered. ⁶⁸ NMS is potentially fatal due to renal failure, sudden cardiorespiratory arrest, disseminated intravascular coagulation, pulmonary emboli, or aspiration pneumonia.

The incidence of NMS is about 0.02% among patients treated with antipsychotic drugs.⁷⁹ Potential risk factors have been investigated in small case control studies, and include dehydration, exhaustion, agitation, catatonia, previous episodes, and high doses of high-potency drugs given parenterally at a rapid rate.⁷⁷ The effect of concurrent administration of multiple antipsychotics, lithium and selective serotonin or serotonin–norepinephrine reuptake inhibitors in enhancing the risk of NMS has been suggested but is unproven. ⁸⁵ Most of these factors are common in psychiatric practice and therefore of limited value in predicting the rare occurrence of NMS, and do not outweigh the value of antipsychotic drugs when indicated.

NMS has been associated with all currently marketed antipsychotic drugs, but correlates with the use of high potency agents. Haloperidol has accounted for about half of all reported cases. The SGAs have also been implicated in isolated case reports, but a few large scale surveys suggest reduced risk compared with FGAs.⁷⁹ There are reports of atypical or milder forms of NMS associated with SGAs, but in fact, NMS has always varied in severity even with FGAs.

Treatment consists of early diagnosis, discontinuing dopamine antagonists, and supportive medical care. Benzodiazepines, dopamine agonists, dantrolene, and ECT have been advocated, but randomized, controlled trials comparing these agents with supportive care have not been done and may not be feasible because NMS is rare, often self-limited after drug discontinuation, and heterogeneous in presentation, course, and outcome. ⁷⁸ We have proposed that these agents may be considered empirically in individual cases, based on symptoms, severity, and duration of the episode. ⁷⁸

TD presents as a polymorphous involuntary movement disorder.^{32, 87, 88} In its most common form, TD is characterized by involuntary, non-rhythmic, repetitive, purposeless hyperkinetic movements. Most often, TD affects orofacial and lingual musculature (“buccolinguomasticatory syndrome”) with chewing or bruxism of the jaw, protrusion, curling or twisting of the tongue, lip smacking, puckering, sucking and pursing, retraction, grimacing or bridling of the mouth, bulging of the cheeks, or eye blinking and blepharospasm. Choreoathetoid movements of the fingers, hands, upper or lower extremities are common. Axial symptoms affecting the neck, shoulders, spine or pelvis may be observed. Dyskinesias can affect breathing, swallowing or speech.

TD may present with other than choreoathetoid symptoms that can be difficult to distinguish from acute EPS. These may co-exist with classic TD symptoms, but represent separate subtypes with increased risk of progression, persistence and severe disability. For example, tardive dystonia, estimated to occur in up to 1% to 4% of treated patients,⁴² may be more generalized and disabling than TD, and may respond to anticholinergic agents. Akathisia, as well as other movement disorders, also occur as tardive variants. ⁸⁹ Dyskinesias increase with emotional arousal, activation or distraction, and diminish with relaxation, sleep, or volitional effort. As a result, symptoms of TD fluctuate over time, such that repeated measurements are necessary for reliable assessment of severity and persistence.

Although different diagnostic schemes for TD have been proposed, criteria proposed by Schooler and Kane have been widely accepted. ⁹⁰ These authors require at least three months of exposure to antipsychotics, ratings of at least moderate severity in one or more body areas, or ratings of at least mild severity in two or more body areas, and the absence of other conditions that might cause dyskinesias. Ratings are obtained using the Abnormal Involuntary Movement Scale (AIMS). ⁹¹

A careful neurologic evaluation is indicated for all patients with new-onset dyskinesias. Clues to neurological causes include family history, sudden onset or progressive course, associated medical or neurological abnormalities, and asymmetry. The differential diagnosis of TD includes acute EPS, transient withdrawal dyskinesias, spontaneous dyskinesias associated with schizophrenia and aging, Wilson’s disease, Huntington’s disease, Sydenham’s chorea, chorea gravidarum, Fahr’s syndrome, systemic lupus, senile chorea, Meige syndrome, edentulous chorea, infarction or lesions of the basal ganglia, post-anoxic and encephalitic states, Tourette’s syndrome, torsion dystonia, spasmodic torticollis, hyperthyroidism, hypoparathyroidism, and conversion disorder. Several drugs and toxins are associated with dyskinesias including caffeine, phenytoin, estrogens, levodopa, antidepressants, antihistamines, stimulants, and manganese poisoning.

The onset of TD occurs insidiously over three months or more of treatment and may begin with tic-like movements or increased eye blink frequency. TD is often suppressed or masked by ongoing antipsychotic treatment, becoming apparent only when treatment is reduced, switched or discontinued.

