In this guideline the term "antipsychotic" refers to multiple medications (Table 2), including the first-generation antipsychotic medications and the second-generation agents clozapine, risperidone, olanzapine, quetiapine, ziprasidone, and aripiprazole. In addition to having therapeutic effects, both first- and second-generation antipsychotic agents can cause a broad spectrum of side effects. Side effects of medications are a crucial aspect of treatment because they often determine medication choice and are a primary reason for medication discontinuation.

Side effects can complicate and undermine antipsychotic treatment in various ways. The side effects themselves may cause or worsen symptoms associated with schizophrenia, including negative, positive, and cognitive symptoms and agitation (772). In addition, these side effects may contribute to risk for other medical disorders (50, 773). Finally, these side effects often are subjectively difficult to tolerate and may affect the patient's quality of life and willingness to take the medication (17).

Most side effects of antipsychotic treatment result from actions on neurotransmitter systems and anatomic regions beyond those involved in mediating the intended therapeutic effects of the medication. Among the antipsychotic medications, differences in the risk of specific side effects are often predictable from the potencies and receptor binding profiles of the various agents. Some side effects result from receptor-mediated effects within the central nervous system (e.g., extrapyramidal side effects, hyperprolactinemia, sedation) or outside the central nervous system (e.g., constipation, hypotension), whereas other side effects are of unclear pathophysiology (e.g., weight gain, hyperglycemia). Side effects that are similar across several classes of agents, including both first- and second-generation antipsychotics, are discussed in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications". These shared side effects include neurological effects (i.e., acute and chronic extrapyramidal effects, neuroleptic malignant syndrome), sedation, cardiovascular effects (i.e., hypotension, tachycardia, and conduction abnormalities), anticholinergic and antiadrenergic effects, weight gain and glucose and lipid metabolic abnormalities, and sexual dysfunction. Side effects unique to particular agents are discussed in the respective sections concerning those agents, as are other unique implementation issues. Suggested approaches for monitoring and clinical management of the side effects of antipsychotic medications are outlined in Table 1.

a) First-generation agents

First-generation antipsychotic agents effect their therapeutic action, as well as their extrapyramidal side effects, primarily by blocking dopamine, subtype 2 (D₂), receptors in mesolimbocortical and nigrostriatal areas of the brain (774).

The evidence supporting the effectiveness of first-generation antipsychotic medications in reducing psychotic symptoms in acute schizophrenia comes from studies carried out in the 1960s (775, 776) as well as numerous subsequent clinical trials (99, 777). Each of these studies compared one or more antipsychotic medications with either placebo or a sedative agent, such as phenobarbital (778), that served as a control. Nearly all of these studies found that the antipsychotic medication was superior for treating schizophrenia. These studies demonstrated the efficacy of first-generation antipsychotic medications for every subtype and subgroup of patients with schizophrenia. Moreover, in reviews of studies that compared more than one first-generation antipsychotic medication, Klein and Davis (779) and Davis et al. (777) found that, with the exception of mepazine and promazine, all of these agents were equally effective, although there were differences in dose, potency, and side effects of the different drugs.

First-generation antipsychotic medications are effective in diminishing most symptoms of schizophrenia. In a review of five large studies comparing an antipsychotic to placebo, Klein and Davis (779) found that patients who received an antipsychotic demonstrated decreases in positive symptoms, such as hallucinations, uncooperativeness, hostility, and paranoid ideation. Patients also showed improvement in thought disorder, blunted affect, withdrawal-retardation, and autistic behavior.

These findings—along with decades of clinical experience with these agents—indicate that first-generation antipsychotic treatment can reduce the positive symptoms (hallucinations, delusions, bizarre behaviors) and secondarily reduce the negative symptoms (apathy, affective blunting, alogia, avolition) associated with schizophrenic psychosis (297). In placebo-controlled comparisons (99, 776), approximately 60% of patients treated with first-generation antipsychotic medication for 6 weeks improved to the extent that they achieved complete remission or experienced only mild symptoms, compared to only 20% of patients treated with placebo. Forty percent of medication-treated patients continued to show moderate to severe psychotic symptoms, compared to 80% of placebo-treated patients. Eight percent of medication-treated patients showed no improvement or worsening, compared to nearly one-half of placebo-treated patients. A patient's prior history of a medication response is a fairly reliable predictor of how the patient will respond to a subsequent trial (780, 781).

Since the advent of second-generation antipsychotic medications, research on first-generation agents has reduced considerably. In recent years, randomized, controlled studies of the efficacy of first-generation agents for acute treatment have focused on dosing strategies and defining the most effective dose range to maximize symptom response and minimize side effects. These studies have consistently found that modest doses of first-generation agents (typically defined in haloperidol doses of less than 10 mg/day or plasma levels <18 ng/ml) are as efficacious or more efficacious than higher doses (782–784). Moderate doses of first-generation agents have been reported to improve comorbid depression (369, 785, 786), whereas higher doses are associated with greater risk of extrapyramidal side effects and dysphoria (785, 787) and may be especially problematic for patients with frontal lobe dysfunction (788).

Empirical research provides relatively little guidance for psychiatrists who are making decisions about medication and dosage during the stabilization phase. The use of first-generation antipsychotic medications during this phase is based on the clinical observation that patients relapse abruptly when medications are discontinued during this phase of treatment.

A large number of studies (789, 790) have compared relapse rates for stabilized patients who continued taking a first-generation antipsychotic medication and for those whose regimen was changed to placebo. During the first year only about 30% of those continuing to take medications relapsed, compared with about 65% of those taking placebo. Even when adherence with medication treatment was ensured by the use of long-acting injectable medications, as many as 24% of patients relapsed in a year (791). Hogarty et al. (792) found that among outpatients maintained with antipsychotic medications for 2–3 years who had been stable and judged to be at low risk of relapse, 66% relapsed in the year after medication withdrawal. Studies in which the medications of well-stabilized patients were discontinued indicate that 75% of patients relapse within 6–24 months (790). Among patients who have experienced a first episode of schizophrenia, a number of carefully designed double-blind studies indicate that 40%–60% of patients relapse if they are untreated during the year after recovery from this initial episode (211, 212, 218).

A critical issue during the stable treatment phase is adherence to the medication regimen. One strategy for improving adherence with first-generation agents is use of the long-acting injectable formulation. Studies with long-acting antipsychotics show a dose-response relationship in prophylactic efficacy, although there is a tradeoff in the relationship between dose and relapse rate on the one hand and side effects on the other (793, 794). The higher the dose used, the lower the relapse rate but the higher the rate of side effects, whereas the reverse is seen with lower doses. Although a small number of randomized trials have assessed the effectiveness of more modest doses of long-acting injectable medications than those typically used in clinical practice, evidence on this question remains inconclusive. Inderbitzen et al. (795) found no loss of clinical effectiveness when the average dose of patients already receiving long-acting injectable fluphenazine was cut gradually by 50% over a 5-month period (from an average of 23 mg every 2 weeks to 11.5 mg every 2 weeks). Similarly, Carpenter et al. (796) found a regimen of 25 mg of fluphenazine decanoate every 6 weeks to be equally effective as the same dose administered every 2 weeks. However, Schooler et al. (219) compared three medication strategies using fluphenazine decanoate: a continuous moderate dose (12.5–50 mg every 2 weeks); a continuous low dose (2.5–10 mg every 2 weeks); and targeted, early intervention (fluphenazine only when the patient was experiencing symptoms). They found that both continuous low-dose and targeted treatment increased the use of rescue medication and the rate of relapse, while only targeted treatment increased the rate of rehospitalization.

Side effects of first-generation antipsychotic medications typically vary with the potency of the agent. High-potency first-generation antipsychotics are associated with a high risk of extrapyramidal effects, a moderate risk of sedation, a low risk of orthostatic hypotension and tachycardia, and a low risk of anticholinergic and antiadrenergic effects. In contrast, low-potency first-generation antipsychotic agents are associated with a lower risk of extrapyramidal effects, a high risk of sedation, a high risk of orthostatic hypotension and tachycardia, and a high risk of anticholinergic and antiadrenergic effects. Although other side effects also vary with the specific medication, in general, the first-generation antipsychotic medications are associated with a moderate risk of weight gain, a low risk of metabolic effects, and a high risk of sexual side effects. With certain agents (thioridazine, mesoridazine, pimozide), a moderate risk of cardiac conduction abnormalities is also present. Neuroleptic malignant syndrome occurs rarely but is likely to be more often observed with first-generation agents (especially high-potency agents) than with second-generation antipsychotic medications. Details on the nature and management of each of these side effects are provided in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications".

Other side effects include seizures, allergic reactions, and dermatological, hepatic, ophthalmological, and hematological effects.

First-generation antipsychotic medications can lower the seizure threshold and result in the development of generalized tonic-clonic seizures (797). The low-potency first-generation antipsychotic medications confer the greatest risk. The frequency of seizures with low-potency antipsychotic medications is dose related, with higher doses associated with greater risk. At usual dose ranges, the seizure rates are below 1% for all first-generation antipsychotic medications, although patients with a history of an idiopathic or medication-induced seizure have a higher risk.

Cutaneous allergic reactions occur infrequently with first-generation antipsychotic medications. Medication discontinuation or administration of an antihistamine is usually effective in reversing these symptoms. Rarely, thioridazine is associated with hyperpigmentation of the skin. Photosensitivity also occurs infrequently and is most common with the low-potency phenothiazine medications; patients should be instructed to avoid excessive sunlight and use sunscreen (99).

Also occurring with this class of medications are elevation of liver enzyme levels and cholestatic jaundice. Jaundice has been noted to occur in 0.1%–0.5% of patients taking chlorpromazine (99). This side effect usually occurs within the first month after the initiation of treatment and generally requires discontinuation of treatment. However, given the relative infrequency of antipsychotic-induced jaundice, other etiologies for jaundice should be evaluated before the cause is judged to be antipsychotic medication.

Pigmentary retinopathies and corneal opacities can occur with chronic administration of the low-potency medications thioridazine and chlorpromazine, particularly at high doses (e.g., more than 800 mg/day of thioridazine). For this reason, patients maintained with these medications should have periodic ophthalmological examinations (approximately every 2 years for patients with a cumulative treatment of more than 10 years), and a maximum dose of 800 mg/day of thioridazine is recommended (797). With the increased use of high-potency medications in the past two decades, there has been virtually no reporting of this side effect (777).

Hematological effects, including inhibition of leukopoiesis, can occur with use of first-generation antipsychotic medications. Such effects include benign leukopenia and the more serious agranulocytosis. The best data exist for chlorpromazine, with which benign leukopenia occurs in up to 10% of patients and agranulocytosis occurs in 0.32% of patients (797).

Issues in implementation of treatment with first-generation antipsychotic medications include route of administration, dosage strategy, and medication interactions.

First-generation antipsychotic medications can be administered in oral forms, as short-acting intramuscular preparations, or as long-acting injectable preparations. Short-acting intramuscular medications reach a peak concentration 30–60 minutes after the medication is administered, whereas oral medications reach a peak in 2–3 hours (798). As a result, the calming effect of the first-generation antipsychotic may begin more quickly when the medication is administered parenterally. However, this calming effect on agitation is different from the true antipsychotic effect of these medications, which may require several days or weeks (779). It is also worth noting that oral concentrates are typically better and more rapidly absorbed than pill preparations and often approximate intramuscular administration in their time to peak serum concentrations.

A single or twice-daily dose of an oral preparation will result in steady-state blood levels in 2–5 days (798). Long-acting injectable first-generation antipsychotic medications (fluphenazine decanoate or enanthate and haloperidol decanoate in the United States) may require up to 3–6 months to reach a steady state (92). As a result, they are seldom used alone during acute treatment, when the psychiatrist is adjusting the dose in accordance with therapeutic effects and side effects.

The advantage of long-acting injectable medications has been best demonstrated in studies such as those conducted by Johnson (799) under conditions that resemble most closely those in community clinics. In these studies, patients with histories of poor adherence were included in the study population and the amount of contact between patients and staff was limited. In the larger, more carefully controlled investigations (791, 800), patients with serious adherence problems—that is, the patients most likely to benefit from treatment with long-acting injectable medications—were commonly not included. Thus, a study by Hogarty et al. (800) showed a reduction in relapse associated with fluphenazine decanoate compared to oral fluphenazine only after 2 years of follow-up, as the effect of the drug on nonadherence and subsequent relapse took time to develop in a study population that was relatively stable and adherent at baseline.

Long-acting injectable medications are thought to be especially helpful in the stabilization and stable phases. Janicak et al. (99) examined six studies that compared the risk of psychotic relapse in patients who were randomly assigned to receive either oral or long-acting injectable medication. The longest of those studies (800) lasted 2 years and showed a relapse rate of 65% for patients taking oral medication and a rate of 40% for patients taking long-acting injectable medication. Although the remaining five studies, all of which lasted 1 year or less, had variable results, a meta-analysis of all six studies showed a significantly lower relapse rate in patients who received long-acting injectable medication (p<0.0002) (99).

The effective dose of a first-generation antipsychotic medication is closely related to its affinity for dopamine receptors (particularly D₂ receptors) and its tendency to cause extrapyramidal side effects (801, 802). Thus, high-potency medications have a greater affinity for dopamine receptors than do low-potency medications, and a much lower dose of high-potency medcations is required to treat psychosis. This relationship can be expressed in terms of dose equivalence (e.g., 100 mg of chlorpromazine has an antipsychotic effect that is similar to that of 2 mg of haloperidol). The dose equivalencies of commonly prescribed medications are listed in Table 2.

High-potency first-generation antipsychotic medications, such as haloperidol and fluphenazine, are more commonly prescribed than low-potency compounds (803). Although these medications have a greater tendency to cause extrapyramidal side effects than the low-potency medications, such as chlorpromazine and thioridazine, their side effects are easier to manage than the sedation and orthostatic hypotension associated with low-potency agents. High-potency medications can more safely be administered intramuscularly, since they seldom cause hypotension. In addition, because of sedation, orthostatic hypotension, and lethargy, the dose of a low-potency medication should be increased gradually, whereas an adequate dose of a high-potency medication can usually be achieved within a day or two. Finding the optimal dose of a first-generation antipsychotic is complicated by a number of factors. Patients with schizophrenia demonstrate large differences in the dose of first-generation antipsychotic they can tolerate and the dose required for an antipsychotic effect. A patient's age may influence the appropriate dose; elderly patients are more sensitive to both the therapeutic and adverse effects of first-generation antipsychotics. In addition, in studies in which dose is not fixed, it is difficult to determine dose by assessing antipsychotic effectiveness, since it may take many days at a therapeutic dose before there is an appreciable decrease in psychosis (778, 780).

