World Psychiatric Association Pharmacopsychiatry Section statement on comparative effectiveness of antipsychotics in the treatment of schizophrenia

1. Introduction 

The past decade witnessed major changes in the practice of antipsychotic therapy around the world. The sequential introduction of eleven “atypical” or second-generation antipsychotics (clozapine, amisulpride, zotepine, risperidone, olanzapine, quetiapine, sertindole, ziprasidone, aripiprazole, perospirone, and paliperidone), led to increased optimism among physicians and patients about what can be achieved with effective antipsychotic therapy in schizophrenia. Like the 51 older “typical” or “conventional” first-generation antipsychotics (FGAs) (Table 1), the eleven newer “atypical” second-generation antipsychotics (SGAs) are at least as effective in reducing the delusional thinking, hallucinatory experiences, and thought disorganization that are the hallmarks of psychosis (positive symptoms). Compared to the older FGAs, however, this new generation of SGAs has a lower liability for inducing extrapyramidal side effects (EPS) resulting in their being termed “atypical” (Moller, 2000, Kapur and Remington, 2001, Meltzer, 2004, Tandon, 2007). SGAs were considered by practitioners to be more effective than FGAs in terms of possessing a broader spectrum of efficacy (particularly with regard to negative, cognitive, and mood symptoms) and having a lower liability to cause acute and long-term motor side-effects. Consequently, expert panels and professional organizations around the world uniformly recommended the use of these newer agents in preference to the older ones (World Psychiatric Associations, 2002, Kane et al., 2003, Miller et al., 2004, American Psychiatric Association, 2004, Falkai et al., 2005).

Table 1. List of approved antipsychotics around the world, 2007

	Typical or conventional or first-generation antipsychotics (FGAs) N=51
	Phenothiazines (N=23)
	Acetophenazine	Butaperazine	Chlorproethazine	Chlorpromazine
	Cyamemazine	Dixyrazine	Fluphenazine	Mesoridazine
	Methotrimeprazine	Perazine	Periciazine	Perphenazine
	Piperacetazine	Pipoptiazine	Prochlorperazine	Promazine
	Propericiazine	Sulforidazine	Thioridazine	Thiopropazate
	Thioproperazine	Trifluoperazine	Triflupromazine
	Non-Phenothiazines (N=28)
	Benperidol	Bromperidol	Chlorprothixene	Clocapramine
	Clopenthixol	Clothiapine	Droperidol	Fluanisone
	Flupenthixol	Fluspirilene	Haloperidol	Loxapine
	Melperone	Molindone	Moperone	Mosapramine
	Nemonapride	Oxypertine	Penfluridol	Pimozide
	Pipamperone	Sulpiride	Sultopride	Thiothixene
	Tiapride	Timiperone	Trifluperidol	Zuclopenthixol
	ATYPICAL OR SECOND GENERATION ANTIPSYCHOTICS (SGAs) N=11
	Amisulpride	Aripiprazole	Clozapine	Olanzapine
	Paliperidone	Perospirone	Quetiapine	Risperidone
	Sertindole	Ziprasidone	Zotepine

Amidst this enthusiasm, governments became concerned about the rapidly escalating costs for this class of medications as global direct expenditures on antipsychotic medications multiplied over 20-fold over the past decade from approximately $ 0.5 billion/year to more than $ 15 billion/year (Hoenberg and Goetz, 2006). This increase was principally driven by the proportionately greater use of the 5-30 fold more costly SGAs at the expense of the older FGAs. Even as many experts believed that SGAs are substantially better than FGAs, some questioned any such advantage. Governments and clinicians also became increasingly skeptical about available comparative information about antipsychotic agents, much of which was derived from pharmaceutical industry-sponsored trials (Melander et al., 2003, Perlis et al., 2005, Lexchin et al., 2003, Montgomery et al., 2004, Heres et al., 2006). It was consequently hoped that results of two large government-sponsored studies [Clinical Antipsychotic Trial of Intervention Effectiveness in schizophrenia in the U.S.A. (CATIE, Lieberman et al., 2005, McEvoy et al., 2006, Stroup et al., 2006a) and Cost Utility of the Latest Antipsychotics in Severe Schizophrenia in the U.K. (CUtLASS, Lewis et al., 2006a, Lewis et al., 2006b, Jones et al., 2006)] would provide definitive answers. Both these trials found SGAs to be neither more effective nor less likely to cause EPS than FGAs (Lieberman et al., 2005, Jones et al., 2006), clozapine to be more effective than other SGAs in refractory schizophrenia (McEvoy et al., 2006, Lewis et al., 2006b), and FGAs to be more cost-effective than SGAs (Rosenheck et al., 2006, Davies et al., 2007). Instead of resolving the controversy about the comparative effectiveness of older and newer antipsychotics, however, findings of CATIE and CUtLASS further fueled disagreement in the field about how SGAs compare to FGAs. While primary investigators of these studies interpreted the results to establish FGAs as being clinically equivalent to SGAs, other experts pointed to study sample and design constraints and suggested the need for caution in interpreting findings of CATIE and CUtLASS. Whereas critics opined that SGAs are clearly superior to FGAs and described these studies as “fundamentally flawed and adding little of value”, proponents suggested that failure to accept the finality of their conclusion that FGAs are at least as effective and safe as SGAs “reflected an inability to face the truth”.

