NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Mental Health (UK). Bipolar Disorder: The Management of Bipolar Disorder in Adults, Children and Adolescents, in Primary and Secondary Care. Leicester (UK): British Psychological Society; 2006. (NICE Clinical Guidelines, No. 38.)

Cover of Bipolar Disorder

Bipolar Disorder: The Management of Bipolar Disorder in Adults, Children and Adolescents, in Primary and Secondary Care.

Show details

8THE MEDICAL AND PHARMACOLOGICAL MANAGEMENT OF BIPOLAR DISORDER – PART I

8.1. INTRODUCTION

Effective pharmacological treatment of bipolar disorder requires treatment of depressive and manic/hypomanic episodes together with long-term treatment to prevent future episodes, both syndromal and sub-syndromal. In recent years the importance of long-term treatment (that is, maintenance treatment) has been emphasised by several guidelines. The need for maintenance treatment is supported by the desire to prevent the costs of future episodes, that is, the intangible suffering to patients and their families and the economic burden of direct and indirect costs. In addition maintenance treatment may reduce long-term impairment associated with the bipolar disorder. There is evidence that functional impairment in patients who have recovered from acute episodes and are asymptomatic is related to the number of previous depressive episodes. The tendency for episodes to become more frequent with time also supports the rationale for maintenance treatment.

In the last decade, evidence from RCTs has accumulated regarding the effectiveness of several ‘new’ agents in the treatment of bipolar disorder. These include valproate (in various forms), lamotrigine and various atypical antipsychotics. The active control arms in several such studies have provided further evidence for the efficacy of ‘older’ treatments, in particular haloperidol (a conventional antipsychotic) in the treatment of mania and lithium in the prophylaxis of mania.

8.1.1. Methodological issues in RCTs

It is important to acknowledge that the RCT evidence base in bipolar disorder has a number of weaknesses which can be summarised as follows:

  • Dominance of industry-sponsored studies: most studies are sponsored by the pharmaceutical industry. Such studies are more likely to report results that favour the sponsor’s product than are independent studies. This may reflect publication bias or design bias.
  • Restricted entry criteria: RCTs tend to have strict inclusion and exclusion criteria. For example, patients with significant suicidal ideation or drug/alcohol misuse are often excluded. A recent study estimated that only 16% of a consecutive series of patients hospitalised for mania or a mixed episode would qualify to enter a standard placebo-controlled RCT in acute mania (Storosum et al., 2004). Consequently it is arguable how representative the results are regarding many patients treated in clinical practice.
  • Short duration: most RCTs in mania range from 3 to 6 weeks in length. In practice many patients take longer to respond.
  • Enriched samples: long-term maintenance studies usually recruit patients in either a depressive phase or manic phase and then randomise those who respond to acute treatment for the maintenance phase of the study. This design inevitably means that the sample is pre-selected for those who are responders to acute treatment. The results cannot therefore be extrapolated to those who start the study drug in a euthymic phase or in an acute illness of the opposite pole.
  • Reliance on statistical as opposed to clinically significant outcome measures.
  • Many aspects of the treatment of bipolar disorder are ‘neglected’ by RCTs: there are few or no RCTs to guide clinical practice in many areas. Examples of ‘neglected’ areas include the treatment of rapid cycling, hypomania and bipolar depression.

8.1.2. Outcomes research in pharmacological therapies for people with bipolar disorder

In the last 10 years there has been a dramatic increase in research in bipolar disorder. Despite this the evidence base remains relatively small and many of the key questions that concern clinicians are addressed inadequately or not at all by the existing research base. For example, there is virtually no RCT data on the treatment of rapid-cycling bipolar disorder, bipolar II disorder or the relative effectiveness of antimanic agents in combination versus monotherapy. Another problem is that the majority of trials deal with the treatment of either mania or major depressive episodes (that is, syndromes seen in bipolar disorder) yet research has clearly demonstrated that subsyndromal symptoms are three times more common than syndromal symptoms during the long-term course of both bipolar I disorder and bipolar II disorder (Judd et al., 2002; Judd et al., 2003).

The issue of the management of resistant mania and depression (that is, episodes that have not responded to first-line treatment) is a common clinical problem but one that is largely ignored by RCTs. The main reason for this neglect is that most RCTs are conducted with the aim of helping a drug gain a licence.

The majority of research has addressed bipolar-I disorder. Trials for mania and bipolar depression are usually short term (3 weeks and up to 10 weeks respectively) and based on changes observed using standard rating scales for example, the Young Mania Rating Scale (YMRS) and the Hamilton Depression Rating Scale. The outcome measures used include the mean change in scores from baseline to end point, response rates and remission rates. Patients entering acute mania trials represent the milder end of the severity spectrum not least because more severely ill patients will be unable to give informed consent. A key issue in trials of treatments for acute mania is that of high attrition rate, with a figure of around 70% not being uncommon in 3-week trials (for example, Zajecka et al., 2002). Another potential concern, relevant to all aspects of bipolar disorder, is publication bias. In industry-sponsored trials, a further bias is that trials may be designed to favour the sponsor’s compound. For example the efficacy of most atypical antipsychotics in mania has been supported by RCTs in which the comparator is haloperidol; haloperidol is a potent blocker of dopamine-2 receptors and is associated with a high risk of extrapyramidal symptoms. It is therefore not surprising that such RCTs show a benefit for the atypical in terms of a lower prevalence of extrapyramidal symptoms (EPS). Extrapolating from data in schizophrenia, it is reasonable to assume that if such studies had been repeated with a low-potency low-dose antipsychotic, then there may have been a less marked advantage or no difference at all in terms of the prevalence of EPS.

In acute mania there are a series of trials that assess the effectiveness of various atypical antipsychotics versus (1) a conventional antipsychotic, usually haloperidol; (2) placebo; and (3) the effectiveness of the atypical plus an antimanic agent (valproate or lithium) versus the antimanic agent alone. Trials that assess long-term efficacy in bipolar disorder usually take patients who have responded to treatment for an acute episode (mania or depression) and then randomise them to different long-term treatment options. A key issue in interpreting long-term studies is the relationship between the drug or drugs that were used in the initial acute treatment phase versus the long-term treatment phase, because selecting patients who are known responders can be an issue (so-called ‘enriched’ samples). A full discussion of issues in research in this area and the state of current research is beyond the scope of this chapter.

8.1.3. Mood stabiliser – a term best avoided

There is confusion over what the term ‘mood stabiliser’ refers to. Although the term is widely used in clinical practice, there is no universally accepted definition. The most rigorous definition of a mood stabiliser is that it is a drug that treats both poles of bipolar disorder and is protective against a return of both poles (Bauer & Mitchner, 2004). A less stringent criterion is that it is an agent that is effective in treating one pole of the disorder and in preventing a recurrence at that pole but that does not increase the risk of the opposite pole of the illness appearing. Adopting either criterion still leaves open the issues of how one (1) defines effectiveness, that is, what additional benefit above placebo is required to regard a drug as effective and (2) how one distinguishes between acute and maintenance treatment.

It has been argued that the most stringent definition of a mood stabiliser (that is, a drug that has short and long-term efficacy at both poles) is only fulfilled by lithium. However, even with lithium the data relating to efficacy in the treatment of bipolar depression is limited and that which does exist indicates only a modest effect. In contrast there is stronger evidence that lithium is effective in the acute and long-term treatment of mania. If one required evidence from at least two RCTs to indicate a clinically significant benefit in the acute and long-term treatment of both phases of bipolar disorder (that is, efficacy in four separate domains) then no drug would fulfil the criteria for a mood stabiliser. For this reason the less stringent criterion of acute and long-term efficacy and prophylaxis at one pole, without evidence of a worsening of the opposite pole, seems a more clinically useful definition and is the one that is adopted by many professionals.

Even this restricted definition of a mood stabiliser is associated with problems. The term may imply that these drugs cause absolute tranquillity of mood in any setting; in reality RCTs indicate that drugs labelled as mood stabilisers are effective, to varying degrees, in the treatment and prophylaxis of a limited number of affective syndromes in bipolar disorder, with most RCTs being limited to major depression or mania. To extrapolate from this to assume that these drugs are effective in sub-syndromal forms of these syndromes plus other affective states is to go beyond the evidence. For some patients the idea of a mood stabiliser, perceived as a drug that ‘flattens’ mood, may be off-putting. Another concern is that ‘mood stabiliser’ becomes a marketing label used by the pharmaceutical industry to promote drugs in bipolar disorder. In this context the term ‘mood stabilisers’ may lead prescribers to conclude all so-called ‘mood stabilisers’ have identical properties rather than to question the evidence for each drug in turn.

Valproate, carbamazepine and lithium have traditionally been regarded as mood stabilisers by clinicians and they fulfil the less stringent definition given above. However, so do other agents (most notably olanzapine and lamotrigine) and it is likely that future trials will enable other agents, particularly other atypical antipsychotics, to be subsumed within this definition of a mood stabiliser. Even without this expansion, it is apparent that existing mood stabilisers are a diverse range of drugs in terms of class, chemical structure, pharmacodynamic action and adverse effects. They also differ in the pole where they exert their main effect; the acute and long-term efficacy of lamotrigine is confined to the depressive pole while all other mood stabilisers are predominantly effective in the acute and long-term treatment of mania.

These pharmacological differences, the ambiguous nature of the term and the potential for it to become a marketing label all raise the question of how useful the term ‘mood stabiliser’ is. Perhaps the best that can be said in support of it is that there is some benefit to a ‘shorthand’ to refer to drugs with combined acute and long-term efficacy without the risk of worsening the course of the illness. However, given the inherent problems, the guideline avoids the term mood stabiliser and uses instead ‘antimanic agent’ or ‘antimanic medication’ to refer to antipsychotics, valproate, carbamazepine and lithium.

8.1.4. Current practice

In the UK, prior to the introduction of the atypical antipsychotics, the mainstay of drug treatment to control manic symptoms was the use of a typical antipsychotic often supplemented with a benzodiazepine. There is now increasing use of atypical antipsychotics and higher-dose valproate salts, particularly since the licensing of valproate (as valproate semisodium) with full dosage information in the summary of product characteristics (Anderson et al., 2004; Lloyd et al., 2003).

Bipolar depression remains a challenging disorder to manage. Studies show a delay of several years between onset of symptoms and diagnosis. The role of antidepressants in the acute treatment of bipolar disorder and in maintenance treatment remains controversial as discussed later in this chapter. A recent survey of prescribing in Greater Manchester, covering the period 2001–2, showed that approximately one third of outpatients seen with bipolar disorder were currently receiving an antidepressant (Anderson et al., 2004) which was significantly higher than in a survey in Newcastle covering the period 2000–2001 (Lloyd et al., 2003). In the Manchester survey, about half of antidepressant prescriptions were for a tricyclic antidepressant (TCA) despite suggestions they are associated with a greater risk of switching to mania than selective serotonin reuptake inhibitors (SSRIs). In the Newcastle survey nearly one third of patients were prescribed two or more prophylactic agents and nearly one fifth lamotrigine. In contrast in the Manchester survey only a quarter of patients received two or more prophylactic agents and there were no patients receiving lamotrigine. The differences between the two surveys may reflect the local availability of a specialist mood disorder service in Newcastle.

In the maintenance phase for relapse prevention, prophylactic medication use has increased and now about 90% of people with bipolar disorder may be on mood stabilisers, although less than a third are as monotherapy. Lithium remains the most commonly prescribed prophylactic agent (about 50%), although there may have been a slight decline in usage, matched by increased use of valproate salts and lamotrigine. Carbamazepine is less popular. Although the evidence base is limited, combinations of prophylactic agents are widely used, perhaps more by tertiary treatment centres.

It should be noted that the treatment of bipolar disorder in the US and UK differs in two main respects. First, at the present time in the US there is less emphasis on the use of antidepressants in treating bipolar depression, with lamotrigine being the favoured treatment. The fact that lamotrigine is licensed for the maintenance treatment of bipolar depression in the US, but not in Britain, may partly explain this difference (lamotrigine was licensed by the US Food and Drug Administration in June 2003). Concerns about the risk of antidepressant-induced switching to mania and cycle acceleration are another reason. The second difference is that over the last decade the US has seen a switch from using lithium to valproate as the prophylactic agent of choice, whereas lithium still remains widely used in Britain. The change in US practice has resulted from concerns about lithium toxicity and long-term side effects, concerns about lithium’s effectiveness and increased attention to the risks of rebound mania following abrupt discontinuation.

Some aspects of clinical practice have not been investigated by RCTs. For example clinicians often use combinations of prophylactic agents (including combinations of lithium, valproate and lamotrigine) but there are no published trials addressing the efficacy of these combinations. This is not to say that such combinations are ineffective, indeed there are good reasons to believe that certain combinations are effective. There is a pressing need for large-scale, well-designed RCTs to investigate combination treatments. Two studies of maintenance treatment are currently underway which are investigating the effectiveness of a combination of lithium and divalproex (valproate semisodium) versus other treatments. These studies are the Bipolar Affective Disorder Lithium/Anticonvulsant Comparative Evaluation (BALANCE) and the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD).

Another important weakness of the existing evidence is that it is almost totally restricted to the treatment of bipolar I disorder. There is very little evidence regarding the treatment of bipolar II disorder and rapid-cycling bipolar disorder.

8.1.5. Issues in the pharmacological management of patients with bipolar disorder

Concordance and compliance

Concordance refers to a process in which a healthcare professional and patient reach a shared agreement about a medication treatment plan. It involves negotiation and reflects an alliance between the two individuals, with the healthcare professional respecting the patient’s views. Compliance is a different concept in that it refers to a behaviour, namely whether a patient follows the treatment plan that is agreed with the health professional. Poor compliance may be unintentional (for example a patient forgetting medication) or intentional (for example because a patient believes they do not need medication to keep well or because they are troubled by adverse effects). Clearly compliance is more likely to follow from a consultation that is concordant.

In all chronic conditions, compliance with medication is poor. Research into compliance is hindered by methodological issues; most studies use proxy measures of compliance such as patient reports, pill counts or medicine containers with a microchip in the lid which records how often the bottle is opened. None of these measures indicates that medication was taken as prescribed. In patients with bipolar disorder, compliance can also be assessed by the number of repeat prescriptions requested and by checking blood levels of drugs such as lithium, valproate and carbamazepine. Prescribers usually prefer to use blood levels to assess compliance in clinical practice when they can.

It is generally assumed that better tolerated medications are associated with better rates of compliance. However, studies comparing atypical and conventional antipsychotics show only marginal advantages for the former in terms of compliance. Patients with early signs of hypomania and mania often stop their medication as they feel well and do not realise that they are ill. Also some patients find that hypomanic and manic symptoms are initially pleasurable and so stop medication in order to experience these symptoms. This is particularly likely to occur when patients are suffering from depression or life circumstances and they perceive that, at least initially, hypomania or mania will be a release. Other patients want to experience hypomania as it facilitates creativity, productivity at work and the possible achievement of goals. Although hypomania can have this positive side to it, mania is, by definition, associated with significant impairment of functioning.

Non-compliance often occurs when a patient perceives that the benefits of medication are outweighed by the disadvantages. However, we would argue that in such cases the patient should not stop medication unilaterally but first consult with their doctor. This allows the clinician to understand the patient’s experience of medication. It also allows the patient to receive the clinician’s views of the benefits and risks of medication, to hear about alternative treatments and ensures that a decision is made in light of the best evidence. Furthermore if medication is stopped it is usually preferable to do so gradually to prevent withdrawal or discontinuation effects, issues that are particularly relevant to antidepressants and lithium.

