Home > Pharmacological interventions in the...

PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Mental Health (UK). Schizophrenia: Core Interventions in the Treatment and Management of Schizophrenia in Primary and Secondary Care (Update) [Internet]. Leicester (UK): British Psychological Society; 2009 Mar. (NICE Clinical Guidelines, No. 82.)

  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

6Pharmacological interventions in the treatment and management of schizophrenia

For the guideline update, all sections of the previous pharmacology chapter were updated, except the section on rapid tranquillisation, which was removed because it was updated by the NICE clinical guideline Violence: the short-term management of disturbed/violent behaviour in in-patient psychiatric settings and emergency departments7. The scope for the update also included updating the NICE technology appraisal (TA43) on the use of newer (atypical) antipsychotic drugs (NICE, 2002). In section 6.9.2 of this chapter, new evidence is presented from economic modelling of pharmacological relapse prevention (the rationale for economic modelling, the methodology adopted, the results and the conclusions from this economic analysis are described in detail in chapter 7). For the guideline update, we use the term ‘first-generation antipsychotics’ or FGAs to refer to drugs that were called in the previous NICE guidance, ‘conventional’ or ‘typical’ antipsychotics. Likewise, we use the term ‘second-generation antipsychotics’ or SGAs to refer to drugs that were previously called ‘atypical’ antipsychotics. This terminology is used here because it is widely used in the literature, it should not be taken to suggest that FGAs and SGAs represent distinct classes of antipsychotics (see section 6.4.1 for further discussion of this issue.

For this chapter, the review of evidence is divided into the following areas:

  • Initial treatment with oral antipsychotic medication (section 6.2)
  • Oral antipsychotics in the treatment of the acute episode (section 6.3)
  • Promoting recovery in people with schizophrenia that is in remission – pharmacological relapse prevention (section 6.4)
  • Promoting recovery in people with schizophrenia whose illness has not responded adequately to treatment (section 6.5)
  • Combining antipsychotic medication with another antipsychotic (section 6.5.10)
  • Treatment with depot/long-acting injectable antipsychotic medication (section 6.6)
  • Side-effects of antipsychotic medication, focusing on metabolic and neurologic adverse events, as these were considered a priority by the GDG and were also highlighted as areas of concern by service users (section 6.7)
  • Effectiveness of antipsychotic medication (section 6.8)
  • Health economics (section 6.9).

Due to the nature of the evidence, all recommendations can be found in section 6.11 at the end of the chapter (rather than after each sub-section), preceded by section 6.10 (from evidence to recommendations) that draws together the clinical and health economic evidence and provides a rationale for the recommendations.

6.1. Introduction

Antipsychotic drugs have been the mainstay of treatment of schizophrenia since the 1950s. Initially used for the treatment of acute psychotic states, their subsequent use to prevent relapse led to these drugs being prescribed for long-term maintenance treatment, either as oral preparations or in the form of long-acting injectable preparations (‘depots’).

Although a number of different classes of drugs have antipsychotic activity, the primary pharmacological action of antipsychotic drugs is their antagonistic effect on the D2 dopamine receptors. Indeed, the potency of a drug’s antipsychotic effect is, at least in part, determined by its affinity for the D2 receptor (Agid et al., 2007; Kapur & Remington, 2001; Snyder et al., 1974), an association that informed the dopamine hypothesis of schizophrenia. It is worth noting, however, that antipsychotic drugs are also of use in the treatment of other psychotic disorders, their dopamine-blocking activity probably again central to their pharmacological efficacy.

Uses of antipsychotics

In the treatment and management of schizophrenia, antipsychotics are currently used for the treatment of acute episodes, for relapse prevention, for the emergency treatment of acute behavioural disturbance (rapid tranquillisation) and for symptom reduction. They are available as oral, intramuscular (IM) and intravenous (IV) preparations, or as medium- or long-acting depot IM preparations. In the UK, clozapine is only licensed for use in people with ‘treatment-resistent’ schizophrenia, defined by the manufacturers’ Summary of Product Characteristics (SPC) as a “lack of satisfactory clinical improvement despite the use of adequate doses of at least two different antipsychotic agents, including an atypical antipsychotic agent, prescribed for adequate duration”.

Antipsychotics are usually prescribed within the recommended SPC dosage range, and there is little evidence to support the use of higher dosage or combination with another antipsychotic if monotherapy proves to be ineffective (Royal College of Psychiatrists, 2006; Stahl, 2004). Antipsychotics are also used in combination with a range of other classes of drugs, such as anticonvulsants, mood stabilisers, anticholinergics, antidepressants and benzodiazepines. Clinicians may augment antipsychotics with such drugs for several reasons:

  • where there is a lack of effective response to antipsychotics alone
  • for behavioural control
  • for the treatment of the side-effects of antipsychotics
  • for the treatment of comorbid or secondary pschiatric problems, such as depression and anxiety.

Although such augmentation strategies are commonly used in clinical practice, they are outside the scope of this guideline. It is anticipated that a future guideline will address the evidence base for these interventions.

Antipsychotic dose

The current British National Formulary (BNF) is the most widely used reference for the prescription of medicines and the pharmacy industry within the UK, and a complete SPC for all the drugs referred to in this guideline can be found in the Electronic Medicines Compendium (http://emc.medicines.org.uk/). The recommended dose ranges listed in the BNF normally echo the information contained in the manufacturers’ SPC, as well as advice from an external panel of experts to ensure that the SPC recommendations on issues such as dose range reflect current good practice (‘standard dosing’). ‘Standard doses’ are identified as doses that fall within the range likely to achieve the best balance between therapeutic gain and dose-related adverse effects. However, with up to a third of people with schizophrenia showing a poor response to antipsychotic medication, there has been a tendency for higher doses to be prescribed: surveys of prescribing practice suggest that doses of antipsychotics exceeding BNF limits, either for a single drug or through combining antipsychotics, continue to be commonly used (Harrington et al., 2002; Lehman et al., 1998; Paton et al., 2008).

In an attempt to increase the rate or extent of response, ‘loading doses’ and rapid dose escalation strategies have been employed (Kane & Marder, 1993); studies have failed to show any advantage for such a strategy in terms of speed or degree of treatment response (Dixon et al., 1995). The Schizophrenia Patient Outcomes Research Team (1998) concluded that in the treatment of acute episodes of schizophrenia, ”massive loading doses of antipsychotic medication, referred to as “rapid neuroleptization,” should not be used”.

Evidence suggests that drug-naïve patients and those experiencing their first-episode of schizophrenia respond to doses of antipsychotic drugs at the lower end of the recommended dosage range (Cookson et al., 2002; McEvoy et al., 1991; Oosthuizen et al., 2001; Remington et al., 1998; Tauscher & Kapur, 2001).

Relapse prevention

For people with established schizophrenia, the chance of relapse when receiving continuous antipsychotic medication appears to be about a third of that on placebo (Marder & Wirshing, 2003). Risk factors for relapse of illness include the presence of persistent symptoms, poor adherence to the treatment regimen, lack of insight, and substance use, all of which can be reasonable targets for intervention.

Stopping antipsychotic medication in people with schizophrenia, especially abruptly, dramatically increases the risk of relapse in the short to medium term, although even with gradual cessation about half will relapse in the succeeding 6 months (Viguera et al., 1997). Low-dose prescribing and the use of intermittent dosing strategies (with medication prompted by the appearance of an individual’s characteristic early signs of relapse) have also been suggested in the past as ways to minimise side effects long-term. However, when these were tested in controlled trials, the risks, particularly in terms of increased relapse, outweighed any benefits (Dixon et al., 1995; Hirsch & Barnes, 1995).

The Schizophrenia Patient Outcomes Research Team (1998) concluded that, “Targeted, intermittent dosage maintenance strategies should not be used routinely in lieu of continuous dosage regimens because of the increased risk of symptom worsening or relapse. These strategies may be considered for patients who refuse maintenance or for whom some other contraindication to maintenance therapy exists, such as side-effect sensitivity”.

Clozapine

The antipsychotic clozapine was introduced in the 1970s, only to be withdrawn soon after because of the risk of potentially fatal agranulocytosis. However, after further research revealed the drug’s efficacy in treatment-resistant schizophrenia (for example, Kane et al., 1988), clozapine was reintroduced in the 1980s with requirements for appropriate haematological monitoring. Clozapine was considered to have a novel mode of action. Its pharmacological profile includes a relatively low affinity for D2 receptors and a much higher affinity for D4 dopamine receptors and for subtypes of serotonin receptors, although it is not clear exactly which aspects are responsible for its superior antipsychotic effect in treatment-resistant schizophrenia.

Side effects

Clinical issues relating to side-effects were summarised by NICE (2002), as follows: ‘All antipsychotic agents are associated with side-effects but the profile and clinical significance of these varies among individuals and drugs. These may include EPS (such as parkinsonism, acute dystonic reactions, akathisia and tardive dyskinesia), autonomic effects (such as blurring of vision, increased intra-ocular pressure, dry mouth and eyes, constipation and urinary retention), increased prolactin levels, seizures, sedation and weight gain. Cardiac safety is also an issue because several antipsychotics have been shown to prolong ventricular repolarisation, which is associated with an increased risk of ventricular arrhythmias. Routine monitoring is a pre-requisite of clozapine use because of the risk of neutropenia and agranulocytosis. Prescribers are therefore required to ensure that effective ongoing monitoring is maintained as alternative brands of clozapine become available.

Individuals with schizophrenia consider the most troublesome side-effects to be EPS, weight gain, sexual dysfunction and sedation. EPS are easily recognised, but their occurrence cannot be predicted accurately and they are related to poor prognosis. Akathisia is also often missed or misdiagnosed as agitation. Of particular concern is tardive dyskinesia (orofacial and trunk movements), which may not be evident immediately, is resistant to treatment, may be persistent, and may worsen on treatment withdrawal. Sexual dysfunction can be a problem, sometimes linked to drug-induced hyperprolactinaemia; it is likely to be an underreported side-effect of antipsychotic treatment, as discussion of this issue is often difficult to initiate.’

Blockade of D2 receptors by antipsychotic drugs is responsible for EPS such as parkinsonism, akathisia, dystonia and dyskinesia, but the therapeutic, antipsychotic effect may occur at a lower level of D2 receptor occupancy than the level associated with the emergence of EPS (Farde et al., 1992). Second-generation antipsychotic drugs (SGAs) were introduced with claims for a lower risk of EPS. The individual SGAs differ in their propensity to cause EPS: for some SGAs (for example, clozapine and quetiapine), acute EPS liability does not differ from placebo across their full dose, while for some others the risk is dose dependent. These differences may reflect individual drug profiles in relation to properties such as selective dopamine D2-like receptor antagonism, potent 5-HT2A antagonism, and rapid dissociation from the D2 receptor, and for aripiprazole, partial agonism at D2 and 5HT1A receptors. Interpretation of the RCT evidence for the superiority of SGAs in respect of acute EPS should take into account the dosage and choice of first-generation antipsychotic (FGA) comparator, most commonly haloperidol which is considered a high potency D2 antagonist with a relatively high liability for EPS.

Raised serum prolactin is also an important adverse effect of antipsychotic medication (Haddad & Wieck, 2004). It can lead to problems such as menstrual abnormalities, galactorrhea and sexual dysfunction, and in the longer term to reduced bone mineral density (Haddad & Wieck, 2004, Meaney et al., 2004). While the propensity for antipsychotic drugs to affect prolactin varies between agents, the extent to which an individual service user will be affected may be difficult to determine in advance of treatment.

Antipsychotic drugs also have strong affinity for a range of other receptors, including histaminergic, serotonergic, cholinergic and alpha-adrenergic types, which may produce a number of other effects such as sedation, weight gain and postural hypotension. As the various antipsychotic drugs possess different relative affinities for each receptor type, each drug will have its own specific profile of side-effects. For example, antipsychotic drugs vary in their liability for metabolic side effects, such as weight gain, lipid abnormalities and disturbance of glucose regulation. These are side effects that have been increasingly recognised as problems that may impact on long-term physical health. Specifically, they increase the risk of the metabolic syndrome, a recognised cluster of features (hypertension, central obesity, glucose intolerance/insulin resistance and dyslipidaemia) (American Diabetes Association et al., 2004; Mackin et al., 2007), that is a predictor of type-2 diabetes and coronary heart disease. Even without antipsychotic treatment, people with schizophrenia may have an increased risk of such problems, which is partly related to lifestyle factors such as smoking, poor diet, lack of exercise, and also, possibly, the illness itself (Brown et al., 1999; Holt et al., 2005; Osborn et al., 2007a, 2007b; Taylor et al., 2005; van Nimwegen et al., 2008). While there is some uncertainty about the precise relationship between schizophrenia, metabolic problems and antipsychotic medication, there is agreement that routine physical health screening of people prescribed antipsychotic drugs long-term is required (Barnes et al., 2007; Newcomer, 2007; Suvisaari et al., 2007) (further information about physical health screening can be found in Chapter 9).

6.2. Initial treatment with antipsychotic medication

6.2.1. Introduction

Evidence published before the previous guideline suggests that drug-naïve patients may respond to doses of antipsychotic medication at the lower end of the recommended range (Cookson et al., 2002; McEvoy et al., 1991; Oosthuizen et al., 2001; Tauscher & Kapur, 2001). This may have particular implications in the treatment of people experiencing their first episode of schizophrenia. Lehman et al. (1998) have suggested that the maximum dose for drug-naïve patients should be 500 mg chlorpromazine equivalents per day. This contrasts with a recommended optimal oral antipsychotic dose of 300–1000 mg chlorpromazine equivalents per day for the routine treatment of an acute episode in non-drug-naïve patients.

6.2.2. Clinical review protocol (initial treatment with antipsychotic medication)

The review protocol, including the primary clinical question, information about the databases searched and the eligibility criteria can be found in Table 10. For the guideline update, a new systematic search was conducted for relevant RCTs published since the previous guideline (further information about the search strategy can be found in Appendix 8).

Table 10. Clinical review protocol for the review of initial treatment with antipsychotic medication.

Table 10

Clinical review protocol for the review of initial treatment with antipsychotic medication.

6.2.3. Studies considered for review8

Nine RCTs (N = 1,801) met the inclusion criteria for the update. Of these, two trials (Emsley1995; Jones1998) were included in the previous guideline, but analysed with the acute treatment trials (that is, non-initial treatment). All included studies are now published in peer-reviewed journals between 1999 and 2008. Further information about both included and excluded studies can be found in Appendix 15.

6.2.4. Antipsychotic drug treatment in people with first-episode or early schizophrenia

Of the nine RCTs included in the meta-analysis, two were multiple-arm trials and, therefore, there were a total of 12 evaluations: three of olanzapine versus haloperidol, one of olanzapine versus quetiapine, three of olanzapine versus risperidone, four of risperidone versus haloperidol, and one of risperidone versus quetiapine (see Table 11 for a summary of the study characteristics). An evidence summary table for each comparison can be found in Chapter 10.