In view of this resulting confusion, the Section of Pharmacopsychiatry in the World Psychiatry Association developed this comprehensive position statement on the comparative effectiveness of different antipsychotic treatments for schizophrenia. The Chairman of the section (HJM) engaged its forty other members from 22 countries over two years in the construction of the statement. Anticipating uncertainty and controversy following the initial publication of CATIE findings in September, 2005 (Lieberman et al., 2005), the Section membership resolved to develop a precise consensus statement about comparative effectiveness of antipsychotics in the treatment of schizophrenia. An initial draft of the paper was distributed in December 2005. Following comments received from Section members, the statement was very substantially revised and a new draft was distributed in June, 2006. Between June–December, 2006, there was an electronic dialogue among members of the Section that led to progressive refinements in the statement. In view of the difficulty in integrating disparate opinions among 41 psychopharmacology experts from around the world, presentation of a majority and minority report was considered; the Section membership uniformly felt, however, that presentation of divergent opinions would only add to confusion and assiduously labored to resolve significant areas of disagreement and develop a statement that all the members could sign on to. As new results from CATIE and CUtLASS continued to be published through 2007 and other findings became available, these were incorporated. The manuscript was revised by the authors in response to the initial editorial review and the updated statement was reviewed by the entire Section membership with additional appropriate changes being made based on input. All members formally approved the final overall statement; a substantial majority approved the entire wording, with four members dissenting on some individual points of emphasis [two of the 41 members of the section felt that the EPS advantage of SGAs over FGAs should have been further emphasized whereas two other members felt that the metabolic disadvantages of SGAs versus FGAs needed additional emphasis]. In addition to providing a unified expert perspective on the comparative effectiveness of antipsychotic treatments, the renewed emphasis on “optimal recovery” in the context of individualized treatment of schizophrenia (Anthony, 1993) also makes this update timely.

2. Comparing effectiveness of different antipsychotic agents in the treatment of schizophrenia 

Although pharmacological findings derived from animal studies and receptor imaging can be useful, only data from clinical trials in schizophrenia can directly provide information about the comparative effectiveness of antipsychotic agents in its treatment. Randomized patient assignment to parallel treatment arms allows one to attribute confidently observed differences in effectiveness between these arms to the effects of the treatments; therefore, well-conducted randomized controlled trials [RCT's] represent the gold standard of clinical information (McAlister et al., 1999, Pocock and Elbourne, 2000). Study findings need to be replicated for both accuracy and generalizability and therefore, multiple RCTs comparing every pair of antipsychotic agents in large samples of schizophrenia patients at various stages of illness across diverse populations would be ideal. The fact that this is not feasible necessitates application of both direct and indirect methods to RCT data in order to elucidate comparative effectiveness of different antipsychotic agents in the treatment of schizophrenia. Utilizing data from the approximately 1,600 RCTs of antipsychotic treatment of schizophrenia (Gilbody et al., 2002, Tharyan, 2005), the three broad approaches (Glenny et al., 2005) include:

Medical literature published in any language on RCTs of antipsychotic treatment of schizophrenia was identified using MEDLINE and EMBASE, supplemented by the reference lists of published articles. Search terms included randomized controlled trial, clinical trial, schizophrenia, antipsychotic, neuroleptic, and generic names of the 62 antipsychotic agents (Table 1). We decided to consider only published studies because they have been subjected to peer-review, are of higher quality, and provide more complete information than meeting abstracts (Dundar et al., 2006, Scherer et al., 2007). We also decided not to seek additional information about unpublished studies from manufacturers of individual antipsychotic agents for the same reason and because of the likely greater retrieval bias in the collection and pooling of such data. We believe the rigor and transparency advantages of this method substantially outweigh the constraint of publication bias (Texier et al., 2007, Martin et al., 2005). We do recognize the significant potential distortions produced by publication bias, however, and in the final section of this report, we discuss strategies to reduce such bias.

3. Meta-analysis 

Meta-analysis is a powerful “averaging” technique that pools data from several clinical trials and then provides a quantitative estimate of treatment effect size. It can be applied to each of the above approaches, but requires the studies whose data are pooled to be ‘similar’ with sufficient information available from each primary study in order to compute a pooled treatment effect size. Common problems affecting meta-analyses include small sample sizes, inadequate power, study heterogeneity, lack of extractable data, lack of interchangeable measurement instruments and definitions of outcomes, and other differences in the design of studies whose data are pooled. This results in multiple tensions between the competing needs of being comprehensive versus preserving the integrity of the meta-analytic method:

Necessary compromises made to resolve these competing needs detract from the internal validity and/or generalizability of the treatment effect size generated by the meta-analysis (Noble, 2006, Mismetti et al., 2007).

4. Applying the indirect method to randomized controlled clinical trials of antipsychotic treatment of schizophrenia 

There are approximately 1600 RCTs of antipsychotic treatment in patients with schizophrenia. For the purpose of registration in different countries, all antipsychotic agents have to be studied over a 4–8 week period among patients with schizophrenia and schizoaffective disorder during an acute psychotic exacerbation, using a randomized, parallel-group design with double-blind assessment; usually the Brief Psychiatric Rating Scale (BPRS, Overall and Gorham, 1962, Guy, 1976) or the Positive and Negative Symptoms Scale (PANSS, Kay et al., 1987) is utilized for symptom assessment. Consequently, studies with the above characteristics provide the best data for the application of indirect approaches to comparing the efficacy of different antipsychotic agents in schizophrenia. Total BPRS and PANSS scores provide reliable measures of overall psychopathology and their positive and negative symptom subscale scores provide reliable measures of positive and negative symptom severity, respectively. This indirect approach can be applied in two ways: (i) “naively”, by comparing change scores for different agents, without subtracting effects of a common comparator; and (ii) “after adjustment”, comparing change scores after subtracting the effects of a common comparator such as placebo or another common agent utilized across studies.

The principal advantage of the naive indirect method of comparing efficacy is that it allows the unbiased pooling of the largest amount of RCT data without the need for data transformation. Since this method does not involve averaging of a “difference from control” effect, all RCT's with specified common criteria can be included, regardless of the nature of the control group (placebo, haloperidol, other). As this allows for a larger pool of RCTs to be available for analysis, greater rigor can be employed in specifying common RCT inclusion characteristics (since using more rigorous inclusion/exclusion criteria reduces the number of studies available for pooled comparison). Pooling estimates of mean change from baseline to endpoint across different agents (without subtracting a common “control” effect) also represents the major shortcoming of this method as it necessitates the assumption that there were no systematic variations in studies across the different agents.