A number of studies (reviewed by Davis et al. [777] and by Baldessarini et al. [95]) provide guidance about the usual doses required for acute treatment. Results of 19 controlled trials suggested that daily doses below 250 mg of chlorpromazine (or 5 mg of haloperidol or fluphenazine) are less adequate for many acutely psychotic patients than are moderate doses, between 300 and 600 mg of chlorpromazine. In the studies, response was typically measured by improvement in the score on the excitement, agitation, or psychosis subscale of the BPRS (6), and the proportions of patients responding to low doses after 1 and 2–10 days were 38% and 50%, respectively; these rates compared unfavorably with the improvement rates of 61% and 56% among patients taking moderate doses for similar periods (95). Davis et al. (777) came to similar conclusions. They found that daily doses between 540 and 940 mg of chlorpromazine were optimal. The findings of clinical trials involve groups of patients; some patients have optimal responses at doses above or below these optimal ranges. Psychiatrists have treated acutely psychotic patients with high doses of high-potency first-generation antipsychotic medications during the first days of treatment. This treatment is based on the belief that higher doses result in a more rapid improvement than that resulting from moderate doses (804). However, studies have revealed that high daily doses (more than 800 mg of chlorpromazine equivalents daily) were no more effective, or faster acting, on average than were moderate doses (500–700 mg/day) (95). After 1 day, 50% of the patients treated with high doses responded, compared to 61% of those who received moderate doses. After 2–10 days, high-dose treatment led to a slightly worse outcome: only 38% of those receiving high doses but 56% of those receiving moderate doses were improved. These studies indicate that higher doses are no more effective for acute treatment than normal doses, but higher doses are associated with a greater incidence of side effects.

Controlled trials have provided similar information regarding the effect of medication dose on outcome during the maintenance phase. In 33 randomized trials in which high doses (mean, 5200 mg/day of chlorpromazine equivalents) were compared to low doses (mean, 400 mg/day) during maintenance treatment, the lower doses were more effective in improving clinical state in more than two-thirds of the trials (95). In addition, in 95% of the studies the higher doses resulted in greater neurological side effects. Studies of doses of less than 200 mg/day of chlorpromazine equivalents tended to show that such doses were less effective than higher doses. An international consensus conference (294) made the reasonable recommendation of a reduction in first-generation antipsychotic dose of approximately 20% every 6 months until a minimal maintenance dose is reached. A minimal dose was considered to be as low as 2.5 mg of oral fluphenazine or haloperidol daily, 50 mg of haloperidol decanoate every 4 weeks, or 5 mg of fluphenazine decanoate every 2 weeks.

Concerns about the side effects of first-generation antipsychotic medications during maintenance treatment and the risk of tardive dyskinesia led to several studies that focused on methods for treating patients with the lowest effective maintenance dose. A number of investigators (19, 805–807) have studied gradual reductions in the amounts of medication given to stabilized patients until the medications are completely discontinued. Each patient was followed closely until there were signs of the beginning of a relapse. At that time, the patient's medication was reinstituted. To make this strategy work, patients and their families were trained to detect the early signs of impending psychotic breakdown. This approach used antipsychotic medications only intermittently to target symptom exacerbations and to avert anticipated exacerbations. Studies of the efficacy of this "targeted medication approach" have produced mixed results, and this approach is not recommended because of the substantial increase in the risk of relapse (19, 219, 805, 806).

Another strategy involves using much lower doses of a long-acting injectable first-generation antipsychotic than are usually prescribed. Several groups have compared low doses to moderate and high doses of fluphenazine decanoate. Initially, studies found that patients receiving very low doses (mean=2.5 mg every 2 weeks) were significantly more likely to relapse over the course of 1 year than were patients receiving standard doses (12.5–50.0 mg every 2 weeks) (56% versus 7%) (794). A subsequent study demonstrated that patients given a slightly higher dose (2.5–10.0 mg every 2 weeks) showed a nonsignificant difference in relapse after 1 year, compared with patients given standard doses (24% versus 14%) (808). Another study found no significant difference in relapse after 1 year between patients who received low doses (mean=5 mg every 2 weeks) and those who received standard doses (25–50 mg every 2 weeks) but did detect a significant difference in relapse rates after 2 years between the low-dose group and the standard-dose group (70% versus 35%) (793). Other studies, however, reported no difference in relapse rates after 2 years between patients who received low doses (mean=3.8 mg every 2 weeks) and those who received standard doses (25 mg every 2 weeks) (809). However, Schooler et al. (219) found that low-dose fluphenazine decanoate (2.5–10 mg every 2 weeks) increased the relapse rate and the use of rescue medication, compared to a continuous moderate dose (12.5–50 mg every 2 weeks). Collectively, these studies indicate that doses of fluphenazine decanoate as low as 5–10 mg every 2 weeks have been shown to be clinically effective, and some patients may respond to even lower doses, but the risk of relapse can increase significantly with these lower doses. However, consideration should be given to judicious reduction in the long-acting injectable dose over time, especially for patients with adverse side effects, in order to evaluate the optimal dose.

In considering the use of low-dose, long-acting injectable first-generation antipsychotics, the beneficial side effect profile associated with the use of lower doses should also be taken into account. Kane et al. (794) found that low-dose users had fewer early signs of tardive dyskinesia after 1 year than did standard-dose users. In a study by Marder et al. (793), lower doses were associated with significantly less discomfort (as measured with the SCL-90-R [810]), psychomotor retardation, and akathisia after 2 years. Hogarty et al. (809) reported that patients receiving minimal doses had less muscle rigidity, akathisia, and other side effects at 1 year and had greater improvements in instrumental and interpersonal role performances at 2 years.

First-generation antipsychotic medications have a very high therapeutic index for life-threatening side effects (780). Consequently, overdoses rarely are fatal unless they are complicated by preexisting medical problems or concurrent ingestion of alcohol or other medications. Symptoms of overdose are generally characterized by exaggerations of the adverse effects, with respiratory depression and hypotension presenting the greatest danger. Treatment is symptomatic and supportive and includes 1) ensuring airway patency and maintenance of respiration; 2) orally administering activated charcoal to decrease absorption and considering gastric lavage; 3) maintaining blood pressure with intravenous fluids and vasopressor agents; and 4) administering anticholinergic agents if needed to counteract extrapyramidal signs (811).

A number of medication interactions can have clinically important effects for patients who are treated with antipsychotic medications (48, 49, 812). Certain heterocyclic antidepressants, most SSRIs, some beta-blockers, and cimetidine may increase antipsychotic plasma levels and increase side effects. On the other hand, barbiturates and carbamazepine decrease plasma levels through effects on cytochrome P450 enzymes.

b) Second-generation agents

The medications discussed in this section are referred to as second-generation antipsychotics primarily because the doses that are effective against the psychopathology of schizophrenia do not cause extrapyramidal side effects. Their therapeutic effects are attributed to central antagonism of both serotonin and dopamine receptors and also possibly to relatively loose binding to D₂ receptors (813–815).

Clozapine is a second-generation antipsychotic with antagonist activity at numerous receptors, including dopamine (D₁, D₂, D_3, D₄, D₅), serotonin (5-HT_1A, 5-HT_2A, 5-HT_2C), muscarinic (M₁, M₂, M₃, M₅), ₁- and ₂-adrenergic, and histamine (H₁) receptors (816–818). Clozapine is an agonist at muscarinic (M₄) receptors (819). Clozapine is also distinguished from other antipsychotic medications by its greater efficacy in treating positive symptoms in patients with treatment-resistant illness and by the absence of extrapyramidal side effects. However, it is associated with several serious and potentially fatal adverse effects, including agranulocytosis in 0.5%–1% of patients, seizures in about 2% of patients, and rare occurrences of myocarditis and cardiomyopathy.

Clozapine has demonstrated superior efficacy for the treatment of general psychopathology in patients with treatment-resistant schizophrenia, compared to the first-generation antipsychotics haloperidol and chlorpromazine in six of eight published double-blind randomized trials (313, 314, 769, 820–824). A meta-analysis pooled the results of five of these studies that categorically defined subjects as "responders" based on clinically meaningful improvement in psychopathology and found that clozapine-treated patients were 2.5 times more likely to meet response criteria than those treated with a first-generation antipsychotic (p=0.001) (87). Clozapine also has demonstrated efficacy in reducing the frequency of suicidal ideation and suicide attempts in a randomized 2-year study of 980 patients with schizophrenia or schizoaffective disorder at high risk for suicide because of previous or current suicidal ideation or behavior (55). In this study patients were randomly assigned to receive either clozapine or olanzapine. Fewer patients who received clozapine attempted suicide (34 subjects), compared to patients who received olanzapine (55 subjects) (p=0.03), and a 24% reduction in risk of suicidal behaviors was found. In light of this evidence, clozapine should be preferentially considered for patients with a history of chronic and persistent suicidal ideation or behaviors. In addition, several studies suggest that clozapine may reduce the severity of hostility and aggression in patients with treatment-resistant symptoms (57, 314, 440, 821, 825–827). Open-label and double-blind studies of clozapine have produced inconsistent results with regard to effects on cognition, with some measures showing improvement and others showing no changes or even decrements in performance (828–840).

There is also preliminary evidence from open-label observational studies that clozapine may reduce risk of relapse in patients with treatment-resistant schizophrenia (841–845). Although these studies are encouraging, they are limited since some included only clozapine responders, while others did not include a comparison group. These studies are supported by the results of a large randomized open trial, in which significantly fewer hospital readmissions were observed for patients treated with clozapine, compared to those treated with usual care in a state hospital system over a 2-year period (822). The only double-blind study that measured readmission rates over a 1-year period failed to show a difference between haloperidol and clozapine for patients with treatment-resistant schizophrenia, although patients treated with clozapine stayed fewer days in the hospital (769). Taken together, the evidence is suggestive that treatment with clozapine is associated with reduced rates of relapse and rehospitalization in patients with treatment-resistant schizophrenia.

Studies of other populations, including patients with first-episode schizophrenia (846) and patients with treatment-responsive schizophrenia or schizoaffective disorder (847), demonstrate only limited or inconsistent superior efficacy for clozapine. In addition, studies comparing clozapine to other second-generation antipsychotics generally show comparable efficacy of clozapine with other second-generation antipsychotics (87, 381, 820, 848). However, since relatively low doses of clozapine were used in these studies, the results must be interpreted with caution.

In summary, a clozapine trial should be considered for patients who have shown a poor response to other antipsychotic medications. Clozapine may also be considered for patients with a history of chronic and persistent suicidal ideation or behaviors. In addition, clozapine may also be considered for patients with persistent hostility and aggression, given that superior efficacy of clozapine has been demonstrated in these patient populations.

Clozapine is associated with a very low risk of acute and chronic extrapyramidal side effects, a high risk of sedation, a high risk of orthostatic hypotension and tachycardia, a low risk of cardiac conduction abnormalities, a high risk of anticholinergic effects, a high risk of weight gain and metabolic abnormalities, and a low risk of prolactin elevation and sexual side effects. Neuroleptic malignant syndrome occurs rarely with clozapine. Details on the nature and management of each of these side effects are provided in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications".

Sialorrhea and drooling occur relatively frequently and are most likely due to decreased saliva clearance related to impaired swallowing mechanisms (849), or possibly as a result of muscarinic cholinergic antagonist activity at the M₄ receptor or to -adrenergic agonist activity (850). Interventions include use of a towel on the pillow at night to reduce discomfort. While there is little systematic information about pharmacological interventions, case reports suggest potential improvement with antimuscarinic agents and receptor agonists (851–853). However, since clozapine also exhibits significant anticholinergic properties, use of agents with added anticholinergic effects must be approached with extreme caution to avoid potential adverse effects such as constipation or cognitive impairment.

Fever (>38° C) may occur during the first few weeks of treatment (854, 855). Generally a clozapine-associated fever is self-limiting and responds to supportive measures. However, fever is a symptom of neuroleptic malignant syndrome, agranulocytosis, and cardiomyopathy, and the presence of fever warrants evaluation for these potentially life-threatening complications of clozapine treatment.

The risk of agranulocytosis (defined as an absolute neutrophil count less than 500/mm³) has been estimated at 1.3% of patients per year of treatment with clozapine (99, 854, 856). The risk is highest in the first 6 months of treatment, and therefore weekly WBC and neutrophil monitoring is required. After 6 months, monitoring may occur every 2 weeks, as the risk of agranulocytosis appears to diminish considerably (an estimated rate of three cases per 1,000 patients). WBC counts must remain above 3000/mm³ during clozapine treatment, and absolute neutrophil counts must remain above 1500/mm³.

In the United States through 1989 there were 149 cases of agranulocytosis, with 48 (32%) fatalities. With the advent of systematic monitoring, fatalities have been greatly reduced (857–859). Between 1989 and 1997, among the 150,409 patients treated with clozapine who were included in the patient registry maintained by the U.S. manufacturer of the drug, 585 cases of agranulocytosis, with nine fatalities, were reported. Thus, awareness of agranulocytosis and the monitoring system have decreased the reported rate of clozapine-induced agranulocytosis to less than 0.5%.

Reports of myocarditis, with resultant cardiomyopathy and fatal heart failure, associated with clozapine use suggest a 17- to 322-fold elevation in risk in clozapine-treated patients. The absolute risk is estimated to range from 1 per 500 treated patients (860) to 1 per 10,000 treated patients (861). An immune mechanism mediated by immunoglobulin E antibodies is suspected because of reports of associated eosinophilia. Most but not all cases have occurred early in treatment, suggesting that the risk of myocarditis may be less after the first few months.

Clozapine is also associated with a dose-related risk of seizures (854). The overall seizure rate is 2.8%; with low-dose treatment (<300 mg/day) the risk is 1%, with medium doses (300–599 mg/day) the risk is 2.7%, and with high doses (>599 mg/day) the risk is 4.4%. The seizure risk for clozapine is also related to rapid increases in dose. Therefore, the rate of titration should not exceed the guidelines described in the subsequent section on implementation of treatment with clozapine.

In addition, there are case reports associating clozapine treatment with several other rare but potentially serious adverse events, including pancreatitis (862, 863), deep vein thrombosis (864, 865), pulmonary embolism, hepatitis (866, 867), and eosinophilia (863). Because of the small number of reports, the causal relationship with clozapine is unclear.