Table 11. Summary of study characteristics for RCTs of antipsychotic drugs in people with first-episode or early schizophrenia.

Table 11

Summary of study characteristics for RCTs of antipsychotic drugs in people with first-episode or early schizophrenia.

6.2.5. Clinical evidence summary

In nine RCTs with a total of 1,801 participants with first-episode or early schizophrenia (including people with a recent onset of schizophrenia and people who have never been treated with antipsychotic medication), the evidence suggested there were no clinically significant differences in efficacy between the antipsychotic drugs examined. Most of the trials were not designed to examine differences in adverse effects of treatment, but metabolic and neurological side effects reported were consistent with those identified in the SPC for each drug.

6.3. Oral antipsychotics in the treatment of the acute episode

6.3.1. Introduction

Early clinical studies established that antipsychotic medications are effective in the treatment of acute schizophrenic episodes (Davis & Garver, 1978), although they proved to be more effective at alleviating positive symptoms than negative symptoms such as alogia or affective blunting. However, no consistent difference between the FGAs was demonstrated in terms of antipsychotic efficacy or effects on individual symptoms, syndromes or schizophrenia subgroups. Accordingly, the choice of drug for an individual was largely dependent on differences in side-effect profiles (Hollister, 1974; Davis & Garver, 1978). The limitations of these first-generation antipsychotic drugs included heterogeneity of response in acute episodes, with a proportion of individuals showing little improvement (Kane, 1987), and a range of undesirable acute and long-term side -effects. The search for better-tolerated and more effective drugs eventually generated a series of second-generation drugs, characterised by a lower liability for EPS (Barnes & McPhillips, 1999; Geddes et al., 2000; Cookson et al., 2002).

6.3.2. Clinical review protocol (acute treatment)

The review protocol, including the primary clinical question, information about the databases searched and the eligibility criteria can be found in Table 12. A new systematic search for relevant RCTs, published since the previous guideline, was conducted for the guideline update (further information about the search strategy can be found in Appendix 8).

Table 12. Clinical review protocol for the review of oral antipsychotics in the treatment of the acute episode.

Table 12

Clinical review protocol for the review of oral antipsychotics in the treatment of the acute episode.

6.3.3. Studies considered for review

In the previous guideline, 180 RCTs were included.9 The update search identified 10 papers providing follow-up or published data for existing trials, and 19 new trials. Two trials (Klieser1996; Malyarov1999) were multi-arm and contributed to more than one comparison. Due to the large volume of evidence, we excluded from the update, open-label studies, head -to-head comparisons of two FGAs, and comparisons with placebo, leaving 72 RCTs (N = 16,556) that met inclusion criteria. Further information about both included and excluded studies can be found in Appendix 15.

6.3.4. Treatment with antipsychotic drugs in people with an acute exacerbation or recurrence of schizophrenia

Because most included studies involved olanzapine or risperidone; we first report comparisons involving these drugs, then comparisons involving other drugs. Twenty-six RCTs compared olanzapine with another antipsychotic (see Table 13 for a summary of the study characteristics), 29 compared risperidone with another antipsychotic (see Table 14). Six RCTs were included in the analysis comparing amisulpride with a FGA, two in the analysis comparing aripiprazole with a FGA, and one compared aripiprazole with ziprasidone (see Table 15), seven compared quetiapine with a FGA and two compared sertindole with a FGA (see Table 16), and seven compared zotepine with a FGA (see Table 17). An evidence summary table for each comparison can be found in Chapter 10.

Table 13. Summary of study characteristics for olanzapine versus another antipsychotic drug (acute treatment).

Table 13

Summary of study characteristics for olanzapine versus another antipsychotic drug (acute treatment).

Table 14. Summary of study characteristics for risperidone versus another antipsychotic drug (acute treatment).

Table 14

Summary of study characteristics for risperidone versus another antipsychotic drug (acute treatment).

Table 15. Summary of study characteristics for amisulpride or aripiprazole versus another antipsychotic drug (acute treatment).

Table 15

Summary of study characteristics for amisulpride or aripiprazole versus another antipsychotic drug (acute treatment).

Table 16. Summary of study characteristics for quetiapine or sertindole versus a FGA (acute treatment).

Table 16

Summary of study characteristics for quetiapine or sertindole versus a FGA (acute treatment).

Table 17. Summary of study characteristics for zotepine versus a FGA (acute treatment).

Table 17

Summary of study characteristics for zotepine versus a FGA (acute treatment).

6.3.5. Clinical evidence summary

In 72 RCTs involving 16,556 participants with an acute exacerbation or recurrence of schizophrenia, there was little evidence of clinically significant differences in efficacy between the oral antipsychotic drugs examined. Metabolic and neurological side effects were consistent with those reported in the SPC for each drug.

6.4. Promoting recovery in people with schizophrenia that is in remission – pharmacological relapse prevention

6.4.1. Introduction

Following their introduction into clinical practice in the early 1950s, chlorpromazine and related drugs rapidly became widely used for both acute treatment of people experiencing symptoms of psychosis and for prevention of relapse. Haloperidol, synthesised in 1959, became by the 1980s the most widely used drug for these purposes in the USA (Davis et al., 1993; Gilbert et al., 1995; Hirsch & Barnes, 1995, Healy, 2002). A meta-analysis (Davis et al., 1993) of 35 double-blind studies compared maintenance treatment using FGAs with placebo in over 3,500 service users. Relapse was reported in 55% of those who were randomised to receive placebo, but in only 21% of those receiving active drugs. Gilbert and colleagues (1995) reviewed 66 antipsychotic withdrawal studies, published between 1958 and 1993, and involving over 4,000 service users. The mean cumulative rate of relapse in the medication withdrawal groups was 53% (follow-up period 6–10 months) compared with 16% (follow-up 8 months) in the antipsychotic maintenance g roups. Over a period of several years, continuing treatment with conventional antipsychotics appears to reduce the risk of relapse by about two-thirds (Kissling, 1991).

When the effects of stopping antipsychotic drugs after an acute psychotic episode or after long-term maintenance treatment were examined, the subsequent rate of relapse seemed to be similar in both situations. Individuals who are well stabilised on maintenance medication show high rates of relapse when their antipsychotic therapy is discontinued (Kane, 1990) or switched to placebo (Hogarty et al., 1976). A recent Cochrane review (Alkhateeb et al., 2007) including 10 trials of chlorpromazine cessation in stable participants (Total N = 1,042) showed that those stopping chlorpromazine had a relative risk of relapse in the short term (up to eight weeks) of 6.76 (95% CI 3.37to 13.54) and in the medium-term (9 weeks to six months) of 4.04 (95% CI 2.81 to 5.8). Relative risk of relapse after six months was 1.70 (95% CI 1.44 to 2.01). Another meta-analysis of data from several large collaborative studies (Davis et al., 1993) suggested that the number of people who survive without relapse after discontinuing drug treatment declines exponentially by around 10% a month.

Whether maintenance drug treatment is required for all people with schizophrenia is uncertain. Around 20% of individuals will only experience a single episode (Möller & van Zerssen, 1995). A recent pragmatic observational study analysing over 4,000 participants who achieved remission in the Schizophrenia Outpatient Health Outcomes study, showed that 25% relapsed over a three-year follow-up period with a constant rate of relapse over this time (Haro et al., 2007). It appears therefore that a proportion of people will experience a relapse despite continued antipsychotic drug treatment. It is unclear whether such people benefit from an increase in antipsychotic dosage during episodes of psychotic exacerbation (Steingard et al., 1994).

Given that there are no consistent, reliable predictors of prognosis or drug response, the previous schizophrenia guideline, as well as other consensus statements and guidelines, generally recommend that pharmacological relapse prevention is considered for every patient diagnosed with schizophrenia (for example, Dixon et al., 1995; Lehman et al., 1998). Possible exceptions are people with very brief psychotic episodes without negative psychosocial consequences, and the uncommon patient for whom all available antipsychotics pose a significant health risk (Fleischhacker & Hummer, 1997).

It is clear from the placebo-controlled RCTs and discontinuation studies cited above that the efficacy of antipsychotics in relapse prevention is established. However, it is also clear from recent pragmatic trials that switching of medication over time is common in clinical practice (Jones et al., 2006; Lieberman et al., 2005). In the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) (Lieberman et al., 2005) study, 74% of participants discontinued their randomised treatment over eighteen months (further information about this trial can be found in section 6.8 on the effectiveness of antipsychotic medication). This may well reflect the need in clinical practice to search collaboratively for the drug that offers the best balance of efficacy and tolerability for the individual patient. The role of depot preparations in contributing to concordance and continuation on medication is discussed in section 6.6.

All the antipsychotics identified for review have established supremacy over placebo in the prevention of relapse, although the evidence that any individual antipsychotic drug, or group of antipsychotics (FGAs and SGAs), has greater efficacy or better tolerability than another is still very uncertain. One of the main aims of antipsychotic drug development in recent decades has been to produce compounds with equivalent antipsychotic efficacy, but without troubling EPS. The doses of haloperidol that came to be used in routine clinical practice by the 1980s and early 1990s were higher than those required for its antipsychotic effect, and EPS were common. The trials conducted in the 1990s comparing SGAs and haloperidol often tested the latter at relatively high doses, arguably above the optimum for at least a proportion of the subjects treated, and highlighted the propensity of haloperidol to cause such side effects in comparison with SGAs. The widespread introduction of SGAs to clinical practice from the mid 1990s onwards thus appeared to offer a genuine therapeutic advance. However, more recent effectiveness (pragmatic) trials have suggested that the claimed advantages of these drugs may have been overstated, especially if their propensity to cause metabolic abnormalities and other side effects is taken into account, and if they are compared with FGAs other than higher dose haloperidol (Geddes et al., 2000; Jones et al., 2006; Lieberman et al., 2005; NICE, 2002). Second-generation antipsychotic drugs are not a homogeneous class and may not deserve a group title. They differ widely in their pharmacology and side effect profile. There are unanswered questions regarding their relative efficacy and tolerability and their use over the long-term compared to FGAs. Their risks of long-term metabolic disturbance are not yet fully quantified and neither is the risk of movement disorders such as tardive dyskinesia compared to FGAs, so any small advantage that may be offered by reduced EPS may be offset by these other adverse consequences not shown by the earlier drugs.

Whilst evaluating each drug against each other would appear superficially the best way of approaching the question posited for this review, in reality the number of possible comparisons and the limited number of studies available would render this a meaningless task. Therefore, the GDG considered that comparing the individual SGAs against all FGA comparators, primarily in terms of relapse, provided the most meaningful analysis of the available data.

Definitions

The definitions of relapse used in this review were those adopted by the individual studies. This definition varied between studies (see sections 6.4.4 and 6.4.5), and therefore, caution should be exercised in the interpretation of the results.

6.4.2. Clinical review protocol (relapse prevention)

The review protocol, including the primary clinical question, information about the databases searched and the eligibility criteria used for this section of the guideline can be found in Table 18. A new systematic search for relevant RCTs, published since the previous guideline, was conducted for the guideline update (further information about the search strategy can be found in Appendix 8 and information about the search for health economic evidence can be found in section 6.9.1).

Table 18. Clinical review protocol for the review of relapse prevention.

Table 18

Clinical review protocol for the review of relapse prevention.

6.4.3. Studies considered for review

In the previous guideline, nine RCTs comparing a SGA with a FGA were included (based on an unpublished review by Leucht and colleagues). Since publication of the previous guideline, Leucht et al. (2003) published their review, including one addition trial and six trials comparing a SGA with placebo that were not included in the previous guideline. For the update, we limited our review to double-blind RCTs of antipsychotics used for relapse prevention, therefore, four studies (Daniel1998; Essock1996; Rosenheck1999; Tamminga1994) included in the previous guideline were excluded from the update. In addition, one trial of an SGA versus another SGA, included in the previous acute treatment review, met the criteria for inclusion in this review (Tran1997). The update search identified four additional RCTs (one comparing a SGA with a FGA, one comparing a SGA with a SGA, and one comparing a SGA with placebo). For the purposes of the health economic model (see section 6.9.2), we included trials of ziprasidone versus placebo because this drug has been compared with a licensed SGA.

In total, 17 RCTs (N = 3,535) met the inclusion criteria for the update. Of these, one was unpublished (STUDY-S029) and the remainder were published in peer-reviewed journals between 1994 and 2007. Further information about both included and excluded studies can be found in Appendix 15.

6.4.4. Second-generation antipsychotics versus placebo in people with schizophrenia that is in remission (relapse prevention)

Eight RCTs were included in the meta-analysis comparing a SGA (amisulpride, aripiprazole, olanzapine, paliperidone, ziprasidone, zotepine) with placebo (see Table 19). An evidence summary table for each comparison can be found in Chapter 10.

Table 19. Summary of study characteristics for RCTs of a SGA versus placebo (relapse prevention).

Table 19

Summary of study characteristics for RCTs of a SGA versus placebo (relapse prevention).

6.4.5. Second-generation antipsychotics versus another antipsychotic drug in people with schizophrenia that is in remission (relapse prevention)

Nine RCTs were included in the meta-analysis comparing a SGA (amisulpride, olanzapine, risperidone) with a FGA (haloperidol) (see Table 20), and two were included in the analysis comparing a SGA (olanzapine) with another SGA (risperidone, ziprasidone) (see Table 21). An evidence summary table for each comparison can be found in Chapter 10.

Table 20. Summary of study characteristics for RCTs of a SGA versus another antipsychotic drug (relapse prevention).

Table 20

Summary of study characteristics for RCTs of a SGA versus another antipsychotic drug (relapse prevention).

Table 21. Summary of study characteristics for RCTs of a SGA versus another SGA (relapse prevention).

Table 21

Summary of study characteristics for RCTs of a SGA versus another SGA (relapse prevention).

6.4.6. Clinical evidence summary

In 17 RCTs including 3,535 participants with schizophrenia, the evidence suggested that when compared to placebo, all of the antipsychotics examined reduced the risk of relapse or overall treatment failure. Although some SGAs show a modest benefit over haloperidol, there is insufficient evidence to choose between antipsychotics in terms of relapse prevention.

6.5. Promoting recovery in people with schizophrenia whose illness has not responded adequately to treatment

6.5.1. Introduction

The phrase ‘treatment-resistant’ is commonly used to to describe people with schizophrenia whose illness has not responded adequately to treatment. The essence of treatment resistance in schizophrenia is the presence of poor psychosocial and community functioning, which persists despite trials of medication that have been adequate in terms of dose, duration and adherence. While treatment resistance is sometimes conceptualised in terms of enduring positive psychotic symptoms, other features of schizophrenia can contribute to poor psychosocial and community functioning, including negative symptoms, affective symptoms, medication side effects, cognitive deficits and disturbed behaviour. Treatment resistance in schizophrenia is relatively common, in that between a fifth and a third of service users show a disappointing response to adequate trials of antipsychotic medication (Brenner et al., 1990; Lieberman et al., 1992; Conley & Buchanan, 1997). In a small proportion of people experiencing their first episode of schizophrenia the illness will be resistant to antipsychotic medication, showing only a limited response, for example, precluding early discharge from hospital (May, 1968; MacMillan et al., 1986; Lieberman et al., 1989, 1992, Lambert et al., 2008), but more commonly the illness become progressively more unresponsive to medication over time (Lieberman et al., 1993; Wiersma et al., 1998).