5. Applying the crude or naive indirect method 

Utilizing the PANSS as the assessment instrument, studies find an average reduction of 14–20 points in the total PANSS score across the various agents over a 4–8 week period; FGAs and SGAs are found to produce a similar range of improvement. In one such analysis (Tandon and Jibson, 2005), the average improvement among approximately 5000 patients across 22 studies of risperidone, olanzapine, and quetiapine was compared. This analysis discovered that while there was considerable variation in the degree of improvement with a particular agent across its different studies, average improvement across agents was similar for all efficacy parameters considered (18.1–18.5 point total PANSS reduction across agents; 5.1–5.7 point PANSS positive score reduction across agents; and 4.4–4.5 point PANSS negative score reduction across agents). When data from amisulpride, sertindole, zotepine, and paliperidone studies were similarly analyzed, average improvements in PANSS total, positive symptom, and negative symptom scores were found to be similar. When data from all the aripiprazole studies are analyzed, a smaller improvement is noted (14.1 points on the PANSS total, 4.2 points on the PANSS positive and 3.2 points on the PANSS negative); similarly, when data from ziprasidone studies are analyzed, improvements of 15.4 points on the PANSS total, 4.4 points on the PANSS positive and 3.7 points on the PANSS negative subscales are noted. Most of these data are, however, derived from 4-week studies; controlling for the duration of trial, no significant difference in mean improvement is noted across agents (Tandon and Fleischhacker, 2005). The only FGA with an adequate amount of relevant PANSS data is haloperidol; average improvements of 15.6 on the PANSS total, 4.7 on the PANSS positive and 3.2 on the PANSS negative subscales are noted over a 4–8 week period.

Utilizing the BPRS as the assessment instrument, studies find an average reduction of 10–17 points in the total BPRS score across the various agents over a 4–8 week period; again, FGAs and SGAs are found to produce a similar range of improvement. The studies that provide the data for such analyses were conducted over a 40-year period and consequently are even more heterogeneous than the ones using the PANSS.

6. Applying the refined or adjusted indirect method 

Subtracting effects of a common comparator, three major meta-analytic comparisons of the efficacy of FGAs versus SGAs have been published (Leucht et al., 1999, Geddes et al., 2000, Davis et al., 2003). Leucht et al. (1999) conducted a meta-analysis of 21 randomized, controlled clinical trials, comparing olanzapine, quetiapine, risperidone, and sertindole to haloperidol and placebo. They observed that all antipsychotics were “about equally more effective than placebo”. Additionally, risperidone and olanzapine (but not sertindole or quetiapine) were found to be “slightly more effective than haloperidol”. Comparing the efficacy of these antipsychotics based on relative effect size versus haloperidol appears valid except that the haloperidol dose with which the different agents were compared to was not the same. Higher comparator doses of haloperidol were employed in many risperidone and olanzapine studies (15–20 mg/day) than in many quetiapine and sertindole studies (10–12 mg/day). Higher doses of haloperidol cause more EPS (Zimbroff et al., 1997, Hugenholtz et al., 2006, Tandon and Nasrallah, 2006) and earlier drop-out from the study, leading to a reduced estimate of efficacy. An examination of the effects of haloperidol in the Leucht analysis indicates that the suggested differences in efficacy among risperidone, olanzapine, quetiapine, and sertindole based on different effect sizes versus haloperidol were, in fact, due to differences in the effects of haloperidol in these comparisons (Tandon and Fleischhacker, 2005).

In a more comprehensive meta-analysis of 52 RCTs of antipsychotic efficacy in schizophrenia (Geddes et al., 2000), “atypical” antipsychotics were found to be marginally superior to conventional antipsychotics in the treatment of schizophrenia but the apparent benefits of SGAs over haloperidol diminished when comparator doses of <12 mg/day of haloperidol were considered. It needs to be emphasized that an advantage for the “atypical” class (albeit smaller) was noted even when compared to <12 mg/day of haloperidol. Geddes and co-workers further noted that “indirect comparison in meta-regression models did not identify any individual atypical antipsychotic as more or less effective when dose of comparator drug was taken into account.”

In the most comprehensive meta-analysis to-date (Davis et al., 2003), Davis et al. utilized data from 124 RCTs comparing one of 10 second generation antipsychotics (SGAs) (amisulpride, aripiprazole, clozapine, olanzapine, remoxipride, risperidone, quetiapine, sertindole, ziprasidone, and zotepine) to a first-generation antipsychotic (FGA). Based on their analysis, the authors concluded that there are two groups of SGAs: group 1 (comprised of amisulpride, clozapine, olanzapine, and risperidone) more efficacious than FGAs; and group 2 (comprised of the other six) not differently efficacious than FGAs. This conclusion is weakened by the fact that the doses of various SGAs that the authors stipulate as equivalent for use in their analyses are questionable (Kane et al., 2003), there were differences in the cohorts of patients who participated in studies of the newer versus the older SGAs, and higher comparator doses of haloperidol were utilized against group 1 than against group 2 SGAs (Tandon and Fleischhacker, 2005, Hugenholtz et al., 2006). Sensitivity analyses performed by the authors (Tandon and Nasrallah, 2006) did not adequately address these limitations.

Whereas these analyses are direct comparisons of FGAs versus SGAs, they provide an adjusted indirect comparison between different SGAs. Comparing FGAs to SGAs, these analyses suggest a modest advantage for SGAs, which diminishes with optimal FGA dosing. With regard to the issue of comparative efficacy among various SGAs, these indirect RCT meta-analyses do not provide undisputed evidence of differential efficacy (Tandon and Fleischhacker, 2005, Motlova et al., 2007).