In some patients with bipolar disorder who would benefit from a long-term antipsychotic, a long-acting intramuscular preparation may aid compliance. However, long-acting injections, like oral tablets, should be chosen as a result of a concordant consultation. Only one atypical antipsychotic is available as a long-acting injection in such a formulation (risperidone); it is licensed for schizophrenia and other psychosis in patients tolerant to oral risperidone. In inpatients in whom compliance is a problem, consideration should be given to the use of medication in liquid formulations or in an oral dispersible formulation. Two atypical antipsychotics (olanzapine and risperidone) are available in oral dispersible formulations.

Advance statements (directives)

The use of advance statements (directives) in bipolar disorder, and psychiatry as a whole, remains low. Advance statements involve ensuring that the patient has a full understanding of the risks and benefits, and the opportunity to weigh them against the risks of the illness, and then making a plan for future care and treatment in light of this knowledge. Lithium should not be started without informed consent from the patient, particularly as early discontinuation can be harmful. Advance directives should be seriously considered after an acute episode, or where one is anticipated. Whilst these can only formally cover medication refusal, they can refer to a patient’s preferred drugs. Whilst there is little direct formal evidence for the outcomes from advance statements, they are widely advocated and evidence is emerging of the benefits of the active involvement of patients in decision making about their care (Henderson et al., 2004).

The pharmacological management of older adults with bipolar disorder

Older adults with bipolar disorder (that is those over 65 years) should be treated in the same way as other adults, but with additional consideration being given to pharmacokinetic differences plus comorbidities and co-prescribed medications both of which are more common in this age group. Mania developing for the first time in older adults is associated with a high rate of medical and neurological disease (Tohen et al., 1994; Young & Klerman, 1992) including subcortical lesions. Consequently all patients presenting with first onset of mania in later life should be carefully screening for contributing medical disorders.

In terms of pharmacokinetics, the elderly tend to have a reduced volume of distribution and reduced renal clearance. Consequently older patients usually require lower drug doses than younger patients. They may also be more susceptible to adverse reactions due to increased end-organ sensitivity. For example the elderly are at higher risk of EPS with antipsychotics, postural hypotension and falls with a variety of drugs, and gastrointestinal bleeding with SSRIs. The tolerability of lithium is also lower in the elderly, who can develop signs of neurotoxicity at plasma concentrations considered ‘therapeutic’ in the general adult population (Sproule et al., 2000). Those taking valproate may be particularly prone to sedation, tremor or gait disturbance.

Licensed indications

At the date of publication the following drugs have UK marketing authorisation for the treatment of mania: carbamazepine, lithium, olanzapine, quetiapine, risperidone, valproate (as semisodium valproate). Medications which do not have a UK marketing authorisation for the indication recommended at the date of publication are marked with an asterisk in the recommendations (*); the summary of product characteristics for current licensed indications should be checked. No drugs have UK marketing authorisation for the treatment of bipolar depression.

Regarding the use of psychotropic medication in children and adolescents under the age of 18, at the date of publication, the only drug to have a UK marketing authorisation is lithium, which is licensed for use in people aged 12 years and over. However, in 2000, the Royal College of Paediatrics and Child Health issued a policy statement on the use of unlicensed medicines, or the use of licensed medicines for unlicensed applications, in children and adolescents. This states that such use is necessary in paediatric practice and that doctors are legally allowed to prescribe unlicensed medicines where there are no suitable alternatives and where the use is justified by a responsible body of professional opinion.12

8.1.6. Clinical practice recommendation

8.1.6.1.

When treating older people with bipolar disorder, healthcare professionals should:

  • be aware of the need to use medication at lower doses
  • be alert to the increased risk of drug interactions when prescribing psychotropic medication to older adults
  • ensure that medical comorbidities have been recognised and addressed.

8.1.7. Topics covered

The following topics in the medical and pharmacological management of bipolar disorder are covered in this chapter:

  • the treatment and management of acute episodes
  • manic, hypomanic and mixed episodes
  • economic evidence for treatment of acute mania
  • depressed episodes
  • acute episodes in the context of rapid cycling
  • economic evidence for the use of electroconvulsive therapy (ECT) in all illness phases
  • rapid tranquillisation.

8.2. OVERVIEW OF CLINICAL EVIDENCE REVIEW

8.2.1. Evidence search

The review team conducted a systematic search for RCTs that assessed the efficacy of pharmacological interventions, transcranial magnetic stimulation and ECT for people with bipolar disorder at all stages of the illness. See Table 23.

Table 23. Databases searched and inclusion/exclusion criteria for clinical effectiveness of pharmacological interventions, transcranial magnetic stimulation and ECT.

Table 23

Databases searched and inclusion/exclusion criteria for clinical effectiveness of pharmacological interventions, transcranial magnetic stimulation and ECT.

8.2.2. Additional inclusion criteria

In addition to the general inclusion criteria for studies reviewed in the guideline (see Appendix 7), the GDG set inclusion criteria specific to studies of pharmacological therapy, as follows:

Dose

The GDG set minimum therapeutic doses for study medication in included studies based on the British National Formulary (BNF).

Study length

The GDG set the minimum length of studies of people in an acute phase of illness at 3 weeks.

8.2.3. Presenting the evidence

Systematic reviews of the evidence are based on the searches described above, supplemented with additional narrative as necessary. Relevant characteristics of all included studies are in Appendix 22, together with a list of excluded studies with reasons for exclusion, and full references for both included and excluded studies. These are presented for each topic covered in this chapter. To aid readability, summaries of the study characteristics are included below, followed by the critical outcomes from the evidence profiles, together with a summary of the evidence profile.

In all of these, studies are referred to by a study ID (primary author in capital letters and date of study publication, except where a study is in press or only submitted for publication, then a date is not used).

Based on the GRADE methodology outlined in chapter 3, the quality of the evidence is summarised in the evidence profiles and summary of the evidence profiles as follows:

  • high = Further research is very unlikely to change our confidence in the estimate of the effect
  • moderate = Further research is likely to have an important impact on our confidence in the estimate of the effect and may change the estimate
  • low = Further research is very likely to have an important impact on our confidence in the estimate of the effect and is likely to change the estimate
  • very low = Any estimate of effect is very uncertain.

8.3. THE PHARMACOLOGICAL TREATMENT OF ACUTE MANIC, HYPOMANIC AND MIXED EPISODES

8.3.1. Introduction

The main aim in treating mania, hypomania and mixed episodes is to achieve rapid control of symptoms. This is particularly important as mania can result in disturbed behaviour that, when extreme, can be a risk to the safety of the patient and others. More commonly mania may cause patients to act in a disinhibited manner and such behaviour may have long-term repercussions for the individual’s career and relationships. Mixed states in which manic and depressive symptoms coexist are reported to be associated with an increased risk for suicide. Various drugs possess antimanic efficacy and these are considered in the following sections.

Lithium

Lithium is held to be effective in acute mania but its onset of action is slower than with antipsychotics. In addition practical considerations mean it is not the treatment of choice in acute mania. For example, prior to starting treatment one needs to establish that baseline renal and thyroid function are normal, but in reality a patient may refuse venepuncture. Furthermore there may be a need to start immediate treatment without waiting for the result of blood tests. Also it is prudent to increase the dose of lithium gradually to minimise the risk of lithium toxicity and this will further increase the time to response.

Lithium is associated with acute and long-term adverse effects and a low therapeutic index. Furthermore abrupt discontinuation can lead to an increased risk of mania and depression over the next few months (that is, ‘rebound’ mania and depression). This phenomenon has been paid relatively little attention but is potentially one of the most serious drawbacks of lithium particularly in those who adhere poorly to treatment. It has been estimated that lithium needs to be continued for 2 years for the detrimental effects associated with abrupt discontinuation to be balanced by the benefits that accrue from its long-term effect (Goodwin, 1994). The risk of rebound appears to be prevented if lithium is withdrawn gradually and several studies suggest that withdrawal over 2 weeks is sufficient (Baldessarini et al., 1997; Faedda et al., 1993).

Adverse effects of lithium that occur at therapeutic plasma levels include polyuria, polydipsia, tiredness, fine tremor, metallic taste and diarrhoea. Toxic levels of lithium (>1.2 mmol/l) cause a range of symptoms including confusion, myoclonic jerks, cardiac arrhythmias, confusion, ataxia, dysarthria, and, as levels rise, further convulsions, coma and death. The diagnosis of lithium toxicity is made on clinical grounds with laboratory confirmation, but toxicity can sometimes occur despite an apparent normal plasma level. Long-term complications of lithium include thyroid abnormalities (particularly hypothyroidism), chronic renal impairment and diabetes insipidus (Macritchie & Young, 2004). Lithium is also associated with an increased rate of malformations particularly of the cardiovascular system.

Antipsychotics

Prior to the introduction of the atypical antipsychotics, the conventional antipsychotics were the standard treatment for mania despite a relative lack of RCTs to support their use. In recent years several atypical antipsychotics agents have been licensed to treat mania, with RCT data underpinning these decisions. Atypical antipsychotics currently licensed in the UK for the treatment of acute mania are olanzapine, risperidone and quetiapine. A major advantage of the atypical antipsychotics over conventional antipsychotics is the lower risk of EPS though this differential has largely been demonstrated in trials where the comparator is haloperidol, a high-potency conventional antipsychotic that is associated with a relatively high incidence of EPS. Some atypical antipsychotics are prolactin sparing, but others raise serum prolactin levels, which can lead to distressing symptoms, including sexual dysfunction, and may also be associated with long-term health risks, particularly if it leads to secondary hypogonadism. Some atypical antipsychotics, in particular olanzapine, are associated with a high risk of significant increase in body weight (≥7% baseline body weight). Concern has only focused on the potential for metabolic complications with the atypical antipsychotics, though high-quality data on this area is lacking.

Anticonvulsants

Valproate is available in various forms including sodium valproate, valproic acid and valproate semisodium, although only valproate semisodium has UK marketing authorisation for the treatment of manic episodes in the context of bipolar disorder. The active element in all formulations is the valproate ion. The guideline uses the generic term ‘valproate’ except when describing the agent used in a particular study, when it uses the name used by the study authors. Carbamazepine is licensed for the treatment of patients with bipolar disorder who are intolerant of lithium or in whom lithium is ineffective. A major complication of carbamazepine is that it can lower the plasma level of concurrently prescribed drugs including antipsychotics. Both carbamazepine and valproate are teratogenic, being associated with an increased risk of neural tube defects. In addition pre-natal exposure to valproate may be associated with an increased risk of developmental problems including reduced cognitive performance.

Calcium channel blockers

Calcium channel-blocking drugs (that is, diltiazem, nifedipine, nimodipine, verapamil) are licensed to treat high blood pressure and angina. Occasionally they are prescribed to treat mania. Side effects include headache, hypotension and tenderness of the gums.

Transcranial magnetic stimulation and ECT

ECT remains a valuable treatment for severe depression where there is an urgent need for a rapid response. Examples include a depressed patient with marked psychomotor retardation who is refusing fluids and food and a depressed patient with persistent and severe suicidal intent. There is a smaller body of evidence which indicates the effectiveness of ECT in acute mania. Transcranial magnetic stimulation is a recently introduced treatment but most of the research undertaken so far is limited to unipolar depression.

Benzodiazepines

Benzodiazepines are often used by clinicians, in conjunction with antimanic agents, for symptomatic control of agitation and insomnia. Sleep deprivation can lead to rapid deterioration in manic episodes and so the short-term use of hypnotics to maintain a circadian rhythm is both reasonable and advisable. Problems include the risks of tolerance, withdrawal symptoms and dependence. These risks become greater when use continues beyond 4 weeks. Other problems include sedation, ataxia and an increased risk of falls, particularly in the elderly. Comorbid anxiety disorders, including generalised anxiety disorder, are common in bipolar disorder. Their treatment should follow similar lines to such disorders that occur in isolation with the proviso that antidepressants should be used with caution due to the risks of cycle acceleration and switching to mania.

Combination therapy

The combination of antipsychotics and antimanic medication (especially lithium and valproate) has been shown to be effective in the treatment of mania. However, there may be an increased risk of adverse effects including weight gain (for example, Casey et al., 2003).

8.3.2. Review strategy

Evidence from RCTs was found for the following treatment strategies:

  • lithium, including lithium compared with placebo and with anticonvulsants and antipsychotics
  • anticonvulsants, including carbamazepine and valproate semisodium (compared with placebo and other drugs), and gabapentin
  • antipsychotics, including antipsychotics compared with placebo, lithium, valproate semisodium and haloperidol
  • miscellaneous pharmacological and non-pharmacological physical interventions, including ECT, transcranial magnetic stimulation and calcium channel blockers

No RCT evidence was found for the use of benzodiazepines in acute mania.

A summary of the evidence for the pharmacological treatment of mania, hypomanic and mixed states in children and adolescents is also provided.

Some trials appear in more than one section where the GDG felt this was appropriate, for example, trials comparing lithium with antipsychotics are included in both sections.

8.3.3. Lithium in the treatment of mania

Studies considered

Twelve trials of lithium in the treatment of mania met inclusion criteria. Excluded studies with reasons for exclusion can be seen in Appendix 22. RCTs published prior to 1973 were not picked up in searches, although several trials published before this date exist. The GDG decided not to include them because of concerns about trial methods and study populations (Schou et al., 1954, Magg, 1963, Goodwin et al., 1969, Stokes et al., 1971).

Included trials compared lithium with placebo, anticonvulsants, antipsychotics and ECT in the treatment of mania. An overview of the included studies is in Table 24, with further details available in Appendix 22. A trial comparing lithium with ECT is considered in section 1.3.6 below.

Table 24. Summary of study characteristics table for lithium in the treatment of mania.

Table 24

Summary of study characteristics table for lithium in the treatment of mania.

Lithium serum levels are referred to as mmol/litre except where the papers refer to meq/litre. The two measures are equivalent.

Summary of evidence for lithium in the treatment of mania

An overview of the results is provided in Table 25, with the full evidence profile in Appendix 23. Three placebo-controlled trials conducted since 1994 indicate that lithium is effective in acute mania. In addition several older trials support the effectiveness of lithium but were excluded from the analysis due to methodological weaknesses common to many trials of the period. The three placebo-controlled RCTs that were analysed all employed a minimum plasma level of ≥0.6 mmol/l. In contrast many UK laboratories quote a lower limit of the therapeutic range of 0.4 to 0.6 mmol/l. There does not appear to be an evidence base to support this as an effective range. In terms of active-comparator trials, one trial indicated that lithium was more effective than valproate and several trials indicated equivalent efficacy to various antipsychotics.

Table 25. Summary of evidence profile for lithium in the treatment of mania.

Table 25

Summary of evidence profile for lithium in the treatment of mania.

8.3.4. Anticonvulsants in the treatment of mania

Studies considered for review

Seven trials met the eligibility criteria set by the GDG. Excluded studies with reasons for exclusion can be seen in Appendix 22. Excluded trials included a study of topiramate, which meant that no high-quality study of this drug was available.

Included trials looked at carbamazepine, valproate semisodium and gabapentin in the treatment of mania. Placebo-controlled trials were available for carbamazepine and valproate.