Before initiating treatment with clozapine, a complete blood count (CBC) with differential should be performed and the patient's general and cardiovascular health status should be evaluated. The cardiovascular side effects of clozapine should be considered in planning treatment for patients with preexisting heart disease. Treatment should be initiated at a low dose (12.5–25 mg once or twice daily) and increased gradually (by no more than 25–50 mg/day) as tolerated until a target dose is reached. Because of the risk of marked hypotension, sedation, and seizures with rapid dose escalation, dose titration should not occur more rapidly. During dose titration the patient's cardiovascular status, including orthostatic pulse, blood pressure, and subjective complaints of dizziness, should be monitored. Since the side effects of clozapine in the initial and dose-adjustment phases may be severe in some patients, admission to the hospital may be justifiable (e.g., for unstable patients who require rapid dose increases to a therapeutic level, patients with a limited social support system, or patients prone to orthostatic hypotension or seizures).

Adequate safety monitoring during treatment is important to minimize the risk of adverse events. The clozapine package label states that WBC and neutrophil counts should be evaluated before treatment is initiated, weekly during the first 6 months of treatment and at least every 2 weeks after 6 months of treatment (854). Clozapine treatment should not be initiated if the initial WBC count is <3500/mm³, if the patient has a history of a myeloproliferative disorder, or if the patient has a history of clozapine-induced agranulocytosis or granulocytopenia.

With maintenance treatment, patients should be advised to report any sign of infection immediately (e.g., sore throat, fever, weakness, lethargy). A WBC count <2000/mm³ or absolute neutrophil count (ANC) <1000/mm³ indicates impending or actual agranulocytosis, and the clinician should stop clozapine treatment immediately, check WBC and differential counts daily, monitor for signs of infection, and consider bone marrow aspiration and protective isolation if granulopoiesis is deficient. A WBC count of 2000–3000/mm³ or ANC of 1000–1500/mm³ indicates high risk of or impending agranulocytosis, and the clinician should stop clozapine treatment immediately, check the WBC and differential counts daily, and monitor for signs of infection. Clozapine may be resumed if no infection is present, the WBC count rises to >3000, and the ANC is >1500 (resume checking WBC count twice a week until it is >3500). If the WBC count is 3000–3500/mm³, if it falls to 3000/mm³ over 1–3 weeks, or if immature WBC forms are present, repeat the WBC count with a differential count. If the subsequent WBC count is 3000–3500/mm³ and the ANC is >1500/mm³, repeat the WBC count with a differential count twice a week until the WBC count is >3500/mm³.

Agranulocytosis is usually reversible if clozapine is discontinued immediately (868). When agranulocytosis develops, clozapine should be immediately discontinued, and patients should be given intensive treatment for the secondary complications, e.g., sepsis. Granulocyte colony stimulating factor has been used to accelerate granulopoietic function and shorten recovery time (869). Lithium has also been considered as a possible treatment for leukopenia or to prevent the development of agranulocytosis in patients who may be susceptible to this adverse effect (870, 871).

Although there have been reports of successful clozapine rechallenge after leukopenia, the risk of recurrence remains high (872). A rechallenge with clozapine should not be undertaken in patients with confirmed cases of agranulocytosis (ANC <500/mm³), as recurrence is almost certain (872). Clinically, rechallenge should only be considered for patients whose WBC count remained greater than 2000/mm³, whose absolute neutrophil count remained greater than 1500/mm³, and for whom trials with multiple other antipsychotics had failed but a good clinical response to clozapine was shown.

In addition, patients should be monitored for weight gain, glucose abnormalities, and hyperlipidemias that may occur during treatment with clozapine. Table 1 outlines suggested monitoring and clinical management of such adverse effects. Patients should also be monitored for other potentially life-threatening adverse effects of clozapine, including fever and other signs of myocarditis. Patients should be advised to report any signs of myocarditis (e.g., fever, fatigue, chest pain, palpitations, tachycardia, respiratory distress, peripheral edema). Immediate clinical evaluation is warranted, and a cardiovascular evaluation is needed if these symptoms are not explained by other causes. A cardiac evaluation is thus recommended for clozapine-treated patients who experience unexplained fever, fatigue, chest pain, palpitations, tachycardia, hypotension, narrowed pulse pressure, respiratory distress, peripheral edema, ST-T wave abnormalities or arrhythmias as shown by ECG, or hypereosinophilia as shown by a CBC, especially if these symptoms are experienced during the first few months of treatment (873).

Controlled trials provide only limited guidance regarding the optimal dose of clozapine for schizophrenia. Since there have been no trials in which patients were randomly assigned to different doses of clozapine, the only available data are based on studies in which psychiatrists used what they considered the most effective dose. Fleischhacker et al. (874) reviewed 16 controlled trials from Europe and the United States. The mean dose from the European trials was 283.7 mg/day, and the U.S. mean was 444 mg/day. Plasma levels may help guide dosing, with studies suggesting that maximal clinical efficacy may be achieved when plasma levels of clozapine are between 200 and 400 ng/ml (typically associated with a dose of 300–400 mg/day) (875–878).

Although most patients whose symptoms respond to clozapine demonstrate maximal clinical improvement during the first 6–12 weeks of treatment, clinical benefits may continue to develop after 6–12 months (87, 879, 880). Twelve-week empirical trials of clozapine appear to be adequate to determine whether a patient is likely to respond to this medication (881, 882).

The elimination half-life of clozapine is approximately 12 hours, indicating that patients are likely to reach a steady-state plasma concentration after 2–3 days (883). Clozapine is metabolized primarily by the CYP1A2 enzyme. Other liver enzymes also contribute to clozapine metabolism, including CYP2C19, CYP2D6, and CYP3A4. Coadministration of drugs that inhibit cytochrome P450 enzymes (e.g., cimetidine, caffeine, erythromycin, fluvoxamine, fluoxetine, paroxetine, sertraline) may lead to a significant increase in clozapine plasma levels; inducers of CYP1A2 (e.g., phenytoin, nicotine, rifampin) can significantly reduce clozapine levels. In particular, changes in smoking status may affect clozapine levels (500, 884). The concomitant use of medications such as carbamazepine can lower the WBC count and increase the potential danger of agranulocytosis; such medications should therefore be avoided. Some cases of respiratory or cardiac arrest have occurred among patients receiving benzodiazepines or other psychoactive medications concomitantly with clozapine. While no specific interaction between clozapine and benzodiazepines has been established, judicious use is advised when benzodiazepines or other psychotropic medications are administered with clozapine (885).

Risperidone is a second-generation antipsychotic with antagonist activity at dopamine (D₁, D₂, D₃, D₄), serotonin (5-HT_1A, 5-HT_2A, 5-HT_2C), ₁- and ₂-adrenergic, and histamine (H₁) receptors (816, 817).

There are numerous published clinical trials comparing the acute efficacy of risperidone with placebo, first-generation antipsychotics (haloperidol or perphenazine), and other second-generation antipsychotics in patients with schizophrenia, schizoaffective disorder, and schizophreniform disorder. Placebo-controlled studies consistently demonstrate that for acutely relapsed patients, risperidone is efficacious in the treatment of global psychopathology and the positive symptoms of schizophrenia (886–888), as well as in increasing the likelihood of clinical response (e.g., 20% improvement on rating scales of global psychopathology). There is less consistent evidence that negative symptoms improve with risperidone treatment, as significant improvement compared with placebo was not found at all doses of risperidone (886, 887, 889). It is likely that the improvements in negative symptoms are due to the decreased likelihood of secondary negative symptoms (e.g., related to parkinsonism or to psychosis). Active-comparator-controlled studies demonstrate comparable or occasionally greater likelihood of clinical response and improvement of global psychopathology and positive symptoms with risperidone, compared with haloperidol (886, 887, 890–894) and perphenazine (895). Meta-analyses of these studies suggest that risperidone may have modestly better efficacy, compared with haloperidol and perphenazine, in decreasing positive symptoms (889, 896, 897) and global psychopathology and increasing the likelihood of response (82, 86, 88, 89, 889, 898–900). These studies also show less consistent evidence that negative symptoms improve with risperidone treatment, with any improvements possibly due to the decreased likelihood of secondary negative symptoms or resulting from comparison with high doses of first-generation antipsychotic agents. One study (271) found risperidone to have similar efficacy to haloperidol in the acute treatment of first-episode schizophrenia, as measured by greater response rates, improvement in global psychopathology, and improvement in positive symptoms.

Several studies have examined the efficacy of risperidone in the treatment of neurocognitive deficits of schizophrenia. Generally, these studies find that global measures of neurocognitive function improved with risperidone, although the magnitude of improvement was similar to that observed with haloperidol in two studies (901, 902). However, in one 14-week trial that included 101 patients, treatment with risperidone resulted in statistically significantly greater improvement in global neurocognition, compared with haloperidol (838). The clinical significance of this effect, however, is unclear. Thus, further investigation is required to determine the magnitude and clinical significance of risperidone effects on neurocognition.

Several studies have examined the efficacy of risperidone in patients with treatment-resistant schizophrenia. In an 8-week double-blind study, risperidone (mean dose=7.5 mg/day, N=34) and haloperidol (mean dose=19.4 mg/day, N=33) demonstrated similar efficacy in the treatment of global psychopathology (903). In a 14-week double-blind trial, treatment with risperidone (mean dose=11.6 mg/day, N=41) resulted in significant improvement in global psychopathology scores but not in positive and negative symptom subscale scores, for which the effects of risperidone were comparable to those of haloperidol (mean dose=25.7 mg/day, N=37) (820). In a 12-week double-blind study, 6 mg/day of risperidone (N=39) was superior to 20 mg/day of haloperidol (N=39) in the treatment of global psychopathology and negative symptoms (904).

Studies comparing risperidone to other second-generation antipsychotics in the treatment of acute episodes have generally found similar efficacy for treatment of psychopatholgy both in patients with treatment-responsive illness and in those with treatment-resistant illness (848, 902, 905–907).

Compared with haloperidol, risperidone has demonstrated superior efficacy in the prevention of relapse in the maintenance phase of treatment. In a study of 397 stable patients with DSM-IV schizophrenia or schizoaffective disorder, haloperidol-treated patients (mean dose=11.7 mg/day) were 1.93 times more likely to relapse than risperidone-treated patients (mean dose=4.9 mg/day) during the 1-year follow-up period (382). In this study, the risperidone-treated patients also had significantly greater improvement in global psychopathology, compared to the haloperidol-treated patients.

Risperidone is associated with a low risk of sedation, a low to moderate risk of extrapyramidal side effects, a moderate risk of orthostatic hypotension and tachycardia, a low risk of anticholinergic effects, a moderate risk of weight gain and metabolic abnormalities, and a high risk of prolactin elevation and sexual side effects. Risperidone slightly alters cardiac conduction but not to a clinically meaningful extent. Neuroleptic malignant syndrome occurs rarely with risperidone. Details on the nature and management of each of these side effects are provided in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications".

Clinical trial data suggest a small increase in the risk of stroke in patients with dementia treated with risperidone, compared with placebo-treated patients. Thus, dementia patients treated with risperidone should be carefully monitored for signs and symptoms of stroke (908). Similar increases in risk of stroke have not been reported in elderly risperidone-treated patients with schizophrenia who do not have dementia.

While the original efficacy studies comparing different doses of risperidone indicated optimal effectiveness at doses of around 6 mg/day, clinical investigations and subsequent studies indicate that for most adult patients optimal doses are between 2 and 6 mg/day, with a minority of patients requiring higher doses. Higher doses often lead to extrapyramidal side effects without greater effectiveness. Patients who develop parkinsonian symptoms are probably receiving too high a dose, and dose reduction is required for these patients.

During the titration and early treatment phase, risperidone-treated patients should be monitored for extrapyramidal side effects, orthostatic hypotension and reflex tachycardia, side effects associated with prolactin elevation, and sedation. In addition, patients should be monitored for weight gain, glucose abnormalities, and hyperlipidemias that may occur during treatment with risperidone. Table 1 outlines suggested strategies for monitoring and clinical management of such adverse effects. Elderly patients, particularly those with dementia, should be monitored for signs and symptoms of stroke.

Risperidone's effectiveness appears to be related to actions of both the parent compound and a major metabolite, 9-hydroxyrisperidone (909). They are therapeutically equipotent, have similar types of pharmacological activity, and, therefore, probably produce similar therapeutic effects. Although risperidone itself has an elimination half-life of only 3 hours, its metabolite has an elimination half-life of about 24 hours. As a result, most patients can be managed with a once-daily dose of risperidone. However, since risperidone can cause orthostatic hypotension, twice-daily dosing may be useful during the titration phase and for patients who may be vulnerable to orthostatic changes, such as elderly patients.

Risperidone is primarily metabolized by the hepatic CYP2D6 enzyme into the 9-hydroxyrisperidone metabolite (910). 9-Hydroxyrisperidone may also be metabolized by the CYP3A4 liver enzyme (500, 911). As a result, inducers of CYP3A4 may decrease risperidone blood levels and thus reduce therapeutic efficacy (912). In contrast, inhibitors of CYP2D6 and CYP3A4 may raise blood levels of risperidone and its active metabolite 9-hydroxyrisperidone and thus produce increased side effects, such as extrapyramidal side effects (913). In 5%–8% of Caucasians and 2%–5% of African Americans and Asians, the activity of the CYP2D6 enzyme is very low or absent. In poor metabolizers, the half-life is 17 hours for risperidone and 30 hours for 9-hydroxyrisperidone, compared to half-lives in extensive metabolizers of 3 hours for risperidone and 21 hours for 9-hydroxyrisperidone. Thus, the relative proportion of risperidone to 9-hydroxyrisperidone will be higher in patients who are slow metabolizers. In addition, drugs that inhibit the CYP2D6 enzyme (e.g., quinidine) will effectively turn extensive metabolizers into poor metabolizers. In terms of CYP liver enzymes other than CYP3A4 and CYP2D6, risperidone does not tend to produce significant inhibition or induction.

Olanzapine is a second-generation antipsychotic with antagonist activity at dopamine (D₁, D₂, D₃, D₄), serotonin (5-HT_2A, 5-HT_2C), muscarinic (M₁, M₂, M₃, M₅), ₁-adrenergic, and histamine (H₁) receptors (818, 914).