7. Applying the direct method: head to head randomized controlled clinical trials 

7.2. CUtLASS 

CUtLASS (Cost Utility of the Latest Antipsychotics in Severe Schizophrenia) was a large multi-center government-sponsored pragmatic trial conducted in the United Kingdom to compare the effectiveness of SGAs versus FGAs (Lewis et al., 2006a, Lewis et al., 2006b, Jones et al., 2006). In band 1 of the study (Jones et al., 2006, Table 2), 227 individuals with schizophrenia poorly responsive to or intolerant of current antipsychotic treatment were randomized to receive either an FGA (any one of 11, including four long-acting preparations, chosen by the treating clinician) or an SGA (either risperidone, olanzapine, quetiapine, or amisulpride, chosen by the clinician) and quality of life was evaluated over one year. No significant differences in quality of life or symptoms were observed among patients assigned to the two groups. Furthermore, FGAs were found to be more cost-effective than SGAs because they were both less costly and more effective in terms of quality-adjusted life-years (Davies et al., 2007).

Table 2. Description of CATIE phase-1 and CUtLASS band-1 studies

		CUtLASS Band 1	CATIE
		Single-phase study	Three-phase study
	SETTING	14 NHS trusts in the UK	57 sites across the USA
	SUBJECTS
	Number	227 subjects	1,460 subjects
	Diagnosis	75% schizophrenia	100% Schizophrenia
	Mean age	41 years	41 years
	Gender	68% male	74% male
	Mental illness duration	14 years	16 years
	First episode patients	Included (13% of total)	Excluded (0% of total)
	Baseline antipsychotic	82% on FGA	15% on FGA
	Baseline PANSS total	72.2	75.7
	METHODS
	When conducted	1999–2003	2001–2004
	Study duration	12 months	18 months
	Randomized assignment to	FGA or SGA Class (50% FGA)	One of 5 agents (4 SGAs, 1 FGA) (20% FGA) in Phase 1
	Antipsychotic switching	All patients switched agents 49% switched class 51% stayed in same class	15% continued on same agent 57% actually switched agents 28% started new agent
	Comparison groups	FGA versus SGA Arms	Five Antipsychotic Arms
	Primary outcome measure	Quality of Life (QOL)	All-Cause Antipsychotic Discontinuation
	Clinical care and primary outcome assessment	Medication blinded to raters but not to patients and physicians	Medication blinded to raters, patients and physicians

At the time the study was conducted, comparing entire classes of FGA versus SGA antipsychotics may have seemed useful; in retrospect, the significant heterogeneity among both FGAs and SGAs likely lessened power and reduced the interpretability of the findings (Tandon et al., 2007, Moller, 2007). Furthermore, in CUtLASS-1, individuals assigned to an FGA could receive either oral or long-acting injectable antipsychotic medication in contrast to those assigned to an SGA who could only receive oral medication. 84 of 227 participants in CUtLASS 1 were receiving a depot antipsychotic at baseline; the vast majority of these individuals was switched to oral agents (FGA or SGA) and did poorly. Only twelve subjects were initially randomized to receive a depot antipsychotic but this number had tripled to 35 by the end of the one-year study. These findings of CUtLASS support the importance of treatment adherence and the potential utility of long-acting injectable antipsychotic agents in the treatment of schizophrenia (Kane, 2006a, Tiihonen et al., 2006, Nasrallah, 2007). A re-assessment of the role of the eleven currently-available long-acting injectable antipsychotic preparations around the world is warranted.

8. CATIE 

Phase-1 and phase-2 results from the U.S.A. government-funded Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) in schizophrenia have been published (Lieberman et al., 2005, McEvoy et al., 2006, Stroup et al., 2006a). In terms of duration (18 months) and numbers of patients (1,460), CATIE is the most comprehensive RCT ever conducted in chronic schizophrenia. In the first phase of this three-phase trial, 1,460 patients with chronic schizophrenia were randomized to receive perphenazine (an FGA) or risperidone, olanzapine, quetiapine, or ziprasidone (four SGAs) (Table 2). At study entry, patients were relatively stable with moderate illness severity on average (CGI-S=4; PANSS total=75.7; (Guy, 1976, Leucht et al., 2005)) and three-quarters of them were receiving an antipsychotic. Patients were randomized to receive perphenazine, risperidone, olanzapine, quetiapine, or ziprasidone (Stroup et al., 2003). Phase-1 study completion rate was 26 percent across the 18-month study- 36% for olanzapine, 26% for risperidone, 25% for perphenazine, 21% for ziprasidone, and 18% for quetiapine. Patients were likely to stay on olanzapine significantly longer than on the other four agents; there were no significant differences between the other four agents in this regard. More recently, the authors reported the presence of broad neurocognitive deficits in a vast majority of CATIE participants at baseline (Keefe et al., 2006) with modest improvement during the course of treatment without any significant differences among patients receiving the five agents (Keefe et al., 2007). While small improvements in psychosocial functioning were observed among study patients over the 18-month study, there were no differences across the five treatment arms (Swartz et al., 2007). There were no differences in quality of life measures between patients in the five treatment arms (Rosenheck et al., 2006). Finally, perphenazine was described by the authors to be the most cost-effective agent because it was less costly but no less effective than the other agents (Rosenheck et al., 2006).

The 1,052 patients who prematurely discontinued phase-1 treatment on an initially-assigned SGA were eligible to participate in phase-2 of the study. Of note, they could choose to enter either the clozapine (efficacy pathway) or the ziprasidone (tolerability pathway) arms of the study regardless of their reason for discontinuing phase-1. Ninety-nine subjects chose to participate in the clozapine-other SGA comparator study (McEvoy et al., 2006); 444 chose to participate in the ziprasidone-other SGA comparator study (Stroup et al., 2006a); and 509 chose to exit the study. In the clozapine pathway, individuals were assigned to receive either clozapine or another SGA which they had not received in phase-1. In the ziprasidone pathway, individuals were assigned to receive ziprasidone or another SGA which they had not received in phase-1. Treatment effectiveness in this phase was again primarily measured by the duration that patients remained on their originally assigned antipsychotic medication. In phase-2, clozapine was found to be more effective than other SGAs among patients who participated in the efficacy arm (McEvoy et al., 2006) whereas risperidone and olanzapine were noted to be more effective than quetiapine and ziprasidone among patients who chose to participate in the tolerability arm (Stroup et al., 2006a).