The natural course of TD is unclear. Early studies showed that withdrawal of antipsychotics may lead to an initial worsening of TD in 33% to 53% of patients (unmasking or withdrawal dyskinesia), but with long-term follow-up, 36% to 55% of patients eventually improve, which led to recommendations for drug reduction or withdrawal. ⁹² However, complete and permanent reversibility beyond the withdrawal period is rare; Glazer et al found that only 2% of patients showed complete reversal of TD after drug discontinuation. ^{93, 94} In a meta-analysis of treatments for TD, Soares and McGrath ⁹⁵ reported that 37.3% of patients assigned to placebo across studies showed some improvement in TD, but concluded that insufficient evidence existed to support drug cessation or reduction as effective treatments for TD, especially when contrasted with robust evidence for the risk of psychotic relapse. ⁹⁶

Data on the change in prevalence of TD during continued treatment have been inconclusive with some studies showing an increase, others a decrease, or no change at all.⁹⁷ However, prevalence rates obscure the dynamics of TD in individual patients. Roughly 50% of patients have persistent TD symptoms, 10% to 30% have a reduction and 10% to 30% show increased symptoms during treatment. ⁹⁸ Long-term studies estimated that from 2% to 23% of patients show loss of observable TD symptoms during treatment with FGAs.⁹⁷ Similarly, studies of SGAs have shown reduction of TD ratings, with some showing greater reductions, less, or no difference, compared with first generation agents. ⁹⁷ Improved outcome of TD correlates in some studies with younger age, lower doses, reduced duration of drug treatment and dyskinesia, and increased length of follow-up.

In the CATIE trial, there was a significant decline in ratings of TD severity among 200 patients with TD at baseline, but there were no significant differences between SGAs in the decline in AIMS scores. ⁹⁷ Fifty-five percent of these patients met criteria for TD at two consecutive visits post-baseline, 76% met criteria at some or all post-baseline visits, 24% did not meet criteria at any subsequent visit, 32% showed ≥50% decrease and 7% showed ≥50% increase in AIMS scores. Thus, similar to past evidence on the course of TD during treatment with FGAs or SGAs, most patients showed either persistence or fluctuation in observable symptoms.

The frequency of TD occurring has been extensively studied but with varied results affected by spontaneous dyskinesias, sensitivity and definitions of diagnosis, susceptibility of the patient sample, fluctuation in TD symptoms, the dynamics of emergence and remission, and the influence of drug treatment on suppression and unmasking of TD. Estimates of the cross-sectional prevalence of TD with FGA antipsychotics range from 3% to 62% with a mean of 24%. ^{87, 99} Several studies have shown a cumulative incidence of TD of about 5% annually.⁸⁷ In studies of first-episode patients with limited prior drug exposure, the incidence of TD was 6% to 12% in the first year even when low doses of antipsychotics were used.^{100, 101} The annual incidence of TD in patients over the age of 45 years was 25% to 30% after one year of treatment.¹⁰²

Previous studies of TD risk have suggested an association with increasing age, female gender, psychiatric diagnosis, longer duration of antipsychotic treatment, higher cumulative drug doses, concomitant drug treatments, higher ratings of negative symptoms and thought disorder, greater cognitive impairments, presence of acute EPS, and diabetes. The patients with TD at baseline in the CATIE trial were found to be significantly older, and had been treated with an antipsychotic significantly longer, were more likely to be currently treated with an FGA and to be currently treated with an anticholinergic agent, had greater neurocognitive impairment, higher levels of psychopathology, and higher ratings of parkinsonism and akathisia compared with non-TD patients.¹⁰³ Gender, race, and ethnicity were not differentially distributed between patients with TD versus those without TD. Patients with diabetes or hypertension did not have higher rates of TD. However, alcohol and drug abuse or dependence were significantly associated with TD.

Differences in liability for TD between FGAs and SGAs have been studied extensively. In contrast to the incidence of TD with FGAs, numerous industry sponsored trials of SGAs found a significant six to 12 fold reduction in risk for TD. ^{12, 27, 104-106} Correll et al, in a meta-analysis, found a 4.6% reduction in the risk of TD for SGAs compared with studies using haloperidol, but later found only a 1.6% difference in attributable risk when studies using mid-potency FGAs were included. ^{12, 27} It remains unlikely that clozapine causes TD. ¹⁰⁷

In contrast to industry studies using haloperidol, there were no significant differences among groups in the CATIE trial receiving perphenazine or four SGAs in the incidence of TD defined by scores of two or more on the AIMS global severity score (Table 1).²⁸ In a second analysis, patients were considered to have persistent TD if they met full Schooler-Kane (S-K) criteria. ^{29, 90} Analyses were also conducted using modified S-K criteria such that meeting the AIMS criteria on only one assessment was required, i.e., “probable” TD. Few patients who had no evidence of TD at baseline met full S-K TD criteria during treatment (1.1% to 4.5% receiving SGAs and 3.3% receiving perphenazine (Table 2). The proportion of patients who met modified S-K TD criteria ranged from 8.3% to 9.6% with SGAs and 11.8% for perphenazine. There were no statistically significant differences between treatment groups on any TD indicator.