An overview of the included studies is in Table 26, with further details available in Appendix 22.

Table 26. Summary of study characteristics table for anticonvulsants in the treatment of mania.

Table 26

Summary of study characteristics table for anticonvulsants in the treatment of mania.

Summary of evidence

An overview of the results is provided in Table 27, with the full evidence profile in Appendix 23. RCTs indicate that carbamazepine is superior to placebo in acute mania. Although valproate semisodium is licensed in the UK and in the USA for the treatment of acute mania, only one placebo-controlled trial of valproate semisodium fulfilled the GDG’s criteria. This showed that valproate semisodium is more effective than placebo (moderate quality evidence). Another RCT showed valproate semi-sodium to be as efficacious as olanzapine, a drug that has been shown to be superior to placebo in two placebo-controlled RCTs. However, valproate semisodium was less efficacious than lithium in another trial (see Table 25). In summary, although the evidence base supporting the efficacy of valproate in mania is relatively small, it appears that valproate is effective in the treatment of mania.

Table 27. Summary of evidence profile for anticonvulsants in the treatment of mania.

Table 27

Summary of evidence profile for anticonvulsants in the treatment of mania.

Valproate is available in several formulations including valproate semisodium, sodium valproate and valproic acid. Most trials in bipolar disorder have used valproate semisodium. There has been debate about whether other forms of valproate share the efficacy of valproate semisodium. There are no good pharmacological reasons why there should be a difference in the efficacy of these different compounds since the valproate ion is the active moiety, although there may be slight tolerability differences for valproate semisodium compared with valproic acid and sodium valproate. In previous years this debate was relevant due to the higher cost of valproate semisodium in comparison with sodium valproate and valproic acid. However, recent changes in pricing mean that there is now little difference in the price of these variations. Given the absence of a significant price differential, the licensed status of valproate semisodium, and the fact that most RCTs of valproate in bipolar disorder used valproate semisodium, it is reasonable to regard valproate semisodium as the compound of choice when valproate is required in mania, although the guideline recommendations refer to valproate throughout and none of the salts is licensed for maintenance treatment.

It cannot be assumed that all anticonvulsants are effective in mania as the results of a trial of gabapentin were classified inconclusive. No trial of topiramate met inclusion criteria.

Carbamazepine is more effective than placebo in the treatment of mania. However, carbamazepine, like valproate, is teratogenic. In addition it induces liver enzymes and so increases the breakdown of various co-prescribed drugs, reducing their effectiveness. Well-recognised examples of this include carbamazepine reducing the plasma level and effectiveness of co-prescribed warfarin, the contraceptive pill and various antipsychotics including haloperidol, olanzapine, risperidone and quetiapine. Uncommon but potentially serious adverse effects include blood dyscrasias and skin rashes. Common adverse effects include reversible blurring of vision, dizziness and ataxia, which are dose-related. Due to these side effects it is usual practice to initiate carbamazepine at a low dose and build this up gradually. This and the potential interaction with antipsychotics both reduce the practicality of carbamazepine as a treatment in acute mania hypomania and bipolar depression.

8.3.5. Antipsychotics in the treatment of mania

Studies considered for review

Twenty-four trials met the eligibility criteria set by the GDG. Excluded studies with reasons for exclusion can be seen in Appendix 22. Characteristics of included studies are in Table 28, with further details in Appendix 22.

Table 28. Summary of study characteristics table for antipsychotics in the treatment of mania.

Table 28

Summary of study characteristics table for antipsychotics in the treatment of mania.

A trial of an antipsychotic with and without ECT is considered with other ECT trials below (section 8.3.6).

Notes on evidence

Where a placebo-controlled trial compared two antipsychotics (that is, a three-arm trial: antipsychotic A, antipsychotic B, placebo) the data from the following arms were entered into the comparison with placebo: SMULEVICH2005 risperidone, MCINTYRER2005 quetiapine. This means that there are no data for haloperidol presented in this analysis.

Due to heterogeneity, KHANNA2005 was removed from the following analyses as an outlier: mania endpoint scores and leaving the study early for any reason.

Summary of evidence for anipsychotics in the treatment of mania

Note that most studies included a proportion of participants with psychotic features and some also included participants experiencing mixed episodes.

An overview of the results is provided in Table 29, with the full evidence profile in Appendix 23.

Table 29. Summary of evidence profile for antipsychotics in the treatment of mania.

Table 29

Summary of evidence profile for antipsychotics in the treatment of mania.

RCTs for several atypical antipsychotics indicate superiority over placebo (aripiprazole, olanzapine, quetiapine, risperidone, ziprasidone) and equivalent efficacy to haloperidol (aripiprazole, olanzapine, quetiapine, risperidone, ziprasidone), lithium (olanzapine) and valproate semisodium (olanzapine) in the treatment of mania. The addition of several atypical antipsychotics to either valproate or lithium in patients with partially treated mania has been shown in RCTs to be more effective than continuing treatment with valproate or lithium alone.

8.3.6. Miscellaneous pharmacological and non-pharmacological physical interventions

Studies considered for review

Four trials met the eligibility criteria set by the GDG. Excluded studies with reasons for exclusion can be seen in Appendix 22. Characteristics of included studies are in Table 30, with further details in Appendix 22.

Table 30. Summary of study characteristics for miscellaneous pharmacological and non-pharmacological physical interventions in the treatment of mania.

Table 30

Summary of study characteristics for miscellaneous pharmacological and non-pharmacological physical interventions in the treatment of mania.

Summary of evidence

An overview of the results is provided in Table 31, with the full evidence profile in Appendix 23. None of the treatments considered (that is, ECT, transcranial magnetic stimulation and calcium channel blockers) was of proven efficacy.

Table 31. Summary of evidence profile for miscellaneous pharmacological and non-pharmacological physical interventions in the treatment of mania.

Table 31

Summary of evidence profile for miscellaneous pharmacological and non-pharmacological physical interventions in the treatment of mania.

8.3.7. The treatment of mania in children and adolescents

Lithium in the treatment of mania in children and adolescents

Studies considered

Three RCTs of lithium in the treatment of mania in children and adolescents were considered (GELLER1998, KAFANTARIS2004, KOWATCH2000). KOWATCH2000 is not a double-blind trial, but was included since evidence in this area is scarce. A summary of study characteristics is in Table 32, with further details in Appendix 22.

Table 32. Summary of study characteristics for lithium in the treatment of mania in children and adolescents.

Table 32

Summary of study characteristics for lithium in the treatment of mania in children and adolescents.

Summary of evidence for lithium in the treatment of mania in children and adolescents

An overview of the results is provided in Table 33, with the full evidence profile in Appendix 23. The KAFANTARIS2004 trial is of children and adolescents given lithium for 4 weeks and then randomised to continue or discontinue treatment. In addition, 25% with psychotic features received antipsychotic treatment. The estimated effect size shows little difference in mania scores, although the 95% CIs are wide making this hard to interpret. In the trial of mania with substance use disorder, lithium was superior to placebo. Given the lack of other trials of pharmacotherapy in children and adolescents, it is necessary to cautiously extrapolate from the adult evidence base.

Table 33. Summary of study characteristics for lithium in the treatment of mania in children and adolescents.

Table 33

Summary of study characteristics for lithium in the treatment of mania in children and adolescents.

Antipsychotics in the treatment of mania in children and adolescents

Three RCTs of antipsychotics in the treatment of mania in children and adolescents were included, only one of which was a double-blind, fully randomised trial. A summary of study characteristics is in Table 34, with further details in Appendix 22.

Table 34. Summary of study characteristics for antipsychotics in the treatment of mania in children and adolescents.

Table 34

Summary of study characteristics for antipsychotics in the treatment of mania in children and adolescents.

Summary of evidence for antipsychotics in the treatment of mania in children and adolescents

An overview of the results is provided in Table 35 with the full evidence profile in Appendix 23. Data for the use of antipsychotics in the treatment of mania in children and adolescents is sparse and of very low quality. Given this it is necessary to extrapolate cautiously from the adult evidence base.

Table 35. Summary of evidence for antipsychotics in the treatment of mania in children and adolescents.

Table 35

Summary of evidence for antipsychotics in the treatment of mania in children and adolescents.

8.3.8. Clinical summary and practical considerations for the treatment of mania

In terms of pharmacological treatment, acute mania has been more studied than any other aspect of bipolar disorder. Drugs that have been shown to be superior to placebo in acute mania include carbamazepine, valproate, lithium and several atypical antipsychotics including aripiprazole, olanzapine, quetiapine, risperidone and ziprasidone. RCTs are not uniformly positive in acute mania. For example, verapamil, a calcium channel antagonist, was not more effective than placebo (JANICAK1998) and gabapentin in combination with valproate or lithium was not more effective than valproate or lithium alone (PANDE2000).

Given the existence of several efficacious agents in mania, comparative efficacy and tolerability should be considered. In terms of comparative efficacy there are few head-to-head trials of single agents in mania. Most such trials (for example, olanzapine versus lithium, olanzapine versus valproate semisodium, and olanzapine versus haloperidol) do not reveal convincing evidence for differential efficacy or tolerability though an exception is a single trial in which lithium was superior to valproate. In view of this non-RCT data, practical issues need to be considered when choosing between monotherapies for mania.

Carbamazepine

Carbamazepine is not a simple drug to use; side-effects include nausea, blurring of vision and unsteadiness. To minimise these it is necessary to gradually increase the dose. Carbamazepine can also cause blood dyscrasias and skin rashes and is teratogenic. It reduces the plasma level of many co-prescribed medications, including antipsychotics, by inducing hepatic enzymes. As a result it cannot be recommended as a first-line treatment for acute mania.

Lithium

Lithium is associated with a number of problems, including long-term risk of chronic renal impairment and hypothyroidism. Lithium can impair renal function in several ways. It inhibits the effect of antidiuretic hormone on the distal tubule and so impairs water reabsorption. This is relatively common, manifests as polyuria and polydipsia and, when severe, can lead to frank diabetes inspidius which occurs in about 10% of patients on long-term lithium treatment (Bendz & Aurell, 1992). Complications include dehydration and an increased risk of toxicity from drugs with a low therapeutic index including lithium. The effects of lithium on renal concentrating ability are usually reversible on stopping treatment. In a small minority of patients, long-term lithium treatment can lead to a reduction in glomerular filtration and a gradual rise in serum creatinine. This may reflect a chronic interstitial or tubulointerstitial nephropathy which can be irreversible, even on stopping lithium. The exact prevalence of this complication is uncertain but it appears to be the exception rather than the rule and progression to end-stage renal failure appears very rare.

Pre- and regular (every 6 months) post-treatment checks of renal and thyroid function are therefore essential. The low therapeutic index of lithium requires that after treatment is commenced the dose is gradually built up to achieve an effective plasma level. In order to achieve satisfactory plasma levels, plasma levels need to be checked one week after initiation (and after every dose change) then every 3 months. Plasma levels of lithium are increased by several medications including diuretics and non-steroidal anti-inflammatory drugs. Lithium is also teratogenic. It must be withdrawn gradually to minimise the occurrence of rebound mania or depression. All these considerations mean that lithium is a rather impractical agent to commence de novo for the treatment of mania. However, in patients whose bipolar illness has previously been controlled by lithium a recurrence of mania associated with a low serum lithium level, perhaps secondary to poor concordance, may be treated by increasing the dose of lithium particularly if symptoms are relatively mild.

Valproate

Valproate is generally well tolerated. More frequent side effects include gastric irritation, nausea, ataxia and tremor. These side effects are often dose related and as the effective dose varies between patients it is usual practice to increase the dose gradually according to clinical response. Plasma level monitoring is not necessary unless there is evidence of ineffectiveness, poor adherence or toxicity, though pre-treatment and post-treatment checks of full blood count and hepatic function are advisable. Valproate can elevate serum levels of carbamazepine and lamotrigine. Its most serious side effects are its association with structural teratogenicity and mental impairment in the newborn when prescribed during pregnancy. Consequently valproate cannot be recommended as a first-line agent in the treatment of mania in women of child-bearing age (unless they are using a highly reliable from of contraception such as an intra-uterine device). In other women and in men valproate can be considered, alongside antipsychotics, as a first-line agent for the treatment of mania.

Polycystic ovary syndrome is characterised by excess androgen levels and ovulatory dysfunction. Several studies have reported a higher prevalence in women with epilepsy and bipolar disorder. Debate has centred on whether this reflect an association with the underlying disorder (that is, epilepsy or bipolar disorder) or the result of antiepileptic drugs, in particular valproate (Rasgon, 2004). Current research is not sufficient to provide a definitive answer but suggests that both pathways may be relevant. In particular a recent small study of reproductive function in women with bipolar disorder found a high rate of menstrual disturbances that in many cases preceded the diagnosis and treatment for the disorder (Rasgon et al., 2005). Furthermore, treatment with valproate was associated with a higher risk of development of menstrual abnormalities than other treatments and valproate was also associated with an increase in testosterone levels over time.

Antipsychotics

Antipsychotics can either be started at a therapeutic antimanic dose on day one of treatment or (in the case of quetiapine and risperidone where titration is necessary to minimise side-effects) rapidly titrated to allow a therapeutic dose to be reached within a few days which is shorter than is the case with titration of lithium and valproate. There is no evidence of teratogenicity with antipsychotics, though one has to be cautious particularly with newly introduced agents such as aripiprazole where experience is more limited. RCTs indicate that the addition of several atypical antipsychotics to pre-existing treatment with either valproate or lithium in patients with partially treated mania is more effective than continuing treatment with valproate or lithium alone. These factors make antipsychotics suitable as first-line treatments for mania.

In choosing between antipsychotics, atypical antipsychotics are preferred to conventional antipsychotics due to the existence of a greater number of supportive RCTs and the lower risk of EPS, at least when the comparator is haloperidol. Several atypical antipsychotics are associated with a low risk of elevation of serum prolactin, another advantage compared with haloperidol. This advantage is not shared by all atypical antipsychotics; both risperidone and amisulpride are commonly associated with elevation of serum prolactin. The atypical antipsychotics licensed for the treatment of mania in the UK are olanzapine, quetiapine and risperidone.

Atypical antipsychotics are not without their drawbacks. All can cause significant weight gain in some patients. The risk is greatest with olanzapine which is associated with about 30% of patients gaining ≥7% of body weight in short-term trials (Lieberman et al., 2005). There is also concern about the potential for metabolic disturbance (elevation of serum glucose and lipids) with the atypical antipsychotics. The absolute and relative risk of metabolic disturbance with atypical antipsychotics is unclear due to the lack of long-term RCTs with metabolic data. What metabolic data is available is often not present for all trial participants and sometimes consists of measures that are not ideal for example, non-fasting serum glucose levels when fasting glucose levels are more relevant. Non-prospective data, which has many methodological limitations, suggests that as a class, atypical antipsychotics are associated with a greater risk of diabetes than conventional antipsychotics. Some evidence suggests that within the atypical class the risk of serum glucose elevation (as opposed to frank diabetes) is greater with olanzapine. Olanzapine may also have more potential to elevate serum cholesterol and triglycerides than other atypical antipsychotics.