Results of this landmark study warrant close scrutiny because of its importance and a range of methodological considerations (Table 2, Moller, 2005, Tandon, 2006a, Kane, 2006b, Tandon, 2006b, Meltzer and Bobo, 2006, Casey, 2006, Kasper and Winkler, 2006, Freedman et al., 2006, Janicak, 2006, Essock et al., 2006, Rosenheck et al., 2007a). The FGA-SGA comparison is compromised by the assignment of the 231 patients with tardive dyskinesia to treatment with SGAs but not the FGA perphenazine. While this may have been clinically desirable because SGAs are associated with a lower risk of tardive dyskinesia than FGAs (Correll et al., 2004, Kane, 2006c, Tarsy and Baldessarini, 2006), patients with tardive dyskinesia are known to suffer from a more severe and less treatment-responsive form of the illness along with a much higher propensity for EPS (Tenback et al., 2006). Consequently, investigators conducted comparisons between perphenazine and the SGAs after excluding patients with tardive dyskinesia. In conjunction with the inclusion-exclusion criteria of the study, this ensured a study sample at relatively low risk for EPS (Tandon, 2006a, Kane, 2006b, Tandon, 2006b, Meltzer and Bobo, 2006) and therefore a reduced assay sensitivity to detect differences in EPS (Tandon and Constantine, 2006, Rosenheck et al., 2007b). This reduced assay sensitivity is confirmed by the fact that while CATIE documented more frequent treatment discontinuation because of EPS among perphenazine-treated patients, it did not detect any differences in various EPS ratings between patients treated with perphenazine or SGAs. Since SGA-FGA differences in spectrum of efficacy are principally driven by the lower liability of SGAs to cause EPS, CATIE's finding of no differences in EPS between perphenazine and the study SGAs might explain why it also found no differences in overall effectiveness, cognition, negative symptoms, and tardive dyskinesia (Tandon and Constantine, 2006; Fig. 1). Since findings of any study apply to populations similar to the one examined, CATIE's findings of no FGA-SGA difference in any measure of efficacy would specifically apply to patients with schizophrenia at low risk for EPS.

View Large Image Download to PowerPoint

Fig. 1. How can first- and second-generation antipsychotics be more effective?

The comparison between the different agents was significantly biased by differences in switching rates across the five agents in the context of the risk of switching antipsychotics among relatively stable, moderately ill schizophrenia patients (on average) as in CATIE. In phase-1 of the study, a significant association between antipsychotic switching rates and 18-month all-cause discontinuation was noted; when these switching effects were controlled for, no significant differences in effectiveness among the five agents remained (Essock et al., 2006). The comparison between different SGAs may have been further compromised by the suboptimal dosing of quetiapine and ziprasidone (and possibly risperidone) relative to olanzapine (Kane et al., 2003).

These methodological issues (Tandon, 2006a, Kane, 2006b, Tandon, 2006b, Meltzer and Bobo, 2006, Casey, 2006, Kasper and Winkler, 2006, Freedman et al., 2006, Janicak, 2006, Essock et al., 2006, Rosenheck et al., 2007a, Tandon and Constantine, 2006, Rosenheck et al., 2007b) do not invalidate the many important lessons from this landmark study, but instead suggest the need to understand its findings in the context of its study design and integrate its results with those of other relevant studies.

9. Optimizing antipsychotic therapy in schizophrenia 

Despite significant advances in our understanding of the nature of the disease, schizophrenia remains one of the most challenging medical conditions of our times. It is characterized by high morbidity and mortality, and available treatments for schizophrenia are incompletely and variably effective and associated with a range of adverse effects. Despite these impediments, however, the individualized provision of a comprehensive array of treatment, rehabilitative and social support services can effectively promote recovery of persons with schizophrenia (Fig. 2, (Tandon et al., 2006)).

View Large Image Download to PowerPoint

Fig. 2. Recovery means optimizing individual outcomes.

While existing antipsychotic treatments for schizophrenia are not completely satisfactory, they can substantially reduce disease burden and make a meaningful difference in the life of each individual patient. Towards this end, critical evaluation of the sizable body of available information about antipsychotics and its individualized application are essential along with efforts to resolve important unanswered clinical questions. In this regard, we need to characterize antipsychotic agents using all the available information, conduct additional requisite studies to better understand and discriminate between them, disseminate maximally current and accurate information about these agents to all who use them, and articulate practice standards that promote appropriate use of these medications.

10. Evaluating the evidence and the science of antipsychotic therapy 

FGAs and SGAs both constitute very heterogeneous classes of antipsychotic medications. The principal distinction between SGAs and FGAs is the ability of the former to provide an equivalent antipsychotic effect with a lower liability to cause EPS, although there is substantial variation within each class in this regard. Clinical trial data collectively indicate that SGAs may demonstrate a broader spectrum of efficacy than FGAs (equivalent efficacy in positive symptoms and variably greater efficacy in negative, mood, and cognitive symptoms), substantially related to their lower propensity to cause motor side-effects. Whereas neither the modestly greater efficacy of SGAs over FGAs nor its significant explanation by the reduced propensity of SGAs versus FGAs to cause EPS is fully established, they are consistent with the preponderance of data. Recent findings further reinforce the importance of achieving an antipsychotic effect without EPS. Clozapine is more effective than other antipsychotic agents in treatment-refractory schizophrenia patients (specifically those with antipsychotic-refractory positive symptoms), ameliorating symptoms in about one-third of such patients; reduced EPS liability does not explain this greater efficacy.