There are several published clinical trials comparing the acute efficacy of olanzapine with placebo, first-generation antipsychotics (haloperidol or chlorpromazine), and other second-generation antipsychotics in patients with schizophrenia, schizoaffective disorder, and schizophreniform disorder. Placebo-controlled studies consistently demonstrate that in acutely relapsed patients olanzapine is efficacious in treating global psychopathology and the positive symptoms of schizophrenia, as well as in increasing the likelihood of clinical response (e.g., 20% improvement on rating scales of global psychopathology) (915–917). The evidence that negative symptoms improve with olanzapine treatment, compared with placebo, is less consistently found across doses of the drug (915–917). It is likely that any improvements in negative symptoms in these studies are due to decreased likelihood of secondary negative symptoms (e.g., related to parkinsonism or to psychosis) rather than to direct effects on primary negative symptoms (323). Active-comparator-controlled studies demonstrate similar or occasionally greater likelihood of clinical response and greater improvement of global psychopathology and positive and negative symptoms with olanzapine, compared to haloperidol (279, 319, 916–920). Meta-analyses of these studies suggest that olanzapine may have modestly better efficacy, compared with haloperidol, in the treatment of global psychopathology and positive and negative symptoms (921) and in increasing the likelihood of response (82, 86, 88). Effects on hostility are mixed, with one study (921) showing greater improvement in hostility with olanzapine than with haloperidol, and another study finding no difference in hostility response (440). In patients with a first episode of schizophrenia, one study (a subanalysis of a Lilly olanzapine database) found significantly greater improvement in global psychopathology, positive and negative symptoms, and response rate after a 6-week trial of olanzapine, compared to haloperidol (272). A second study found that a significantly larger proportion of olanzapine-treated patients, compared with haloperidol-treated patients, remained in the trial and completed the first 12 weeks of treatment (279). In addition, the study found that the olanzapine-treated patients had slight but significant improvements in global psychopathology and negative symptoms and were more likely to meet the response criteria, although this difference only approached significance (p=0.06).

Four studies have examined the efficacy of olanzapine in the treatment of neurocognitive deficits of schizophrenia. Two of these studies found significant improvement in neurocognition as measured by a global index in olanzapine-treated patients, compared to haloperidol-treated patients (838, 902). One 12-week analysis of treatment effects in first-episode patients found significant improvement with olanzapine, compared to haloperidol, in global neurocognition assessed with a measure derived from a principal-component analysis, but the difference only approached significance when an empirically derived a priori measure of global neurocognition was used (922). The fourth study did not find differences between haloperidol and olanzapine in effects on global neurocognition (923). Olanzapine significantly improved motor function (838, 902), verbal fluency, nonverbal fluency and construction, immediate recall (902), general executive function (838, 923), and perceptual function and attention (838). Although a relatively consistent finding is that olanzapine has beneficial effects on neurocognition in schizophrenia, findings for the specific domains affected and the clinical significance of these effects are less clear. Further study is needed to determine the magnitude and clinical significance of the effects of olanzapine on neurocognition.

Several studies have examined the efficacy of olanzapine in patients with treatment-resistant illness (i.e., patients who have shown little or no response to adequate trials of other antipsychotics). In an 8-week double-blind study, 25 mg/day of olanzapine (N=42) and 1200 mg/day of chlorpromazine (N=39) demonstrated similar efficacy in the treatment of global psychopathology (924). In a 14-week double-blind trial, treatment with a mean dose of 30.4 mg/day of olanzapine (N=39) resulted in significantly greater improvement in global psychopathology and negative symptoms, compared with a mean dose of 25.7 mg/day of haloperidol (N=37) (820). A third study that used a Lilly clinical trial database to retrospectively identify patients without response to first-generation antipsychotics found that olanzapine-treated patients, compared to haloperidol-treated patients, had significantly greater improvements in global psychopathology and positive, negative, and mood symptoms; higher response rates; and higher completion rates (925). Although higher doses of olanzapine (doses up to 60 mg/day) are being used clinically for patients with treatment-resistant illness, current evidence of improved efficacy at higher doses is inconclusive (820, 926, 927).

Studies comparing olanzapine to other second-generation antipsychotics in the treatment of acute episodes have generally found similar efficacy for treatment of psychopathology both in patients with treatment-responsive symptoms and in those with treatment-resistant symptoms (381, 820, 905, 906), with some exceptions in which olanzapine was found to be superior (820, 902).

In terms of treatment during the stabilization and stable phases, analysis of data from the double-blind extension phase of 6-week acute treatment trials suggests that olanzapine may reduce the risk of relapse, compared to haloperidol. Pooling data across three studies, the investigators found that 19.7% of olanzapine-treated patients relapsed during the 1-year follow-up period, compared to 28% of haloperidol-treated patients (p<0.04) (928).

Olanzapine is associated with a low risk of extrapyramidal side effects, a low risk of sedation, a low risk of orthostatic hypotension and tachycardia, a low risk of cardiac conduction abnormalities, a moderate risk of anticholinergic effects, a high risk of weight gain and metabolic abnormalities, and a low risk of prolactin elevation and sexual side effects. Neuroleptic malignant syndrome occurs rarely with olanzapine. Details on the nature and management of each of these side effects are provided in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications".

Olanzapine is an effective antipsychotic when administered in doses of 10–20 mg/day in the acute phase of schizophrenia, although higher doses, up to 60 mg/day, have been reported to be used for patients with treatment-resistant schizophrenia (820, 926, 927). With the possible exception of akathisia, parkinsonian symptoms are infrequent at any dose of olanzapine.

During the titration and early treatment phase olanzapine-treated patients should be monitored for extrapyramidal side effects, orthostatic hypotension and reflex tachycardia, and sedation. Orthostatic hypotension may be more likely if benzodiazepines are coadministered (929). Evening administration may improve tolerance of the sedation that is common early in treatment. In addition, patients should be monitored for weight gain, glucose abnormalities, and hyperlipidemias that may occur during treatment with olanzapine. Table 1 outlines suggested monitoring and clinical management of such adverse effects.

Patients are typically managed with a single daily dose of olanzapine since the elimination half-life of olanzapine is 33 hours (ranging from 21 to 54 hours) (929). Olanzapine is primarily metabolized by the hepatic CYP1A2 enzyme, with a minor metabolic pathway involving the CYP2D6 enzyme. Inducers of the CYP1A2 enzyme (such as tobacco use) may reduce olanzapine plasma levels, and so changes in smoking status may affect efficacy and side effects at a given dose (500). There is some evidence to suggest differential metabolism of olanzapine by gender, with women exhibiting higher plasma concentrations than men at equivalent doses (509).

Quetiapine is a second-generation antipsychotic with antagonist activity at dopamine (D₁, D₂), serotonin (5-HT_1A, 5-HT_2A, 5-HT_2C), ₁-adrenergic, and histamine (H₁) receptors (818, 930).

There are several published clinical trials comparing the acute efficacy of quetiapine with that of placebo, first-generation antipsychotics (haloperidol or chlorpromazine), and other second-generation antipsychotics in patients with schizophrenia, schizoaffective disorder, and schizophreniform disorder. Placebo-controlled studies consistently demonstrate that in acutely relapsed patients quetiapine is efficacious in the treatment of global psychopathology, in improving the likelihood of clinical response (e.g., 20% improvement on rating scales of global psychopathology), and in improving the positive symptoms of schizophrenia (318, 931, 932). The evidence that negative symptoms improve with quetiapine treatment is less clear, as significant improvement with quetiapine, compared with placebo, is less consistently found across different doses of the drug and different studies (318, 931, 932). It is likely that the improvements in negative symptoms in these studies are due to decreased likelihood of secondary negative symptoms (e.g., related to parkinsonism or to psychosis) rather than to direct effects on primary negative symptoms. Active-comparator-controlled studies demonstrate comparable or occasionally greater improvement of global psychopathology and positive and negative symptoms, as well as an increased likelihood of clinical response with quetiapine, compared to haloperidol or chlorpromazine (933–935). Meta-analyses of these studies suggest that the efficacy of quetiapine is similar to that of first-generation antipsychotics (82, 86, 88).

In terms of relapse prevention, one 4-month randomized, open-label study compared the efficacy of quetiapine (mean dose=254 mg/day, N=553) to that of risperidone (mean dose=4.4 mg/day, N=175) (907). This study found both antipsychotics to have similar effects on global psychopathology and positive symptoms and negative symptoms, with marginally significant greater improvement in depressive symptoms in the quetiapine-treated patients. While this study lends preliminary evidence for the efficacy of quetiapine in preventing relapse, further studies using blinded methods are needed before definitive conclusions can be made.

One study compared the efficacy of 600 mg/day of quetiapine (N=143) to that of 20 mg/day of haloperidol (N=145) in patients with treatment-resistant illness and found that a significantly greater proportion of the quetiapine-treated patients met the response criteria (52%, compared to 38% of the haloperidol-treated patients) (936). However, the mean changes in global psychopathology, positive symptoms, and negative symptoms were similar for both groups.

Two studies found beneficial effects on neurocognition for quetiapine, compared to first-generation antipsychotics. In a 6-month randomized, double-blind study, significant improvement in global cognition, verbal reasoning and fluency, and immediate recall was found for subjects treated with 300–600 mg/day of quetiapine (N=13) but not for subjects treated with 10–20 mg/day of haloperidol (N=12) (937). Similarly, in a 24-week double-blind, randomized study, significantly greater improvement in global cognition, executive function, attention, and verbal memory was found for subjects treated with 600 mg/day of quetiapine, compared to subjects treated with 12 mg/day of haloperidol (938).

Quetiapine is associated with a very low risk of extrapyramidal side effects, a high risk of sedation, a moderate risk of orthostatic hypotension and tachycardia, a low risk of cardiac conduction abnormalities, a low risk of anticholinergic effects, a moderate risk of weight gain and metabolic abnormalities, and a low risk of prolactin elevation and sexual side effects. Neuroleptic malignant syndrome occurs rarely with quetiapine. Details on the nature and management of each of these side effects are provided in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications".

Preclinical studies in beagles found associations between quetiapine and increased risk of cataracts, prompting the FDA to suggest routine screening ophthalmological examinations before and every 6 months during quetiapine treatment. This risk has not been confirmed in humans, and there is no indication from postmarketing reporting of an association between increased cataract risk and quetiapine use (939).

Quetiapine is an effective antipsychotic when administered in doses of 300–800 mg/day in the acute phase of schizophrenia. Evidence suggests that the higher doses in this range (and perhaps doses greater than 800 mg/day) may be more efficacious (318). Even at doses above 800 mg/day, there are virtually no extrapyramidal side effects, with the possible exception of akathisia. During the titration and early treatment phase, quetiapine-treated patients should be monitored for orthostatic hypotension, reflex tachycardia, and sedation. Patients are typically managed with twice-daily dosing of quetiapine, since the elimination half-life is 6 hours (940). However, uneven dosing with the larger dose given at bedtime may improve tolerance of the sedation that is common early in treatment. In addition, patients should be monitored for weight gain, glucose abnormalities, and hyperlipidemias, which may occur during treatment with quetiapine. Table 1 outlines suggested strategies for the monitoring and clinical management of such adverse effects. Quetiapine is primarily metabolized by the hepatic cytochrome P450 CYP3A4 enzyme. Metabolism of the drug is minimally altered in patients with renal disease, but it may be significantly altered in patients with liver disease. Smoking does not affect the metabolism of quetiapine (500, 940). However, coadministration of phenytoin with quetiapine has been demonstrated to increase the clearance of quetiapine up to fivefold (941). Similarly potent inducers of CYP3A4 are likely to produce similar decreases in quetiapine levels, which may lead to loss of therapeutic efficacy.

Ziprasidone is a second-generation antipsychotic with antagonist activity at dopamine (D₂), serotonin (5-HT_2A, 5-HT_2C, and 5-HT_1B/1D), ₁-adrenergic, and histamine (H₁) receptors. In addition, ziprasidone has partial agonist activity at serotonin 5-HT_1A receptors and inhibits neuronal reuptake of serotonin and norepinephrine (942).

There are several published clinical trials comparing the efficacy of ziprasidone with placebo and with first-generation antipsychotics in the acute treatment of patients with schizophrenia, schizoaffective disorder, and schizophreniform disorder. Placebo-controlled studies consistently demonstrate that in acutely relapsed patients ziprasidone is efficacious in the treatment of global psychopathology and the positive symptoms of schizophrenia, as well as in increasing the likelihood of clinical response (e.g., 20% improvement on rating scales of global psychopathology) (943, 944). The evidence that negative symptoms improve with ziprasidone treatment is less clear, as significant improvement with ziprasidone, compared with placebo, is less consistently found across doses of drug and across studies (943–945). Active-comparator-controlled studies of ziprasidone compared with haloperidol demonstrate comparable improvement in positive and negative symptoms and in global psychopathology, as well as comparable likelihood of clinical response (945, 946). It is likely that the improvements in negative symptoms in these studies are due to a decreased likelihood of secondary negative symptoms (e.g., related to parkinsonism or to psychosis) rather than to direct effects on primary negative symptoms. However, in an additional placebo-controlled study of stable, residually symptomatic patients, the time course of improvement in negative symptoms was consistent with a therapeutic effect on primary negative symptoms (947). Nonetheless, this study was conducted with environmentally deprived persons who received more attention than usual by participating in the study, which may explain (in part) the observed improvements in negative symptoms.

One 52-week study demonstrated that ziprasidone is effective in reducing risk of relapse, compared with placebo, during the maintenance phase of treatment (947). Relapse risk was 43%, 35%, and 36% for patients receiving 40 mg/day, 80 mg/day, and 160 mg/day of ziprasidone, respectively, compared to 77% for placebo-treated patients.

Several studies have demonstrated the efficacy of intramuscular administration of ziprasidone for the treatment of acute agitation in relapsed patients with schizophrenia or schizoaffective disorder (76, 948, 949).

Ziprasidone is associated with a low risk of extrapyramidal side effects, a low risk of sedation, a low risk of orthostatic hypotension and tachycardia, a moderate risk of cardiac conduction abnormalities, a low risk of anticholinergic effects, a low risk of weight gain and metabolic abnormalities, and a low risk of prolactin elevation and sexual side effects. Neuroleptic malignant syndrome occurs rarely with ziprasidone. Details on the nature and management of each of these side effects are provided in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications".

While short-term clinical trials do not report insomnia as an adverse event with ziprasidone, there is some evidence from a longer-term outpatient study to suggest that stable outpatients whose medication is switched to ziprasidone may experience insomnia (945). This insomnia appears early in treatment, is typically transient, and most often responds to usual sedative-hypnotics (e.g., zolpidem, trazodone).

Ziprasidone is an effective antipsychotic when administered in doses of 80–200 mg/day in the acute phase of schizophrenia. There is emerging evidence that doses up to 320 mg/day may be safe, although there are no published data suggesting improved efficacy at high doses. Stable patients whose medication is switched to ziprasidone may report insomnia, which usually is transient and responsive to sedative-hypnotics.