Despite controversy about the relative risk of metabolic disturbance with different agents, there is a consensus that patients prescribed atypical antipsychotics should have pre- and post-treatment checks of plasma glucose, cholesterol and triglycerides. At the present time, given the limitations of the evidence, it is not possible to make recommendations regarding choice between individual atypical antipsychotics licensed for the treatment of acute mania on the basis of safety and tolerability other than to state that weight gain is more common with olanzapine and raised prolactin is more common with risperidone. Should the uncertainty about the risk of metabolic disturbance with different atypical antipsychotics be resolved and robust evidence emerge of a differential effect, then this situation may change. This would also be the case should evidence emerge of a differential efficacy.

Summary

In summary, for women of child-bearing potential not currently taking medication, the treatment of choice for acute mania is an atypical antipsychotic. In other women or those using a highly reliable form of contraception such as an intra-uterine device, and in men, the treatment of choice is either an atypical antipsychotic or valproate. In a patient with mania already taking lithium (with a suboptimal plasma level) or taking valproate, one can consider increasing the dose of the existing antimanic agent. However if symptoms are severe, such dose increases will often need to be done in conjunction with starting an antipsychotic. For patients already taking lithium or valproate where it is not appropriate to increase the dose, treatment options include adding either lithium or valproate or adding an antipsychotic.

8.4. ECONOMIC EVIDENCE FOR THE TREATMENT OF ACUTE MANIC EPISODES

8.4.1. Review overview

The literature search identified six economic studies that assessed the cost-effectiveness of pharmacological agents for the treatment of acute manic episodes associated with bipolar disorder. Four of the studies (TOHEN1999A; ZAJECKA2002; TOHEN2002A; REVICKI2005) were conducted alongside RCTs, whereas the other two analyses (KECK1996; BRIDLE2004) were based on decision-analytic modelling.

Full references, characteristics and results of all studies included in the economic review are presented in the form of evidence tables in Appendix 14.

8.4.2. Olanzapine versus placebo

One study evaluated the improvements in clinical symptoms and health-related quality of life (HRQOL), as well as direct medical cost implications from a third party payer’s perspective (insurance), associated with olanzapine treatment in patients diagnosed with bipolar I disorder experiencing a manic or mixed episode (TOHEN1999A). The study was undertaken in the US. Improvements in clinical symptoms and HRQOL were measured using the YMRS and the Medical Outcomes Study 36-item short-form Health Survey (SF-36) respectively. The time horizon of the study was one year (52 weeks), comprising a 3-week acute phase in which patients were randomised to receive either olanzapine or placebo, followed by a 49-week open-label extension; during the second phase, use of lithium and fluoxetine was permitted for patients experiencing breakthrough symptoms. Direct costs were estimated only for the 49-week open-label extension, and were compared with respective costs incurred over 12 months prior to the study.

In both the acute phase and the open-label extension, olanzapine was associated with overall improvement in clinical and HRQOL outcomes. Compared with costs incurred 12 months prior to the study, olanzapine therapy resulted in monthly savings of almost $900 per patient (1995 prices) during the 49-week open-label extension, largely driven by inpatient cost reductions over that period. The authors concluded that olanzapine had a significant impact on the treatment of mania and could be considered a cost-effective treatment option for use in patients with bipolar I disorder. However, the main drawback of the economic analysis was that it utilised findings of a within-group comparison (before-after study), which may have potentially introduced systematic bias and confounding in the analysis.

8.4.3. Valproate semisodium versus lithium

REVICKI2005 compared the effectiveness and medical costs between valproate semi-sodium and lithium for the treatment of patients with bipolar I disorder hospitalised for an acute manic or mixed episode. The analysis was based on a pragmatic, multi-centre, open-label RCT carried out in the US. The trial was performed under conditions of routine psychiatric care without blinding; patients entering the study continued to receive their usual medical care. Other medications for treating bipolar disorder were administered as clinically necessary, although concurrent treatment with both study medications was strongly avoided, unless clinically required. The time horizon of the analysis was 12 months, comprising an acute phase, extending from randomisation to hospital discharge, and a maintenance phase, starting from the day after discharge and lasting until the total 12-month duration of the trial. The primary clinical outcomes were the mean number of months without DSM-IV level manic or depressive symptoms, as well as the patient functioning and well-being measured using the Mental Component Summary (MCS) and Physical Component Summary (PCS) of the SF-36. The perspective of the analysis was that of the health service, therefore only direct medical costs were estimated. The results of the analysis demonstrated that there were no significant differences between valproate semisodium and lithium in terms of effectiveness and medical costs. Total medical costs per patient were $28,911 for valproate semisodium and $30,666 for lithium (p = 0.693, 1997 prices).

KECK1996 assessed the one-year direct medical costs associated with valproate semisodium or lithium monotherapy for the management of patients with bipolar I disorder hospitalised for a manic or mixed episode, or rapid cycling. The study, carried out in the US, was based on a deterministic decision-analytic model. The model incorporated rates of response to treatment (defined as >50% improvement in clinical symptoms), length of initial hospitalisation, relapse rates and rates of adverse events, and subsequently estimated the total costs incurred over a year of acute and long-term treatment associated with each of the agents compared. Total costs consisted of costs of medication, initial hospitalisation, long-term treatment and treatment of relapses, including laboratory test costs and costs of treating side effects. Initial response rates to treatment were calculated as a weighted mean of response rates reported in studies identified in a literature review; other effectiveness and resource-use parameters were based on published and unpublished data and an expert panel opinion.

The total annual cost per patient was estimated to be $39,643 for valproate semi-sodium, and $43,400 for lithium (1994 prices). Overall savings associated with valproate semisodium, despite its higher acquisition costs, were attributed to the lower cost of initial hospitalisation, resulting from a shorter length of stay (14.3 days with valproate semisodium versus 18.4 days with lithium). As expected, cost results were highly sensitive to the length of initial hospitalisation: a 30% increase in the length of stay for valproate semisodium or a 37% decrease in the length of stay for lithium would lead to total costs of treatment being equal for the two drugs. Although no effectiveness measure was explicitly used in the economic analysis, it can be inferred from the base-case results that, since valproate semisodium was associated with a higher response rate compared with lithium (0.59 versus 0.49 respectively), and the two drugs were considered to have equivalent relapse rates (0.56), valproate semisodium was likely to be a more cost-effective option than lithium, as it was associated with lower costs and better outcomes.

8.4.4. Valproate semisodium versus olanzapine

A 12-week, double-blind, multi-centre RCT compared the clinical, HRQOL and economic outcomes of valproate semisodium and olanzapine for the treatment of acute mania in patients with bipolar I disorder (ZAJECKA2002). The study was conducted in the US. Clinical symptoms were measured using the Mania Rating Scale (MRS) from the Schedule for Affective Disorders and Schizophrenia (SADS) Change Version, and the Hamilton Rating Scale for Depression (HRSD). HRQOL was measured using the Quality of Life Enjoyment and Satisfaction Questionnaire (Q-LES-Q) and the number of days with restricted activity. Economic outcomes consisted of direct medical costs and a health service perspective was adopted. The results of the analysis showed that there were no significant differences between the two agents in terms of clinical, HRQOL and economic outcomes over the 12-week period. Valproate semi-sodium was associated with significantly lower outpatient costs compared with olanzapine; nevertheless, total direct medical costs related to the two agents were similar (valproate semisodium $13,703 per patient, olanzapine $15,180 per patient, p = 0.88).

Another study comparing valproate semisodium and olanzapine in terms of direct treatment costs and various clinical outcomes (including YMRS scores, remission rates, median times to remission etc) was also performed in the US (TOHEN2002A). The economic analysis was conducted alongside a 47-week, double-blind, multi-centre RCT. The study population consisted of patients with bipolar I disorder hospitalised for an acute manic or mixed episode and the study adopted a health service perspective. Overall, clinical outcomes were reported to be favouring olanzapine, while total direct medical costs were found to be similar for the two pharmacological agents compared (valproate semisodium $15,801 per patient, olanzapine $14,967 per patient, p > 0.05).

8.4.5. Assessment of newer drugs for the treatment of mania (quetiapine, olanzapine, valproate semisodium)

BRIDLE2004 was the only study included in the review that was carried out in the UK; therefore, it is discussed in more detail. The economic analysis was conducted as part of an NHS Health Technology Assessment; the objective of the report was to evaluate the clinical and cost-effectiveness of quetiapine, olanzapine and valproate semisodium in the treatment of mania associated with bipolar disorder. For this purpose a systematic review of the literature was undertaken. Based on the results of the review, the authors concluded that, regarding clinical effectiveness, all three agents were superior to placebo in reducing manic symptoms. In comparison to lithium, no significant differences were found for any of the three drugs in terms of clinical effectiveness. The clinical effectiveness of quetiapine and olanzapine was found to be comparable to that of haloperidol; however, olanzapine resulted in fewer negative outcomes regarding HRQOL relative to haloperidol. All three agents were associated with side effects.

Regarding cost-effectiveness, no safe conclusions could be drawn from the findings of the literature review, due to a number of important limitations characterising already published studies, such as lack of direct comparisons between the drugs under evaluation, as well as differences between the studies regarding the study designs and assumptions used. Moreover, the economic studies included in the review had been carried out in the US; consequently their results referred to a setting that was likely to differ substantially compared with the UK healthcare setting in terms of resource utilisation and costs. Therefore, a new probabilistic economic model was developed, in order to overcome these limitations and assist the decision-making process in the context of the NHS.

The model estimated total medical costs and benefits resulting from treating acute manic episodes associated with bipolar disorder. Effectiveness data were derived from a meta-analysis that incorporated seven of the studies included in the systematic review, based on additional selection criteria. The availability of effectiveness data from the systematic review determined the choice of treatments included in the economic analysis; the following pharmacological agents were thus analysed: quetiapine, olanzapine, valproate semisodium, haloperidol and lithium. The primary measure of benefit used in the economic analysis was the number of responders to treatment; response was defined as ≥50% improvement in patient’s manic symptoms, expressed in changes in YMRS scores. The time horizon was equal to 3 weeks in the base-case analysis, to reflect the most commonly reported length of follow-up for which effectiveness data were provided in the clinical trials. Estimated costs, expressed in 2001–2002 prices, included direct medical costs from the NHS perspective; these consisted of hospitalisation and drug-acquisition costs, as well as costs of diagnostic and laboratory tests required for the monitoring of patients. Costs of treating adverse events were not included in the analysis, due to lack of relevant data reported in the literature. However, the authors’ opinion was that the majority of adverse events associated with the agents compared were unlikely to have significant resource implications in the 3-week time horizon of the model. Hospitalisation costs were estimated to be the same for all drug treatment options, as all patients were assumed to be hospitalised at the start of the model and to remain hospitalised for the total 3-week period, regardless of response to treatment.

The base-case results of the analysis showed that mean response rates for olanzapine (0.54) and haloperidol (0.52) were higher than for lithium (0.50), quetiapine (0.47), and valproate semisodium (0.45). Haloperidol had the lowest mean total costs per patient (£3,047) in comparison to valproate semisodium (£3,139), olanzapine (£3,161), lithium (£3,162), and quetiapine (£3,165). In terms of cost-effectiveness, lithium, valproate semisodium, and quetiapine were dominated by haloperidol, that is, they were both less effective and more costly, and therefore were excluded from further analysis. Compared with haloperidol, olanzapine was both more effective and more costly, demonstrating an incremental cost-effectiveness ratio (ICER) equal to £7,179 per additional responder. The authors concluded that if decision makers were prepared to pay less than £7,179 per additional responder, then haloperidol was the optimal decision; however, if they were prepared to pay at least £7,179 per additional responder, then olanzapine became the most cost-effective option.

One-way sensitivity analyses were undertaken, to explore whether using alternative model assumptions would have an impact on the base-case results. In terms of dominance, results were robust to the majority of alternative assumptions tested (such as discharge of non-responders at a later time than responders, treatment of non-responders with second and third-line pharmacological therapies, reductions in diagnostic and laboratory costs, inclusion of effectiveness data for patients initially excluded from analysis according to a modified intention-to-treat approach, and inclusion of treatment costs for extrapyramid symptoms due to haloperidol use). The ICER of olanzapine compared with haloperidol ranged between £1,236 (when longer hospitalisation was assumed for non-responders) and £7,165 (when second and third-line treatment was assumed for non-responders, with a low cost estimate used) per additional responder. Base-case results were sensitive only to the entire exclusion of diagnostic and laboratory costs from the analysis, which constituted a rather extreme scenario. In this case lithium became the cheapest option. Valproate semisodium and quetiapine were dominated by both lithium and haloperidol. The ICERs of olanzapine and haloperidol relative to lithium were £7,109 and £103 per additional responder respectively in this scenario.

In order to incorporate the uncertainty within the model related to both cost estimates and response rates, a probabilistic analysis was carried out using Monte Carlo simulation techniques. Instead of mean values, costs and response rates were assigned a range of values based on probability distributions; results were presented in the form of Cost-Effectiveness Acceptability Curves (CEACs), which demonstrated the probability of each agent being the most cost-effective option after taking into account the underlying uncertainty in model input parameters. Probabilities were estimated for a range of potential maximum cost values that the health service was willing to pay in order to achieve an additional unit of benefit (that is, an additional responder). The CEACs demonstrated that, for a willingness to pay (WTP) equal to £20,000 per additional responder, the probabilities of treatments under assessment being cost-effective were: olanzapine 0.44, haloperidol 0.37, lithium 0.16, quetiapine 0.029 and valproate semisodium 0.01. The probability that olanzapine was cost-effective increased as the WTP increased: for a maximum WTP £10,000 per additional responder this probability was 0.42, increasing to 0.45 if the maximum WTP rose to £40,000. At the extreme of a zero value placed on the WTP for an additional responder, haloperidol was proved to be the most cost-effective option (with probability equalling 1).

Although the above analysis was well conducted, it was characterised by a number of limitations, as acknowledged by the authors. The effectiveness rates utilised in the model were based on indirect evidence; this fact may have introduced bias in the analysis. Moreover, systematic searches for all possible comparators were not undertaken, which means that there may have been additional indirect evidence that was not incorporated in the meta-analysis that provided the effectiveness rates for the economic model. In addition, no combination therapies or 2nd and 3rd line treatments were considered in the analysis, due to lack of available data. Finally, the results of the meta-analysis demonstrated that the effectiveness rates were similar between the drugs assessed (as shown by overlapping 95% CIs around mean response rates), which means that their relative cost-effectiveness in practice depended on differences in associated costs only. Cost differences between pharmacological agents were found to be very small (varying between £3,047 and £3,165), and were attributed exclusively to differences in acquisition and monitoring costs, as no evidence was available to suggest a difference in the length of hospital stay relating to each pharmacological treatment examined. However, potential differences between drugs in terms of associated length of hospitalisation may affect significantly their relative cost-effectiveness, as inpatient care is the major driver of total medical costs associated with treatment of acute mania.

In terms of model structure and assumptions used, the short time frame of the analysis limits the value of the results, as potential differences between pharmacological agents in terms of overall length of hospitalisation – beyond the time frame of 3-weeks may strongly affect their relative cost effectiveness. On the other hand, it was assumed that all patients (both responders and non-responders) were hospitalised over the 3-week time frame of the model; if some of the agents are actually associated with lower length of hospital stay for responders, this assumption has underestimated their cost-effectiveness, probably in a significant degree.

Finally, the use of response rates as the primary outcome measure instead of a generic measure of outcome such as Quality Adjusted Life Years (QALYs) due to lack of appropriate data, and the exclusion of costs and quality of life aspects of adverse events from the analysis due to the same reason, were also acknowledged by the authors as further limitations of their analysis.