Indirect comparisons of efficacy across studies and direct comparisons in the relatively few randomized, controlled head-to-head studies between other SGAs suggest that they are essentially similar with regards to overall efficacy and efficacy in treating positive and negative symptoms; if differences exist, these are small. The occasional observation of the superior efficacy of some SGAs may be explained by the fact that optimal dose ranges for olanzapine, risperidone, and amisulpride are somewhat better defined than those for quetiapine, ziprasidone, and aripiprazole. Additionally, the need for multiple-daily dosing and administration with food may complicate the real-life optimal dosing of ziprasidone.

In view of the significant individual variability in drug pharmacokinetics and treatment responsivity, it should also be emphasized that equivalent overall efficacy across patient groups does not translate into equal efficacy in each individual patient. There is no best agent or a best dose of any agent for all patients. It is not currently possible to prospectively predict which antipsychotic medication might be optimal for a given patient. Decisions about antipsychotic therapy consequently entail a trial and error process with careful monitoring of clinical response and adverse effects and an ongoing risk-benefit assessment.

Substantial differences in the adverse effect profiles of these medications are very well documented and different individuals are differentially tolerant of different adverse effects. Generally, SGAs have a lower propensity to induce EPS than FGAs although there are differences among both groups of agents with regard to the ease and consistency with which an adequate antipsychotic effect can be achieved without EPS (Tandon and Jibson, 2002, Weiden, 2007). Metabolic side-effects which increase the risk of ischemic heart disease (weight gain, dyslipidemia, diabetes mellitus) have recently received particular attention (Newcomer, 2005, Franciosi et al., 2005). In view of their likely contribution to the increased mortality of persons with schizophrenia, they warrant close attention. Although SGAs are generally associated with metabolic adverse effects to a greater extent than FGAs, there are variations among both SGAs and FGAs with regard to their liability to cause these side-effects. CATIE, for example, found olanzapine to cause more weight gain and related metabolic side-effects than perphenazine, risperidone and quetiapine to cause equivalent weight gain to perphenazine, while ziprasidone caused fewer metabolic problems than perphenazine. Other adverse effects differ among FGAs and SGAs as well (Tandon et al., 2006); the impact of these adverse effects on both the safety and tolerability of these agents differs across individuals. A minimum protocol for monitoring adverse effects should be implemented and this needs to be customized to the patient's specific vulnerabilities/needs and the agent selected (Marder et al., 2004, Constantine et al., 2006, Table 3).

Table 3. Minimum Health Monitoring Recommendations (adapted from Mount Sinai Conference (Marder et al., 2004) and State of Florida (Constantine et al., 2006) Guidelines)

		Baseline	4 Weeks	8 Weeks	12 Weeks	Quarterly	Annually
	History	X					X
	Weight (BMI)	X	X	X	X	X
	Waist circumference	X			X		X
	Blood pressure	X			X		X
	Fasting Gluc & Hb A1c	X			X		X
	Fasting lipid profile	X			X		X
	Sexual Function	X	X		X	X
	EPS and TD	X	X	X	X	X
	Ocular evaluation	X					X

The frequency of monitoring may need to be increased if the patient has additional vulnerability to that adverse effect or the agent utilized is more likely to cause it. Other monitoring based on agent and patient. Agent-specific monitoring (eg., WBC monitoring with clozapine).

If bothersome or hazardous adverse effects develop, informed treatment adjustments should be made collaboratively by the physician-patient team (Seale et al., 2006). Switching antipsychotics because of inadequate efficacy or poor tolerability has been found to be useful (Weiden et al., 2003, Tandon et al., 2006), but the risks of discontinuing a partially effective treatment need to be weighed against the benefits of switching to a possibly more effective one (Tandon, 2006a, Davis et al., 2006, Edlinger et al., 2005); both the risks and benefits were evident in CATIE and to a more limited extent in CUtLASS.

Antipsychotic treatment has a significant impact on the long-term course of schizophrenic illness and can significantly facilitate recovery (Fig. 2). Early intervention has much promise in reducing the substantial decline in function that frequently accompanies the onset of schizophrenic illness, but effects of early antipsychotic treatment and differences between agents are inadequately characterized. Extrapolating effectiveness findings from trials conducted in chronically ill subjects as in CATIE and CUtLASS to first-episode patients warrant great caution. While previous studies suggest benefits of SGAs over FGAs in this population (Schooler, 2007), results of the recently completed European First-Episode Schizophrenia Trial (EUFEST, Fleischhacker et al., 2005) are awaited with great interest. In later stages of schizophrenia, high rates of treatment nonadherence result in incomplete remissions and frequent relapses; a variety of pharmacological, psychosocial, and disease-management strategies can be utilized to improve treatment adherence and promote recovery (Kane, 2006a, Tandon et al., 2006).

11. The individual with schizophrenia, the physician providing antipsychotic treatment, and the treatment system 

It is the responsibility of every individual patient, his or her doctor, and the clinic in which the patient is receiving treatment to ensure that the best possible treatment is being provided to enable the best possible outcome to be achieved (Anthony, 1993, Tandon et al., 2006, Tunner and Salzer, 2006, Committee on Crossing the Quality Chasm, 2005).

Collaboratively, the person with schizophrenia and their doctor need to evaluate antipsychotic treatment options, weigh issues of efficacy and safety/tolerability in the context of available information and the patient's preferences, and choose a suitable treatment. Treatment targets should be explicitly defined, the impact of treatment on the individual carefully monitored, and adjustments in treatment made based on response. Data related to reliable measures of symptomatology, side-effects, risk factors for metabolic and other diseases, appropriate laboratory tests, measurement of mild EPS, and other relevant individualized risk-benefit data should be obtained in a timely manner, included in the medical records and utilized to assist decision-making. Fiscal implications should be considered in conjunction with the various clinical considerations.

The treatment system needs to ensure that adequate resources are available to clinicians and patients to optimize individual recovery. A full range of treatment options along with a culture of recovery and accountability are crucial. Necessary information should be provided to the patient and suitable tools provided to the treating psychiatrist to promote best-possible collaborative treatment. The system can promote the appropriate use of evidence-based treatments by clearly enunciating principles of good practice, requiring outcome measurement, monitoring actual practice, and providing feedback to clinicians about their practice (Table 4).