Before treatment with ziprasidone is initiated in patients with preexisting cardiovascular disease and those who are at risk for electrolyte disturbances (e.g., patients taking diuretics and those with chronic diarrhea), the safety of using the medication should be evaluated. This evaluation should include laboratory assessment of electrolytes and an ECG. Preexisting prolonged QT syndrome, persistent findings of QTc interval >500 msec, history of arrhythmia, recent acute myocardial infarction, or uncompensated heart failure are contraindications to use of ziprasidone. The value of a screening ECG in apparently healthy persons to reliably detect congenital prolonged QT syndrome is not established and is of questionable utility, given the normal variability of the QT interval. During the maintenance phase of treatment, regular monitoring of electrolytes should be done for patients who are also treated with diuretics or who may be at risk for electrolyte disturbances. Patients should be monitored regularly for symptoms of possible arrhythmia, including dizziness, syncopal episodes, and palpitations. Patients with such symptoms should be referred for cardiovascular evaluation. Patients should also be warned about concomitant treatment with other drugs that also may affect the QT interval.

Patients are typically treated with twice-daily dosing of ziprasidone, since the elimination half-life is 7 hours, and steady state is reached after 1–3 days. Food increases the absorption of ziprasidone; under fasting conditions only 60% of ziprasidone will be absorbed. Two-thirds of ziprasidone is metabolized by aldehyde oxidase, and one-third by the cytochrome P450 system, primarily by the liver CYP3A4 enzyme, and to a lesser extent by CYP1A2 (950). Sex, age, smoking, and the presence of renal failure have not been found to affect the metabolism of ziprasidone, but liver disease potentially affects metabolism of the drug (951, 952). Ziprasidone has little effect on other liver enzyme systems and has not been found to affect the metabolism of other drugs.

Aripiprazole is pharmacologically distinct from other second-generation antipsychotic medications. It has partial agonist activity at dopamine (D₂) and serotonin (5-HT_1A) receptors and antagonist activity at dopamine (D₃), serotonin (5-HT_2A, 5-HT_2C, 5-HT₇), ₁-adrenergic, and histamine (H₁) receptors. In addition, aripiprazole inhibits neuronal reuptake of serotonin to a modest extent (953, 954).

There are several published clinical trials comparing the acute efficacy of aripiprazole with placebo and first-generation antipsychotics in patients with schizophrenia, schizoaffective disorder, and schizophreniform disorder. Placebo-controlled studies consistently demonstrate that in acutely relapsed patients, aripiprazole is efficacious in the treatment of global psychopathology, in improving the likelihood of clinical response (e.g., 20% improvement on rating scales of global psychopathology), and in improving the positive symptoms of schizophrenia (955). The evidence that negative symptoms improve with aripiprazole treatment is less clear, as significant improvement with aripiprazole, compared with placebo, is less consistently found across doses of drug and studies (955). It is possible that the improvements in negative symptoms in these studies are due to decreased likelihood of secondary negative symptoms (e.g., related to parkinsonism or to psychosis) rather than to direct effects on primary negative symptoms. Active-comparator-controlled studies demonstrate comparable or occasionally greater improvement of global psychopathology and positive and negative symptoms, as well as an increased likelihood of clinical response, with aripiprazole, compared to haloperidol or chlorpromazine (955, 956).

Two studies have found aripiprazole effective in reducing risk of relapse. In a 26-week randomized, double-blind trial that included stable patients with schizophrenia or schizoaffective disorder, the time to relapse was significantly longer for patients treated with aripiprazole (15 mg/day) than for those who received placebo, and a greater proportion of patients who received placebo (57%) than of aripiprazole-treated patients (34%) met the relapse criteria (957). In a 52-week randomized, double-blind trial that included patients with acute exacerbation of schizophrenia or schizoaffective disorder, the response rate and time to discontinuation for any reason were significantly greater for patients treated with aripiprazole (30 mg/day, N=647) than for those treated with haloperidol (10 mg/day, N=647) (unpublished 2003 manuscript of R.D. McQuade et al.). A greater proportion of aripiprazole-treated patients (43%) than of the haloperidol-treated patients (30%) completed the 52-week trial.

Aripiprazole is associated with a low risk of extrapyramidal side effects, a moderate risk of sedation, a low risk of orthostatic hypotension and tachycardia, a low risk of cardiac conduction abnormalities, a low risk of anticholinergic effects, a low risk of weight gain and metabolic abnormalities, and a low risk of prolactin elevation and sexual side effects. There have been no reports to date of neuroleptic malignant syndrome with aripiprazole. Details on the nature and management of each of these side effects are provided in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications".

Aripiprazole received FDA approval for use in the treatment of schizophrenia in late 2002, and thus experience with the drug in clinical settings and knowledge of rare side effects are limited. Insomnia was not reported as an adverse event in treatment trials involving patients with acutely exacerbated symptoms (955, 956). Trials that included stable patients found transient insomnia and acute agitation early in treatment, but these side effects typically resolved after several weeks (955, 956). While there are no systematic studies, it is reasonable to treat aripiprazole-associated insomnia with sedative-hypnotics (e.g., zolpidem, trazodone, antihistamines) and agitation with benzodiazepines.

Aripiprazole is an effective antipsychotic when administered in doses of 10–30 mg/day in the acute phase of schizophrenia. With the possible exception of akathisia, parkinsonian symptoms rarely occur within the usual dose range.

Stable patients whose medication was switched to aripiprazole may report insomnia that usually is transient and responsive to sedative-hypnotics.

Patients are typically treated with once-daily dosing of aripiprazole since the elimination half-life is 75 hours (94 hours for the active metabolite dehydro-aripiprazole), and steady state is reached after 14 days. Aripiprazole is metabolized by the cytochrome P450 system, primarily by the liver enzymes CYP2D6 and CYP3A4 (958). Sex, age, smoking, and the presence of renal or hepatic failure have not been found to significantly affect the metabolism of aripiprazole (958). Aripiprazole has little effect on other liver enzyme systems and has not been found to affect the metabolism of other drugs.

c) Shared side effects of antipsychotic medications

This section provides information on the side effects that are shared among multiple antipsychotic medications.

Neurological side effects of antipsychotic medications include acute extrapyramidal side effects such as medication-induced parkinsonism, dystonia, and akathisia; chronic extrapyramidal side effects such as tardive dyskinesia and tardive dystonia; and neuroleptic malignant syndrome.

Extrapyramidal side effects are especially common in patients treated with the first-generation antipsychotics and occur to varying extents with several of the second-generation agents, especially higher doses of risperidone. Of the spectrum of adverse effects of first-generation antipsychotic medications, the neurological side effects are the most common and the most troublesome (959, 960). Extrapyramidal side effects can broadly be divided into acute and chronic categories. Acute extrapyramidal side effects are signs and symptoms that occur in the first days and weeks of antipsychotic medication administration, are dose dependent, and are reversible with medication dose reduction or discontinuation. The three types of acute extrapyramidal side effects are parkinsonism, dystonia, and akathisia (961–964). Chronic extrapyramidal side effects are signs and symptoms that occur after months and years of antipsychotic medication administration, are not clearly dose dependent, and may persist after medication discontinuation. Chronic extrapyramidal side effects include tardive dyskinesia and tardive dystonia. Detailed descriptions and differential diagnoses of the extrapyramidal side effect syndromes are provided in the "Medication-Induced Movement Disorders" section of DSM-IV-TR. More than 60% of patients who receive acute treatment with first-generation antipsychotic medications develop clinically significant extrapyramidal side effects in one form or another (959, 960, 965). Some patients may develop more than one form at the same time. Second-generation drugs as a group cause fewer or no extrapyramidal side effects, relative to first-generation drugs. Studies using multiple doses of risperidone (886, 887, 890) have shown that risperidone causes a dose-related increase in extrapyramidal side effects, with risk highest in doses greater than 6 mg/day (82, 899). In any individual patient, it is likely that the maximally clinically effective dose of risperidone is lower than the dose that will cause extrapyramidal side effects. Thus, first-line intervention for extrapyramidal side effects due to risperidone should be to gradually lower the dose until symptoms resolve. The other second-generation drugs cause few or no extrapyramidal side effects, with the possible exception of akathisia. However, younger patients (children, adolescents, and young adults) may be more prone to extrapyramidal side effects from second-generation medications (unpublished 2003 manuscript of L. Sikich et al.).

Medication-induced parkinsonism is characterized by the symptoms of idiopathic Parkinson's disease (rigidity, tremor, akinesia, and bradykinesia) and is the most common form of extrapyramidal side effect caused by first-generation antipsychotics (787, 964). These symptoms arise in the first days and weeks of antipsychotic medication administration and are dose dependent. Medication-induced parkinsonism generally resolves after discontinuation of antipsychotic medication, although some cases of persisting symptoms have been reported (966, 967).

Akinesia or bradykinesia is a feature of medication-induced parkinsonism that affects both motor and cognitive function. A patient with this condition appears to be slow moving, less responsive to the environment, apathetic, emotionally constricted, and cognitively slowed. This effect has been noted alone or with other extrapyramidal side effects in almost one-half of patients treated with first-generation antipsychotics. In very severe cases, it may mimic catatonia. Akinesia is subjectively unpleasant and may be associated with poor medication adherence (968, 969). Depressive symptoms can also be present in patients with akinesia, in which case the syndrome is termed "akinetic depression" (970, 971). Symptoms of medication-induced parkinsonism, in particular the cognitive and emotional features, need to be carefully distinguished from the negative symptoms of schizophrenia. Furthermore, it is noteworthy that patients may experience these emotional and cognitive symptoms of parkinsonism in the absence of detectable motor symptoms.

The first approach to treatment of parkinsonism associated with first-generation antipsychotics should be to lower the antipsychotic dose to the EPS threshold (dose where minimal rigidity is detectable in a physical examination), since studies indicate that doses above the EPS threshold are unlikely to yield further clinical benefits (94). If dose reduction does not sufficiently improve symptoms, then a switch to a second-generation antipsychotic should be considered. Medications with anticholinergic (e.g., benztropine) or dopamine agonist (e.g., amantadine) activity often reduce the severity of parkinsonian symptoms. However, dopamine agonists carry a potential risk of exacerbating psychosis, and anticholinergic drugs can cause anticholinergic side effects. Thus, excessive doses and chronic use of these agents should be avoided or minimized (972, 973).

Acute dystonia is characterized by the spastic contraction of discrete muscle groups. Dystonic reactions occur in up to 10% of patients beginning therapy with high-potency first-generation antipsychotic agents. Although precise estimates of the incidence of dystonic reactions are not available, they appear to be less common with treatment with low-potency first-generation antipsychotic agents and relatively rare with second-generation antipsychotics. In addition to the use of high-potency medications, other risk factors for dystonic reactions include young age, male gender, high doses, and intramuscular administration. Dystonic reactions frequently arise after the first few doses of medication (90% occur within the first 3 days) (974). They can occur in various body regions but most commonly affect the muscles of the neck, larynx, eyes, and torso (963). The specific name of the reaction is derived from the specific anatomic region that is affected. Hence, the terms "torticollis,""laryngospasm,""oculogyric crisis," and "opisthotonos" are used to describe dystonic reactions in specific body regions (975). These reactions are sudden in onset, are dramatic in appearance, and can cause patients great distress. For some patients, these conditions, e.g., laryngospasm, can be dangerous and even life-threatening.

Acute dystonic reactions respond dramatically to the administration of anticholinergic or antihistaminic medication. Parenteral administration will have a more rapid onset of action than oral administration. Short-term maintenance treatment with an oral regimen of anticholinergic antiparkinsonian medication prevents the recurrence of acute dystonic reactions.

Akathisia is characterized by somatic restlessness that is manifest subjectively and objectively in up to 30% of patients treated with first-generation antipsychotics (961, 970). Although precise estimates of the incidence of akathisia are not available, it appears to be less common with low-potency first-generation antipsychotics and even more infrequent with second-generation antipsychotic agents. Patients characteristically complain of an inner sensation of restlessness and an irresistible urge to move various parts of their bodies. Objectively, this appears as increased motor activity. With mild akathisia, the patient may control body movements; in more severe forms, the patient may rock from foot to foot while standing, pace, and have difficulty sitting still. Even in mild forms in which the patient is able to control most movements, this side effect is often extremely distressing to patients, is a frequent cause of nonadherence with antipsychotic treatment, and, if allowed to persist, can produce dysphoria. Case reports suggest that akathisia may also be a possible contributor to aggressive or suicidal behavior (409). Intervention includes dose reduction or switching to a second-generation antipsychotic with less risk of akathisia. In this regard, however, it is important to note that risperidone may cause akathisia at the higher end of the dose range (887).

Effective treatments for akathisia include centrally acting beta-blockers such as a low dose of propranolol (30–90 mg/day) (972, 976). When these medications are administered, blood pressure and pulse rate should be monitored with dose changes. Benzodiazepines such as lorazepam and clonazepam are also effective in decreasing symptoms of akathisia (977). In contrast, anticholinergic antiparkinsonian medications have limited efficacy in treating akathisia (972). While there has been little systematic study, akathisia induced by risperidone or other second-generation antipsychotics is treated similarly to akathisia associated with first-generation antipsychotic treatment.

A common problem that arises in assessing patients with akathisia is distinguishing this side effect from psychomotor agitation associated with the psychosis. Mistaking akathisia for psychotic agitation and raising the dose of antipsychotic medication usually leads to a worsening of the akathisia and thus the agitation. When the etiology of agitation is unclear, the nonspecific effects of benzodiazepines on akathisia and agitation can be useful, although the dose necessary for therapeutic effects on psychotic agitation usually is higher than that required for akathisia (978).

Given the high rate of acute extrapyramidal side effects among patients receiving first-generation antipsychotic medications, and to a lesser extent risperidone, the prophylactic use of antiparkinsonian medications may be considered. The benefit of this approach has been demonstrated in several studies. For example, Hanlon et al. (979) found that only 10% of patients taking perphenazine with an antiparkinsonian medication developed an extrapyramidal side effect, in contrast to 27% of patients taking perphenazine without an antiparkinsonian medication. The risk is that some patients may be treated unnecessarily with these medications, risking anticholinergic side effects (978). However, schizophrenia is a long-term illness, and the development of a therapeutic alliance is of paramount importance. The minimization of uncomfortable, painful, and unnecessary side effects can contribute significantly to establishing such an alliance. Thus, prophylactic antiparkinsonian medication may be considered for all patients with a prior history of susceptibility to extrapyramidal side effects and for patients for whom antipsychotic agents known to induce these effects (e.g., first-generation agents, high doses of risperidone) are prescribed.