The definition of the term ‘treatment-resistant schizophrenia’ varies considerably in the studies covered in this review. Kane et al. (1988) introduced rigorous criteria involving aspects of the clinical history, cross-sectional measures and prospective assessments One trend has been a move towards broader definitions of treatment resistance that allow a larger number of individuals to be viewed as clinically eligible for treatment with clozapine. For example, Bondolfi and colleagues (1998) included in their trial people with chronic schizophrenia who “had previously failed to respond to or were intolerant of at least two different classes of antipsychotic drugs given in appropriate doses for at least 4 weeks each”. Others have adopted an even wider clinical notion of ‘incomplete recovery’(Pantelis & Lambert, 2003), which acknowledges the presence of lasting disability in functional and psychosocial aspects despite psychotherapeutic and psychosocial as well as pharmacological interventions while also recognising the potential for improvement.

6.5.2. Treatment-resistant schizophrenia and antipsychotic medication

High dosage antipsychotic medication is commonly used for treatment-resistant schizophrenia, although there is little evidence to suggest any significant benefit with such a strategy (Royal College of Psychiatrists, 2006). Clinicians may also try switching to another antipsychotic, although similarly the research evidence on the possible value of such a strategy is not consistent or promising (Kinon et al., 1993; Lindenmayer et al., 2002; Shalev et al., 1993). An alternative strategy has been to try to potentiate antipsychotics by combining them, either with each other (see section 6.5.3), or with other classes of drugs. Possible adjuncts to antipsychotic treatment include mood stabiliser and anticonvulsant drugs such as lithium, carbamazepine, sodium valproate, lamotrigine, antidepressants and benzodiazepines (Barnes et al., 2003; Chong & Remington, 2000; Durson & Deakin, 2001). The use of such adjunctive treatments to augment the action of antipsychotics is beyond the scope of this guideline.

Kane and colleagues (1988, 2001) established the efficacy of clozapine over FGAs in strictly-defined TRS, and subsequent meta-analyses have confirmed the superiority of clozapine in terms of reducing symptoms and the risk of relapse (Chakos et al., 2001; Wahlbeck et al., 1999). However, Chakos and colleagues concluded from their meta-analysis that the evidence for clozapine when compared to the SGAs tested was inconclusive. Even with optimum clozapine treatment, the evidence suggests that only 30–60% of TRS will show a satisfactory response (Iqbal et al., 2003). As clozapine is associated with severe and potentially life-threatening side-effects, particularly the risk of agranulocytosis, the SPC states that drug should only be considered where there has been a lack of satisfactory clinical improvement despite adequate trials, in dosage and duration, of at least two different antipsychotic agents, including a SGA.

Monitoring plasma clozapine concentration may be helpful in establishing the optimum dose of clozapine in terms of risk–benefit ratio, and also in assessing adherence (Gaertner et al., 2001; Llorca et al., 2002; Rostami-Hodjegan et al., 2004), particularly for service users showing a poor therapeutic response or experiencing significant side-effects despite appropriate dosage. An adequate trial will involve titrating the dosage to achieve a target plasma level, usually considered to be above 350μg/l, although response may be seen at lower levels (Dettling et al., 2000; Rostami-Hodjegan et al., 2004). If the response to clozapine monotherapy is poor, augmentation strategies may be considered (see section 6.5.3 for a review of the evidence).

A number of patient-related factors have been reported to increase the variability of plasma clozapine concentrations, with gender, age and smoking behaviour being the most important (Rostami-Hodjegan et al. 2004). Smoking is thought to increase the metabolism of clozapine by inducing the cytochrome P450 1A2 (CYP1A2) and other hepatic enzymes (Flanagan, 2006; Ozdemir et al., 2002). The metabolism of clozapine is mainly dependent on CYP1A2. This has several clinical implications. First, there is some evidence that smokers are prescribed higher doses by clinicians to compensate for higher clozapine clearance (Tang et al., 2007). Secondly, plasma concentrations of clozapine and its active metabolite, norclozapine, vary considerably at a given dosage, and this variation may be greater in heavy smokers receiving lower doses of clozapine, increasing the risk of sub-therapeutic concentrations (Diaz et al., 2005). Thirdly, prompt adjustment of clozapine dosage in patients who stop smoking during treatment is important, to avoid the substantially elevated clozapine concentrations and increased risk of toxicity that would otherwise be expected (Flanagan, 2006; McCarthy, 1994; Zullino et al., 2002).

6.5.3. Combining antipsychotic drugs

In clinical practice, the prescription of combined antipsychotics is relatively common. A multi-centre audit of the prescription of antipsychotic drugs for inpatients in 47 mental health services in the UK, involving over 3,000 inpatients, found that nearly half were receiving more than one antipsychotic drug (Harrington et al., 2002). Similarly, prescription surveys in the UK by Taylor et al. (2000 (2002) and the Prescribing Observatory for Mental Health (Paton et al., 2008) have confirmed a relatively high prevalence of combined antipsychotics for people with schizophrenia, including co-prescription of first- and second-generation antipsychotics.

The reasons for such prescriptions include as required (‘p.r.n.’) medication, a gradual switch from one antipsychotic drug to another, and adding an oral antipsychotic to depot treatment to stabilise illness. A common rationale for combining antipsychotics is to achieve a greater therapeutic response when there has been an unsatisfactory response to a single antipsychotic. In this respect, there is little supportive evidence for superior efficacy (Chan & Sweeting, 2007; Chong & Remington, 2000), and Kreyenbuhl et al. (2007) reported that psychiatrists perceive antipsychotic polypharmacy to be generally ineffective for persistent positive psychotic symptoms. The concerns with combined antipsychotics include prescribing higher than necessary total dosage and an increased risk of side effects. If there is clinical benefit, one problem is the attribution of this to the combination rather than one or other of the individual antipsychotics, and thus uncertainty about the implications for optimal pharmacological treatment longer term.

For treatment resistant schizophrenia that has proved to be unresponsive to clozapine alone, adding a second antipsychotic would seem to be a relatively common strategy. The prevalence of this augmentation strategy in people with schizophrenia on clozapine ranges from 18% to 44%, depending on the clinical setting and country (Buckley et al., 2001; Potter et al., 1989; Taylor et al., 2000).

The mechanisms that might underly any increase in therapeutic effect with combined antipsychotics have not been systematically studied (McCarthy & Terkelsen, 1995). However, in relation to the strategy of adding an antipsychotic to clozapine, it has been hypothesised that any pharmacodynamic synergy might be related to an increased level of D2 dopamine receptor occupancy, above a threshold level (Chong & Remington, 2000; Kontaxakis et al., 2005). However, such an increase might also be expected to be associated with an increased risk of EPS. An alteration of the interaction between serotonin (5-hydroxytryptamine) and D2 activity has also been suggested as a relevant mechanism (Shiloh et al., 1997). Further, pharmacokinetic interactions might play a part, although there is no consistent evidence that adding an antipsychotic leads to increased clozapine plasma levels (Honer et al., 2006; Josisassen et al., 2005; Yagcioglu et al., 2005).

Randomised controlled trials and open studies have reported clozapine augmentation with a second antipsychotic to be relatively well tolerated. The main treatment-emergent side effects have been predictable from pharmacology of the augmenting drug, with EPS and prolactin elevation amongst the most common problems. However, with risperidone as the augmenting antipsychotic there are isolated reports of problems such as agranulocytosis, atrial ectopics and possible NMS with (Chong et al., 1996; Godleski & Serynak, 1996; Kontaxakis et al., 2002) and with aripiprazole as the second antipsychotic, reports mention nausea, vomiting, insomnia, headache and agitation in the first 2 weeks (Ziegenbein et al., 2006) and also modest weight loss (Karunakaran et al., 2006; Ziegenbein et al., 2006).

6.5.4. Clinical review protocol (people with schizophrenia whose illness has not responded adequately to treatment)

The clinical review protocol, including the primary clinical questions, information about the databases searched and the eligibility criteria can be found in Table 22. A new systematic search for relevant RCTs, published since the previous guideline, was conducted for the guideline update (further information about the search strategy can be found in Appendix 8).

Table 22. Clinical review protocol for the review of interventions for people with schizophrenia whose illness has not responded adequately to treatment.

Table 22

Clinical review protocol for the review of interventions for people with schizophrenia whose illness has not responded adequately to treatment.

6.5.5. Studies considered for review

In the previous guideline included, 19 RCTs were included in the review of antipsychotic medication for people with schizophrenia whose illness has not responded adequately to treatment. The update search identified five papers providing follow-up data or published versions of existing trials, and eight new trials (one trial [LIBERMAN2002] provided no useable outcome data and was excluded from the analysis). In addition, six trials (Altamura1999; Breier2000; Conley1998a; Emsley1999; Heck2000; Kern1998) previously analysed as acute phase studies were now included in this review, and three (Essock1996a; Gelenberg1979b; Wahlbeck2000) previously included were now excluded. In total, 26 trials (N = 3,932) met the inclusion criteria for the update. Further information about both included and excluded studies can be found in Appendix 15.

A new analysis, not conducted for the previous guideline, examined RCTs of antipsychotic medication in people with persistent negative symptoms of schizophrenia. Three trials (Boyer1990; Lecrubier1999; Murasaki1999) included in the previous review of acute treatment are now included here, but excluded from the updated acute treatment review. One trial (OLIE200610) excluded from the previous guideline is now included. One trial (Speller1997) included in the relapse prevention review also met the inclusion criteria for this review. The update search also identified five new RCTs that are included in this review, and one trial (HERTLING2003) that reported no appropriate data and so was excluded from the analysis. In total, 10 RCTs (N = 1,200) met the inclusion criteria for the update. Further information about both included and excluded studies can be found in Appendix 15.

For the review of clozapine augmentation, an existing systematic review and meta-analysis (Paton et al., 2007), published since the previous guideline, was used as the basis for an updated meta-analysis. This published review focused on the augmentation of clozapine with another SGA and included 4 RCTs. The update search identified two further RCTs. In total, six trials (N = 252) met the inclusion criteria for the update. In addition, two small studies (Assion et al., 2008; Mossaheb et al., 2006) with less than 10 participants in either arm were excluded, and one trial of clozapine plus amisulpride versus clozapine plus quetiapine (Genc et al., 2007) was excluded. Further information about both included and excluded studies can be found in Appendix 15.

6.5.6. Clozapine versus another antipsychotic drug in people with schizophrenia whose illness has not responded adequately to treatment

Seven RCTs were included in the analysis comparing clozapine with a FGA in people with schizophrenia whose illness has not responded adequately to treatment (see Table 23), and 10 RCTs were included in the analysis of clozapine versus another SGA (see Table 24). An evidence summary table for each comparison can be found in Chapter 10.

Table 23. Summary of study characteristics for RCTs of clozapine versus a FGA in people with schizophrenia whose illness has not responded adequately to treatment.

Table 23

Summary of study characteristics for RCTs of clozapine versus a FGA in people with schizophrenia whose illness has not responded adequately to treatment.

Table 24. Summary of study characteristics for RCTs of clozapine versus another SGA in people with schizophrenia whose illness has not responded adequately to treatment.

Table 24

Summary of study characteristics for RCTs of clozapine versus another SGA in people with schizophrenia whose illness has not responded adequately to treatment.

6.5.7. Second-generation antipsychotic drugs (other than clozapine) versus FGAs in people with schizophrenia whose illness has not responded adequately to treatment

Ten RCTs were included in the analysis comparing clozapine with another antipsychotic in people with schizophrenia whose illness has not responded adequately to treatment (see Table 25). An evidence summary table for each comparison can be found in Chapter 10.

Table 25. Summary of study characteristics for RCTs of SGAs versus FGAs in people with schizophrenia whose illness has not responded adequately to treatment.

Table 25

Summary of study characteristics for RCTs of SGAs versus FGAs in people with schizophrenia whose illness has not responded adequately to treatment.

6.5.8. Second-generation antipsychotic drugs (other than clozapine) versus SGAs in people with schizophrenia whose illness has not responded adequately to treatment

Three RCTs were included in the analysis comparing a SGA (olanzapine, risperidone) with another SGA in people with schizophrenia whose illness has not responded adequately to treatment (see Table 26). An evidence summary table for each comparison can be found in Chapter 10.

Table 26. Summary of study characteristics for RCTs of SGAs versus SGAs in people with schizophrenia whose illness has not responded adequately to treatment.

Table 26

Summary of study characteristics for RCTs of SGAs versus SGAs in people with schizophrenia whose illness has not responded adequately to treatment.

6.5.9. Second-generation antipsychotic drugs (other than clozapine) versus another antipsychotic in people who have persistent negative symptoms

Five RCTs were included in the analysis comparing a SGA (amisulpride, olanzapine, quetiapine, risperidone) with another SGA in people who have persistent negative symptoms (see Table 27). Five RCTs were included in the analysis comparing a SGA (amisulpride, olanzapine, quetiapine, risperidone) with another SGA in people who have persistent negative symptoms (see Table 28). An evidence summary table for each comparison can be found in Chapter 10.

Table 27. Summary of study characteristics for RCTs of SGAs versus a FGA in people who have persistent negative symptoms.

Table 27

Summary of study characteristics for RCTs of SGAs versus a FGA in people who have persistent negative symptoms.

Table 28. Summary of study characteristics for RCTs of SGAs versus another SGA in people who have persistent negative symptoms.

Table 28

Summary of study characteristics for RCTs of SGAs versus another SGA in people who have persistent negative symptoms.

6.5.10. Combining antipsychotics (augmentation of clozapine with another SGA)

One trial was included in the analysis comparing clozapine plus aripiprazole with clozapine plus placebo, four trials compared clozapine plus risperidone with clozapine plus placebo, and one trial compared clozapine plus sulpiride with clozapine plus placebo (see Table 29). An evidence summary table for each comparison can be found in Chapter 10.

Table 29. Summary of study characteristics for trials of clozapine augmentation.

Table 29

Summary of study characteristics for trials of clozapine augmentation.

6.5.11. Clinical evidence summary

In 18 RCTs including 2,554 participants whose illness had not responded adequately to treatment, clozapine had the most consistent evidence for efficacy over the FGAs included in the trials. Further evidence is required to establish equivalence between clozapine and any other SGA, and to establish whether there are differences between any of the other antipsychotic drugs. Side effects were consistent with those reported in the SPC for each drug.