Table 4. Strategies to encourage evidence-based antipsychotic therapy (adapted from Constantine et al., 2006)

	•Promulgate clear guidelines for use.
	– Articulate principles of use [Defined objectives]
	– Indicate how to use [Dosing, trial duration, sequence]
	– Specify protocols and tools for monitoring [Efficacy, physical health]
	•Provide point-of-care aids to treating physician.
	– Information about prescription refills, etc.
	•Track actual prescribing practices and use a hierarchy of interventions to guide them.
	– Monitor practice and provide feedback.
	– Targeted interventions for unusual practices.

The psychiatrist should provide disciplined and targeted pharmacological treatment whose impact (both beneficial and adverse effects along with their functional implications) is reliably measured. The advisability of any pharmacological treatment is determined by balancing the expected benefits against the potential risks of that treatment, considering the costs of the treatment. Pharmacological treatment decisions made by the psychiatrist in conjunction with the individual receiving treatment for schizophrenia include initial choice of antipsychotic, dosing strategy of chosen agent, determination of specific treatment target and assessment of response, and choosing between available options (e.g., allowing more time, changing the dose, adding another medication, switching antipsychotics) in the event of an unsatisfactory response.

The approach of formulating explicit rules for sequential decision-making should be complemented by efforts to carefully scrutinize pharmacological practices “outside the norm”. The norm is generally defined by a combination of clinical trial evidence and expert consensus; outlier analysis is the tool employed to identify pharmacological practices outside these normative parameters (Constantine et al., 2006). A proximate objective of antipsychotic treatment of schizophrenia is to achieve a maximum reduction of positive symptoms while avoiding EPS or using anticholinergic medications. Although it is possible to achieve this goal with suitably-dosed FGA treatment in a fair number of individuals, SGA treatment may provide a better likelihood of more predictably accomplishing this objective (Fig. 1). If the initially selected treatment does not accomplish this goal, appropriate treatment adjustments need to be made. Other key objectives include optimizing reduction of other symptom domains while minimizing adverse effects considering individual patient vulnerabilities and preferences. Since cardiovascular disease is the principal contributor to the doubling of age-standardized mortality among persons with schizophrenia, metabolic and other risk factors for ischemic heart disease warrant particular attention (Newcomer, 2005, Franciosi et al., 2005).

12. Government and the pharmaceutical industry 

For maximization of the public good with regards to the treatment of schizophrenia, the 9th leading cause of medical disability in the world, governments bear the responsibility for promoting the best possible treatment of this devastating illness. Towards the goal of optimal pharmacotherapy, governments oversee the drug registration process, ensure broad public access to necessary treatments, review drug labeling, and promote dissemination of good quality information about treatments. Although governments around the world are beginning to recognize the enormous individual and public costs of schizophrenia, they lag in efforts at reducing the stigma (Schomerus et al., 2006) and addressing other barriers to optimal recovery. Although existing treatments for schizophrenia are not completely satisfactory, they are fairly effective and can enable affected individuals to lead more productive and satisfying lives. SGAs, by virtue of their equivalent efficacy for positive symptoms with a lower propensity to cause EPS than FGAs, are generally preferred antipsychotic agents for many patients but their moderately greater value needs to be considered in the context of the cost differential (Patel et al., 2007). Current pharmaco-economic studies in this regard are, however, of limited value at this time (Hansen et al., 2006, Polsky et al., 2006). Just as relative prescribing costs of antipsychotic agents differ substantially across countries, mental health delivery systems vary as well. Given the substantial individual variation in antipsychotic response, it is useful to have a broad range of antipsychotic treatment options available in all settings. Each system should ensure effective and efficient utilization of all treatment and rehabilitative options to maximally promote individual recovery among the largest number of people receiving mental health services in the context of its resource constraints. Reimbursement strategies should also encourage psychosocial and disease management best practices that can substantially improve outcomes in schizophrenia.

Stronger initiatives to promote development of more effective treatments are necessary. Through mandates and incentives, government also needs to ensure the timely conduction of high quality clinical trials that address important questions of clinical practice. As vital questions remain unanswered, increased funding for relevant clinical trials is essential (Stroup et al., 2006b, Honer et al., 2007). A more effective pharmaco-epidemiological surveillance system needs to be developed in countries across the world and mechanisms to efficiently share data should be strengthened. Robust efforts to promote best-possible treatment need to be made and better incentives to study critical clinical questions provided. Additionally, government should monitor dissemination of accurate information about available treatments.

Pharmaceutical companies should be commended for accepting mandatory registration of all their clinical trials; they should now consistently comply with this registration process (Zarin et al., 2005, Kane et al., 2007). Comprehensive data from all clinical trials should be made available in a timely manner. The industry should be encouraged to develop more effective treatments for schizophrenia. The high costs and high risks associated with drug development (particularly for a broadly characterized entity such as schizophrenia) are disincentives to the development of truly innovative treatments for schizophrenia- but that is what is clearly needed! Existing treatments also can be better utilized and pharmaceutical companies need to be encouraged and/or required to more fully characterize their products and describe strategies for their optimal use. For example, it has been clear that original dosing strategies recommended for virtually all SGAs were suboptimal- risperidone being overdosed and olanzapine, quetiapine, and ziprasidone being underdosed. Yet the manufacturers of these products have not conducted comprehensive studies to better elucidate proper dosing or initiated serious efforts to get their product labels suitably changed. Similarly, no RCTs comparing effectiveness of long-acting risperidone microspheres to long-acting injectable FGAs have been initiated despite the importance of this topic. Information relevant to many areas of antipsychotic practice in schizophrenia is inadequate; high-quality trials addressing questions of clinical import need to be conducted in a timely manner.