8.4.6. Summary of economic evidence

Overall, the existing economic evidence suggests that it is likely that there is no significant difference in the relative cost-effectiveness between the various pharmacological agents used for the management of acute manic episodes associated with bipolar disorder. REVICKI2005 concluded that there were no significant differences between valproate semisodium and lithium in terms of effectiveness and costs. KECK1996 stated that valproate semisodium was likely to be a more cost-effective option than lithium; however, the results of this analysis were highly sensitive to the length of initial hospitalisation for the treatment of mania. ZAJECKA2002 and TOHEN2002A demonstrated that valproate semisodium and olanzapine were associated with similar overall treatment costs; the findings of the two studies differed in terms of effectiveness: the first study concluded that the two treatments were equally effective, while the second one reported a higher rate of effectiveness for olanzapine.

The BRIDLE2004 economic analysis, the only one that utilised UK cost data, showed that lithium, valproate semisodium and quetiapine were more costly and less effective than haloperidol. Olanzapine was found to be more effective and more costly than haloperidol, with an ICER equal to £7,179 per additional responder. However, the results of the meta-analysis conducted as part of the same study suggested that treatments did not differ significantly in terms of clinical effectiveness. Moreover, the differences in total medical costs between the drugs assessed were shown to be very small. This was expected, as hospitalisation costs, the major component of costs associated with the treatment of acute mania, were estimated to be equal for all agents, due to lack of evidence of a differential effect of treatments on the length of hospitalisation following an acute manic episode.

Based on the results of the literature review it can be concluded that if all medications for the treatment of acute mania have a similar effect on the length of hospital stay, then their relative cost-effectiveness is likely to be comparable. Future research, exploring whether pharmacological treatments used for the management of acute mania in patients with bipolar disorder have a differential impact on the length of hospital stay, may determine their relative cost-effectiveness with more certainty.

8.4.7. Clinical practice recommendations for the treatment of acute mania

General advice

8.4.7.1.

As soon as practicable after initial presentation of a patient with bipolar disorder, healthcare professionals should:

  • establish the patient’s smoking status and alcohol use
  • perform thyroid, liver and renal function tests, blood pressure, and measure full blood count, blood glucose, lipid profile
  • measure weight and height
  • consider EEG, CT scan or MRI scan if an organic aetiology or a relevant comorbidity is suspected
  • consider drug screening, chest X-ray and ECG if suggested by the history or clinical picture.
8.4.7.2.

Contraception, and the risks of pregnancy (including the risks of relapse, damage to the fetus, and the risks associated with stopping or changing medication) should be discussed with all women of child-bearing potential, regardless of whether they are currently planning a pregnancy. They should be encouraged to discuss pregnancy plans with their doctor.

8.4.7.3.

To help reduce the negative consequences of manic symptoms, healthcare professionals should consider advising patients to avoid excessive stimulation, to engage in calming activities, to delay important decisions, and to establish a structured routine (including a regular sleep pattern) in which the level of activity is reduced.

8.4.7.4.

If a patient is taking an antidepressant at the onset of an acute manic episode, the antidepressant should be stopped. This may be done abruptly or gradually, depending on the patient’s current clinical need and previous experience of discontinuation/withdrawal symptoms, and the risk of discontinuation/withdrawal symptoms of the antidepressant in question.

8.4.7.5.

People experiencing a manic episode, or severe depressive symptoms, should normally be seen again within a week of their first assessment, and then regularly at appropriate intervals, for example, every 2–4 weeks in the first 3 months and less often after that, if response is good.

Pharmacological management of acute mania for those not currently taking antimanic medication

8.4.7.6.

If a patient develops acute mania when not taking antimanic medication, treatment options include starting an antipsychotic, valproate or lithium. When making the choice, prescribers should take into account preferences for future prophylactic use, the side-effect profile, and consider:

  • prescribing an antipsychotic if there are severe manic symptoms or marked behavioural disturbance as part of the syndrome of mania
  • prescribing valproate or lithium if symptoms have responded to these drugs before, and the person has shown good compliance
  • avoiding valproate in women of child-bearing potential
  • using lithium only if symptoms are not severe because it has a slower onset of action than antipsychotics and valproate.
8.4.7.7.

In the initial management of acute behavioural disturbance or agitation, the short-term use of a benzodiazepine (such as lorazepam*) should be considered in addition to the antimanic agent.

8.4.7.8.

If treating acute mania with antipsychotics, olanzapine, quetiapine or risperidone should normally be used, and the following should be taken into account:

  • individual risk factors regarding side effects (such as the risk of diabetes)
  • the need to initiate treatment at the lower end of the therapeutic dose range recommended in the summary of product characteristics and titrate according to response
  • that if an antipsychotic proves ineffective, augmenting it with valproate or lithium should be considered
  • that older people are at greater risk of sudden onset of depressive symptoms after recovery from a manic episode.
8.4.7.9.

Carbamazepine* should not be routinely used for treating acute mania, and gabapentin*, lamotrigine*, topiramate* are not recommended.

Pharmacological management of acute mania for those currently taking antimanic medication

8.4.7.10.

If a patient already taking an antipsychotic experiences a manic episode, the dose should be checked and increased if necesssary. If there are no signs of improvement, the addition of lithium or valproate should be considered.

8.4.7.11.

If a patient, already taking lithium experiences a manic episode, plasma lithium levels should be checked. If levels are suboptimal (that is, below 0.8 mmol per litre), the dose should normally be increased to a maximum blood level of 1.0 mmol per litre. If the response is not adequate, augmenting lithium with an antipsychotic should be considered.

8.4.7.12.

If a patient, already taking valproate* experiences a manic episode, the dose should be increased until:

  • symptoms start to improve, or
  • side effects limit further dose increase.
    If there are no signs of improvement, the addition of olanzapine, quetiapine, or risperidone should be considered. Patients on doses higher than 45 mg per kilogram should be monitored carefully.
8.4.7.13.

For patients who present with severe mania when already taking lithium or valproate*, adding an antipsychotic should be considered at the same time as gradually increasing the dose of lithium or valproate.

8.4.7.14.

For patients who present with mania when already taking carbamazepine, the dose should not routinely be increased. Adding an antipsychotic should be considered, depending on the severity of mania and the current dose of carbamazepine. Interactions with other medication are common with carbamazepine, and doses should be adjusted as necessary.

The management of acute mixed episodes
8.4.7.15.

Prescribers should consider treating patients with an acute mixed episode as if they had an acute manic episode, and avoid prescribing an antidepressant.

8.4.7.16.

Prescribers should monitor patients with an acute mixed episode closely (at least weekly), particularly for suicide risk.

The management of acute manic episodes in children and adolescents
8.4.7.17.

When prescribing medication for children or adolescents with an acute manic episode, the recommendations for adults with bipolar disorder should be followed except drugs should be initiated at lower doses. In addition, at initial presentation:

  • height and weight should be checked (and monitored regularly afterwards) – for example, monthly for 6 months then every 6 months
  • prolactin levels should be measured
  • when considering an antipsychotic, the risk of increased prolactin levels with risperidone* and weight gain with olanzapine* should be considered
  • where there is an inadequate response to an antipsychotic, adding lithium or valproate* should be considered. Valproate should normally be avoided in girls and young women because of risks during pregnancy and because of the risk of polycystic ovary syndrome.

8.5. THE PHARMACOLOGICAL TREATMENT OF ACUTE EPISODES IN THE CONTEXT OF RAPID CYCLING

8.5.1. Introduction

Rapid cycling occurs in approximately 20% of patients with bipolar disorder. It is usually defined as the occurrence of four or more distinct mood episodes (that is, depression, mania, hypomania or mixed states) within one year. Some patients who would be regarded clinically as rapid cycling may not meet the somewhat arbitrary criteria in a standard diagnostic manual such as DSM-IV. For example in DSM-IV two consecutive mood episodes in the same direction need to be interrupted by 8 weeks of euthymia, or substantial improvement, to qualify towards rapid cycling.

Since rapid cycling was identified by researchers in the 1970s, further subtypes have been recognised including ‘ultra rapid cycling’ in which episodes last less than 24 hours and ‘ultra ultra rapid cycling’ in which several switches of mood can occur in a 24-hour period. Rapid cycling is a course specifier rather than a specific type of bipolar disorder i.e. any person with bipolar disorder can develop a rapid cycling pattern but then revert to a non-rapid cycling course. Rapid cycling is more prevalent in bipolar II subtype, and those with thyroid abnormalities (Calabrese et al., 2001; Kupka et al., 2003). Data from a cohort of 456 bipolar probands suggests rapid cycling is not more common in females than males (Baldassano et al., 2005), as had been indicated by previous studies in smaller samples.

Key elements in the treatment of rapid cycling are to eliminate any maintaining factors, particularly substance misuse and antidepressant use. With regard to the treatment of mania in rapid cycling, secondary analyses from several placebo-controlled studies of mania including valproate semisodium, olanzapine, and risperidone (Nemeroff, 2000; Bowden et al., 1994, Sachs et al., 2002) have demonstrated no difference between those with and those without a rapid-cycling course. This suggests that the short-term efficacy of other antimanic agents may not differ according to the cycling status of patients.

The predominant mood state in most rapid-cycling patients is depression. Lamotrigine may be useful in rapid-cycling bipolar disorder. The only RCT in the maintenance treatment of rapid-cycling bipolar disorder (Calabrese et al., 2000) demonstrated that lamotrigine was superior to placebo in time to discontinuation for any reason, and in percentage of patients completing the trial (see Chapter 9).

8.5.2. Clinical practice recommendations

8.5.2.1.

Acute episodes in patients with rapid-cycling bipolar disorder should normally be managed in secondary mental health services. Treatment should be as for manic and depressive episodes, but in addition healthcare professionals should do the following.

  • Review the patient’s previous treatments for bipolar disorder, and consider a further trial of any that were inadequately delivered or adhered to.
  • Focus on optimising long-term treatment rather than on treating individual episodes and symptoms; trials of medication should usually last at least 6 months.
  • Adopt a psychoeducational approach and encourage patients to keep a regular mood diary to monitor changes in severity and frequency of symptoms, and the impact of interventions.

8.6. THE PHARMACOLOGICAL TREATMENT OF ACUTE DEPRESSED EPISODES

8.6.1. Introduction

The treatment of bipolar depression has tended to be ignored, with far more interest focusing on mania. This is paradoxical as the average patient with bipolar disorder experiences more depressive than manic episodes, on average depressive episodes last longer than manic ones, the suicide risk is higher in bipolar depression than in mania and the number of depressive episodes correlates more strongly than the number of manic episodes with social functioning when patients are in between episodes.

The main aims of the treatment of bipolar depression are resolution of symptoms and return to a premorbid level of social functioning. What differentiates the treatment of unipolar and bipolar depression is that in the latter one has to ensure that treatment does not cause switching to mania/hypomania or acceleration of cycling. There is controversy among guidelines as to what appropriate first-line treatments for bipolar depression are. The first-line treatments listed in the American Psychiatric Association bipolar guidelines are lithium and lamotrigine. In contrast there has been a tradition in Europe to use antidepressants as first-line treatments. The situation has recently become more complex, with preliminary data indicating that quetiapine, at a moderate dose, may be an effective treatment for bipolar depression.

8.6.2. The role of antidepressants in bipolar disorder

There are widely differing views on the roles of antidepressants in bipolar disorder. On one hand, there is the view that antidepressants have significant risks and limited benefits in bipolar disorder and as a result their use should be restricted (for example, Ghaemi et al., 2003). On the other, there is the view that the risks have been somewhat exaggerated and that antidepressants have a useful role (for example, Moller & Grunze, 2000; Altshuler et al., 2004). In more detail these opposing arguments centre on five issues, though some of these overlap.

The risk of antidepressant-induced switching

Opponents of antidepressants cite high switch rates to mania/hypomania. The opposing argument is that switch rates are fewer with SSRIs than TCAs and can be reduced further by concomitant use of antimanic medication. The subject of antidepressant-induced switching to mania/hypomania is therefore controversial. This largely reflects the paucity of randomised prospective studies investigating the phenomenon. Most authorities regard switching as a proven risk, but a minority dispute this and regard ‘switching’ as representing the natural course of the bipolar disorder. Among those who accept the reality of antidepressant-induced switching, debate centres on the magnitude of the risk and to what extent treatment and patient variables influence it.

Estimates of the incidence of switching vary greatly. In a detailed review, Goldberg and Truman (2003) concluded that in bipolar disorder switch rates associated with anti-depressant treatment varied from 20% to 40% and that while mood stabilisers may reduce the risk they do not totally eradicate it. In contrast, in a recent meta-analysis of 12 RCTs, in which patients with bipolar disorder were randomised to receive an antidepressant or placebo, the switch rate did not differ significantly between the antidepressant and placebo groups (Gijsman et al., 2004). This finding may reflect the fact that 75% of patients were receiving a mood stabiliser which may reduce the risk of switching.

Individual studies differ in their estimates of the degree of protection that anti-manic medication offers. A small naturalistic study (Henry et al., 2001) found that mood switches were less frequent in patients receiving lithium (15%, 4/26) than in patients not treated with lithium (44%, 8/18; p = 0.04) but that anticonvulsants did not protect against switching. In contrast, several studies have reported a general protective effect from antimanic agents including carbamazepine and valproate (Boerlin et al., 1998; Bottlender et al., 2001).

Antidepressant-induced switching has been reported in both bipolar I and II disorder and also in children (Briscoe et al., 1995). A pharmacoepidemiological study from the United States, which assessed switching rates in children, adolescents and young adults with an anxiety or non-bipolar mood disorder, concluded that patient age affected the risk of antidepressant-associated manic switching (Martin et al., 2004). In this study, treatment with antidepressants was associated with the highest switch rate among children aged 10 to 14 years.

Other factors reported to be associated with higher risk of switching include a high score on the hyperthymia component of the Semistructured Affective Temperament Interview (p = 0.008) (Henry et al., 2001) and a history of a greater number of past manic episodes (p < 0.023) (Boerlin et al., 1998). Goldberg and Truman (2003) have suggested that a family history of bipolar disorder, a history of previous antidepressant-induced mania and onset of bipolar illness in adolescence or young adulthood may all confer an increased risk. These risk factors can be considered as indicating an increased genetic loading for bipolar disorder and switching.

More controversial is the suggestion by Goldberg and Truman (2003) that exposure to multiple antidepressant trials is associated with a higher switch rate. In other words, antidepressant use may carry a long-term risk in bipolar disorder. However it could be argued that patients with bipolar disorder who have received multiple anti-depressant trials are more ‘cyclical’ than other patients and therefore switching would be more likely to be observed even if there was not a causal relationship with anti-depressant use. It is interesting to note that psychiatrists with more interest in bipolar disorder are more cautious about using antidepressant in patients with bipolar disorder (Ghaemi et al., 2003). It is unclear whether female gender is a risk factor for switching (Goldberg & Truman, 2003).