The various medications used to treat acute extrapyramidal side effects are listed in Table 5. The major differences among the anticholinergic medications are in their potencies and durations of action. Patients who are very sensitive to anticholinergic side effects (e.g., dry mouth, blurred vision, constipation) may require lower doses or less potent preparations (e.g., trihexyphenidyl, procyclidine hydrochloride). The need for anticholinergic medications should be reevaluated after the acute phase of treatment is over and whenever the dose of antipsychotic medication is changed. If the dose of antipsychotic medication is lowered, anticholinergic medication may no longer be necessary or may be given at a lower dose.

Tardive dyskinesia is a hyperkinetic abnormal involuntary movement disorder caused by sustained exposure to antipsychotic medication; tardive dyskinesia can affect neuromuscular function in any body region but is most commonly seen in the oral-facial region (980, 981). (For a description of tardive dyskinesia and its differential diagnosis, see DSM-IV-TR.) Evaluation of the risk of tardive dyskinesia is complicated by the fact that spontaneous dyskinesias are clinically indistinguishable from tardive dyskinesia and have been described in up to 20% of never-medicated patients with chronic schizophrenia, as well as in elderly patients (982, 983). Thus dyskinetic movements are part of the natural history of schizophrenia. Tardive dyskinesia occurs at a rate of approximately 4%–8% per year in adult patients treated with first-generation antipsychotics (980, 984). Various factors are associated with greater vulnerability to tardive dyskinesia, including older age, antipsychotic-induced parkinsonian symptoms, female gender combined with postmenopausal status, diagnosis of affective disorder (particularly major depressive disorder), concurrent general medical disease such as diabetes, and use of high doses of antipsychotic medications (982, 985–987). Studies comparing intermittent, targeted first-generation antipsychotic drug treatment with maintenance antipsychotic treatment have found increased risk of tardive dyskinesia with targeted treatment strategies (988).

Tardive dyskinesia has been reported after exposure to any of the available antipsychotic medications, although the risk appears to be substantially less (approximately 10-fold) with the second-generation antipsychotics, compared to first-generation antipsychotics (83, 319, 382, 989–992). One study summarizing available longitudinal clinical trial data with risperidone reports an annual risk of 0.3%, which is substantially less than the expected risk with first-generation antipsychotics of approximately 5% per year (989, 992, 993). In a 9-month study of older patients (mean age=66 years), substantially more patients treated with haloperidol (32%), compared with risperidone-treated patients (5%), developed tardive dyskinesia (535, 990). In these studies the mean dose of both antipsychotics was low, and the rates of tardive dyskinesia in the haloperidol-treated subjects were similar to those reported for older patients in other studies (987). For olanzapine, analyses of longitudinal double-blind data from multiple studies find a 12-fold lower risk of tardive dyskinesia with olanzapine treatment, compared to haloperidol treatment (0.05% and 7.45%, respectively) (319, 992). There are few systematic data concerning quetiapine and risk of tardive dyskinesia. In a 52-week open-label study of quetiapine that included 184 patients age >65 years, there was no change in the severity of dyskinetic movements, as evaluated by rating scales (994). In addition, emerging results from studies of other second-generation antipsychotics suggest that low risk of tardive dyskinesia may be found with drugs such as quetiapine that have a low risk of extrapyramidal effects.

With clozapine, although long-term prospective incidence studies are lacking, controlled short- and long-term trials generally find that the severity of dyskinetic movements improves with clozapine treatment, compared to treatment with first-generation antipsychotics (769, 995).

Although the majority of patients who develop tardive dyskinesia have mild symptoms, a proportion (approximately 10%) develop symptoms of moderate or severe degrees. An often severe variant of tardive dyskinesia is tardive dystonia, which is characterized by spastic muscle contractions in contrast to choreoathetoid movements (996). Tardive dystonia is often associated with great distress and physical discomfort. Patients receiving antipsychotic medication treatment on a sustained basis (for more than 4 weeks) should be evaluated at a minimum of every 3 months for signs of dyskinetic movements. The occurrence of dyskinetic movements warrants a neurological evaluation (980).

Treatment options for tardive dyskinesia occurring in the context of treatment with first-generation antipsychotic agents include switching to a second-generation antipsychotic or reducing the dose of the first-generation antipsychotic. An initial increase in dyskinetic symptoms may occur after conversion to a second-generation drug or antipsychotic dose reduction (withdrawal-emergent dyskinesia). With sustained first-generation antipsychotic exposure without dose reduction after the development of tardive dyskinesia, the likelihood of reversibility diminishes but is not lost. In some patients dyskinetic movements can persist despite long periods of time without medication. Despite the fact that continued treatment with antipsychotic medication increases the chances for the persistence of tardive dyskinesia symptoms, in many patients the severity of tardive dyskinesia does not increase over time at steady, moderate doses. The documentation in the clinical record should reflect that, despite mild tardive dyskinesia, a risk-benefit analysis favored continued maintenance of antipsychotic treatment to prevent the likelihood of relapse.

A large number of agents have been evaluated as possible treatment for tardive dyskinesia with few positive results. Although not consistent, there is some evidence that vitamin E may reduce the risk of development of tardive dyskinesia (225, 226). Given the low risk of side effects associated with vitamin E, patients may be advised to take 400–800 I.U. daily as prophylaxis. Small clinical trials have investigated the potential benefits of benzodiazepines, anticholinergic agents, calcium channel blockers (997), γ-aminobutyric acid agonists (998), essential fatty acids, estrogen, and insulin, with no studies yet producing convincing data to suggest any of these agents may be effective treatments for tardive dyskinesia (225, 999–1002).

Neuroleptic malignant syndrome is characterized by the triad of rigidity, hyperthermia, and autonomic instability, including hypertension and tachycardia (962). In addition, neuroleptic malignant syndrome is often associated with an elevated level of serum creatine kinase. In patients treated with second-generation antipsychotic medications, this classic triad of symptoms is generally although not invariably present (1003, 1004). The prevalence of neuroleptic malignant syndrome is uncertain, but this effect probably occurs in less than 1% of patients treated with first-generation antipsychotic medications (1005–1007) and is even more rare among patients treated with second-generation antipsychotic medications (1003, 1004, 1008–1012).

Neuroleptic malignant syndrome is frequently misdiagnosed and can be fatal in 5%–20% of patients if untreated (1013). It can be sudden and unpredictable in its onset and usually occurs early in the course of treatment, often within the first week after treatment is begun or the dose is increased. Risk factors for neuroleptic malignant syndrome include acute agitation, young age, male gender, preexisting neurological disability, physical illness, dehydration, rapid escalation of antipsychotic dose, use of high-potency medications, and use of intramuscular preparations (1014, 1015). Other diagnostic considerations in patients presenting with rigidity, hyperthermia, autonomic instability, or elevated levels of serum creatine kinase include neuroleptic-induced heat stroke, lethal catatonia, serotonin syndrome (in patients also taking serotonergic drugs such as SSRIs), anticholinergic syndrome, "benign" elevations in the level of serum creatine kinase, and fever in association with clozapine treatment (855, 1015–1018).

Since neuroleptic malignant syndrome is rare, most evidence regarding treatment comes from single case reports or case series. Antipsychotic medications should always be discontinued, and supportive treatment to maintain hydration and to treat the fever and cardiovascular, renal, or other symptoms should be provided. Some case series suggest that, compared with supportive treatment alone, treatment with dopamine agonists such as bromocriptine and amantadine or with dantrolene, which directly reduces skeletal muscle rigidity, may improve the symptoms of neuroleptic malignant syndrome (1019). Based on the overlap in symptoms between catatonia and neuroleptic malignant syndrome (1020), treatment with benzodiazepines, such as lorazepam, may also be helpful (1021, 1022). In patients with severe and treatment-resistant neuroleptic malignant syndrome, ECT is reported to improve symptoms (107, 1016, 1023). After several weeks of recovery, patients may be retreated with antipsychotic medication cautiously (1024). Generally, when treatment is resumed, doses are increased gradually, and a medication other than the precipitating agent is used (usually a second-generation antipsychotic or a first-generation antipsychotic medication of lower potency).

Sedation is a very common side effect of first-generation antipsychotic medications, as well as several of the second-generation agents, including clozapine, risperidone, olanzapine, and quetiapine. This effect may be related to antagonist effects of those drugs on histamine, adrenergic, and dopamine receptors (777, 1025, 1026). Most patients experience some sedation, particularly with the low-potency first-generation agents such as chlorpromazine, but it occurs to some extent with virtually all antipsychotic medications. With clozapine, sedation is very common, and in many patients it may be persistent and severe. Quetiapine has a high risk of sedation that may be maximal at the low end of the dose range (e.g., maximal by 100–200 mg/day). Olanzapine has a moderate dose-related risk of sedation. Risperidone produces dose-related sedation (890); within the usual dose range (<6 mg/day) the risk of sedation is relatively low, compared to the risk with other first-generation and second-generation (e.g., olanzapine, clozapine, quetiapine) antipsychotics.

Sedation is most pronounced in the initial phases of treatment, since most patients develop some tolerance to the sedating effects with continued administration. For agitated patients, the sedating effects of these medications in the initial phase of treatment can have therapeutic benefits. However, persistent sedation, including daytime drowsiness and increased sleep time, can interfere with social, recreational, and vocational function. Lowering of the daily dose, consolidation of divided doses into one evening dose, or changing to a less sedating antipsychotic medication may be effective in reducing the severity of sedation.

There are no systematic data on specific pharmacological interventions for sedation, but caffeine is a relatively safe option (1027). Some forms of psychostimulants (e.g., modafinil) have also been used to treat daytime drowsiness (1028). However, there have been case reports of clozapine toxicity associated with modafinil and other stimulant treatments of sedation, and thus this drug combination should be carefully considered and used with caution (1029, 1030).

Cardiovascular effects include orthostatic hypotension, tachycardia, and QTc prolongation.

Hypotension is related to the antiadrenergic effects of antipsychotic medications. With clozapine treatment initiation and dose escalation, there is a high risk of orthostatic hypotension and compensatory tachycardia, with rare (one of 3,000 patients treated) reports of cardiovascular collapse (854). These side effects typically limit the rate of titration, and orthostatic vital signs should be regularly monitored with dose escalation. When orthostatic hypotension is severe, it can cause dizziness and syncopal episodes. Patients who experience severe postural hypotension must be cautioned against getting up quickly and without assistance. Elderly patients are particularly prone to this adverse effect, and syncopal episodes may contribute to an increased risk of falls and related hip fractures in elderly patients. Risperidone has high affinity and quetiapine has moderate affinity for -adrenergic receptors and thus can produce orthostatic hypotension and reflex tachycardia. Clozapine has the highest affinity and greatest propensity to cause hypotension. Gradual dose titration starting with a low dose minimizes risk. Management strategies for orthostatic hypotension include decreasing or dividing doses of antipsychotic or switching to an antipsychotic without antiadrenergic effects. Supportive measures include the use of support stockings, increased dietary salt, and, as a last resort, administration of the salt/fluid-retaining corticosteroid fludrocortisone to increase intravascular volume.

Tachycardia can result from the anticholinergic effects of antipsychotic medications but may also occur as a result of postural hypotension. While healthy patients may be able to tolerate some increase in resting pulse rate, this may not be the case for patients with preexisting heart disease. Tachycardia unrelated to orthostatic blood pressure changes that result from anticholinergic effects may occur in up to 25% of patients treated with clozapine. Because of the cardiovascular side effects of clozapine, extreme care should be taken in initiating a clozapine trial in patients with heart disease. Tachycardia due to anticholinergic effects without hypotension can be managed with low doses of a peripherally acting beta-blocker (e.g., atenolol) (1031, 1032).

The length of time required for the heart ventricles to repolarize is measured by the QT interval on the electrocardiogram. The QT interval varies with heart rate; thus, a QT interval corrected for heart rate (the "QTc") is routinely used clinically. Prolongation of the QTc interval above 500 msec is associated with increased risk for a ventricular tachyarrhythmia, "torsades de pointes." Torsades de pointes is associated with syncopal episodes and may lead to life-threatening consequences (e.g., ventricular fibrillation, sudden death).

Among the first-generation antipsychotic agents, thioridazine, mesoridazine, pimozide, and high-dose intravenous haloperidol have been associated with risk of QTc prolongation (1033). Because of the clinically significant risk of torsades de pointes–type arrhythmias and the potential for related sudden death (1033), the FDA recommends that thioridazine should be used only when patients have not had a clinically acceptable response to other available antipsychotics (885). This safety warning is available online at http://www.fda.gov/medwatch/safety/2000/mellar.htm and at http://www.medsafe.govt.nz/Profs/PUarticles/thioridazine.htm.

Ziprasidone is associated with an average increase of 20 msec in the QTc interval; however, the clinical effects of this magnitude of QT prolongation are uncertain (1034). Unlike drugs that prolong the QTc interval to a greater extent (e.g., thioridazine) (1035), ziprasidone has not been reported to be associated with arrhythmias or sudden death (1034). Patients treated with ziprasidone should be monitored for other risk factors for torsades de pointes, including congenital prolonged QT syndrome, bradycardia, hypokalemia, hypomagnesemia, heart failure, and factors that might increase levels of a drug associated with QTc prolongation (e.g., hepatic or renal failure, overdose of ziprasidone or other drugs known to prolong the QTc interval). Concomitant treatment with other drugs known to significantly prolong the QTc interval at normal clinical doses should be avoided. A list of such drugs is available at http://www.torsades.org. Given the normal variability of the QT interval (about 100 msec), an ECG is of questionable value in screening for congenital prolonged QT syndrome or in evaluating the effects of ziprasidone on the QTc interval in individual patients. Glassman and Bigger (1036) have reviewed the literature on prolonged QTc interval, torsades de pointes, and sudden death with antipsychotic drugs, including ziprasidone.

The anticholinergic effects of first-generation antipsychotic medications (along with the anticholinergic effects of antiparkinsonian medications, if concurrently administered) can produce a variety of peripheral side effects, including dry mouth, blurred vision, constipation, tachycardia, urinary retention, and thermoregulatory effects. Anticholinergic side effects may occur in 10%–50% of treated patients (980, 1037). These effects are also common with the second-generation agent clozapine. Although most anticholinergic side effects are mild and tolerable, these side effects can be particularly troublesome for older patients (e.g., older men with benign prostatic hypertrophy) (1037). In rare instances, serious consequences of anticholinergic effects can occur. For example, death can result from ileus of the bowel if it is undetected. In addition, some patients can develop hyperthermia, particularly in warm weather.