13. In conclusion 

Different antipsychotic agents pose different challenges in terms of balancing efficacy and safety-tolerability and there is considerable individual variability in antipsychotic response and vulnerability to different adverse effects. Antipsychotic treatment needs to be individually tailored to promote optimal recovery and this requires careful monitoring and ongoing joint decision-making by the clinician-patient team about choice of antipsychotic agent, dosing, continuation/switching, and augmentation. Although existing antipsychotic treatments for schizophrenia are not completely satisfactory, they can meaningfully reduce the devastating effects of the illness. Even as we try to develop better treatments, we need to better characterize and compare existing treatments. A balanced understanding of our extensive data repository on treatments for schizophrenia is essential and effective mechanisms for fair dissemination of this information are necessary. A meticulous application of this understanding can reduce the significant gap between what we know and how we provide antipsychotic pharmacotherapy for persons with schizophrenia. Collectively, we can do much to better utilize existing treatments to optimize individual outcomes and reduce the considerable morbidity and mortality associated with the illness.

Role of the funding source 

The analysis was independently conducted and the manuscript independently prepared by the members of the WPA Section of Pharmacopsychiatry. There was no external funding or support for the development of this statement.

Contributors 

Rajiv Tandon, M.D.,1 Hans-Juergen Moeller, M.D., Ph.D.,2 R.H. Belmaker, M.D.,3 Wagner, F. Gattaz, M.D., Ph.D.,4 Juan J. Lopez-Ibor, Jr., M.D., Ph.D.,5 Ahmed Okasha, M.D., Ph.D.,6 Bruce Singh, M.B.B.S., Ph.D.,7 Dan J. Stein, M.D., Ph.D.8 Jean-Pierre Olie, M.D., Ph.D.,9 and Wolfang Fleischhacker, M.D.,10 for the Section of Pharmacopsychiatry, World Psychiatry Association

1. Chief of Psychiatry, State of Florida Program of Mental Health, Tallahassee, Florida, USA.

2. Professor and Chair, Psychiatry Department, University of Munich, Munich, Germany.

3. Professor and Chair, Department of Psychiatry, Ben Gurion University of the Negev, Beersheva, Israel.

4. Professor and Chair, Institute of Psychiatry, University of Sao Paulo, Sao Paulo, Brazil.

5. Professor and Chair, Institute of Psychiatry, Complutense University, Madrid, Spain.

6. Professor and Chair, Institute of Psychiatry, Ain Shams University, Cairo, Egypt.

7. Cato Professor and Chair, Department of Psychiatry, University of Melbourne, Melbourne, Australia.

8. Professor and Chair, Department of Psychiatry, University of Cape Town, Cape Town, South Africa.

9. Professor and Chair, Department of Psychiatry, Centre-Hospitale Saint-Anne, Paris, France.

10. Professor of Psychiatry and Vice-Chair, Department of Psychiatry, University of Innsbruck, Innsbruck, Austria.

Conflict of interest 

Disclosure statement

This statement was independently developed by the Section of Pharmacopsychiatry (Chair: Hans-Juergen Moeller) of the World Psychiatric Association. While fully endorsed by the Section, the statement does not purport to be an official document of the Association. Whereas the 41 members of the Section have varying degrees of potential conflicts of interest, relevant disclosure of the ten authors is provided herein. Drs. Belmaker, Lopez-Ibor, Okasha, and Tandon report no current conflicts relevant to this article. Within the previous five years, however, Dr. Tandon has been a consultant for and received speakership honoraria supported by AstraZeneca, Bristol Myers-Squibb, Eli Lilly, Janssen, and Pfizer. The content of the article is not part of the purview of Dr. Tandon's current employment by the State of Florida which bears no responsibility its contents. Dr Moeller is a consultant for and on the speakership bureaus of AstraZeneca, Bristol-Myers Squibb, Eisai, Eli Lilly, GlaxoSmithKline, Janssen, Lundbeck, Organon, Pfizer, Sanofi-Aventis, Sepracor, and Wyeth. Dr. Gattaz has received research support and speakership honoraria from AstraZeneca, Bristol Myers-Squibb, Eli Lilly, and Janssen-Cilag. Dr. Olie is a consultant for Eli Lilly, Janssen, Pierre Fabre, Servier, and Sanofi-Aventis. Dr Singh is a consultant for and has received research support from AstraZeneca, Bristol-Myers Squibb, Eli Lilly, and Janssen-Cilag. Dr. Stein has received research grants and/or speakership honoraria from AstraZeneca, Eli Lilly, GlaxoSmithKline, Johnson and Johnson, Lundbeck, Orion, Pfizer, Pharmacia, Roche, Servier, Solvay, Sumitomo, Tikvah, and Wyeth. Dr. Fleischhacker has received research grants from, is a consultant for and on the speakership bureaus of AstraZeneca, Bristol-Myers Squibb, Eli Lilly, Janssen Cilag, Otsuka, Pfizer, and Wyeth.

Acknowledgement 

Members of the WPA Section on Pharmacopsychiatry:

A. Carlo Altamura, David S. Baldwin, David Baron, Michael Bauer, R.H. Belmaker, Pierre Blier, Patrice Boyer, William E. Bunney, Graham Burrows, Wolfang Fleischhacker, Daniel Flores, Wagner F. Gattaz, Guy Goodwin, Gerardo Heinze, Ian Hindmarch, Hans Hippius, Cyril Hoschl, Siegfried Kasper, Per Kragh-Sorensen, Juan J. Lopez-Ibor, Ulrik Malt, Bruno Millet, Sung Kil Min, Hans-Juergen Moeller (Chair), Jaime Monti, Bruno Muller-Oerlinghausen, Franz Muller-Spahn, David J. Nutt, Ahmed Okasha, Jean Pierre Olie, Eugene S. Paykel, Giorgio Racagni, Perry Renshaw, Raben Rosenberg, Bernd Saletu, Bruce Singh, Dan J. Stein, Rajiv Tandon, Marcio Versiani, Eduard Vieta, and Joseph Zohar.