In 10 RCTs including 1,200 participants with persistent negative symptoms, there was no evidence of clinically significant differences in efficacy between any of the antipsychotic drugs examined. Careful clinical assessment to determine whether such persistent features are primary or secondary is warranted, and may identify relevant treatment targets such as drug-induced parkinsonism, depressive features or certain positive symptoms.

In six RCTs including 252 participants with schizophrenia whose illness has not responded adequately to clozapine treatment, there is some evidence that clozapine augmentation with a second antipsychotic may improve both total and negative symptoms if administered for an adequate duration.

6.6. Treatment with depot/long-acting injectable antipsychotic medication

6.6.1. Introduction

The introduction of long-acting injectable formulations (‘depot’) of antipsychotic medication in the 1960s was heralded as a major advance in the treatment of established schizophrenia outside hospital. At the time it was hoped that depot preparations would lead to improved outcomes from antipsychotic pharmacotherapy. Consistent drug delivery and avoidance of the bioavailability problems that occur with oral preparations (such as gut wall and hepatic first-pass metabolism) were felt to be important factors. Other benefits include eliminating the risk of deliberate or inadvertent overdose. In the subsequent decades, the main practical clinical advantage to emerge has been the avoidance of covert non-adherence (both intentional and nonintentional)11 to antipsychotic drug treatment, where there is close nursing supervision and documentation of clinic attendance (Barnes & Curson, 1994; Patel & David, 2005). Service users who are receiving depot treatment and who decline their injection or fail to receive it through forgetfulness (or any other reason) can be immediately identified, allowing appropriate intervention, bearing in mind that poor adherence to the medication can be both a cause and consequence of worsening of illness. In practice, the use of depot drugs does not guarantee good treatment adherence, with a significant number who are prescribed maintenance treatment with depot preparations after discharge from hospital, failing to become established on the injections (Crammer & Eccleston, 1989; Young et al., 1989, 1996). But for those who continue with long-acting injections, there may be some adherence advantage over oral antipsychotics, indicated by a longer time to medication discontinuation (Zhu et al., 2008). There is also some evidence to suggest a better global outcome with depot as compared to oral antipsychotics (Adams et al., 2001) with a reduced risk of rehospitalisation (Schooler, 2003, Tiihonen et al., 2006). In 2002, a long acting formulation of a SGA, risperidone, became available, offering the same advantages of convenience and the avoidance of covert non-adherence (Hosalli & Davis, 2003).

Information on the use of long-acting antipsychotic injections has been limited (Adams et al., 2001), but relevant surveys and audits of antipsychotic prescription in the UK suggest that between a quarter and a third of psychiatric patients prescribed an antipsychotic may be receiving a long-acting injection, depending on the clinical setting (Barnes et al., 2009; Foster et al., 1996; Paton et al., 2003).

6.6.2. Use of long-acting antipsychotic injections

Long-acting injectable antipsychotic formulations generally consist of an ester of the drug in an oily solution. Another way of formulating such a preparation is to use microspheres of the drug suspended in aqueous solution. These drugs are administered by deep intramuscular injection, and are then slowly released from the injection site, giving relatively stable plasma drug levels over long periods, allowing the injections to be given every few weeks. However, this also represents a potential disadvantage, as there is a lack of flexibility of administration, with adjustment to the optimal dosage being a protracted and uncertain process. The controlled studies of low-dose maintenance treatment with depot preparations suggest that any increased risk of relapse consequent upon a dose reduction may take months or years to become manifest. Another disadvantage is that, for some people, receiving the depot injection is an ignominious and passive experience. Further, there have been reports of pain, oedema, pruritus and sometimes a palpable mass at the injection site. In some cases, these concerns may lead service users to take active steps to avoid these injections and may even disengage with services altogether rather than receive medication through this route. Nevertheless, a substantial proportion of people receiving regular, long-acting antipsychotic injections prefer them to oral therapy, largely because they consider them to be more convenient (Patel & David, 2005; Walburn et al., 2001).

6.6.3. Clinical review protocol (depot/long-acting injectable antipsychotics)

The review protocol, including the primary clinical questions, information about the databases searched and the eligibility criteria can be found in Table 30. A new systematic search for relevant RCTs, published since the previous guideline, was conducted for the guideline update (further information about the search strategy can be found in Appendix 8).

Table 30. Clinical review protocol for the review of depot/long-acting injectable antipsychotics.

Table 30

Clinical review protocol for the review of depot/long-acting injectable antipsychotics.

6.6.4. Studies considered for review

In the previous guideline, the review of depot antipsychotic medication was based on a meta-review of five Cochrane reviews (David & Adams, 2001), which included 13 RCTs of flupentixol decanoate, 48 RCTs of fluphenazine decanoate, 11 RCTs of haloperidol decanoate, 10 RCTs of pipothiazine palmitate, and 3 RCTs of zuclopenthixol decanoate.

Since publication of the previous guideline, the review of fluphenazine decanoate (David et al., 2004) was updated and now includes 70 trials. The review of pipothiazine palmitate (Dinesh et al., 2004) was also updated and now includes 18 trials. In addition, one SGA (long-acting injectable risperidone) has been licensed for use as a depot. A Cochrane review of this medication for people with schizophrenia was published in 2003 (Hosalli & Davis, 2003). The update search identified no additional trials that met the eligibility criteria. Because of the volume of evidence for FGA depots, the GDG checked the updated Cochrane reviews were consistent with the previous guideline and then focused on the evidence for long-acting risperidone, which had not previously been reviewed. In total, two trials (N = 1,042) met inclusion criteria (one trial of long-acting risperidone versus placebo, and one trial of long-acting risperidone versus oral risperidone). Both trials were published in peer-reviewed journals between 2003 and 2005. Further information about the included studies can be found in Appendix 15.

6.6.5. Long-acting risperidone injection versus placebo or oral risperidone

One RCT was included in the analysis comparing long-acting risperidone injection with placebo injection, and one RCT was included in the analysis comparing long-acting risperidone with oral risperidone plus placebo injection (see Table 31). An evidence summary table for each comparison can be found in Chapter 10.

Table 31. Summary of study characteristics for RCTs of long-acting risperidone versus placebo or oral risperidone.

Table 31

Summary of study characteristics for RCTs of long-acting risperidone versus placebo or oral risperidone.

6.6.6. Clinical evidence summary

The update search did not identify any new evidence for the efficacy and safety of depot FGAs beyond that included in the updated Cochrane reviews (utilised in the previous guideline). These reviews did not indicate robust new evidence that would warrant changing the existing recommendations for depot antipsychotic medication.

Since publication of the previous guideline, the first depot SGA (risperidone) was licensed for use in the UK. However, there is currently only limited evidence from two double-blind RCTs regarding the efficacy and safety of long-acting injectable risperidone compared to placebo or oral antipsychotic medication (risperidone). The placebo controlled trial suggests that 25–75 mg of long-acting risperidone may improve the chance of response and produce a clinically significant reduction in the symptoms of schizophrenia, but larger doses carry an increased risk of neurological side effects. There is no evidence to suggest that long-acting risperidone has either greater efficacy or greater risk of adverse effects when compared to oral risperidone. However, as suggested by the trial authors, the trial was only designed to investigate the short-term switching of participants from oral medication to long-acting risperidone; further studies are needed to understand the effect of continuous delivery of this medication.

6.7. Side effects of antipsychotic medication

6.7.1. Introduction

Given that for some antipsychotics there was a paucity of side effect data, the GDG decided to pool data, where appropriate, from the studies included in the other meta-analyses reported in this chapter and from any other relevant clinical trial. The review focused on metabolic and neurological side effects.

6.7.2. Studies considered for review

All RCTs included in the efficacy reviews (except studies of depot/long-acting antipsychotics) were included in the overall side effect meta-analysis. In addition, two trials (ATMACA2003; LIEBERMAN2003) did not meet the inclusion criteria for any of the efficacy reviews, but reported relevant side effect data and so were included here.

6.7.3. Second-generation antipsychotic drugs versus another antipsychotic drug (overall analysis of side effects)

As shown in Table 32, thirteen RCTs were included in the analysis of amisulpride against haloperidol (k = 6), a non-haloperidol FGA (k = 2), or a SGA (k = 5). Seven trials were included in the analysis of aripiprazole against haloperidol (k = 2), a non-haloperidol FGA (k = 1), or a SGA (k = 4). Sixteen trials were included in the analysis of clozapine against haloperidol (k = 4), a non-haloperidol FGA (k = 4), or a SGA (k = 9). Thirty-nine trials were included in the analysis of olanzapine against haloperidol (k = 17), a non-haloperidol FGA (k = 5), or a SGA (k = 18). Three trials were included in the analysis of paliperidone against a SGA (k = 3). Thirteen trials were included in the analysis of quetiapine against haloperidol (k = 5), a non-haloperidol FGA (k = 2), or a SGA (k = 7). Thirty-seven trials were included in the analysis of risperidone against haloperidol (k = 20), a non-haloperidol FGA (k = 4), or a SGA (k=16). Three trials were included in the analysis of sertindole against haloperidol (k = 2), or a SGA (k =1). Seven trials were included in the analysis of zotepine against haloperidol (k = 5), a non-haloperidol FGA (k = 1), or a SGA (k = 1). An evidence summary table for each comparison can be found in Chapter 10.

Table 32. Summary of studies included in the overall analysis of side effects.

Table 32

Summary of studies included in the overall analysis of side effects.

6.7.4. Clinical evidence summary

Pooling data from 138 evaluations of one antipsychotic versus another antipsychotic, did not reveal metabolic and neurological side effects that were inconsistent with those reported in the SPC for each drug. Because most trials were of relatively short duration and not designed to prospectively examine side effects, these trials provide little insight into the longer-term adverse effects of treatment or whether there are clinically significant differences between antipsychotic drugs.

6.8. Effectiveness of antipsychotic medication

6.8.1. Introduction

The RCT is widely recognised as the ‘gold standard’ for evaluating treatment efficacy, but some methodological issues may compromise the generalisability of the findings of research to the ordinary treatment setting. Nevertheless, it is still recognised that RCTs are an indispensable first step in the evaluation of interventions in mental health and provide the most valid method for determining the impact of two contrasting treatment conditions (treatment efficacy), while controlling for a wide range of participant factors, including the effects of spontaneous remission.

Once an approach has been demonstrated as efficacious under the stringent conditions of an RCT, a next step is to examine its effectiveness in ordinary treatment conditions, including large-scale effectiveness (pragmatic) trials (very few of which were available when the previous guideline was developed).

In addition, the use of RCTs and other studies in the evaluation of interventions in the treatment of schizophrenia is limited in many cases by the absence of important outcome measures. For example, few trials report evidence on quality of life or satisfaction with services, despite the fact that services users and carers view these measures as very important. Effectiveness studies address this issue by focusing on patient-important outcomes.

6.8.2. Effectiveness (pragmatic) trials

Given the large scope of the guideline update, the GDG decided to focus on effectiveness trials that included a comparison between a SGA and a FGA. To ensure that the evidence was from high quality research and reduce the risk of bias, studies were included only if they used a randomised design with an intention-to-treat analysis and at least independent rater-blinding (that is, the clinicians doing the assessment of outcome were independent and blind to treatment allocation). All studies identified during the searches for other sections of this chapter were considered for inclusion.

Two studies published since the previous guideline met the inclusion criteria for this review. These were the CATIE study (Lieberman et al., 2005; Stroup et al., 2003), funded by The National Institute of Mental Health, and the Cost Utility of the Latest Antipsychotic Drugs in Schizophrenia Study (CUtLASS 1) (Jones et al., 2006; Lewis et al., 2006), funded by the NHS Research and Development Health Technology Assessment Programme.

In the initial phase of CATIE (phase 1), which was conducted at 57 clinical sites in the US, 1,493 participants with chronic schizophrenia were randomised (double-blind) to one of four SGAs or a FGA (perphenazine) (see Table 33). Participants with current tardive dyskinesia could enrol, but were not able to be randomised to perphenazine. For the purposes of the guideline update, the GDG focused on the primary outcome (discontinuation of treatment for any reason), tolerability, and both metabolic and neurological side effects. An evidence summary table for these outcomes can be found in Chapter 10.

Table 33. Summary of study characteristics for the initial phases of CATIE and CUtLASS.

Table 33

Summary of study characteristics for the initial phases of CATIE and CUtLASS.

In the initial phase of CUtLASS (Band 1), 227 participants with schizophrenia (or related disorder) were randomised to a FGA or SGA (choice of individual drug was made by psychiatrist responsible for care of patient). The study was conducted in 14 NHS trusts in England and was specifically designed to test effectiveness in routine NHS practice. For the purposes of the guideline update, the GDG focused on the primary outcome (the Quality of Life Scale; Heinrichs et al., 1984), tolerability, and neurological side effects. An evidence summary table for these outcomes can be found in Chapter 10.

Further analysis of cost-effectiveness, including Band 2 of the CUtLASS trial can be found in section 6.9.

6.8.3. Clinical evidence summary

Two trials involving 1,720 participants, failed to establish clinically significant differences in effectiveness between the oral (non-clozapine) antipsychotic drugs examined. Although both trials have limitations (for further information see Carpenter & Buchanan, 2008; Kasper & Winkler, 2006; Möller, 2008; Lieberman, 2006), it is clear that more effective medication is needed. Furthermore, neither study included participants experiencing their first episode of schizophrenia or examined depot/long-acting antipsychotic medication.

With regard to adverse effects of treatment, the diverse side effect profiles seen in the efficacy trials reported elsewhere in this chapter were supported by CATIE and CUtLASS; primarily confirming differential metabolic effects. However, there was no consistent clinically significant differences between antipsychotics in terms of treatment-emergent EPS. It should be noted that the various FGAs tested (such as perphenazine and sulpiride) were generally not high-potency antipsychotics and were prescribed in standard doses. Further analyses of baseline data from CATIE also confirms other reports that people with schizophrenia are undertreated for metabolic disorders (Nasrallah et al., 2006).

6.9. Health economics

6.9.1. Systematic literature review

The systematic search of the economic literature undertaken for the guideline update identified 33 eligible studies on pharmacological treatments for people with schizophrenia. Of these, one study assessed oral antipsychotic medications for initial treatment of schizophrenia (Davies & Lewis, 2000); 15 studies examined oral drug treatments for acute psychotic episodes (Alexeyeva et al., 2001; Almond & O’Donnell, 2000; Bagnall et al., 2003; Beard et al., 2006; Bounthavong et al., 2007; Cummins et al., 1998; Edgell et al., 2000; Geitona et al., 2008; Hamilton et al., 1999; Jerrell, 2002; Lecomte et al., 2000; Nicholls et al., 2003; Palmer et al., 1998 & 2002; Rosenheck et al., 2003); 8 studies assessed oral antipsychotic medications aimed at promoting recovery (Davies et al., 1998; Ganguly et al., 2003; Knapp et al., 2008; Launois et al., 1998; Oh et al., 2001; Rosenheck et al., 2006; Tunis et al., 2006; Vera-Llonch et al., 2004); 4 studies examined pharmacological treatments aiming at promoting recovery in people with schizophrenia whose illness has not responded adequately to treatment (Rosenheck et al., 1997; Tilden et al., 2002; Lewis et al., 2006 & Davies et al., 2008); and 6 studies evaluated depot antipsychotic treatments (Chue et al., 2005; De Graeve et al., 2005; Edwards et al., 2005; Heeg et al., 2008; Laux et al., 2005; Oh et al., 2001). Details on the methods used for the systematic review of the economic literature in the guideline update are described in Chapter 3; references to included and excluded studies and evidence tables for all economic evaluations included in the systematic literature review are provided in Appendix 14.