Central anticholinergic effects include impaired learning and memory and slowed cognition. Symptoms of anticholinergic toxicity include confusion, delirium, somnolence, and hallucinations (1038, 1039). Such symptoms are more likely to occur with medications that have more potent anticholinergic effects (e.g., chlorpromazine, thioridazine) or from administration of anticholinergic antiparkinsonian medications and in elderly or medically debilitated patients. Clozapine is frequently associated with anticholinergic side effects, including constipation and urinary retention (1040, 1041). Rarely, these effects have been severe, resulting in fecal obstruction and paralytic ileus and enduring impairment of bladder function (1042). Because of these anticholinergic effects, patients with preexisting prostate hypertrophy require careful monitoring of urinary function, and clozapine is contraindicated in patients with narrow-angle glaucoma (1031, 1032). Olanzapine has moderate affinity for muscarinic receptors and acts as an antagonist at the M₁, M₂, M₃, and M₅ receptors; however, anticholinergic effects are infrequent. The rarity of these effects is believed to be due to a difference between the drug's in vitro binding affinities and its actions in vivo. Constipation is occasionally associated with olanzapine treatment, but generally there is a low risk of anticholinergic side effects with olanzapine. Quetiapine has moderate affinity for muscarinic receptors. Constipation and dry mouth are occasionally associated with quetiapine treatment, and elderly and medically debilitated patients may be more sensitive to its anticholinergic side effects.

Anticholinergic side effects are often dose-related and thus may improve with lowering of the dose or administration of the anticholinergic antiparkinsonian drug in divided doses. In cases of anticholinergic delirium, parenteral physostigmine (0.5–2.0 mg i.m. or i.v.) has been used to reverse the symptoms, although this treatment should be provided only under close medical monitoring.

Weight gain occurs with most antipsychotic agents. Up to 40% of patients treated with first-generation agents gain weight, with the greatest risk associated with the low-potency antipsychotics (797). The most notable exception is molindone, which may not cause significant weight gain (1043). The risk of weight gain with clozapine is thought to be the highest of all antipsychotics (1043), with studies reporting that between 10% and 50% of clozapine-treated patients are obese (1044, 1045). Typically, weight gain is progressive over the first 6 months of treatment, although some patients continue to gain weight indefinitely. In a meta-analysis of available studies, the mean weight gain after 10 weeks of treatment with clozapine was estimated at 4.45 kg (1043). Weight gain also is common in patients treated with risperidone and olanzapine. With risperidone, mean weight gain is estimated at 2.1 kg over the first 10 weeks of treatment (1043) and 2.3 kg after 1 year (382). With olanzapine, mean weight gain is estimated at 4.2 kg after 10 weeks of treatment (1043), and one study observed a mean weight gain of 12.2 kg after 1 year of treatment with olanzapine (918). No appreciable weight gain was observed with ziprasidone after 10 weeks (1043) or 1 year (947). Few studies have characterized the extent of weight gain with quetiapine or aripiprazole.

While studies have not systematically examined the health consequences of antipsychotic-related weight gain, the risk of cardiovascular disease, hypertension, cancers, diabetes, osteoarthritis, and sleep apnea is likely similar to that in idiopathic obesity. The association of high cholesterol and triglycerides with weight gain further increases the risk of cardiovascular disease (1046–1052). Adolescents may be particularly vulnerable to these side effects (1053).

Prevention of weight gain should be a high priority, since weight loss is difficult for many patients. Efforts should be made to intervene proactively, since obese persons rarely lose more than 10% of body weight with weight loss regimens. When weight gain occurs, clinicians should suggest or refer patients to diet and exercise interventions (1054). If the patient has not had substantial clinical benefits of the antipsychotic medication that outweigh the health risks of weight gain, a trial of an antipsychotic with lower weight-gain liability should be considered. Few systematic studies have been done to evaluate the effectiveness of specific interventions to prevent antipsychotic-induced weight gain or to promote weight loss, although potential strategies include diet and exercise programs (1055, 1056). No pharmacological interventions have proven efficacy in treating weight gain associated with second-generation antipsychotics, although uncontrolled studies have reported possible benefit from amantadine (1057, 1058), topiramate (1059–1063), the H₂ histamine antagonist nizatidine (1064, 1065), and noradrenergic reuptake inhibitor antidepressants (1066).

Uncontrolled studies and case series suggest that clozapine and olanzapine are associated with increased risk of hyperglycemia and diabetes (1050–1052, 1067–1073). While controlled studies are lacking, one prospective study found that 30 of 82 (36%) clozapine-treated outpatients developed diabetes during the 5-year follow-up period (1050). Complicating the evaluation of antipsychotic-related risk of diabetes is that schizophrenia is associated also with increased diabetes risk (1074). In some patients obesity may contribute to diabetes risk. Other mechanisms may also be involved. For example, insulin resistance may develop early in treatment with olanzapine and contribute to abnormal regulation of glucose and subsequent diabetes (1075, 1076).

Further, some of the second-generation antipsychotic agents, olanzapine and clozapine in particular, have been associated with diabetic ketoacidosis and nonketotic hyperosmolar coma, relatively rare complications of diabetes that are extremely dangerous if untreated (1077–1081). Numerous case reports have described scenarios in which diabetic ketoacidosis appears acutely in the absence of a known diagnosis of diabetes (1082). Diabetic ketoacidosis can present with mental status changes that can be attributed to schizophrenia. The treating psychiatrist must be aware of the possibility of diabetic ketoacidosis, given its potential lethality and its often confusing presentation. The overall prevalence and mechanism of diabetic ketoacidosis associated with antipsychotics and the differential risk of specific antipsychotic agents to cause this side effect are at present unknown.

Given the rare occurrence of extreme hyperglycemia, ketoacidosis, hyperosmolar coma, or death and the suggestion from epidemiological studies of an increased risk of treatment-emergent adverse events with second-generation antipsychotics, the FDA has requested all manufacturers of second-generation antipsychotic medications to include a warning in their product labeling regarding hyperglycemia and diabetes mellitus.

There is also suggestive evidence that certain antipsychotic medications, particularly clozapine and olanzapine, may increase the risk for hyperlipidemias. Most of the evidence is derived from case reports and other uncontrolled studies (1048–1050, 1067, 1070, 1083–1087). Pharmacological treatment with lipid-lowering drugs should be considered in patients with hyperlipidemia.

Table 1 lists suggested strategies for monitoring and clinical management associated with weight gain, glucose abnormalities, and hyperlipidemias in patients with schizophrenia.

Disturbances in sexual function can occur with a number of antipsychotic agents, including first- and second-generation agents (1088). Several mechanisms contribute to the genesis of sexual side effects with these medications. Prolactin elevation is very common in patients treated with first-generation antipsychotics as well as risperidone (1089). Female patients appear to be more sensitive to prolactin elevation than male patients (1090). All first-generation antipsychotic medications increase prolactin secretion by blocking the inhibitory actions of dopamine on lactotrophic cells in the anterior pituitary. This prolactin elevation may be even greater with risperidone than with first-generation antipsychotics. The reason for the propensity of risperidone to elevate prolactin may be due to risperidone's relative difficulty in crossing the blood-brain barrier, with the pituitary, which is outside the blood-brain barrier, exposed to higher peripheral levels of risperidone (1091).

Effects of hyperprolactinemia may include breast tenderness, breast enlargement, and lactation. Since prolactin also regulates gonadal function, hyperprolactinemia can lead to decreased production of gonadal hormones, including estrogen and testosterone. In women decreased gonadal hormone production may disrupt or even eliminate menstrual cycles. In both men and women prolactin-related disruption of the hypothalamic-pituitary-gonadal axis can lead to decreased sexual interest and impaired sexual function (1088).

The long-term clinical consequences of chronic elevation of prolactin are poorly understood. There is some epidemiological evidence, however, that postmenopausal women may have an increased risk of breast cancer if exposed to medications that potentially elevate levels of prolactin (1092). Chronic hypogonadal states may increase risk of osteopenia and osteoporosis (1093–1097), but increased risk of these disorders has not been directly linked to antipsychotic-induced hyperprolactinemia.

If a patient is experiencing clinical symptoms of prolactin elevation, the dose of antipsychotic may be reduced or the medication regimen may be switched to an antipsychotic with less effect on prolactin (e.g., any of the second-generation antipsychotics with the exception of risperidone). When the antipsychotic must be maintained, dopamine agonists such as bromocriptine (2–10 mg/day) or amantadine may reduce prolactin levels and thus the symptoms of hyperprolactinemia (1058).

The association between the other second-generation antipsychotic medications (clozapine, olanzapine, quetiapine, ziprasidone, and aripiprazole) and sexual dysfunction is less clear. Sexual interest and function may be reduced in both men and women receiving clozapine, but generally to a lesser extent than with first-generation antipsychotics (1098, 1099). Sexual dysfunction may also occur in patients treated with olanzapine and quetiapine (1100, 1101), but there is no prospective study that might indicate whether a causal relationship exists.

Erectile dysfunction occurs in 23%–54% of men treated with first-generation medications (812). Other effects can include ejaculatory disturbances in men and loss of libido or anorgasmia in women and men. In addition, with specific antipsychotic medications, including thioridazine and risperidone, retrograde ejaculation has been reported, most likely because of antiadrenergic and antiserotonergic effects (886). Dose reduction or discontinuation usually results in improvement or elimination of symptoms. A 25–50-mg dose of imipramine at bedtime may be helpful for treating retrograde ejaculation induced by thioridazine (1102). If dose reduction or a switch to an alternative medication is not feasible, yohimbine (an ₂-antagonist) or cyproheptadine (a 5-HT₂ antagonist) can be used (797). Because retrograde ejaculation is annoying rather than dangerous, psychoeducation may also help the patient tolerate this side effect. Priapism is very rarely associated with clozapine (1103, 1104), risperidone (1104), olanzapine (1105, 1106), quetiapine (1107), and ziprasidone (1108, 1109). There have been no reports to date of priapism associated with aripiprazole.

A wide variety of medications, including additional antipsychotics, have been added to antipsychotic medications, either to enhance their efficacy for the treatment of symptoms of schizophrenia or to treat other symptoms often associated with the illness. Targets of these added medications have included residual positive symptoms, negative symptoms, cognitive deficits, depression, agitation and aggression, obsessions and compulsions, and anxiety. Some medications (e.g., antidepressants) have been used for more than one symptom cluster (e.g., depression, obsessions and compulsions).

a) Anticonvulsants

A number of studies of the efficacy of carbamazepine and valproate in schizophrenia have been done. Excluding findings suggesting their use in treating patients whose illness has strong affective components, the evidence is quite convincing that neither agent, used alone, is of significant value in the long-term treatment of schizophrenia. Recent studies have tended to concentrate on use of anticonvulsants in combination with antipsychotics.

With carbamazepine, studies examining the effects of the drug in combination with first-generation antipsychotics have had negative findings (236, 1110). For valproate, on the other hand, both negative and positive results have been noted (102, 235–237, 1111). Most studies have included relatively few patients, but the study by Casey et al. (237) included 242 subjects with acutely exacerbated symptoms who were randomly assigned to receive risperidone or olanzapine, each combined with placebo or divalproex. Compared with the placebo group, the divalproex group improved significantly more rapidly over the first 2 weeks of treatment. Both groups were equally improved by the end of the study at 4 weeks. This intriguing result warrants further study and replication to establish whether divalproex augmentation shortens the time to discharge and to determine the value of longer-term divalproex augmentation.

There are generally no additional side effects from the combination of anticonvulsant and antipsychotic medications beyond those of the individual medications themselves. Carbamazepine is not recommended for use with clozapine, because of the potential of both medications to cause agranulocytosis.

For patients with schizophrenia, these medications are generally used in the same therapeutic dose ranges and blood levels that are used for the treatment of seizure disorders and bipolar disorder. Studies to determine dosing in schizophrenia have not been reported. A complicating factor is the fact that carbamazepine can decrease the blood levels of antipsychotic medications by induction of hepatic enzymes (1112–1114).

b) Antidepressants

Studies of antidepressants in schizophrenia broadly subdivide into those that have examined these agents as treatment for depression and those that have tested their efficacy for other symptoms, such as negative symptoms. These areas will be reviewed separately.

Good clinical practice dictates that clinicians be alert to the occurrence of depression in a broad spectrum of psychiatric and medical disorders and treat it when it is diagnosed. Earlier work (1115) indicated the effectiveness of a tricyclic antidepressant for symptoms of depression in schizophrenia, and 12-month follow-up showed the advantage of maintenance treatment (1116). One study found the effects of an SSRI (sertraline) to be equal to those of imipramine for treatment of postpsychotic depression (1117) and another noted positive effects of citalopram (540), but the only placebo-controlled study of an SSRI for treatment of depressed patients with schizophrenia showed a large placebo effect and no difference between groups (1118). Although the evidence is most strong for patients who meet the syndromal criteria for depression, two reviews have noted the paucity of evidence for the efficacy of antidepressants in schizophrenia (222, 1119). For clinicians, a further question, not addressed in the literature, is whether failure of an antidepressant to improve depression in a person with schizophrenia is an indication for changing antidepressants or changing antipsychotics.

A number of studies have tested the efficacy of antidepressants in treating the negative symptoms of schizophrenia. The overlap between depressive and negative symptoms has complicated study design and interpretation. In five placebo-controlled studies of SSRIs for negative symptoms, one reported a modest advantage of fluoxetine added to long-acting injectable antipsychotic medication (1120), while four found no advantage for SSRIs, compared with placebo, in patients receiving clozapine (1121) or first-generation antipsychotics (1122–1124). Several studies of adjunctive fluvoxamine have demonstrated positive results (1125–1127). An open-label study of selegiline found beneficial effects on negative symptoms (1128), but in a placebo-controlled trial both the selegiline and placebo groups improved, and there was no difference between them (1129). Overall, the evidence for efficacy of antidepressants for negative symptoms of schizophrenia is very modest. Since most of the studies have been done in combination with first-generation antipsychotics, it is possible that the findings might be different with second-generation antipsychotics, although this possibility seems unlikely.

In terms of treating other symptoms that are sometimes observed in patients with schizophrenia, two small studies found efficacy of clomipramine and fluvoxamine in treating obsessive-compulsive symptoms in schizophrenia (221, 223). In a small crossover study in which citalopram or placebo was added to first-generation antipsychotics, patients with a history of aggression had significantly fewer incidents while taking citalopram (1130).