Although antidepressant-induced switching has been reported with all the major antidepressant classes, there is a general consensus that the risk is higher with TCAs than with SSRIs (Calabrese et al., 1999). Peet (1994) calculated the rate of treatment-emergent switch into mania from all available clinical trial data on four SSRIs (fluoxetine, fluvoxamine, paroxetine, sertraline) relative to comparative groups treated with TCAs or placebo. In patients with bipolar disorder a higher switch rate was observed with TCAs (11.2%) than with SSRIs (3.7%) or placebo (4.2%). In contrast in unipolar depressives, the rate of manic switch was less than 1% and differences between drugs and placebo were statistically but not clinically significant. Mania has been reported in association with use of hypericum (St John’s wort) (Nierenberg et al., 1999).

Prior to starting an antidepressant, clinicians should warn patients of the risk of antidepressant-induced switching. To minimise the risk, where at all possible, TCAs should be avoided in bipolar disorder and antidepressants only prescribed in conjunction with an antimanic agent.

The risk of antidepressant-induced cycle acceleration

It has been suggested that antidepressants may cause an increase in the cycling rate of bipolar disorder (sometimes referred to as ‘mood destabilisation’). This possibility has received less attention than antidepressant-induced switching although, if it occurs, it is potentially a greater problem. Unfortunately, there are no published RCT data to inform the debate. One often-quoted naturalistic study that suggests that TCAs can cause cycle acceleration is Wehr and colleagues (1988). However, this study was based on an atypical group of patients in that they all had treatment-resistant bipolar disorder and were assessed in a specialist mood service. An unpublished randomised study suggests that there may be some increase in affective morbidity in patients with bipolar disorder who remain on antidepressants long term (Ghaemi et al., 2005).

In contrast are the results of a non-randomised 1-year study of long-term anti-depressant use, versus early antidepressant discontinuation, after remission of bipolar depression (Altshuler et al., 2004). All subjects continued with ongoing antimanic medication throughout. In this study the risk of depressive relapse was significantly associated with discontinuing antidepressants soon after remission. Furthermore, there was no evidence of manic switching in association with continued antidepressant use. In summary, this study did not show any association with cycle acceleration and suggested a beneficial effect of antidepressant use. The authors argued that maintenance of antidepressant treatment in combination with a mood stabiliser may be warranted in some patients with bipolar disorder. A major weakness of the study is that subjects were not randomised to the antidepressant continuation and discontinuation groups.

The effectiveness of antidepressants in acute depression

Although there is RCT evidence to support the use of antidepressants in the acute treatment of bipolar depression, the data are limited and opponents argue that equally good evidence is available for lamotrigine. However, the beneficial effect of lamotrigine is delayed due to the need to gradually increase the dose in order to reduce the risk of skin rashes.

The effectiveness of antidepressants in prophylaxis of bipolar depression

Proponents of antidepressants have argued that they have a long-term effect in bipolar depression. The Altshuler and colleagues study (2004) supports this view but is potentially flawed as it is non-randomised. Other studies have failed to demonstrate the effectiveness of antidepressants in preventing relapse in patients with bipolar disorder (Ghaemi et al., 2001; Zablotsky et al., 2005). Of particular note is the Zablotsky and colleagues study (2005) which is an interim analysis of patients with bipolar disorder who had recovered from a depressive episode following treatment with a mood stabiliser and an antidepressant who were then randomised to continue or discontinue the antidepressant and then followed up for 1 year. After adjusting for potential confounders, the time that patients remained in remission did not differ significantly between the two groups. This study is the first RCT to assess the efficacy of antidepressants in the prevention of bipolar disorder and suggests that such efficacy is lacking. Further interim data from this same study suggested that in patients with non-rapid-cycling bipolar disorder long-term antidepressant treatment may increase affective morbidity (Ghaemi et al., 2005). The final element relevant to this issue is that opponents of antidepressants argue that there is RCT evidence that lamotrigine can prevent recurrence of bipolar depression and that lithium also has a long-term antidepressant effect, albeit weaker, and that neither lithium nor lamotrigine carry the risks of switching or cycle acceleration that occur with antidepressants.

The effectiveness of antidepressants in reducing suicide

Opponents of antidepressants point out that there is no data to indicate that antidepressants reduce suicide in bipolar disorder but evidence does suggest that lithium can do this. (See next section for further discussion).

The conflicting evidence regarding the long-term effect of antidepressants in bipolar depression contrasts with the situation in unipolar depression. In unipolar depression there is a strong evidence base supporting the routine use of continuation treatment after symptomatic recovery from an acute depressive episode and also supporting the use of maintenance treatment in those with recurrent depression.

8.6.3. LITHIUM AND SUICIDE

Tondo and colleagues (2001) reported studies on lithium treatment in patients with bipolar disorder, major affective disorder and schizoaffective disorder that examined data on suicide rates. Twenty-two studies on a total of 5,647 patients with 33,473 patient years were included. Only three of the studies were randomized. For comparison, data from 13 studies reported suicide rates on patients who were not receiving lithium treatment over a mean period of 5 years. The suicide rate during lithium treatment was significantly less than for those without lithium treatment. Results consistent with a protective effect against suicide were also found when the analysis was restricted to the three randomized trials. The antisuicidal effects of lithium in mood disorders has been recently further supported by two large long-term studies – a retrospective one by Kessing and colleagues (2005) and a prospective one by Angst and colleagues (2005). A naturalistic study by Brodersen and colleagues (2000) reported on the suicide rate in a cohort of patients with mood disorders who were discharged from hospital after commencement of lithium. Over 16 years of follow-up, the SMR for suicide was significantly increased compared with the general population. Within the first 2 years of the study, the suicide rate was higher in those who were non-compliant versus those who were compliant with lithium. However, the suicide SMR was elevated in both the compliant and non-compliant groups indicating that although compliance with lithium reduced the risk of suicide it did not eliminate it. The above data make a strong case for an antisuicidal effect of lithium but cannot separate whether the effect is specific or non-specific (that is, only found because patients who are compliant with long-term lithium are patients who are less likely to die by suicide).

A meta-analysis of 32 RCTs comparing lithium with placebo or other compounds used in long-term treatment for mood disorders provides some support for a specific anti-suicidal effect of lithium (Cipriani et al., 2005). In the 32 trials, 1,389 patients were randomly assigned to receive lithium and 2,069 to receive other compounds. Patients who received lithium were less likely to die by suicide than were patients treated with other compounds. The combined outcome measure of suicide plus self-harm was also lower in patients who received lithium. The authors concluded that lithium is effective in the prevention of suicide and self-harm in patients with mood disorders. However, there were insufficient data to discern whether this effect was also seen in bipolar disorder.

In a narrative review, Muller-Oerlinghausen and colleagues (2005) summarised the evidence for lithium’s effects against suicide and suicidal behaviour in affective disorders, concentrating on bipolar disorder. Two interesting questions were raised: is the putative anti-suicidal effect of lithium specific to lithium and does it only occur with improvements in affective symptoms?

In discussing whether the anti-suicidal effect of lithium is shared by other psychotropic agents, the authors cited two trials. Firstly, the Multicenter Study of Affective Psychoses (MAP) (Thies-Flechtner et al., 1996) compared lithium with carbamazepine treatment for 2.5 years in bipolar and schizoaffective patients. There were significantly more suicides and suicide attempts in the carbamazepine group. Secondly, Goodwin and colleagues (2003) conducted a retrospective review of suicide risk in patients on lithium compared with those on valproate and those on carbamazepine. The adjusted suicide risk was 2.7 times higher in valproate treated patients in comparison to that of the lithium group. The study was insufficiently powered to allow similar comparisons with carbamazepine.

In order to determine whether lithium has a specific antisuicidal effect, independent of its effects in episode treatment and prevention, a sub-analysis of data from the International Group for the Study of Lithium-treated Patients (IGSLI) (Ahrens and Muller-Oerlinghausen, 2001) selected patients with recurrent affective disorders and at least one suicide attempt prior to the onset of lithium treatment (a total of 176) and categorised them depending on their response to lithium. A statistically significant reduction in suicide attempts was found in all three groups, even those who were considered to have responded poorly to treatment. Although clearly not conclusive, these data support an independent antisuicidal an effect for lithium.

The weight of all of this evidence underscores the importance of compliance measures particularly early in the course of lithium treatment and a probable long-term benefit not shared by other prophylactic agents. However, specific studies in bipolar disorder are needed to examine how generalisable this effect is and to investigate the role of factors associated with good compliance in the mediation of this effect.

8.6.4. Treatments considered

Studies were found to enable the following treatments to be considered in the treatment of acute depression, both as monotherapy and in combination therapy: anti-depressants, antipsychotics, anticonvulsants, and transcranial magnetic stimulation. In addition, a single study of ECT was found (DALY2001), but was excluded.

Antidepressants

As discussed in the previous section, a key issue in the use of antidepressants for bipolar depression is that of inducing switching to mania and the possibility of cycle acceleration. In the UK no antidepressant is specifically licensed for use in bipolar depression.

Lithium

There are no RCTs of lithium monotherapy as treatment for bipolar depression. Several cross-over studies were completed prior to 1980 and it is largely on these that lithium is recommended as a first-line treatment. A practical disadvantage of lithium as an antidepressant is the long delay before an antidepressant effect begins to be seen (usually 6–8 weeks).

8.6.5. Review strategy

Evidence from RCTs was found for the following treatment strategies:

  • Antidepressants, including antidepressants compared with placebo, another anti-depressant an antipsychotic, lithium or valproate semisodium
  • Antipsychotics, including antipsychotic compared with placebo, and antipsychotic-antidepressant combination
  • Anticonvulsants, including lamotrigine and valproate semisodium
  • Transcranial magnetic stimulation compared with sham-transcranial magnetic stimulation.

A summary of the evidence for the pharmacological treatment of depression in children and adolescents is also provided.

There are no RCTs of lithium in the treatment of acute bipolar depression.

8.6.6. Antidepressants in the treatment of acute depression

Studies considered

Nine trials met the eligibility criteria set by the GDG. Excluded studies with reasons for exclusion can be seen in Appendix 22. Efficacy data could not be extracted from three studies (AMSTERDAM2005, COHN1989, HIMMELHOCH1991). Data were therefore available to compare imipramine and paroxetine (in patients maintained on lithium, both high (>0.8 mmol/litre) and low serum levels (=<0.8 mmol/litre)) with both placebo and each other, imipramine with moclobemide, and idazoxan13 with bupropion.14 In addition, data were available to compare amitriptyline with the antipsychotic l-sulpiride. (A comparison of fluoxetine combined with an antipsychotic against an antipsychotic is considered below.) In most studies, some patients were maintained on lithium. See Table 36 for outline study characteristics and Appendix 22 for more detailed information.

Table 36. Summary of study characteristics for antidepressants in the treatment of acute depression.

Table 36

Summary of study characteristics for antidepressants in the treatment of acute depression.

Summary of evidence for antidepressants in the treatment of acute depression

An overview of the results is provided in Table 37, with the full evidence profile in Appendix 23.

Table 37. Summary of evidence profile for antidepressants in the treatment of acute depression.

Table 37

Summary of evidence profile for antidepressants in the treatment of acute depression.

Clinical summary for antidepressants in the treatment of acute depression

The evidence for the use of antidepressants in bipolar depression is weak. In patients maintained on lithium, there is no difference in the efficacy of paroxetine or imipramine on any outcome measure. Both are more efficacious than placebo, but the difference is neither statistically nor clinically significant. However, when the data are analysed by lithium serum levels, with >0.8 mmol/litre classified high and <= 0.8 mmol/litre as low, there is a clinically important advantage for both paroxetine and imipramine compared with placebo in the low-serum group but not for the high-serum group. Moclobemide is superior to imipramine, although nearly half the patients in the comparison were taking lithium, which makes the data hard to interpret. In addition, there are concerns regarding the risk of interaction with serotonergic agents and certain foodstuffs, and these risks are likely to be higher in patients with bipolar disorder compared with those with unipolar disorder because of the risk of mixed affective states and mania, which may impair judgement.

There is no significant difference between idazoxan and bupropion on any outcome measure or between imipramine and tranylcypromine. There is only inconclusive evidence of efficacy, acceptability and tolerability for an antidepressant compared with an antipsychotic. The study of lithium or valproate semisodium plus paroxetine versus lithium or valproate semisodium alone does not report extractable efficacy outcomes. All acceptability and tolerability measures were inconclusive. In a comparison between amitriptyline and l-supiride, there is no significant difference between the two. All patients were on lithium in this study.

Based on evidence from trials in unipolar depression (NCCMH, 2004), SSRIs are the safest treatment for depression because of their increased tolerability and safety compared with TCAs and monoamine oxidase inhibitors. In addition, in bipolar depression data suggests a lower risk of switching compared with TCAs.

8.6.7. Antipsychotics in the treatment of acute depression

Studies considered

Three trials met the eligibility criteria set by the GDG. Excluded studies with reasons for exclusion can be seen in Appendix 22. See Table 38 for outline study characteristics and Appendix 22 for more detailed information.

Table 38. Summary of study characteristics for antipsychotics in the treatment of acute depression.

Table 38

Summary of study characteristics for antipsychotics in the treatment of acute depression.

Summary of evidence

The efficacy data for olanzapine and quetiapine (300 mg and 600 mg) could not be combined in meta-analysis because of extreme heterogeneity (I2 = 91%). They were therefore analysed separately. There was not such difficulty with tolerability/acceptability data. One study (CALABRESE2005) also included both patients with bipolar II disorder and those with bipolar I disorder. The data for those with bipolar I disorder were analysed in a sub-analysis.

An overview of the results is provided in Table 39, with the full evidence profile in Appendix 23.

Table 39. Summary of evidence profile for antipsychotics in the treatment of acute depression.

Table 39

Summary of evidence profile for antipsychotics in the treatment of acute depression.

Clinical summary

There is some evidence for the use of atypical antipsychotics in acute bipolar depression, although none is licensed in the UK for the treatment of bipolar depression. However, the single study of olanzapine has quality issues, in particular a high attrition rate and apparently different results for different measures of depression. In addition, weight gain was five times more likely to be reported by those taking olanzapine than those on placebo. This is supported by a review of weight gain in patients with schizophrenia taking antipsychotics, which showed mean weight gain after 10 weeks’ treatment with olanzapine was more marked than with other antipsychotics, with the exception of clozapine (Allison et al., 1999). Quetiapine was more effective than placebo, although there was no difference between high (600 mg/day) and low (300 mg/day) doses of the drug on efficacy measures. However, people on low-dose quetiapine were less likely to leave treatment early or to leave because of side effects. There was also a high attrition rate from this study.

8.6.8. Anticonvulsants in the treatment of acute depression

Studies considered for review

Three trials met inclusion criteria. See Table 40 for outline study characteristics, and Appendix 22 for more detailed information.

Table 40. Summary of study characteristics for anticonvulsants in the treatment of depression.

Table 40

Summary of study characteristics for anticonvulsants in the treatment of depression.

Summary of evidence

An overview of the results is provided in Table 41 with the full evidence profile in Appendix 23.

Table 41. Summary of evidence profile for anticonvulsants in the treatment of acute depression.

Table 41

Summary of evidence profile for anticonvulsants in the treatment of acute depression.

Clinical summary

Of the two trials of anticonvulsant monotherapy in acute depression in people with bipolar disorder that met the inclusion criteria, one was of lamotrigine and the other valproate semisodium. Neither drug is licensed in the UK for the treatment bipolar depression. The overall risk-benefit ratios in both trials were unclear since in both the acceptability and tolerability data were inconclusive, although there is evidence of efficacy versus placebo for valproate semisodium. In the lamotrigine trial, the risk of switching to mania did not differ significantly between the 50 mg/day, 200 mg/day and placebo groups.

