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

National Collaborating Centre for Mental Health (UK). Depression: The Treatment and Management of Depression in Adults (Updated Edition). Leicester (UK): British Psychological Society; 2010. (NICE Clinical Guidelines, No. 90.)

12THE PHARMACOLOGICAL AND PHYSICAL MANAGEMENT OF DEPRESSION THAT HAS NOT ADEQUATELY RESPONDED TO TREATMENT, AND RELAPSE PREVENTION

In this chapter, sections marked by asterisks (**__**) are from the previous guideline and have not been updated except for style and minor clarification.

12.1. INTRODUCTION

**Despite major developments in the management of mood disorders, in clinical practice the problem of incomplete, or lack of, response to treatment continues to be problematic. Numerous outcome studies have demonstrated that approximately one-third of patients treated for depression do not respond satisfactorily to first-line anti-depressant pharmacotherapy. Follow-up observations reveal that a considerable number of patients have a poor prognosis with as many as 20% remaining unwell 2 years after the onset of illness (Keller et al., 1986). Even after multiple treatments, up to 10% of patients remain depressed (Nierenberg & Amsterdam, 1990). A range of studies suggests that between 10 and 20% of patients with depression have a long-term poor outcome (Lee & Murray, 1988; Winokur et al., 1993).

It is difficult, however, to evaluate the true degree of poor response to treatment for depression from these figures. Although poor response is relatively common in clinical practice, a major problem has been the inconsistent way in which it has been characterised and defined, limiting systematic research. In recent years there have been attempts to agree definitions of ‘treatment resistance’ in order to improve the characterisation of the phenomenon, although there is still disagreement on some of the items. The key parameters that have been used to characterise and define treatment resistance include the basic criteria used to specify the diagnosis, response to treatment, previous treatment trials and the adequacy of treatment (Nierenberg & Amsterdam, 1990).**

While it is important to be able to describe these parameters, this guideline update, as discussed in Chapter 2, does not use the term ‘treatment-resistant depression’, which was defined in the previous guideline as depression that had not responded adequately to two courses of antidepressants (of adequate dose and length). The term implies that following two antidepressant-treatment failures, depression enters a new ‘difficult-to-treat’ category. Furthermore, the term may be taken by both clinicians and patients as a pejorative label. It is also not supported by the evidence. For example, it does not take into account the fact that there are different degrees of improvement and stage of illness, or the possible impact of other treatments including psychosocial treatments and non-antidepressant augmenting agents. The GDG for this guideline update preferred to approach the problem of inadequate response from the direction of ‘next-step’ treatment options rather than categorising by patient response.

12.2. APPROACH TO THE REVIEWS

The major reviews undertaken for the previous guideline are represented here and updated with new studies where these were available. Previously, studies had been categorised ‘treatment-resistant’ where participants had been recruited because their depression had not responded to two sequential antidepressant drugs prescribed in an adequate dose for an adequate duration of time, and ‘acute-phase non-responder’ where participants’ depression had not adequately responded to one antidepressant. These distinctions were not made in the present review, although the studies were coded for the number of antidepressant courses ‘failed’ both historically and prospectively (for example, H2P1 denoted that participants had inadequately responded to two antidepressants historically and one prospectively). In addition, studies of augmentation strategies that had not recruited people specifically because their depression had not responded to at least one previous treatment were removed from the analyses. A few studies used an open-label design. Since there are relatively few data on this topic, these were analysed separately and described narratively.

The electronic databases searched for published trials are given in Table 102. Details of the search strings used are in Appendix 8. In total, 11 new trials were found to supplement the previous reviews. Data were available to examine the following next-step strategies:

Table 102. Databases searched and inclusion/exclusion criteria for clinical effectiveness of pharmacological treatments.

Table 102

Databases searched and inclusion/exclusion criteria for clinical effectiveness of pharmacological treatments.

In addition, narrative reviews of evidence for transcranial magnetic stimulation (TMS) and vagal nerve stimulation (VNS) were included (Section 12.5).

There were no new data for augmentation with lithium, anticonvulsants, pindolol or benzodiazepines, but augmentation was part of a topic that was restructured for the guideline update. Sections on acute-phase non-responders and treatment-resistant depression in the previous guideline became ‘next-step treatments’ in this guideline update and some of the sections have been redrafted.

**The above strategies were reviewed because there is sufficient evidence to come to a conclusion about efficacy and/or there is significant clinical usage of such strategies in the UK. There is, however, a wide range of other strategies used where first-line treatment has not been effective, for which either the evidence base is so weak or the clinical usage so low that the GDG did not include them in this review. Examples of these latter strategies include the use of stimulants or glucocorticoid antagonists either alone or to augment antidepressants.