Initial treatment with antipsychotic medication

One study that assessed oral antipsychotics for the treatment of people with a first episode of schizophrenia was included in the systematic economic literature review (Davies & Lewis, 2000). The study, which was conducted in the UK, was a cost-utility analysis based on a decision-analytic model in the form of a decision tree. The antipsychotic treatments assessed were olanzapine, risperidone, chlorpromazine, haloperidol and clozapine. All drugs, with the exception of clozapine, were assessed as 1st, 2nd, 3rd or 4th line of treatment, whereas clozapine was assessed as 3rd or 4th line of treatment only. According to the model structure, people switched to the next line of treatment when an antipsychotic was not acceptable to them; treatment unacceptability was defined as treatment intolerance (development of non-treatable or unacceptable side effects), inadequate response or non-compliance. People who found treatment acceptable were transferred to maintenance therapy. If they experienced a relapse during acceptable treatment over the time frame of the analysis, they were treated with the same antipsychotic. Acceptable side effects were treated without change in antipsychotic therapy. The adverse events considered in the analysis were EPS (except tardive dyskinesia which was considered separately), tardive dyskinesia, neuroleptic malignant syndrome, hepatic dysfunction and agranulocytosis. Clinical efficacy data were derived from a systematic literature review and meta-analysis. The perspective of the analysis was that of health and social care services including expenses of people with schizophrenia. Resource use was based on published literature, other national sources and further assumptions. Prices were taken from national sources. The time horizon of the analysis was 3 years.

Results were reported separately for different scenarios regarding sequence of antipsychotic treatments. Olanzapine and haloperidol were dominated by chlorpromazine when used as any line of treatment. Risperidone was more effective than chlorpromazine, but always at an additional cost, which reached £34,241 per Quality Adjusted Life Year (QALY) when 1st line treatment was assessed. Clozapine dominated olanzapine and risperidone when used as 3rd or 4th line of treatment. It was shown to yield the highest number of QALYs out of all antipsychotics included in the analysis. Its Incremental Cost-Effectiveness Ratio (ICER) versus chlorpromazine was £35,689 and £47,980 per QALY, when they were compared as 3rd and 4th line treatments, respectively.

The results of the analysis were statistically significant and indicated that olanzapine and haloperidol were not cost-effective options compared with the other antipsychotic drugs assessed for the treatment of people with a first episode of schizophrenia. The authors concluded that clozapine (as 3rd or 4th line treatment) and risperidone might be more effective than chlorpromazine, but at a higher cost. However, they recognised that because multiple comparisons of costs and QALYs had been made, some statistically important differences might have occurred by chance rather than reflected real differences. Moreover, they recognised the limited availability of clinical data used in the model.

An additional limitation of the analysis was that efficacy data for each antipsychotic medication were apparently derived from “naive” addition of data across relevant treatment arms of all RCTs included in the systematic literature review. This method treats the data as if they came from a single trial and practically breaks the randomisation: data from treatment arms not directly relevant to the analysis are not taken into account and between-trial variance is completely ignored (Glenny et al., 2005). Glenny and colleagues argue that such method of combining trial data is liable to bias, is highly unpredictable and also produces over-precise answers. They conclude that results of such analysis are completely untrustworthy, and therefore naive comparisons should never be made.

Furthermore, utility data used in the base-case analysis by Davis and Lewis (2000) were based on published utility values of 7 people with schizophrenia in Canada (Glennie, 1997), which appeared to be favouring FGAs and clozapine. Overall, the conclusions of this analysis should be interpreted with caution.

Oral antipsychotics in the treatment of the acute episode

The systematic review of the economic literature considered 15 studies evaluating oral antipsychotic medications for the management of acute psychotic episodes (Alexeyeva et al., 2001; Almond & O’Donnell, 2000; Bagnall et al., 2003; Beard et al., 2006; Bounthavong & Okamoto, 2007; Cummins et al., 1998; Edgell et al., 2000; Geitona et al., 2008; Hamilton et al., 1999; Jerrell, 2002; Lecomte et al., 2000; Nicholls et al., 2003; Palmer et al., 1998 and 2002; Rosenheck et al., 2003). Of these, 4 were conducted in the UK (Almond & O’Donnell, 2000; Bagnall et al., 2003; Cummins et al., 1998; Nicholls et al., 2003) and are described in more detail. Of the rest 11 studies, 7 were conducted in the US (Alexeyeva et al., 2001; Bounthavong & Okamoto, 2007; Edgell et al., 2000; Hamilton et al., 1999; Jerrell, 2002; Palmer et al., 1998; Rosenheck et al., 2003), 1 in Germany (Beard et al., 2006), 1 in Belgium (Lecomte et al., 2000), 1 in Mexico (Palmer et al., 2002) and 1 in Greece (Geitona et al., 2008).

Bagnall and colleagues (2003), using the same economic model structure as Davies and Lewis (2000), evaluated the cost-effectiveness of SGAs for the treatment of acute episodes in people with schizophrenia in the UK. Ten antipsychotic medications were included in a cost-utility analysis: olanzapine, risperidone, quetiapine, amisulpride, zotepine, sertindole, ziprasidone, clozapine, chlorpromazine and haloperidol. Clinical data were based on a systematic literature review and meta-analysis, and other published literature. The study adopted the perspective of health and social care services. Resource use was based on published literature and further assumptions. National unit costs were used. Outcomes were expressed in QALYs. Utility values in the base-case analysis were also taken from Glennie (1997). The time horizon of the analysis was one year.

Results were reported separately for 1st, 2nd, 3rd and 4th line of treatment. The authors performed comparisons between each SGA and the rest medications. Ziprasidone and amisulpride were associated with the highest costs and QALYs. According to the authors, amisulpride was the most cost-effective SGA drug if ziprasidone remained unlicensed. Risperidone was the most effective and costlier drug following ziprasidone and amisulpride. Olanzapine was the least costly and least effective antipsychotic. The authors suggested that sertindole, zotepine and quetiapine were not superior to other SGAs in terms of cost-effectiveness. However, the cost and the effectiveness results were characterised by high uncertainty. In addition, clinical data for haloperidol and chlorpromazine were taken from the control arms of SGA trials, as no systematic review of the literature was undertaken for FGAs; this methodology may have introduced bias in the analysis. A further limitation of the study was that analysis of efficacy data utilised the “naive” method for data pooling, as described earlier, and therefore the analysis is subject to bias. For all these reasons, no clear conclusions on the relative cost-effectiveness of SGAs can be drawn from this analysis, and this was also the authors’ conclusion.

Cummins and colleagues (1998) used the results of a RCT comparing olanzapine with haloperidol for acute treatment of people with schizophrenia (TOLLEFSON1997) to inform a decision-tree that was constructed to assess the relative cost effectiveness of the two antipsychotic drugs in the UK. According to the model structure, people in acute episode were started on one of the two evaluated drugs and were followed up for one year. Those who did not respond to treatment, or withdrew, or relapsed following response were switched to haloperidol (if they had been started on olanzapine) or fluphenazine (if they had been started on haloperidol). The perspective of the analysis was that of the NHS. Resource use was based on published literature and further assumptions. Prices were taken from national sources. Outcomes were expressed in QALYs. Utility values were estimated using the index of health-related quality of life (IHRQ), a generic measure designed to capture social, psychological and physical functioning.

Olanzapine was found to dominate haloperidol, as it produced more QALYs (0.833 versus 0.806) and resulted in lower costs (£26,200 versus £31,627). The results were robust in a number of sensitivity analyses carried out. Limitations of the analysis, as stated by the authors, were the weak evidence on longer-term effects of antipsychotics, which led to a number of assumptions in the model, and the simplicity of the model structure, which did not capture all events related to treatment of acute episodes with antipsychotics.

Almond and O’Donnell (2000) conducted an economic analysis to compare the costs and benefits associated with olanzapine, risperidone, and haloperidol in the treatment of acute psychotic episodes in the UK. Analysis was based on decision-analytic modelling. The economic model considered cycles of acute episodes, remission and relapse over a period of 5 years. Efficacy data were taken from 2 clinical trials (TOLLEFSON1997 and TRAN1997). The outcomes of the analysis were the percentage of people with BPRS score below 18 and the percentage of people without relapse over the time frame of the analysis. The study adopted the NHS perspective. Resource use estimates were based on published literature and further assumptions. UK national prices were used.

Olanzapine was reported to be less costly than both risperidone and haloperidol (costs of olanzapine, risperidone and haloperidol were £35,701, £36,590 and £36,653 respectively). In addition, it was found to be more effective (percentages of people with score below 18 over 5 years for olanzapine, risperidone and haloperidol were 63.6%, 63.0%, and 52.2%, respectively; percentages of people without relapse over 5 years were 31.2%, 29.3% and 18.2%, respectively). These figures show that olanzapine and risperidone dominated haloperidol (olanzapine was more effective at a lower cost; risperidone was more effective at a similar cost). Olanzapine also dominated risperidone (it was slightly more effective at a lower cost). Cost results were sensitive to daily dosages, relapse rates and drop-out rates. The authors reported as limitations of their analysis the assumptions needed in order to estimate resource utilisation and the omission of some categories of cost, such as costs of monitoring drug therapy, owing to lack of relevant data.

Nicholls and colleagues (2003) performed a cost-minimisation analysis alongside an international, multicentre clinical trial that compared amisulpride with risperidone over a 6 month treatment period (LECRUBIER2000). The trial had demonstrated that amisulpride and risperidone had similar effectiveness, as measured using PANSS, BPRS and CGI scores. The economic analysis, which adopted the perspective of the NHS, utilised resource use estimates from the trial and UK unit costs.

Amisulpride was found to be overall less costly than risperidone by £2,145, but the result was not statistically significant (95% CI: -£5,379 to £1,089). The findings of the study are not directly applicable to the UK setting, as resource use was based on settings other than the UK, where clinical practice is likely to be different. For example, part-time hospitalisations were recorded in some settings; the authors stated that this type of care was not universally recognised in the NHS, and for this reason respective UK unit costs were not available and needed to be based on assumptions.

Of the rest 11 studies included in the systematic review of cost effectiveness of oral antipsychotics in the management of acute psychotic episodes, 9 involved comparisons between olanzapine, risperidone and haloperidol. Relative cost effectiveness between olanzapine and risperidone cannot be established with certainty from the results of these studies: Beard and colleagues (2006) suggested that olanzapine was dominant over risperidone, as it was shown to be more effective at a lower cost. The analysis, which was conducted from the perspective of the German healthcare system, was based on decision-analytic modelling. Other models of similar structure replicated this result in other countries: olanzapine dominated risperidone in the US (Palmer et al., 1998) and in Mexico (Palmer et al., 2002). On the other hand, the modelling studies by Bounthavong & Okamoto (2007) in the US and Lecomte and colleagues (2000) in Belgium indicated that risperidone might be marginally dominant over olanzapine, as it was associated with better or similar outcomes at similar or slightly lower costs. Two economic analyses conducted alongside clinical trials in the US (Edgell et al., 2000; Jerrell, 2002) could also not draw certain conclusions: in both trials, olanzapine appeared to be less costly than risperidone, but cost results were not statistically significant. In one of the trials, olanzapine was associated with longer maintenance of response and lower EPS rates (Edgell et al., 2000) but the other trial (Jerrell, 2002) failed to demonstrate a superiority of olanzapine over risperidone in terms of clinical effectiveness.

With respect to the cost effectiveness between olanzapine and haloperidol, there was smaller variety in the study results: two modelling studies (Bounthavong & Okamoto, 2007; Palmer et al., 1998) and one economic analysis undertaken alongside a clinical trial (Hamilton et al., 1999) demonstrated that olanzapine dominated haloperidol in the US, as it was more effective at a lower cost. Another multicentre RCT conducted in the US (Rosenheck et al., 2003) showed that olanzapine had similar effectiveness to haloperidol, measured by BPRS scores, and lower akathisia rates. It was more expensive than haloperidol but cost results were not statistically significant. Finally, two modelling studies suggested that olanzapine was more effective than haloperidol at an additional cost that approximated £3 per day with minimum symptoms and toxicity in Belgium (Lecomte et al., 2000) and £11,350 per relapse avoided in Mexico (Palmer et al., 2002). Overall, these results suggest that olanzapine may be more cost-effective than haloperidol in the treatment of acute episodes.

Two of the comparisons of risperidone versus haloperidol showed that risperidone was the dominant option in the US (Bounthavong & Okamoto, 2007) and in Belgium (Lecomte et al., 2000), while one economic model used to assessed the relative cost effectiveness of the two antipsychotics in 2 different countries found risperidone to be more effective than haloperidol at an additional cost that reached $2,100/QALY in the US (Palmer et al., 1998) and about £13,900 per relapse avoided in Mexico (Palmer et al., 2002). These findings suggest that risperidone may be more cost-effective than haloperidol.

Finally, of the remaining two studies included in the systematic economic literature review of acute treatment for people with schizophrenia, the one conducted by Alexeyeva and colleagues (2001) compared the cost effectiveness between olanzapine and ziprasidone in the US; the study, which was based on decision-analytic modelling, utilised published and unpublished clinical data and concluded that olanzapine dominated ziprasidone, as it was more effective at a similar total cost. The other study (Geitona et al., 2008) assessed the cost effectiveness of paliperidone relative to risperidone, olanzapine, quetiapine, aripiprazol and ziprasidone from the perspective of the Greek healthcare system. The study, which was also based on decision-analytic modelling, utilised efficacy data from selected placebo-controlled trials and other published sources. Resource utilisation estimates were based on expert opinion. According to the authors’ conclusions, paliperidone was the most cost-effective drug as it dominated all other treatment options assessed. This finding was reported to be robust in sensitivity analysis. However, dominance of paliperidone over olanzapine was only marginal (paliperidone resulted in 0.3 additional days free of symptoms per year and an annual extra saving of €4 compared with olanzapine).

It must be noted that the results of most modelling studies were sensitive to changes in response and drop-out rates, drug acquisition costs, and hospitalisation rates for an acute episode. Most of these studies did not maintain randomisation effects, as they used (and in some cases combined) efficacy data from arms of different trials for each antipsychotic drug evaluated, using a “naive” method of pooling. The impact of side effects on the HRQoL was not explored in the majority of them.