The most serious (but rare) side effect of lamotrigine is Stevens-Johnson syndrome. In early clinical trials in epilepsy the incidence of serious rash was 0.3% in adults and 1% in children (Guberman et al., 1999; Messenheimer et al., 1998). However, the introduction of a slow dose titration reduced the risk of serious rash to 0.01% in adults, which is similar to that seen with other anticonvulsants (Calabrese et al., 1999). Rashes can occur at any point during treatment but are most common early on in treatment and when lamotrigine is co-prescribed with valproate. Particular concern attaches to a rash that is widespread or associated with fever, sore throat or mucosal involvement. Patients prescribed lamotrigine must be advised to see their psychiatrist or GP if any rash appears. If it has any of the features listed previously, lamotrigine must be stopped. Lamotrigine may have limited use in the treatment of acute depression because of its slow dose titration and onset of action.

8.6.9. Transcranial magnetic stimulation in the treatment of acute depression

Studies considered

Two trials were identified from searches of electronic databases, with one meeting inclusion criteria. See Table 42 for a summary of the study characteristics and Appendix 22 for more detailed information.

Table 42. Summary of study characteristics for transcranial magnetic stimulation in the treatment of depression.

Table 42

Summary of study characteristics for transcranial magnetic stimulation in the treatment of depression.

Summary of evidence

An overview of the results is provided in Table 43, with the full evidence profile in Appendix 23.

Table 43. Summary of evidence profile for transcranial magnetic stimulation in the treatment of depression.

Table 43

Summary of evidence profile for transcranial magnetic stimulation in the treatment of depression.

Clinical summary

There is no difference in the efficacy of transcranial magnetic stimulation compared with sham transcranial magnetic stimulation on reducing depression symptoms, but participants receiving sham treatment had lower endpoint mania symptom scores. No one left treatment early and the data on side effects are not reported.

8.6.10. Fatty acids in the treatment of acute depression

Studies considered for review

One trial of omega 3 fatty acids in the treatment of acute depression met inclusion criteria. This compared omega 3 fatty acids with placebo. An overview of the study is in Table 44, with further details available in Appendix 22.

Table 44. Study information table for omega 3 fatty acids in the treatment of acute depression.

Table 44

Study information table for omega 3 fatty acids in the treatment of acute depression.

Summary of evidence

Only one trial met inclusion criteria. This shows a significant effect on depression scores for omega 3 fatty acids (at both 1 g and 2 g doses) compared with placebo in patients with moderate depression. The effect on mania scores was hard to judge since the 95% CIs were wide. However, the estimated effect sizes are small, indicating little effect on manic symptoms. Despite this, since the trial is small (around 25 in each arm), there is insufficient evidence to warrant recommending omega 3 at present. An overview of the results is provided in Table 45, with the full evidence profile in Appendix 23.

Table 45. Summary evidence profile for omega 3 fatty acids in the treatment of acute depression.

Table 45

Summary evidence profile for omega 3 fatty acids in the treatment of acute depression.

8.6.11. The treatment of treatment-resistant depression

Studies considered for review

One trial (NIERENBERG2006) met the eligibility criteria set by the GDG. There were no excluded studies. See Table 46 for outline study characteristics and Appendix 22 for more detailed information. The trial, which is part of STEP-BD, used equipoise randomisation whereby participants could choose which two of three available study treatments (risperidone, lamotrigine, inositol) they preferred in addition to existing antimanic medication (lithium, valproate or carbamazepine) and antidepressant(s). They were then randomised to one of their chosen acceptable treatments. This created four strata: risperidone versus lamotrigine, risperidone versus inositol, lamotrigine versus inositol. The study is analysed according to these strata. However, three participants were willing to be randomised to any treatment and these data are included twice. Since the number involved was small, data from different strata are combined in meta-analysis.

Table 46. Summary of study characteristics for treatment-resistant depression.

Table 46

Summary of study characteristics for treatment-resistant depression.

Summary of evidence for the treatment of treatment-resistant depression

Treatment-resistance depression (that is, no response to treatment in the first 12 weeks of standard or randomised care pathway for bipolar depression or a well-documented failure of at least two antidepressants or an antidepressant plus antimanic agent). All the calculated effect sizes were inconclusive since the 95% CIs were too wide to draw firm conclusions. This is probably because the study lacked power since the randomisation method led to low numbers in each treatment group. However, most effect sizes indicated little difference between treatments, although that for non-responders taking lamotrigine compared with inositol approached both clinical and statistical significance favouring lamotrigine, but that for lamotrigine compared with risperidone did not. Also, that for inositol versus risperidone favoured inositol. Although data for switching to mania/hypomania were available, the number of events was too low to draw firm conclusions. An overview of the results is provided in Table 47, with the full evidence profile in Appendix 23.

Table 47. Summary evidence profile for treatment-resistant depression.

Table 47

Summary evidence profile for treatment-resistant depression.

8.6.12. Economic evidence for the pharmacological treatment of acute depressive episodes

No economic studies on the cost-effectiveness of pharmacological treatment for acute depressive episodes in patients with bipolar disorder were identified.

8.6.13. Clinical practice recommendations

8.6.13.1.

A patient who is prescribed antidepressant medication should also be prescribed an antimanic drug. The choice of antimanic drug should be compatible with decisions about future prophylactic treatment, the likely side effects and whether the patient is a woman of child-bearing potential.

8.6.13.2.

When initiating antidepressant treatment for a patient who is not already taking antimanic medication, prescribers should explain the risks of switching to mania and the benefits of taking an adjunctive antimanic agent. People who are not willing to take antimanic medication should be monitored carefully. Treatment should begin at a low dose and be increased gradually if necessary.

8.6.13.3.

If a person has an acute depressive episode when taking antimanic medication, prescribers should first check they are taking the antimanic agent at the appropriate dose and adjust the dose if necessary.

8.6.13.4.

For patients with acute mild depressive symptoms, a further assessment should be arranged, normally within 2 weeks (‘watchful waiting’) if:

  • previous episodes of mild depression have not developed into chronic or more severe depression in this patient, or
  • the patient is judged not to be at significant risk of developing a more severe depression.

If the patient is judged to be at significant risk of worsening or on review continues to be unwell, they should be managed as for moderate or severe depression, particularly if functional impairment is evident.

8.6.13.5.

For patients with moderate or severe depressive symptoms, prescribers should normally consider:

  • prescribing an SSRI antidepressant (but not paroxetine in pregnant women) because these are less likely than tricyclic antidepressants to be associated with switching, or
  • adding quetiapine, if the patient is already taking antimanic medication that is not an antipsychotic
8.6.13.6.

If a trial of drug treatment at an adequate dose and with adequate compliance does not produce a significant improvement for moderate depressive symptoms, a structured psychological treatment should be considered. This should focus on depressive symptoms, problem solving, promoting social functioning, and education about medication.

Starting antidepressant treatment and monitoring risk

8.6.13.7.

Antidepressants should be avoided for patients with depressive symptoms who have:

  • rapid-cycling bipolar disorder
  • a recent hypomanic episode
  • recent functionally impairing rapid mood fluctuations.

Instead consider increasing the dose of the antimanic agent or the addition of a second antimanic agent (including lamotrigine*).

8.6.13.8.

Patients’ concerns about taking antidepressants should be addressed. For example, they should be advised that craving and tolerance do not occur, and that taking medication should not be seen as a sign of weakness.

8.6.13.9.

When antidepressant treatment is started, patients should be told about:

  • the possibility of manic or hypomanic switching
  • the delay in onset of effect, and the gradual and fluctuating nature of improvement
  • the need to take medication as prescribed and the risk of discontinuation/withdrawal symptoms
  • the need to monitor for signs of akathisia, suicidal ideation (normally anyone under 30 should be reviewed within 1 week of initiation of treatment), and increased anxiety and agitation (particularly in the initial stages of treatment)
  • the need to seek help promptly if these side effects are distressing.
8.6.13.10.

If a patient with bipolar disorder develops marked and/or prolonged akathisia or agitation while taking an antidepressant, the use of the drug should be reviewed urgently.

8.6.13.11.

Care should be taken when prescribing SSRIs* to people – particularly older people – taking other medication that can cause intestinal bleeding, such as non-steroidal anti-inflammatory drugs. The use of a gastroprotective drug may be considered.

Stopping antidepressants after the treatment of acute depression

8.6.13.12.

When a patient is in remission from depressive symptoms (or symptoms have been significantly less severe for 8 weeks), stopping the antidepressant medication should be considered, to minimise the risk of switching to mania and increased rapid cycling. The dose of antidepresant should be gradually reduced over several weeks, while maintaining the antimanic medication. Particular care is needed when discontinuing paroxetine and venlafaxine because they are associated with a higher risk of discontinuation/withdrawal symptoms.

Additional guidance on specific treatments

8.6.13.13.

The following treatments should not be routinely used for acute depressive episodes in people with bipolar disorder:

  • lamotrigine* as a single, first-line agent in bipolar I disorder
  • transcranial magnetic stimulation*.

Incomplete response to the treatment for acute depression

8.6.13.14.

When a patient’s depressive symptoms do not fully respond to an antidepressant, the patient should be reassessed for evidence of substance abuse, psychosocial stressors, physical health problems, comorbid disorders, such as anxiety or severe obsessional symptoms, and inadequate adherence to medication. Prescribers should then consider:

  • increasing the dose of the antidepressant within ‘British national formulary’ (‘BNF’) limits
  • individual psychological therapy focused on depressive symptoms
  • switching to an alternative antidepressant (for example, mirtazapine or venlafaxine)
  • adding quetiapine* or olanzapine if the patient is not already taking one of these, or
  • adding lithium if the patient is not already taking it.

Management of treatment-resistant depression in bipolar disorder

8.6.13.15.

If a patient’s depressive symptoms have failed to respond to at least three courses of treatment for depression of adequate dose and duration, seeking the advice of, or referral to, a clinician with a specialist interest in treating bipolar disorder shold be considered.

Management of concurrent depressive and psychotic symptoms

8.6.13.16.

For patients with a diagnosis of bipolar disorder experiencing concurrent depressive and psychotic symptoms, prescribers should consider augmenting the current treatment plan with antipsychotic medication, such as olan-zapine, quetiapine, or risperidone, or the use of electroconvulsive therapy if the depressive illness is severe.

Management of persistent depressive symptoms

8.6.13.17.

For patients with persistent depressive symptoms and no history of recent rapid cycling, including those who have declined an antidepressant, structured psychological therapy may be considered. This should focus on depressive symptoms, problem solving, improving social functioning, and further discussion of medication concordance.

8.6.14. Research recommendations

Treatments for patients in partial remission from depressive symptoms in bipolar disorder

8.6.14.1.

A randomised placebo-controlled trial should be conducted to investigate the efficacy and cost effectiveness of adding an antidepressant (SSRI) to an existing antimanic agent for patients with bipolar disorder in partial remission from a depressive episode. The trial would ideally recruit from both primary and secondary care and outcome measures would include time to recovery from depression, time to prevention of the next episode and social functioning with a 2-year follow-up period.

Why this is important

The treatment of severe mental illness is a key priority in the National Service Framework for mental health, and depression of all kinds is a major cause of long-term disability and unemployment. People with bipolar disorder suffer more depressive episodes than manic ones. Partial remission from symptoms is common, so successful treatment would greatly improve functioning and quality of life. But there is little evidence on which to base recommendations for treatment of bipolar depression, and none on treatment after partial remission.

Treatments for depression in bipolar disorder, and their risks

8.6.14.2.

A sequenced set of randomised controlled trials should be undertaken to investigate the efficacy and cost effectiveness of antidepressants, in the presence of an antimanic medication, in treating bipolar depression. The studies should address the different stages of depression (acute, continuation and maintenance) and also evaluate the risks, particularly switching to mania and cycle acceleration, associated with antidepressant treatment. Patients with bipolar I and II disorder should be recruited. Outcome measures would include time to recovery from depression, time to prevention of the next episode and social functioning.

Why this is important

People with bipolar disorder suffer more depressive episodes than manic episodes. Depression is the major cause of suicide and the rate of suicide is very high among patients with bipolar disorder (10–15%). Therefore successful treatment would greatly improve functioning and quality of life. There is little evidence on the treatment of bipolar depression, particularly in different phases of the illness. Reducing depression could contribute to meeting the national targets to reduce suicide in bipolar disorder, and to reduce depression as a major cause of long-term disability and unemployment.

8.6.15. The treatment of acute depression in children and adolescents

There are no studies of the management of acute depression in children and adolescents with bipolar disorder. It is necessary to extrapolate cautiously from the adult database and to refer to the evidence on the treatment of unipolar depression in children and adolescents. The NICE guideline Depression in Children and Young People (NCCMH, 2005) summarises the evidence for the use of antidepressants in under 18 year olds with unipolar depression. Fluoxetine is the only SSRI antidepressant where clinical trials demonstrated that the risk benefit ratio was positive for unipolar depression in young people under 18 years. However, the issue of whether SSRIs are linked to increased suicide risk is unclear. Whilst there is evidence that the rates in those taking SSRIs are no different to those taking TCAs (Jick et al., 2004; Martinez et al., 2005), the risk may also be due to discontinuation of treatment due to adverse events. However, there is evidence of increased risk of non-fatal self-harm in those taking SSRIs under 18 (Martinez et al., 2005). The addition of CBT may reduce this risk. Antidepressants should normally first be prescribed only in conjunction with psychological therapies and the child or adolescent should be monitored for agitation, suicidal ideation and self-harm. If treatment with fluoxetine is unsuccessful or not tolerated because of side effects, consideration should be given to the use of another antidepressant. In the case of children and adolescents, sertraline or citalopram are the recommended second-line treatments. In the UK none of these drugs is licensed for use in under 18 year olds but can be prescribed unlicensed by a specialist practitioner.

8.6.16. Clinical practice recommendations

Depression in children and adolescents

8.6.16.1.

Children and adolescents with bipolar disorder experiencing mild depressive symptoms assessed as not requiring immediate treatment should be monitored weekly and offered additional support, for example at home and in school.

8.6.16.2.

Children or adolescents with depressive symptoms needing treatment should normally be treated by specialist clinicians (based in at least Tier 3 services). Treatment should be as for adults with bipolar disorder except that a structured psychological therapy aimed at treating depression should be considered in addition to prophylactic medication.

8.6.16.3.

If there has been no response to psychological therapy for depression combined with prophylactic medication after 4 weeks, prescribers should consider:

If there is still no response, advice should be sought from a specialist in affective disorders.

8.6.16.4.

For developmentally advanced adolescents with depressive symptoms, the recommendations on managing depression in adults with bipolar disorder should be followed.

8.7. ELECTROCONVULSIVE THERAPY (ECT) (ALL ILLNESS PHASES)

8.7.1. Clinical evidence

No study of ECT in acute bipolar depression met inclusion criteria, although there is some evidence for its effectiveness compared with an antipsychotic in acute mania. The recommendations below are adapted from the NICE technology appraisal of ECT (NICE, 2003).