Although the side effects of antidepressants are no different when administered to patients with schizophrenia than to patients with other disorders, combinations of antipsychotics and antidepressants have the potential for adverse, even dangerous, pharmacokinetic and pharmacodynamic interactions. In addition to prior history of response to antidepressant treatment, potential drug-drug interactions should be taken into account in selecting an antidepressant agent. Of particular concern with regard to drug toxicity are the inhibitory effects of some antidepressants on clozapine metabolism, leading to increased serum levels and risk of seizures. Fluvoxamine can cause large increases in clozapine serum levels, and the combination of the two drugs should be avoided. Some other SSRIs and nefazodone may also cause clinically significant increases in clozapine serum levels and should be used carefully in clozapine-treated patients. Clozapine serum levels should be monitored after adding one of the antidepressants discussed earlier to the medication regimen of patients treated with clozapine. Because bupropion itself is associated with a risk of seizures, a pharmacodynamic interaction with clozapine exists. Therefore, the combination of clozapine and bupropion should be avoided. There are many sources of information about drug-drug interactions. A useful, frequently updated web site maintained by D. Flockhart at Indiana University is available at http://medicine.iupui.edu/flockhart. Another useful drug interaction computer program maintained by J. Oesterheld and D. Osser is available at http://www.mhc.com/Cytochromes.

Use of antidepressants in schizophrenia generally has been studied by using the doses and titration schedules that are usually used when the agents are administered by themselves. There is no reason to think that dosing should be modified on the basis of coexisting schizophrenia. As noted earlier, however, the potential for drug-drug interactions suggests that close monitoring of side effects is warranted. Monitoring of the blood levels of the antipsychotic at baseline and after several weeks of antidepressant treatment may be helpful, particularly for clozapine, where there is evidence that high blood levels are associated with increased risk of seizures and low levels may be ineffective. The same considerations apply when an antidepressant is being discontinued.

c) Antipsychotics

Most reports on the combination of antipsychotics describe the effects of combinations with clozapine. The only randomized, controlled trial used sulpiride, a dopamine receptor antagonist similar to first-generation antipsychotics that is available in Europe but not in North America. Shiloh et al. (1131) added placebo or sulpiride, titrated up to a dose of 600 mg/day, to clozapine for 10 weeks in the treatment of 28 partially responsive patients who were taking stable doses of clozapine and who had BPRS scores >42. The sulpiride group had significantly greater decreases in BPRS (15%), Scale for the Assessment of Negative Symptoms (10%), and Scale for the Assessment of Positive Symptoms (12%) scores.

Case series show improvements in residual positive symptoms with the addition of a number of other antipsychotics to clozapine. These agents include loxapine (233), pimozide (234), and risperidone (232).

Although the quality of the evidence for augmentation of clozapine with another antipsychotic is modest, this strategy seems reasonable in treating patients whose response to clozapine is fair at best. Before taking this step, however, the clinician should be sure that the clozapine treatment has been of sufficient duration and that the patient's blood level of clozapine indicates a sufficient dose. The other alternatives—switching to monotherapy with a different antipsychotic not already tried or combining two other antipsychotics—have even less evidence to support them than does augmentation of clozapine.

Combinations of two or more antipsychotics, neither of which is clozapine, are also used frequently for treatment of schizophrenia (1132). Some of this use reflects periods of cross-titration in the transition from one antipsychotic to another, but much of it represents long-term treatment. Evidence for (or against) this practice is minimal, as there are no controlled studies in the literature. The largest case series includes six persons with inadequate responses to 20–40 mg/day of olanzapine, who had average decreases in BPRS and PANSS scores of 35% after addition of 60–600 mg/day of sulpiride for at least 10 weeks (1133). Without a control group, such results are difficult to evaluate. Moreover, sulpiride is not available in the United States, and there is no way to know if similar results might be found with other antipsychotics.

The absence of evidence for combinations of antipsychotics does not mean that there are no patients who are best treated with such a combination. However, their use should be justified by strong documentation that the patient is not equally benefited by monotherapy with either component of the combination. Practitioners should be aware of the problems inherent in combination therapies, including increased side effects and drug interactions as well as increased costs and decreased adherence (1132).

d) Benzodiazepines

Benzodiazepines have been evaluated as monotherapy for schizophrenia and as adjuncts to antipsychotic medications. Wolkowitz and Pickar (224) reviewed double-blind studies of benzodiazepines as monotherapy and found that positive effects (reductions in anxiety, agitation, global impairment, or psychotic symptoms) were reported in nine of 14 studies. Six of 10 studies that specifically examined psychotic symptoms showed greater efficacy for benzodiazepines than placebo. In a study comparing diazepam, fluphenazine, and placebo as treatments for impending psychotic relapse in patients who were taking no antipsychotic medications, the effects of diazepam and fluphenazine were equal, and both were superior to placebo (1134).

Double-blind studies evaluating benzodiazepines as adjuncts to antipsychotic medications were also reviewed by Wolkowitz and Pickar (224). Seven of 16 studies showed some positive effect on anxiety, agitation, psychosis, or global impairment; five of 13 showed efficacy in treating psychotic symptoms specifically. The reviewers concluded that benzodiazepines may improve the response to antipsychotic medications.

Some studies indicate that the effectiveness of benzodiazepines as adjuncts to antipsychotic medications is limited to the acute phase and may not be sustained. Altamura et al. (1135) found that clonazepam plus haloperidol, but not haloperidol alone or placebo, produced significant lowering of total BPRS scores after 1 week. This reduction, which was primarily due to decreases in anxiety and tension, disappeared by the end of the 4-week study. Csernansky et al. (1136) also found that when alprazolam was added to antipsychotic medication, there was a significant reduction in the BPRS withdrawal/retardation subfactor score after the first week, but this reduction disappeared by study end at week 5.

Benzodiazepines are commonly used alone or in combination with an antipsychotic for acutely agitated patients in emergency department settings. One study compared the effects of lorazepam with those of haloperidol over the first 4 hours of treatment (1137). The compounds were equal in efficacy, and the authors suggested that lorazepam may be preferable, in that delayed extrapyramidal symptoms can occur with haloperidol. Another study compared lorazepam and haloperidol alone with the combination of both over 12 hours (75). Combination treatment was modestly more effective during the first 3 hours, and there were no significant differences between groups at later times. The haloperidol alone group needed more injections and had more extrapyramidal symptoms.

Benzodiazepines are effective for treatment of acute catatonic reactions, whether associated with schizophrenia or other disorders (137, 140, 142, 1138–1141). Although most studies have used lorazepam (1–2 mg i.v. or i.m. or 2–4 mg p.o., repeated as needed over 48–72 hours), beneficial effects have also been found with clonazepam and oxazepam. One report has questioned the value of benzodiazepines in treating chronic catatonia, although patients were maintained on antipsychotic treatment during the study, and the contribution of tardive dystonia to the observed behaviors was uncertain (1142).

Benzodiazepines have some limitations in schizophrenia. Their common side effects include sedation, ataxia, cognitive impairment, and a tendency to cause behavioral disinhibition in some patients. This last side effect can be a serious problem in patients who are being treated for agitation. Reactions to withdrawal from benzodiazepines can include psychosis and seizures. In addition, patients with schizophrenia are vulnerable to both abuse of and addiction to these agents.

Evidence relating to the choice of a specific benzodiazepine is limited, since few studies have compared the effectiveness of more than one. Important considerations in selection include abuse potential and severity of withdrawal symptoms if treatment is prolonged. In general, longer-acting agents have lower abuse potential. Withdrawal of alprazolam seems more likely to be associated with seizures, compared to withdrawal of other benzodiazepines.

e) Beta-blockers

Beta-blocking agents are often used for treatment of drug-induced akathisia, discussed in Section V.A.1.c, "Shared Side Effects of Antipsychotic Medications". There are also a few controlled studies of the combination of beta-blockers with antipsychotics to treat aggression. Pindolol in a dose of 5 mg t.i.d. reduced aggression scores significantly more than placebo in a double-blind crossover study that included 30 male patients with schizophrenia in a maximum-security facility (100). In a psychiatric intensive care setting, 80–120 mg/day of nadolol had initial beneficial effects on psychosis scores and extrapyramidal symptoms, compared with placebo (101). The difference in extrapyramidal symptoms persisted over the 3 weeks of the study. In both of these studies, most patients were taking first-generation agents. Replication of the findings with aggressive patients taking second-generation agents would be helpful. As noted earlier, clozapine is indicated as a treatment for persistently aggressive, psychotic patients.

f) Cognition enhancers

Cognitive deficits are characteristic of schizophrenia, and several studies have examined the efficacy of adding acetylcholinesterase inhibitors developed for use in dementia to treat patients with schizophrenia. One case report found substantial cognitive benefits from donepezil, compared with placebo (247), and an uncontrolled study observed positive results with donepezil on a variety of cognitive measures (249). However, a randomized, placebo-controlled trial of donepezil in 34 patients with chronic schizophrenia reported no group differences (248). As such, there is currently insufficient evidence to support the usefulness of these agents in improving cognitive performance in schizophrenia.

g) Glutamatergic agents

Because phencyclidine, which blocks ion channels associated with NMDA-type glutamate receptors, can produce a clinical state with psychotic and negative symptoms resembling schizophrenia, agents with glutamatergic properties have been tested in schizophrenia (1143). The agents that have been tested are glycine, d-cycloserine, and d-serine. Of these, only d-cycloserine is available for medicinal human use in the United States, as an antituberculosis treatment.

Five randomized, controlled trials have examined the effects of glycine in doses ranging from 0.4 to 0.8 g/kg. Most have reported beneficial effects of glycine on negative symptoms, with decreases of 15%–40% in negative symptom measures (239, 240, 242, 1144). Little effect on positive symptoms has been found in most studies. In a group of 30 patients who were taking clozapine, glycine did not produce any significant symptom changes, compared with placebo (241), confirming the result of an earlier case series report (1145). Javitt et al. (242) did, however, report robust negative symptom improvements in four patients who received clozapine plus glycine.

Results with d-cycloserine are more variable. The usual dose is 50 mg/day. Modest, but significant, decreases in negative symptoms were found by some investigators (244–246) but not others (1146, 1147). In a study of d-cycloserine added to clozapine, there was no benefit of the combination (243), which may be related to its dose-response curve.

A report by Tsai et al. (238) on adjunctive d-serine noted significant decreases in positive and negative symptoms in patients stabilized with a first-generation antipsychotic or risperidone. The same group later reported no benefit from adding d-serine to clozapine (1148).

Overall, the evidence for glutamatergic agents is encouraging, except as additions to clozapine. Most studies have used first-generation antipsychotics, risperidone, or clozapine. It remains to be seen if combinations of glutamatergic agents with other second-generation antipsychotics are helpful. Although the data seem most positive for glycine, studies directly comparing these agents are needed to determine if their effects actually differ.

h) Lithium

Lithium as a sole treatment has limited effectiveness in schizophrenia and is inferior to treatment with antipsychotic medications (1149–1152).

Earlier reports indicated that when added to antipsychotic medications, lithium augmented the antipsychotic response, in general, and improved negative symptoms specifically (1153, 1154). Other evidence indicated benefits of lithium for patients with schizophrenia with affective symptoms and for patients with schizoaffective disorder (1155–1159).

More recent literature, however, has not reported robust effects (1160). Relatively low doses of lithium over an 8-week period improved anxiety symptoms more than did placebo, but effects in other areas of psychopathology were not found (1161). Patients who had not responded to 6 months of treatment with fluphenazine decanoate showed no more improvement than the placebo group after 8 weeks of lithium augmentation at therapeutic levels (1162). There have been no reported controlled trials of lithium combined with second-generation antipsychotics. Since at least some of these agents have evidence for effects on depression, anxiety, and mood stabilization, the potential value of combining lithium with them may be limited.

The side effects of lithium include tremor, gastrointestinal distress, sedation or lethargy, impaired coordination, weight gain, cognitive problems, nephrogenic diabetes insipidus with associated polyuria and polydipsia, renal insufficiency, hair loss, benign leukocytosis, acne, and edema. These have been reviewed in detail in APA's Practice Guideline for the Treatment of Patients With Bipolar Disorder (1163) (included in this volume). The combination of an antipsychotic medication and lithium may increase the possibility of the development of neuroleptic malignant syndrome. However, the evidence for this association comes mainly from some debated reports of cases or series of cases, rather than from quantitative data. Most reported cases of neuroleptic malignant syndrome in patients treated with lithium plus antipsychotic medication have occurred in cases of high lithium blood levels associated with dehydration.

Generally, lithium is added to the antipsychotic medication that the patient is already receiving, after the patient has had an adequate trial of the antipsychotic medication but has reached a plateau in the level of response and has persisting residual symptoms. The dose of lithium is that required to obtain a blood level in the range of 0.8–1.2 meq/liter. Response to treatment usually appears promptly; a trial of 3–4 weeks is adequate for determining whether there is a therapeutic response, although some investigators have noted that improvements may emerge only after 12 weeks or more (1160). Patients should be monitored for adverse effects that are commonly associated with lithium (e.g., polyuria, tremor) and with its interaction with an antipsychotic medication (e.g., extrapyramidal side effects, confusion, disorientation, other signs of neuroleptic malignant syndrome) (266), particularly during the initial period of combined treatment. Given the toxicity of lithium in overdose, prescription of conservative quantities should be considered for patients at increased risk for suicidal behaviors.

i) Monoaminergic agents

Some studies have examined the efficacy of adjunctive dopaminergic and noradrenergic agents in schizophrenia. High doses of oral tyrosine added to molindone produced no clinical effects different from placebo in a crossover study of 11 patients, even though there was physiological evidence that the tyrosine had CNS effects (1164). Clonidine added to 20 mg/day of haloperidol reduced psychotic symptoms more than placebo in a small study of 12 patients (1165). The lack of an effect of clonidine on chronic polydipsia in schizophrenia has recently been reported (459).

j) Polyunsaturated fatty acids

Based on hypotheses concerning membrane stability and second messenger dysfunction in schizophrenia, several investigators have tested the efficacy of polyunsaturated fatty acids in the illness. The bulk of the evidence comes from studies of eicosapentaenoic acid (EPA). One study also examined docosahexaenoic acid and found it had no effects (1166).

In separate studies, Peet et al. (1166) found that EPA added to a stable dose of antipsychotic improved total and positive symptoms more than placebo and that EPA alone was more effective as sole treatment for unmedicated schizophrenia patients than placebo alone. By contrast, Fenton et al. (1167) found no benefit of EPA, compared with placebo, in a study of more than 80 patients. Emsley et al. (1168), in a South African cohort, noted greater decreases in PANSS and dyskinesia scores with EPA added to stable antipsychotic dose. The two phenomena were correlated; those with dyskinesia improvements were most likely to also have symptom improvements. Patients taking clozapine did not benefit from EPA treatment.

The EPA data are intriguing but far from definitive. Most data come from studies of combination therapy with first-generation antipsychotics. The compound appears to be free of side effects other than initial mild gastrointestinal upset in some patients.