Promoting recovery in people with schizophrenia that is in remission – pharmacological relapse prevention

Eight studies that were included in the systematic economic literature review assessed oral antipsychotic medications for relapse prevention (Davies et al., 1998; Ganguly et al., 2003; Knapp et al., 2008; Launois et al., 1998; Oh et al., 2001; Rosenheck et al., 2006; Tunis et al., 2006; Vera-Llonch et al., 2004). None of the studies was undertaken in the UK.

The most relevant study to the UK context was that by Knapp and colleagues (2008); it evaluated the cost effectiveness of olanzapine versus a number of other antipsychotic medications (including risperidone, quetiapine, amisulpride, clozapine, as well as oral and depot FGAs) using clinical and resource use data from a multicentre prospective observational study conducted in outpatient settings in 10 European countries. The analysis adopted the health service payer’s perspective; costs were estimated by applying UK national unit cost data to recorded healthcare resource use. Outcomes were expressed in QALYs, estimated by recording and analysing participants’ EQ-5D scores and linking them to respective UK population tariffs in order to determine utility values. The time horizon of the analysis was 12 months.

The study made separate comparisons of olanzapine with each of the rest antipsychotic medications considered; no direct comparisons were made between the rest antipsychotic medications. According to the performed comparisons, olanzapine dominated quetiapine and amisulpride; it was more effective than risperidone and clozapine at an additional cost reaching £5,156 and £775 per QALY, respectively. Compared with oral and depot FGAs, olanzapine was more effective and more costly, with an ICER of £15,696 and £23,331 per QALY, respectively (2004 prices). However, FGAs were analysed together, as a class, and no results from comparisons between olanzapine and specific FGAs were reported. Probabilistic sensitivity analysis conducted using bootstrap techniques revealed that the probability of olanzapine being more cost-effective than quetiapine was 100% at a willingness-to-pay lower than £5,000/QALY; the probability of olanzapine being cost-effective when compared with risperidone and amisulpride was 100% at a willingness-to-pay around £18,000/QALY; at a willingness-to-pay equalling £30,000 per QALY, the probability of olanzapine being more cost-effective than clozapine, oral FGAs and depot FGAs was 81%, 98% and 79% respectively.

The results of the analysis indicated that olanzapine had a high probability of being cost-effective relative to each of the other options assessed. However, no formal incremental analysis across all comparators was performed, as all comparisons involved olanzapine versus each of the other antipsychotics included in the analysis. The study conclusions may have limited applicability in the UK, as reported healthcare resource use reflected average routine clinical practice in European countries, and only unit costs were directly relevant to the UK health service.

The rest of the economic studies on pharmacological relapse prevention included mainly comparisons between olanzapine, risperidone and haloperidol. Two modelling studies, one in Australia (Davies et al., 1998) and one in Canada (Oh et al., 2001) concluded that risperidone was more cost-effective than haloperidol, as it was more effective at a lower cost. One modelling US study reported that risperidone was more effective and also more expensive than haloperidol (Ganguly et al., 2003). The measure of outcome was the number of employable persons in each arm of the analysis; employability was determined by a PANSS score reduction of at least 20% from baseline and a WCST-Cat score of ≥ 3.5. The ICER of risperidone versus haloperidol was estimated at $19,609 per employable person.

An economic analysis undertaken alongside an open label trial in the US (Tunis et al., 2006) showed that olanzapine was associated with better outcomes and lower costs compared to risperidone in people with chronic schizophrenia, but results were statistically insignificant. Another study based on mainly unpublished data and employing Markov modelling techniques (Vera-Llonch et al., 2004) came to different conclusions: according to this study risperidone led to lower discontinuation rates, had overall lower side effect rates, and was less costly than olanzapine. A modelling study carried out in France (Launois et al., 1998) reported that sertindole dominated olanzapine and haloperidol; between olanzapine and haloperidol, the former was the cost-effective option. Results of modelling studies were overall sensitive to changes in response rates, compliance rates, and hospital discharge rates.

Finally, Rosenheck and colleagues (2006) performed an economic analysis alongside a large effectiveness trial in the US (CATIE, Lieberman et al., 2005). The study compared olanzapine, quetiapine, risperidone, ziprasidone and perphenazine in people with chronic schizophrenia. It was demonstrated that perphenazine dominated all other antipsychotic medications, as it was significantly less costly than the other antipsychotics and had similar effectiveness, expressed in QALYs (perphenazine was significantly more effective than risperidone at the 0.005 level in intention-to-treat analysis). Differences in total healthcare costs were mainly caused by differences in drug acquisition costs between perphenazine and the rest of the antipsychotic drugs.

Promoting recovery in people with schizophrenia whose illness has not responded adequately to treatment (treatment resistance)

Four studies examining pharmacological treatments aiming at promoting recovery in people with schizophrenia whose illness has not responded adequately to treatment were included in the systematic review (Rosenheck et al., 1997; Tilden et al., 2002; Lewis et al., 2006 & Davies et al., 2008).

Tilden and colleagues (2002) constructed a Markov model to assess the cost effectiveness of quetiapine versus haloperidol in people with schizophrenia only partially responsive to FGAs, from the perspective of the UK NHS. The model was populated with clinical data taken from various sources: rates of response to treatment were taken from a multicentre RCT, comparing the two antipsychotic treatments in people with schizophrenia partially responsive to FGAs (EMSLEY1999). In this study, response to treatment was defined as an improvement in PANSS total score of at least 20% between the start and the end of the trial. Compliance rates in the economic model were estimated by linking non-compliance with the presence of EPS. Relapse rates were estimated by linking relapse with non-response to treatment. Other clinical data were derived from published literature. Resource use estimates were based on published studies and further assumptions; national unit costs were used. The measure of outcome for the economic analysis was the average number of relapses and the expected duration of time in response per person with schizophrenia, over the time horizon of the analysis, which was 5 years.

Quetiapine was found to be more effective than haloperidol, at a slightly lower cost. Sensitivity analysis revealed that cost results were sensitive to differences in response rates between the two antipsychotic drugs, to the risk of relapse in non-responding and non-compliant individuals, and to the proportion of people requiring hospitalisation following relapse.

Rosenheck and colleagues (1997) assessed the cost effectiveness of clozapine relative to haloperidol in people with schizophrenia refractory to treatment and a history of high level use of inpatient services in the US, using a societal perspective. The analysis was based on clinical and resource use evidence from a multicentre RCT carried out in 14 Veterans Affair Centres. Clinical outcomes included PANSS scores, Quality of Life Scale scores, side effect rates and compliance rates. Clozapine resulted in significantly lower mean PANSS scores, better compliance rates and lower rates of EPS compared to haloperidol. The total medical cost associated with clozapine was lower than the respective cost of haloperidol, but the difference in costs was not statistically significant.

In addition to the above two studies, Lewis and colleagues (2006) described two effectiveness trials conducted in the UK, which aimed at determining the clinical and cost effectiveness of SGAs versus FGAs and clozapine versus SGAs in people with schizophrenia responding inadequately to, or having unacceptable side effects from, their current medication (CUtLASS, Bands 1 and 2). The studies would be normally excluded from the systematic review of the economic literature, because they treated SGAs and FGAs as classes of antipsychotic medications; no data relating to specific antipsychotic drugs were reported. However, these studies were directly relevant to the UK context and their findings could lead to useful conclusions supporting formulation of guideline recommendations. Therefore, their methods and economic findings are discussed in this section.

Both trials were conducted in adult mental health settings in 14 NHS trusts in Greater Manchester, Nottingham and London. Participants in Band 1 (N=227) were randomised to either a SGA (olanzapine, risperidone, quetiapine or amisulpride) or a FGA in oral or depot form. Participants in Band 2 (N=136) were randomised to either clozapine or one of the 4 SGAs named above. The primary clinical outcome of the analyses was the Quality of Life Scale (QLS), with secondary outcomes PANSS scores, side effects from medication and participant satisfaction. The measure of outcome in economic analyses was the number of QALYs gained. QALYs were estimated by recording and analysing participants’ EQ-5D scores and subsequently linking them to respective UK population tariffs in order to determine utility values. Costs were estimated from the perspective of health and social care services, and included medication, hospital inpatient and outpatient services, primary and community care services and social services. The time horizon of the analyses was 12 months.

According to the results for Band 1, FGAs dominated SGAs as they resulted in better outcomes at a lower total cost, but the results were not statistically significant. Bootstrap analysis of costs and QALYs, including imputed values for missing observations and censored cases, demonstrated that FGAs resulted in 0.08 more QALYs and net savings of £1,274 per person compared with SGAs (2001/02 prices). In univariate sensitivity analyses, FGAs dominated SGAs or had an ICER lower than £5,000 per QALY. Probabilistic sensitivity analysis (employing bootstrap techniques) showed that at a zero willingness-to-pay, FGAs had a probability of being cost-effective 65%; this probability rose up to 91% at a willingness-to-pay equalling £50,000 per QALY. At a willingness-to-pay of £20,000 per QALY, the probability of FGAs being more cost-effective than SGAs was roughly 80%. The results of the economic analysis indicate that FGAs are likely to be more cost-effective than SGAs at the NICE cost-effectiveness threshold of £20,000-£30,000 per QALY (NICE, 2008B).

According to the results for Band 2, clozapine resulted in a statistically significant improvement in symptoms, but not in quality of life. Total costs associated with clozapine were also significantly higher than respective costs of SGAs. Updated bootstrap analysis of costs and QALYs showed that clozapine yielded 0.07 more QALYs per person relative to SGAs, at an additional cost of £4,904 per person (Davies et al., 2007). The ICER of clozapine versus SGAs was estimated at £33,240 per QALY (2005/06 prices). This value ranged from approximately £23,000 to £70,000 per QALY in univariate sensitivity analyses. Probabilistic sensitivity analysis showed that at a zero willingness-to-pay, clozapine had a 35% probability of being cost-effective compared to SGAs; this probability reached 50% at a willingness-to-pay ranging between £30,000 and £35,000 per QALY. Results indicate that clozapine is unlikely to be cost-effective at the NICE cost-effectiveness threshold of £20,000-£30,000 per QALY (NICE, 2008B).

Analysis of costs in both trials revealed that the vast majority of costs (approximately 90% of total costs) were incurred by psychiatric hospital attendances; only 2–4% of total costs constituted drug acquisition costs. Overall, there was great variance in the use of health services and associated costs among study participants. The significant difference in cost between clozapine and SGAs was caused by great difference in psychiatric hospital costs between the two arms, possibly reflecting the licensing requirement for inpatient admission for initiation of therapy with clozapine at the time of the study. Nowadays such requirements are not in place anymore; therefore, at present, the cost effectiveness of clozapine versus SGAs is likely to be higher than that demonstrated in the analysis.

Treatment with depot/long-acting injectable antipsychotic medication

The systematic review of the economic literature identified 6 studies assessing the cost effectiveness of depot antipsychotic medications for people with schizophrenia (Chue et al., 2005; De Graeve et al., 2005; Edwards et al., 2005; Heeg et al., 2008; Laux et al., 2005; Oh et al., 2001). All studies were conducted outside the UK and employed modelling techniques.

According to the results of these studies, long-acting risperidone was dominant over haloperidol depot in Belgium (De Graeve et al., 2005), Germany (Laux et al., 2005), Portugal (Heeg et al., 2008), Canada (Chue et al., 2005), and the US (Edwards et al., 2005). It was dominant over olanzapine in Belgium (De Graeve et al., 2005), Germany (Laux et al., 2005), and the US (Edwards et al., 2005). It was dominant over oral risperidone in Portugal (Heeg et al., 2008), Canada (Chue et al., 2005), and the US (Edwards et al., 2005). Finally, it was shown to dominate quetiapine, ziprasidone and aripiprazole in the US (Edwards et al., 2005). In all the above studies, the cost effectiveness of long-acting risperidone was largely determined by its estimated higher compliance compared with oral antipsychotics. However, in most studies, the methodology used to estimate compliance as well as other clinical input parameters was not clearly described; a number of economic models were populated with estimates based to a great extent on expert opinion.

Oh and colleagues (2001) using data from published meta-analyses and expert opinion, reported that both haloperidol depot and fluphenazine depot were dominated by oral risperidone in Canada. Although the methodology adopted was clearly reported, the main limitation of this study was that randomisation effects from clinical trials were not maintained, as clinical input parameters were estimated by pooling data from different clinical trials for each drug (“naive” method of synthesis).

Overall, the quality of evidence on depot antipsychotic medications was rather poor and of limited applicability to the UK context, given that no study was conducted in the UK.

The impact of compliance with antipsychotic treatment on healthcare costs incurred by people with schizophrenia

The systematic search of economic literature identified a number of studies that assessed the impact of non-adherence to antispychotic medication on healthcare costs incurred by people with schizophrenia. Although these studies did not evaluate the cost effectiveness of specific pharmacological treatments and therefore do not form part of the systematic review of economic evidence, they are described in this section because they provide useful data on the association between compliance, risk of relapse and subsequent healthcare costs. This information was considered by the GDG at formulation of the guideline recommendations.

Knapp and colleagues (2004) analysed data from a national survey of psychiatric morbidity among adults living in institutions in the UK, conducted in 1994. Approximately 67% of the population surveyed had a diagnosis of schizophrenia. According to the data analysis, non-adherence was one of the most significant factors that increased health and social care costs. Non-adherence predicted an excess annual cost reaching £2,500 per person for inpatient services and another £2,500 for other health and social care services, such as outpatient and day care, contacts with community psychiatric nurses, occupational therapists and social workers, and sheltered employment (2001 prices).

A modelling exercise that simulated the treated course of schizophrenia assessed the impact of compliance on health benefits and healthcare costs in people with schizophrenia in the UK over a period of 5 years (Heeg et al., 2005). The study considered people experiencing a second or third episode of schizophrenia and took into account factors such as gender, disease severity, potential risk of harm to self and society, and social and environmental factors. Other factors such as number of psychiatric consultations, presence of psychotic episodes, symptoms and side effects were also incorporated into the model structure. People with a first episode of schizophrenia were excluded from analysis. The analysis demonstrated that a 20% increase in compliance with antipsychotic treatment resulted in cost-savings of £16,000 and in prevention of 0.55 psychotic episodes per person with schizophrenia over 5 years. Cost-savings were almost exclusively attributed to the great reduction in hospitalisation costs following improved compliance. Higher levels of compliance were also associated with increased time between relapses, decreased symptom severity and improved ability of people to take care of themselves.

With regards to people with a first episode of schizophrenia, Robinson and colleagues (1999B) assessed the rates of relapse following response to antipsychotic treatment in 104 people with a first episode of schizophrenia or schizoaffective disorder. The authors reported that, after initial recovery, the cumulative first-relapse rate was 82% over 5 years. Discontinuation of pharmacological treatment increased the risk of relapse by almost 5 times. The authors concluded that the risk of relapse within 5 years of recovery from a first episode of schizophrenia or schizoaffective disorder was high, but could be diminished with maintenance antipsychotic drug therapy. Although the study did not assess the costs associated with non-compliance, its results indicate that compliance with treatment can reduce healthcare costs considerably, by reducing rates of relapse which leads to high hospitalisation costs.