8.7.2. Health economics evidence

A recent NHS Health Technology Assessment (Greenhalgh et al., 2005) aimed at evaluating the clinical and cost effectiveness of ECT for depressive illness, schizophrenia, catatonia and mania. The report failed to identify any economic studies relating to ECT fulfilling the inclusion criteria set for the systematic review. In addition, although decision-analytic models were developed in order to assess the cost effectiveness of ECT for the management of depressive illness and schizophrenia, it was not possible to construct any economic models for either mania or catatonia, due to lack of published data relating to these disease areas.

Only one economic study relating to ECT for the management of acute bipolar episodes was identified and included in the systematic review carried out for this guideline (KUTCHER1995). The study, which was conducted in Canada, compared ECT to standard pharmacological inpatient care for the management of adolescents and young adults (16–22 years old) with treatment-resistant acute bipolar episodes, either manic or depressive, and was based on a retrospective comparative design. The main drawback of the analysis was the method of group assignment, which was based on self-selection: patients eligible for ECT chose whether they wanted to undergo ECT; those agreeing comprised the ECT group, while those denying received standard inpatient care. Costs consisted of hospitalisation costs only; intervention costs were not estimated. Clinical outcomes included improvement in clinical symptoms as measured using the BPRS, length of hospital stay, and rate of side effects in the ECT group. The time horizon of the analysis was the period until hospital discharge.

The analysis demonstrated that both groups were characterised by significant improvement in BPRS scores over time, with the ECT group score at discharge being significantly lower than that of the control group. Side effects of ECT were reported to be generally mild, occurring in 28% of the ECT sessions. The length of hospital stay was shorter for the ECT group, resulting in lower hospitalisation costs compared with the control group ($58,608 versus $143,264 per admission respectively). Overall, ECT appeared to be more cost-effective than standard care; however, the design of the study, which was subject to major selection bias, as well as the estimation of hospitalisation costs only and the omission of all other relevant costs from the analysis, limit considerably the scope of it; consequently, no robust conclusions on the cost-effectiveness of ECT can be drawn based on this study.

Full reference, characteristics and results of the above study are presented in the evidence tables for economic studies in Appendix 14.

8.7.3. Clinical practice recommendations

8.7.3.1.

Electroconvulsive therapy (ECT) is recommended only to achieve rapid and short-term improvement of severe symptoms after an adequate trial of other treatment options has proven ineffective and/or when the condition is considered to be potentially life-threatening, in individuals with:

  • severe depressive illness
  • catatonia
  • a prolonged or severe manic episode.15
8.7.3.2.

The decision as to whether ECT is clinically indicated should be based on a documented assessment of the risks and potential benefits to the individual, including:

  • the risks associated with the anaesthetic
  • current comorbidities
  • anticipated adverse events, particularly cognitive impairment
  • the risks of not having treatment.15
8.7.3.3.

When using ECT to treat bipolar disorder, prescribers should consider:

  • stopping or reducing lithium or benzodiazepines before giving ECT
  • monitoring the length of fits carefully if the patient is taking anticonvulsants
  • monitoring mental state carefully for evidence of switching to the opposite pole.

8.8. THE MANAGEMENT OF ACUTE BEHAVIOURAL DISTURBANCE

8.8.1. Introduction

Acute behavioural disturbances in the context of bipolar disorder typically occur during periods of hospitalisation, but may arise at any time during the course of the illness, as part of an exacerbation. A person may be agitated, aggressive or violent towards others, secondary to psychotic symptoms (for example, grandiose delusions, command hallucinations) or non-psychotic symptoms (for example, high levels of arousal). However, in the Royal College of Psychiatrists’ guideline on the management of imminent violence (1998), environmental factors, including overcrowding, lack of privacy, lack of activities and long waiting times to see staff, are identified as playing an important role in increasing the likelihood of aggression or violence. Also, social factors, such as poor communication between patients and staff and weak clinical leadership, may contribute to feelings of frustration and tension amongst all parties. Dealing with these issues in advance may reduce the risk of violence and aggression (Royal College of Psychiatrists, 1998).

The initial response should be to provide structure, reduce stimulation and try to verbally reassure and calm the person (Osser & Sigadel, 2001). Staff need to be trained to anticipate possible violence and to de-escalate the situation at the earliest opportunity. Physical means of restraint or seclusion should be resorted to ‘only after the failure of attempts to promote full participation in self-care’. In this context, the use of drugs to control disturbed behaviour (rapid tranquillisation) is often seen as a last resort, for use where appropriate psychological and behavioural approaches have failed or are inappropriate. However, patients who participated in discussion groups on this topic have expressed the view that when they behaved violently, their preference was for medication rather than seclusion or prolonged physical restraint (Royal College of Psychiatrists, 1998).

The aim of drug treatment in such circumstances is to calm the person and reduce the risk of violence and harm, rather than treat the underlying psychiatric condition. An optimal response would be a reduction in agitation or aggression without sedation, allowing the patient to participate in further assessment and treatment. Ideally, the drug should have a rapid onset of action and a low level of side effects. Psychiatrists, and the multidisciplinary team, who use rapid tranquillisation should be trained in the assessment and management of patients specifically in this context: this should include assessing and managing the risks of drugs (benzodiazepines and antipsychotics), using and maintaining the techniques and equipment needed for cardiopulmonary resuscitation, and prescribing within therapeutic limits and using flumazenil (a benzodiazepine antagonist).

8.8.2. Current practice

In clinical practice in the UK, the most common choice of drug for rapid tranquillisation has been an antipsychotic, such as haloperidol or chlorpromazine, often in combination with a benzodiazepine, such as diazepam or lorazepam (Pilowsky et al., 1992; Cunnane, 1994; Mannion et al., 1997; Simpson & Anderson, 1996). Zuclopenthixol acetate, a short-acting depot antipsychotic, has also been commonly used, although doubts as to its value have recently been raised (Fenton et al., 2000). Previously published guidelines have generally recommended that drug treatment should be administered orally, if the person is willing to accept this (Atakan & Davies, 1997; Kerr & Taylor, 1997; Royal College of Psychiatrists, 1997). Liquid or rapidly dissolving preparations may be particularly useful.

However, current practice is not underpinned by a strong evidence base. There are relatively few controlled studies on which to base prescribing decisions. Further, in studies of drug treatment of acute behavioural disturbance, there may be some doubt about the generalisability of the patient samples, in that many of the individuals who require rapid tranquillisation would be too disturbed to give informed consent to participate. Placebo-controlled studies would be necessary to assess the overall advantage of treatment using the various drug regimens, but such studies are likely to be regarded as unethical given that rapid tranquillisation is usually indicated only after other non-pharmacological approaches have failed. In any event, a major concern in this situation is to minimise the risk of harm to the individual, the staff and others.

8.8.3. Safety considerations in rapid tranquillisation

The benefit of reducing the risk of harm to the individual or others must be balanced against the risk of adverse effects associated with such drug regimens. In a survey of around 100 incidents of rapid tranquillisation, Pilowsky and colleagues (1992) found few adverse events, although those reported were potentially serious cardiovascular and cardiorespiratory events. The cardiovascular effects of antipsychotics in such situations have become a source of growing concern. Osser and Siagdel (2001) recommend that chlorpromazine is avoided because of its greater risk of hypotension. Droperidol, an antipsychotic widely used for rapid tranquillisation, was voluntarily withdrawn in 2001 because of reports of QT prolongation, serious ventricular arrhythmia and sudden death. A change in the rate-corrected QT interval (QTc) on the ECG associated with an antipsychotic may be an indicator of cardio toxicity. An increased risk of QT interval prolongation has been reported with several antipsychotic drugs (Royal College of Psychiatrists, 1997). In a naturalistic study in the UK (Reilly et al., 2000), a prolonged QTc was associated with both thioridazine and droperidol. Partly on the basis that QTc prolongation may be a marker of risk of arrhythmia, the use of thioridazine has been restricted in the UK since the end of 2000 and would not be appropriate for rapid tranquillisation.

Acute EPS, such as akathisia, parkinsonism and dystonia, are commonly observed with intramuscular conventional antipsychotic drugs. Dystonic reactions can be particularly severe and unpleasant (Royal College of Psychiatrists, 1997) and both akathisia and dystonia could exacerbate disturbed behaviour. If an antipsychotic is to be used alone in people at high risk for dystonia, such as males under 35 years of age, long-term antiparkinsonian drug treatment should be considered (Osser & Sigadel, 2001).

Benzodiazepines, although considerably safer, can cause cognitive impairment, behavioural disinhibition, over-sedation and, most seriously, respiratory depression with the administration of high doses (Mendelson, 1992). The benzodiazepine partial antagonist flumazenil can counter these effects, but it is not easy to give in an emergency situation by those inexperienced in its use: flumazenil has a shorter half-life than the majority of benzodiazepines, meaning that repeated doses may be required. It can also induce seizures in people who have been receiving regular benzodiazepines.

8.8.4. Definition and aim of rapid tranquillisation

Rapid tranquillisation means the use of drug treatments to achieve rapid, short-term behavioural control of extreme agitation, aggression and potentially violent behaviour that places the individual or those around them at risk of physical harm (Broadstock, 2001). The term rapid tranquillisation is usually restricted to parenteral forms of medication. The aim of rapid tranquillisation is to achieve sedation sufficient to minimise the risk posed to the person themselves or to others. The individual should be able to respond to spoken messages throughout the period of sedation (Royal College of Psychiatrists, 1998).

8.8.5. Studies considered for review

This review is based on a new systematic search for RCTs of pharmacological agents used in rapid tranquillisation. Since there are very few studies in specific bipolar populations, inclusion criteria allowed studies with people with other severe mental illness. The search was undertaken in May 2005; 30 trials were identified from searches of electronic databases, with five meeting the eligibility criteria set by the GDG. Excluded studies with reasons for exclusion can be seen in Appendix 22.

Important characteristics of the included studies are in Table 48, with fuller details of studies in Appendix 22.

Table 48. Summary of study characteristics for rapid tranquillisation in the management of acute behavioural disturbance.

Table 48

Summary of study characteristics for rapid tranquillisation in the management of acute behavioural disturbance.

8.8.6. Outcomes

The following outcomes were extracted at 2 and 24 hours after administration of study drugs: scores on the Positive and Negative Symptom Scale (PANSS) excited subscale and total scale, overt aggression scale, agitated behaviour scale, YMRS, and BPRS, plus the number who received benzodiazepines. Scores on the PANSS excited subscale and agitated behaviour scales were considered primary outcomes. Also, the number leaving the study early for any reason and because of side effects were extracted, together with side effects including parkinsonian side effects, akathisia, EPS, dystonia, nausea, vomiting, hyperkinesias [EPS or movement disorders] agitation, somnolence and QTc interval change. Parkinsonian side effects, akathisia, EPS and QTc interval change were also considered primary outcomes.

8.8.7. Clinical summary

Only six studies met inclusion criteria. Of these, only one included 100% bipolar patients (MEEHAN2001), with two others including a proportion of people with bipolar disorder (FOSTER1997, 35%; CURRIER2004, 8%). The majority of people in the dataset had a diagnosis of schizophrenia or related disorders. There was also some uncertainty about the quality of the included trials, with only one (CURRIER2004) describing an adequate randomisation process. The others failed to report how randomisation was carried out. Among other things, these aspects led to downgrading of evidence.

On critical outcomes there was evidence (moderate quality) that IM olanzapine and IM lorazepam are effective in reducing aggressive behaviour compared with placebo, as well as evidence (low quality) for the efficacy of IM haloperidol compared with placebo. There was inconclusive evidence (very low quality) of any difference in efficacy between IM olanzapine (10 mg) and IM haloperidol (7.5 mg), but some evidence that IM olanzapine is more effective than IM lorazepam (moderate quality). There were no critical outcomes reported for IM haloperidol compared with IM lorazepam and the available evidence was graded very low quality. In the one study including oral medication (oral risperidone + oral lorazepam versus IM risperidone + oral lorazepam) (CURRIER2004), there was no difference between the two drug combinations (moderate quality).

Turning to side effects, there was no difference on critical side effects between olanzapine and placebo (low quality evidence), and some evidence that haloperidol caused more side effects than placebo (low quality evidence). Critical outcomes were not reported by the study comparing lorazepam with placebo. When olanzapine was compared with haloperidol, there was some evidence of a more favourable side-effect profile for olanzapine (moderate quality evidence). Relevant side effects were not reported in the study of olanzapine compared with lorazepam. Evidence was inconclusive (very low quality) for oral risperidone + oral lorazepam versus IM risperidone + oral lorazepam.

An overview of the results is provided in Table 49, with the full evidence profile in Appendix 23.

Table 49. Summary of evidence profile for rapid tranquillisation in the management of acute behavioural disturbance.

Table 49

Summary of evidence profile for rapid tranquillisation in the management of acute behavioural disturbance.

8.8.8. Clinical recommendations

8.8.8.1.

If a patient with bipolar disorder exhibits seriously disturbed behaviour, or is judged to be at risk of doing so, healthcare professionals should:

  • place the patient in the least stimulating and confrontational, and most supportive environment available
  • review the patient’s safety and physical status, including hydration levels, and take appropriate action
  • consider using distraction techniques and diverting the patient’s energy into less risky or more productive activities to prevent or reduce behavioural disturbance.

Pharmacological management of severe behavioural disturbance in people with bipolar disorder

8.8.8.2.

Severe behavioural disturbance in people with bipolar disorder should normally be treated first with oral medication, such as lorazepam* or an antipsychotic, or a combination of an antipsychotic and a benzodiazepine. Risperidone and olanzapine are available in orodispersible formulations that are easier for patients to take and are more difficult to spit out.

8.8.8.3.

If a severely disturbed patient with bipolar disorder cannot be effectively managed with oral medication and rapid tranquilisation is needed, intramuscular olanzapine (10 mg), lorazepam* (2 mg) or haloperidol (2–10 mg) should be considered, wherever possible as a single agent. When making the choice of drug, prescribers should take into account:

  • that olanzapine or lorazepam* are preferable to haloperidol because of the risk of movement disorders (particularly dystonia and akathisia) with haloperidol
  • that olanzapine and benzodiazepines should not be given intramuscularly within 1 hour of each other
  • that repeat intramuscular doses can be given up to 20 mg per day (olanzapine), or 4 mg per day (lorazepam*) or 18 mg per day (haloperidol) –total daily dose including concurrent oral medication should not normally exceed ‘BNF’ limits
  • the patient’s previous response and tolerability, their current regular medication, and the availability of flumazenil.
8.8.8.4.

Intravenous preparations of any psychotropic drug, intramuscular diazepam*, intramuscular chlorpromazine, paraldehyde* and zuclopenthixol acetate are not recommended for routine use for managing behavioural disturbances in people with bipolar disorder.

Footnotes

12

Joint Royal College of Paediatrics and Child Health/Neonatal and Paediatric Pharmacists Group Standing Committee on Medicines (2000) The Use of Unlicensed Medicines or Licensed Medicines for Unlicensed Applications in Paediatric Practice - Policy Statement. London: Royal College of Paediatrics and Child Health.

13

Does not have UK marketing authorisation.

14

Does not have UK marketing authorisation for the treatment of depression.

15

This recommendation is from NICE technology appraisal 59 and has been incorporated into this guideline in line with NICE procedures for the development of guidelines.

Copyright © 2006, The British Psychological Society & The Royal College of Psychiatrists.

All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Enquiries in this regard should be directed to the British Psychological Society.

Bookshelf ID: NBK55369

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.5M)
  • Disable Glossary Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...