Since there is no uniformly effective treatment for TD, it is important to minimize the risk of TD by early detection (Figure 1). Some preventive principles are to confirm the indication for antipsychotics, use conservative doses opting for lower potency agents, inform patients and caregivers of risk, assess on a regular basis for incipient signs, consider differential diagnosis and reconsider drug treatment if symptoms emerge.

The first step in treatment is the decision on antipsychotic treatment. Although drug withdrawal had been recommended in the past as increasing the odds of resolution of TD, about 33% to 53% of patients will experience worsening of dyskinesias initially, 36% to 55% may show improvement over time,⁹² but few will show complete resolution of symptoms,⁹⁴ and the risk of psychotic relapse is 53% within nine months.⁹⁶

A second option in a patient with good control of psychotic symptoms is to decide not to change the antipsychotic, try to gradually reduce the dose, inform patients and caregivers of risks, document the decision and monitor carefully. In most cases, TD is not progressive even with continued antipsychotic treatment, although symptoms may worsen in a few cases (7% in the CATIE trial).⁹⁷ Another alternative is to switch antipsychotics; more potent antipsychotics, like haloperidol, can be used to suppress disabling symptoms of TD in about 67% of patients. ⁹⁸ SGAs have also been associated with reduction of TD symptoms; ^{14, 15} results of the CATIE trial indicated that while most patients treated with SGAs show a persistent (34%) or fluctuating course (42%), 24% did not meet criteria on any visit at follow-up, and 32% showed greater than 50% reduction in AIMS scores.⁹⁷ Clozapine has been recommended for suppressing tardive dystonia. While there have been speculations that SGAs may increase the possibility of remission on an active or passive basis, existing data are inconclusive whether recovery rather than suppression occurs during treatment with SGAs or FGAs.

After discontinuing or optimizing antipsychotic therapy, there are a large number of agents that have been tested in the treatment of TD (Figure 1).^{98, 108} These include dopamine depleting agents, dopamine agonists, noradrenergic agonists and antagonists, GABAergic drugs (benzodiazepines, valproate, baclofen, levetiracetam), lithium, calcium channel blockers, serotonergic drugs, vitamins (vitamin E), branched-chain amino acids, neuropeptides, cholinergic precursors and cholinesterase inhibitors, ECT, and botulinum toxin or surgical intervention (for tardive dystonia). Anticholinergic agents can worsen TD (expect for tardive dystonia) with improvement noted in 60% of cases after discontinuation.⁹⁸

Several hypotheses have been proposed to explain the pathophysiology of TD including dopamine receptor hypersensitivity, GABA insufficiency, and structural damage resulting from increased catecholamine metabolism and oxidative free radical production. ¹⁰⁹ Another hypothesis proposes that TD results from damage to striatal cholinergic interneurons due to the loss of dopamine mediated inhibition. ¹¹⁰ If correct, this implies that cholinesterase inhibitors or cholinergic agonists may be effective in suppressing TD by directly enhancing post-synaptic cholinergic activity, thereby compensating for the loss of pre-synaptic cholinergic neurons. Several preliminary trials have explored the use of cholinesterase inhibitors with mixed results. ^111-113 However, this hypothesis is supported by animal and human evidence correlating cholinergic mechanisms with the delay in onset, irreversibility, age-related risk, worsening with anticholinergic drugs, and reduced risk with the newer SGA drugs, suggesting further investigation of cholinergic mechanisms may be worthwhile.

In summary of specific EPS, dystonia is less frequent than before but may be encountered with high potency drugs. Parkinsonism is quite debilitating but is more likely to be seen when antipsychotics are used in elderly patients. Akathisia remains a problem even with newer SGA drugs. TD is most often mild but irreversible with limited options for treatment. Catatonia is rare and associated with high potency drugs. NMS is extremely rare, but early diagnosis is crucial for a positive outcome. EPS are less likely to occur with SGAs compared with haloperidol, but clozapine and quetiapine, perhaps because of fast dissociation of dopamine receptor binding, are least likely to induce EPS.