Finally, two published reviews examined the impact of compliance with antipsychotic therapy on healthcare costs incurred by people with schizophrenia (Thieda et al., 2003; Sun et al., 2007). The reviews analysed data from 21 studies in total and concluded that antipsychotic non-adherence led to an increase in relapse and, subsequently, hospitalisation rates and hospitalisation costs.

Summary of findings and conclusions from systematic economic literature review

The economic literature review included 31 economic evaluations of specific antipsychotic treatments for the management of people with schizophrenia, plus 2 effectiveness trials conducted in the UK, which assessed antipsychotic medications grouped in classes. Twenty-two studies were based on decision-analytic modelling and were characterised by varying quality with respect to sources of clinical and utility data and methods of evidence synthesis. Clinical data were derived from a variety of sources, ranging from published meta-analyses and RCTs to unpublished trials and expert opinion. Even when data were taken from meta-analyses of trial data, the effects of randomisation were not retained, because data were simply pooled (by using weighted mean values) from the respective trials evaluating the drug under assessment. This “naive” method is likely to have introduced strong bias in the analyses, and therefore is inappropriate for evidence synthesis of trial data (Glenny et al., 2005). The impact of side effects on the HRQoL was explored in few trials, and even in these cases it was the decrement in HRQoL owing to the presence of EPS that was mostly considered. The impact of other side effects on HRQoL was not explored. The majority of the studies were funded by industry, which may have resulted in additional bias.

The included studies reported a variety of findings. The results of modelling exercises were sensitive, as expected, to a number of parameters, such as response and drop-out rates, as well as rates and/or length of hospitalisation. Most of the cost results derived from clinical studies were statistically insignificant. With the exception of a few studies, the majority of economic evaluations included a very limited number of antipsychotic medications for the treatment of people in schizophrenia, mainly olanzapine, risperidone and haloperidol; however, a wider variety of antipsychotic medications has been shown to be clinically effective and is available in the market. Results of comparisons between the 3 most examined drugs were in some cases contradictory. Nevertheless, overall findings of the systematic review seem to suggest that olanzapine and risperidone may be more cost-effective than haloperidol. Similarly, there is evidence that long-acting risperidone may lead to substantial cost-savings and higher clinical benefits compared with oral forms of antipsychotic medication, due to higher levels of adherence characterising long-acting injectable forms. However, evidence on long-acting injectable forms comes from non-UK modelling studies that are characterised by unclear methods in estimating a number of crucial input parameters (such as levels of adherence).

The results of non-UK studies are not directly applicable to the UK context, and therefore, although they may be indicative of trends in relative cost effectiveness of different antipsychotic drugs worldwide, they should not be used exclusively to inform decisions in the UK context. On the other hand, the results of UK studies were overall characterised by high uncertainty and several important limitations.

The results of the economic analyses alongside effectiveness trials in the UK (Lewis et al., 2006; Davies et al., 2008) suggest that hospitalisation costs are the drivers of total costs associated with treatment of people with schizophrenia. Drug acquisition costs are only a small part of total costs, and are unlikely to affect significantly the cost effectiveness of antipsychotic medications. It could be hypothesised that in the short term, and for people with schizophrenia treated as inpatients (for example during an acute episode), there are no big differences in total costs between antipsychotic medications, unless there are differences in the length of hospital stays. It might be reasonable to argue that antipsychotic drugs that reduce the rate and length of hospital admissions (for example drugs that reduce the rate of future relapses and/or the length of acute episodes) are cost-saving options in the long term, despite of potentially high acquisition costs. A related factor affecting the magnitude of healthcare costs and subsequently the cost effectiveness of antipsychotic medications is the level of adherence: according to published evidence, high levels of adherence to antipsychotic treatment can greatly reduce the risk of relapse and subsequent hospitalisation costs.

Details of the methods and the results of all economic evaluations described in this section are provided in Appendix 14.

6.9.2. Economic modelling

A decision-analytic model was developed to assess the relative cost effectiveness of antipsychotic medications aiming at promoting recovery (preventing relapse) in people with schizophrenia that is in remission. The rationale for economic modelling, the methodology adopted, the results and the conclusions from this economic analysis are described in detail in chapter 7. This section provides a summary of the methods employed and the results of the economic analysis.

Overview of methods

A Markov model was constructed to evaluate the relative cost effectiveness of a number of oral antipsychotic medications over two different time horizons, that is, 10 years and over lifetime. The antipsychotic drugs assessed were olanzapine, amisulpride, zotepine, aripiprazole, paliperidone, risperidone and haloperidol. The choice of drugs was based on the availability of relapse prevention data identified in clinical evidence review (see section 6.4). The study population consisted of people with schizophrenia that is in remission. The model structure considered events such as relapse, discontinuation of treatment due to intolerable side effects and switch to another antipsychotic drug, discontinuation of treatment due to other reasons and move to no treatment, development of (tolerable) side effects such as acute EPS, weight gain, diabetes and glucose intolerance, complications related to diabetes, and death. Clinical data were derived from studies included in the guideline systematic review of clinical evidence and other published literature. Where appropriate, clinical data were analysed using Mixed Treatment Comparison or standard meta-analytic techniques. The measure of outcome in the economic analysis was the number of QALYs gained. The perspective of the analysis was that of health and personal social care services. Resource use was based on published literature, national statistics and, where evidence was lacking, the GDG expert opinion. National UK unit costs were used. The cost year was 2007. Two methods were employed for the analysis of input parameter data and presentation of the results. First, a deterministic analysis was undertaken, where data were analysed as point estimates and results were presented in the form of Incremental Cost-Effectiveness Ratios (ICERs) following the principles of incremental analysis. A probabilistic analysis was subsequently performed, in which most of the model input parameters were assigned probability distributions. This approach allowed more comprehensive consideration of the uncertainty characterising the input parameters and captured the non-linearity characterising the economic model structure. Results of probabilistic analysis were summarised in the form of cost effectiveness acceptability curves, which express the probability of each intervention being cost-effective at various levels of willingness-to-pay per QALY gained (that is, at various cost effectiveness thresholds).

Overview of results

Results of deterministic analysis demonstrated that zotepine dominated all other treatment options, as it was less costly and resulted in a higher number of QALYs, both at 10 years and over lifetime of antipsychotic medication use. After zotepine, olanzapine and paliperidone appeared to be the second and third most cost-effective drugs, respectively, in both time horizons of 10 years and over lifetime. Paliperidone and olanzapine dominated all other drugs (except zotepine) at 10 years; the ICER of paliperidone versus olanzapine was approximately £150,000/QALY. Over lifetime, olanzapine was shown to be the least effective and least costly intervention among those examined, but according to incremental analysis it was still ranked as second most cost-effective option following zotepine, using a cost effectiveness threshold of £20,000/QALY (note that adopting a threshold of £30,000/QALY would result in paliperidone being ranked second most cost-effective option and olanzapine third, as the ICER of paliperidone versus olanzapine was just above the £20,000/QALY threshold, at £20,872/QALY). According to sensitivity analysis, results were highly sensitive to the probability of relapse attached to each antipsychotic drug, but were not driven by the estimated probabilities of developing each of the side effects considered in the analysis.

Probabilistic analysis revealed that zotepine had the highest probability of being the most cost-effective option among those assessed, but this probability was rather low, roughly 27-30%, reflecting the uncertainty characterising the results of the analysis. This probability was practically independent of the cost effectiveness threshold and the time horizon examined. The other antipsychotic medications had probabilities of being cost-effective that ranged from approximately 5% (haloperidol) to 16% (paliperidone). Again, these probabilities were rather unaffected by different levels of willingness-to-pay and consideration of different time horizons.

The results of the economic analysis are characterised by substantial levels of uncertainty as illustrated in probabilistic analysis, indicating that no antipsychotic medication can be considered clearly cost-effective compared to the other options included in the assessment. Moreover, it needs to be emphasised that the evidence base for the economic analysis was in some cases limited, as clinical data in the area of relapse prevention for 3 medications, namely, zotepine, paliperidone and aripiprazole, came from 3 single placebo-controlled trials, respectively.

6.10. From evidence to recommendations

In the previous guideline (which incorporated the recommendations from the NICE technology appraisal of second-generation antipsychotics, NICE, 2002), in some situations, SGAs were recommended as first-line treatment, primarily because they were thought to carry a lower potential risk of EPS. However, evidence from the updated systematic reviews of clinical evidence presented in this chapter, particularly with regard to other adverse effects such as metabolic disturbance, together with new evidence from effectiveness (pragmatic) trials suggest that choosing the most appropriate drug and formulation for an individual may be more important than the drug group.

Moreover, design problems in the individual trials continue to make interpretation of the clinical evidence difficult. Such problems include: a) high attrition from one or both treatment arms in many studies; b) differences between treatment arms in terms of medication dose; c) small numbers of studies reporting the same outcomes for some drugs.

For people with schizophrenia whose illness has not responded adequately to antipsychotic medication, clozapine continues to have the most robust evidence for efficacy. In addition, evidence from the effectiveness studies (CATIE, Phase 2; CUtLASS, Band 2) suggests that in people who have shown a poor response to non-clozapine SGAs, there is an advantage in switching to clozapine rather than another SGA. Nevertheless, even with optimum clozapine treatment it seems that only 30–60% of treatment-resistant illnesses will respond satisfactorily (Chakos et al., 2001, Iqbal et al., 2003).

The systematic review of the economic literature identified a number of studies of varying quality and relevance to the UK setting. Results were characterised, in most cases, by high uncertainty. The majority of studies assessed the relative cost effectiveness between olanzapine, risperidone and haloperidol. Although study findings are not consistent, they seem to indicate that, overall, olanzapine and risperidone might be more cost-effective than haloperidol.

In the area of antipsychotic treatment for first episode or early schizophrenia the economic evidence is limited and characterised by important limitations, and therefore no safe conclusions on the relative cost effectiveness of antispychotic medications can be drawn.

The amount of economic evidence is substantially higher in the area of pharmacological treatment for people with an acute exacerbation or recurrence of schizophrenia. However, the number of evaluated drugs is very limited and does not cover the whole range of drugs licensed for treatment of people with schizophrenia in the UK. In addition, existing studies are characterised by a number of limitations and, in may cases, by contradictory results. Available evidence indicates that olanzapine and risperidone may be more cost-effective options than haloperidol for acute exacerbation or recurrence of schizophrenia.

The economic literature in the area of relapse prevention is characterised by similar methodological limitations and also limited number of drugs assessed. Olanzapine and risperidone have been suggested to be more cost-effective than haloperidol in preventing relapse, but these conclusions are based on results from analyses conducted outside the UK. On the other hand, evidence from CATIE suggests that perphenazine may be more cost-effective than a number of SGAs (that is, olanzapine, quetiapine, risperidone, ziprasidone) in the US.

For people with schizophrenia whose illness has not responded adequately to treatment, sparce data on the cost effectiveness of specific antipsychotic medications are available. Evidence from CUtLASS, although not providing data on the cost effectiveness of individual drugs, provides useful insight into the factors that affect total costs incurred by people with schizophrenia. According to economic FINAL VERSION findings from CUtLASS, psychiatric inpatient care costs are the drivers of total healthcare costs incurred by people with schizophrenia, with drug acquisition costs being only a small fraction of total costs.

CUtLASS Band 2 found that clozapine was more effective than SGAs in the treatment of people with inadequate response to, or unacceptable side effects from, current medication, but at a higher cost that reached £33,000/QALY (ranging between £23,000-£70,000/QALY in univariate sensitivity analysis). It was suggested that the significant difference in cost between clozapine and SGAs might have been caused by great difference in psychiatric hospital costs between clozapine and SGAs, possibly reflecting the licensing requirement for inpatient admission for initiation of therapy with clozapine at the time of the study. Nowadays clozapine can be initiated in an outpatient setting; therefore, currently, the cost effectiveness of clozapine versus SGAs for people with inadequate reponse to treatment or unacceptable side effects is likely to be higher than that esimated at the time CUtLASS Band 2 was conducted.

Regarding depot/long-acting injectable antipsychotic medication, there is evidence that long-acting risperidone may lead to substantial cost-savings and greater clinical benefits compared with oral forms of antipsychotic medication, due to higher levels of adherence characterising long-acting injectable forms. However, this evidence comes from non-UK modelling studies that are characterised by unclear methods in estimating a number of crucial input parameters.

The economic analysis undertaken for this guideline estimated the cost effectiveness of oral antipsychotic medications for relapse prevention in people with schizophrenia. The results of the analysis suggest that out of the oral antipsychotic drugs included in the model, zotepine is potentially the most cost-effective option. However, results were characterised by high uncertainty and probabilistic analysis showed that no antipsychotic medication could be considered clearly cost-effective compared with the other treatment options assessed: according to results of probabilistic analysis, the probability of each drug being cost-effective ranged from roughly 5% (haloperidol) to about 27-30% (zotepine), and was independent of the cost effectiveness threshold used and the time horizon of the analysis (that is, 10 years or lifetime). The probability of 27-30% assigned to zotepine, although indicative, is rather low and inadequate to lead to a safe conclusion on zotepine’s superiority over the other antipsychotics assessed in terms of cost effectiveness. Moreover, clinical data for zotepine in the area of relapse prevention were exclusively derived from one small placebo-controlled RCT. Similarly, clinical data for paliperidone and aripiprazole were taken from 2 placebo-controlled trials, respectively. It must be noted that the economic analysis did not examine the cost effectiveness of quetiapine and any FGAs apart from haloperidol, owing to lack of respective clinical data in the area of relapse prevention.

An interesting finding of the economic analysis was that drug acquisition costs did not affect the cost effectiveness of antipsychotic medications: in fact haloperidol, which has the lowest price in the UK among those assessed, appeared to have the lowest probability of being cost-effective, of about 5%, at any level of willingness-to-pay. On the other hand, zotepine, which had the lowest average relapse rate across all evaluated treatments, dominated all other options in deterministic analysis and demonstrated the highest probability of being cost-effective in probabilistic analysis; this f inding together with results of sensitivity analysis indicate that the effectiveness of an antipsychotic drug in preventing relapse is the key-determinant of its relative cost effectiveness, apparently because relapse prevention, besides clinical improvement, leads to a substantial reduction in hospitalisation rates and respective costs.

Hospitalisation costs have been shown to drive healthcare costs incurred by people with schizophrenia, both in published evidence and in the economic analysis carried out for this guideline. It might be reasonable to argue that antipsychotic drugs that reduce the rate and length of hospital admissions (for example drugs that reduce the rate of future relapses and/or the length of acute episodes) are cost-saving options in the long term, despite potentially high acquisition costs. This hypothesis is supported by published evidence, which shows that increased adherence to antipsychotic treatment is associated with a significant decrease in healthcare costs incurred by people with schizophrenia, through a reduction in the risk of relapse and subsequent need for hospitalisation.

The GDG considered all clinical and economic evidence summarised in this section in order to formulate recommendations. In therapeutic areas where clinical and/or economic evidence on specific antipsychotic medications was lacking, as in the case of quetiapine and FGAs other than haloperidol in the area of relapse prevention, the GDG made judgements on the clinical and cost effectiveness of antipsychotic medications by extrapolating existing evidence and conclusions from other therapeutic areas.

Taking into account the findings from the systematic reviews of both the clinical and health economic literature, and the uncertainty characterising the results of economic modelling undertaken for this guideline, the evidence does not allow for any general recommendation for one antipsychotic to be preferred over another, but does support a specific recommendation for clozapine for people whose illness does not respond adequately to other antispsychotic medication.

Finally, the GDG noted that the following are the key points to be considered before initiating an antipsychotic medication in an acute episode of schizophrenia. First, there may be some lack of insight into the presence of a mental illness, and the relevance of drug treatment. Careful explanation is needed regarding the rationale for antipsychotic medication and its modes of action. People with schizophrenia will usually accept that they have been stressed, suffering from insomnia and not eating well so the acceptance of a tranquillising medication to help reduce stress and improve sleep and appetite might be acceptable. It can also be explained, if the patient is insightful enough, that the medication is antipsychotic and can help reduce the severity of distressing hallucinations, delusions and thought disorder.

Secondly, if possible, medication should always be started at low dose after a full discussion of the possible side effects. Starting at low dose allows monitoring for the early emergence of side effects such as EPS, weight gain or insomnia. The dose can then be titrated upwards within the BNF treatment range. Although polypharmacy with antipsychotic medications is not recommended, it is equally important not to under treat the acute psychotic episode.

Thirdly, people with schizophrenia should be consulted on their preference for a more or less sedative medication option. Medication is ideally started following a period of antipsychotic free assessment within an acute ward setting or under the supervision of a crisis home treatment team, early intervention in psychosis team or assertive outreach team.

6.11. Recommendations

6.11.1. Initiation of treatment (first episode)

6.11.1.1.

For people with newly diagnosed schizophrenia, offer oral antipsychotic medication. Provide information and discuss the benefits and side-effect profile of each drug with the service user. The choice of drug should be made by the service user and healthcare professional together, considering:

  • the relative potential of individual antipsychotic drugs to cause extrapyramidal side effects (including akathisia), metabolic side effects (including weight gain) and other side effects (including unpleasant subjective experiences)
  • the views of the carer if the service user agrees.

6.11.2. How to use oral antipsychotic medication

6.11.2.1.

Before starting antipsychotic medication, offer the person with schizophrenia an electrocardiogram (ECG) if:

  • specified in the SPC
  • a physical examination has identified specific cardiovascular risk (such as diagnosis of high blood pressure)
  • there is personal history of cardiovascular disease, or
  • the service user is being admitted as an inpatient.
6.11.2.2.

Treatment with antipsychotic medication should be considered an explicit individual therapeutic trial. Include the following:

  • Record the indications and expected benefits and risks of oral antipsychotic medication, and the expected time for a change in symptoms and appearance of side effects.
  • At the start of treatment give a dose at the lower end of the licensed range and slowly titrate upwards within the dose range given in the British National Formulary (BNF) or SPC.
  • Justify and record reasons for dosages outside the range given in the BNF or SPC.
  • Monitor and record the following regularly and systematically throughout treatment, but especially during titration:
    • – efficacy, including changes in symptoms and behaviour
    • – side effects of treatment, taking into account overlap between certain side effects and clinical features of schizophrenia, for example the overlap between akathisia and agitation or anxiety
    • – adherence
    • – physical health.
  • Record the rationale for continuing, changing or stopping medication, and the effects of such changes.
  • Carry out a trial of the medication at optimum dosage for 4-6 weeks.
6.11.2.3.

Treatment with antipsychotic medication should be considered an explicit individual therapeutic trial. Include the following:

  • Record the indications and expected benefits and risks of oral antipsychotic medication, and the expected time for a change in symptoms and appearance of side effects.
  • At the start of treatment give a dose at the lower end of the licensed range and slowly titrate upwards within the dose range given in the British National Formulary (BNF) or SPC.
  • Justify and record reasons for dosages outside the range given in the BNF or SPC.
  • Monitor and record the following regularly and systematically throughout treatment, but especially during titration:
    • – efficacy, including changes in symptoms and behaviour
    • – side effects of treatment, taking into account overlap between certain side effects and clinical features of schizophrenia, for example the overlap between akathisia and agitation or anxiety
    • – adherence
    • – physical health.
  • Record the rationale for continuing, changing or stopping medication, and the effects of such changes.
  • Carry out a trial of the medication at optimum dosage for 4-6 weeks.
6.11.2.4.

Discuss any non-prescribed therapies the service user wishes to use (including complementary therapies) with the service user, and carer if appropriate. Discuss the safety and efficacy of the therapies, and possible interference with the therapeutic effects of prescribed medication and psychological treatments.

6.11.2.5.

Discuss the use of alcohol, tobacco, prescription and non-prescription medication and illicit drugs with the service user, and carer if appropriate. Discuss their possible interference with the therapeutic effects of prescribed medication and psychological treatments.

6.11.2.6.

‘As required’ (p.r.n.) prescriptions of antipsychotic medication should be made as described in recommendation 6.11.2.2. Review clinical indications, frequency of administration, therapeutic benefits and side effects each week or as appropriate. Check whether ‘p.r.n.’ prescriptions have led to a dosage above the maximum specified in the BNF or SPC.

6.11.2.7.

Do not use a loading dose of antipsychotic medication (often referred to as ‘rapid neuroleptisation’).

6.11.2.8.

Do not initiate regular combined antipsychotic medication, except for short periods (for example, when changing medication).

6.11.2.9.

If prescribing chlorpromazine, warn of its potential to cause skin photosensitivity. Advise using sunscreen if necessary.

6.11.3. Acute treatment recommendations

6.11.3.1.

For people with an acute exacerbation or recurrence of schizophrenia, offer oral antipsychotic medication. The choice of drug should be influenced by the same criteria recommended for starting treatment (see section 6.11.1). Take into account the clinical response and side effects of the service user’s current and previous medication.

6.11.4. Rapid tranquillisation

6.11.4.1.

Occasionally people with schizophrenia pose an immediate risk to themselves or others during an acute episode and may need rapid tranquillisation. The management of immediate risk should follow the relevant NICE guidelines (see recommendations 6.11.4.2 and 6.11.4.5).

6.11.4.2.

Follow the recommendations in ‘Violence’ (NICE clinical guideline 25) when facing imminent violence or when considering rapid tranquillisation.

6.11.4.3.

After rapid tranquillisation, offer the person with schizophrenia the opportunity to discuss their experiences. Provide them with a clear explanation of the decision to use urgent sedation. Record this in their notes.

6.11.4.4.

Ensure that the person with schizophrenia has the opportunity to write an account of their experience of rapid tranquillisation in their notes.

6.11.4.5.

Follow the recommendations in ‘Self-harm’ (NICE clinical guideline 16) when managing acts of self-harm in people with schizophrenia.

6.11.5. Early post-acute period

6.11.5.1.

Inform the service user that there is a high risk of relapse if they stop medication in the next 1–2 years.

6.11.5.2.

If withdrawing antipsychotic medication, undertake gradually and monitor regularly for signs and symptoms of relapse.

6.11.5.3.

After withdrawal from antipsychotic medication, continue monitoring for signs and symptoms of relapse for at least 2 years.

6.11.6. Promoting recovery recommendations

6.11.6.1.

The choice of drug should be influenced by the same criteria recommended for starting treatment (see section 6.11.2).

6.11.6.2.

Do not use targeted, intermittent dosage maintenance strategies12 routinely. However, consider them for people with schizophrenia who are unwilling to accept a continuous maintenance regimen or if there is another contraindication to maintenance therapy, such as side-effect sensitivity.

6.11.6.3.

Consider offering depot/long-acting injectable antipsychotic medication to people with schizophrenia:

  • who would prefer such treatment after an acute episode
  • where avoiding covert non-adherence (either intentional or unintentional) to antipsychotic medication is a clinical priority within the treatment plan.

6.11.7. How to prescribe depot/long-acting injectable antipsychotic medication

6.11.7.1.

When initiating depot/long-acting injectable antipsychotic medication:

  • take into account the service user’s preferences and attitudes towards the mode of administration (regular intramuscular injections) and organisational procedures (for example, home visits and location of clinics)
  • take into account the same criteria recommended for the use of oral antipsychotic medication (see section 6.11.2), particularly in relation to the risks and benefits of the drug regimen
  • initially use a small test dose as set out in the BNF or SPC.

6.11.8. Interventions for people with schizophrenia who have an inadequate or no response to pharmacological or psychological treatment

6.11.8.1.

For people with schizophrenia whose illness has not responded adequately to pharmacological or psychological treatment:

  • review the diagnosis
  • establish that there has been adherence to antipsychotic medication, prescribed at an adequate dose and for the correct duration
  • review engagement with and use of psychological treatments and ensure that these have been offered according to this guideline. If family intervention has been undertaken suggest CBT; if CBT has been undertaken suggest family intervention for people in close contact with their families
  • consider other causes of non-response, such as comorbid substance misuse (including alcohol), the concurrent use of other prescribed medication or physical illness.
6.11.8.2.

Offer clozapine to people with schizophrenia whose illness has not responded adequately to treatment despite the sequential use of adequate doses of at least two different antipsychotic drugs. At least one of the drugs should be a non-clozapine second-generation antipsychotic.

6.11.8.3.

For people with schizophrenia whose illness has not responded adequately to clozapine at an optimised dose, healthcare professionals should consider recommendation 1.4.6.1 (including measuring therapeutic drug levels) before adding a second antipsychotic to augment treatment with clozapine. An adequate trial of such an augmentation may need to be up to 8–10 weeks. Choose a drug that does not compound the common side effects of clozapine.

6.11.9. Research recommendations

6.11.9.1.

More long-term, head-to head RCTs of the efficacy and safety/tolerability and-patient acceptability of the available antipsychotic drugs are required, in individuals in their first episode of schizophrenia, testing the risk-benefit of dosage at the lower end of the recommended dosage range.

6.11.9.2.

Large-scale, observational, survey-based studies, including qualitative components, of the experience of drug treatments, for available antipsychotics, should be undertaken. Studies should include data on service-user satisfaction, side-effects, preferences, provision of information and quality of life.

6.11.9.3.

Quantitative and qualitative research is required investigating the utility, acceptability and safety of available drugs for urgent sedation/control of acute behavioural disturbance (including benzodiazepines and antipsychotics), employing larger samples, in settings that reflect current clinical practice, and systematically manipulating dosage and frequency of drug administration.

6.11.9.4.

Further work is required on the nature and severity of antipsychotic drug discontinuation phenomena, including the re-emergence of psychotic symptoms, and their relationship to different antipsychotic withdrawal strategies.

6.11.9.5.

Direct comparisons between available oral antipsychotics are needed to establish their respective risk/long-term benefit, including effects upon relapse rates and persistent symptoms, and cost-effectiveness. Trials should pay particular attention to the long-term benefits and risks of the drugs, including systematic assessment of side effects: metabolic effects (including weight gain), EPS (including tardive dyskinesia), sexual dysfunction, lethargy and quality of life.

6.11.9.6.

Further RCT-based, long -term studies are needed to establish the clinical and cost-effectiveness of available depot/long-acting injectable antipsychotic preparations to establish their relative safety, efficacy in terms of relapse prevention, side-effect profile and impact upon quality of life.

6.11.9.7.

Further RCT-based, long -term studies are needed to establish the clinical and cost-effectiveness of augmenting antipsychotic monotherapy with an antidepressant to treat persistent negative symptoms.

6.11.9.8.

Controlled studies are required to test the efficacy and safety of combining antipsychotics to treat schizophrenia that has proved to be poorly responsive to adequate trials of antipsychotic monotherapy.

6.11.9.9.

A randomised placebo-controlled trial should be conducted to investigate the efficacy and cost-effectiveness of augmentation of clozapine monotherapy with an appropriate second antipsychotic where a refractory schizophrenic illness has shown only a partial response to clozapine.

6.11.9.10.

A randomised placebo-controlled trial should be conducted to investigate the efficacy and cost-effectiveness of augmentation of antipsychotic monotherapy with lithium where a schizophrenic illness has shown only a partial response. The response in illness with and without affective symptoms should be addressed.

6.11.9.11.

A randomised placebo-controlled trial should be conducted to investigate the efficacy and cost-effectiveness of augmentation of antipsychotic monotherapy with sodium valproate where a schizophrenic illness has shown only a partial response. The response of illness in relation to behavioural disturbance, specifically persistent aggression, should be specifically addressed to determine if this is independent of effect on potentially confounding variables, such as positive symptoms, sedation, or akathisia

6.11.9.12.

Further controlled studies are required to test the claims that clozapine is particularly effective in reducing hostility and violence, and the inconsistent evidence for a reduction in suicide rates in people with schizophrenia.

Footnotes

7
8

Here and elsewhere in this chapter, each study considered for review is referred to by a study ID, with studies included in the previous guideline in lower case and new studies in upper case (primary author and date or study number for unpublished trials). References for included studies denoted by study IDs can be found in Appendix 15.

9

Of these, 146 trials came from the following existing sources: NICE TA43 (NICE, 2002) and the Cochrane reviews of benperidol (Leucht & Hartung, 2002), loxapine (Fenton et al., 2002b), pimozide (Sultana & McMonagle, 2002), sulpiride (Soares et al., 2002) and thioridazine (Sultana et al., 2002). New systematic reviews were conducted for chlorpromazine, flupentixol, fluphenazine, oxypertine, pericyazine, perphenazine, prochlorperazine, promazine, trifluoperazine, and zuclopenthixol dihydrochloride. Data from poor quality trials, placebo comparisons and drugs not available in the UK were excluded.

10

In the previous guideline this trial was labelled as ‘Study 128–305’.

11

Further information about medicines concordance and adherence to treatment can be found in the NICE guideline on this topic (see http://www​.nice.org.uk).

12

Defined as the use of antipsychotic medication only during periods of incipient relapse or symptom exacerbation rather than continuously.

Copyright © 2009, National Collaborating Centre for Mental Health.
Cover of Schizophrenia
Schizophrenia: Core Interventions in the Treatment and Management of Schizophrenia in Primary and Secondary Care (Update) [Internet].
NICE Clinical Guidelines, No. 82.
National Collaborating Centre for Mental Health (UK).
Leicester (UK): British Psychological Society; 2009 Mar.

NICE (National Institute for Health and Care Excellence)

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...