Details of the available information about these strategies (for example, case reports, open studies, expert opinion) can be found elsewhere (Thase & Rush, 1997; Price et al., 2001; Bauer et al., 2002b). These papers also include details of the pharmacological issues associated with these strategies. Wide varieties of new treatments to augment antidepressants are being developed or are in pilot trial phase. These are beyond the scope of this review and details can be found elsewhere (Tamminga et al., 2002).

MAOIs have been used extensively in the management of ‘treatment-resistant’ depression for 4 decades but there is no randomised data on which to base recommendations. Most information and experience is with phenelzine. McGrath and colleagues (1987b) treated patients in a cross-over design with high doses of phenelzine (maximum 90 mg), imipramine (maximum 300 mg) or placebo and found that of the non-responders only four of the 14 patients responded to a tricyclic crossover with 17 of the 26 patients responding to an MAOI cross-over. There was some evidence of a preferential response in treatment-resistant patients with atypical symptoms of depression, but Nolen and colleagues (1988) subsequently showed that not only patients with atypical depressive symptoms but also patients with depression and melancholia responded to MAOIs, in particular tranylcypramine. It does not appear that moclobemide has the same spectrum of efficacy in treatment resistance as the classical MAOIs. Nolen and colleagues (1994) switched patients with resistant depression stabilised on tranylcypromine to moclobemide. About 60% of the patients showed deterioration and one-third relapsed.**

12.3. PHARMACOLOGICAL ‘NEXT-STEP’ TREATMENT FOR DEPRESSION THAT HAS NOT ADEQUATELY RESPONDED TO TREATMENT

12.3.1. Increasing the dose

Introduction

When depression does not respond adequately, a common treatment strategy is to increase the dose of the antidepressant within the licensed dosage range. There is little objective evidence to support higher response rates with increasing dose (within the licensed dosage range) for the majority of antidepressants, but this does not preclude the possibility of a beneficial effect being seen in individual patients. Any beneficial effect is likely to be at least partially determined by individual differences in hepatic metabolising enzymes.

Studies considered183

Nine studies were found that compared drugs at different doses following lack of response to the initial dose (see Table 103), of which one was found in the update search (WHITMYER2007), but only two included a treatment group that remained on the previous dose after an adequate trial of the initial treatment (summary study characteristics of these two studies are in Table 104, with full details of the studies in Appendix 17c). Only one study (Licht2002) used a licensed dose for all patients in the initial phase, allowed adequate time to respond to this dose, and then randomised patients to remain on this dose or receive a higher dose.

Table 103. Studies (RCTs) comparing antidepressants at different doses in people whose depression is resistant to treatment.

Table 103

Studies (RCTs) comparing antidepressants at different doses in people whose depression is resistant to treatment.

Table 104. Summary study characteristics of included studies of dose escalation in people whose depression had failed to respond adequately to treatment.

Table 104

Summary study characteristics of included studies of dose escalation in people whose depression had failed to respond adequately to treatment.

Clinical evidence

There was evidence that increasing the dose led to small improvements in outcomes compared with continuing with the current dose, although these are not clinically important. However, there are few randomised trials (see Table 105) for the summary evidence profile. The full evidence profile and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 105. Summary evidence profile for dose escalation following inadequate treatment response.

Table 105

Summary evidence profile for dose escalation following inadequate treatment response.

Clinical summary

There is little objective evidence that increasing the dose improves outcomes, although there are very few randomised studies. It is known that there are genetically determined differences in the activity of several hepatic enzymes that are involved in the metabolism of antidepressant drugs. Fast or extensive metabolisers may therefore need higher doses. Until further data are available, it is reasonable to consider increasing the dose of an antidepressant within the SPC recommended range, particularly where there has been a partial response and side effects are not problematic.

12.3.2. Switching to another antidepressant

Introduction

**Approximately 20 to 30% of patients with depression do not respond to the first antidepressant prescribed (assuming an adequate dose, duration of treatment and compliance with medication; Cowen, 1998). It is normal clinical practice at this point to increase the dose to the maximum tolerated (within licensed limits; see section 12.3.1) and, if there is still no or minimal response, to switch to an alternative antidepressant (Anderson et al., 2008). Most prescribers select an antidepressant from a different class to the ‘failed’ drug (Fredman et al., 2000). Randomised studies of switching are difficult to interpret as they either include patients who may be expected to fare poorly on one of the treatments (for example, patients with atypical depression in a study with an MAOI and TCA arm; McGrath et al., 1993) or employ a cross-over design (Thase et al., 1992; McGrath et al., 1993). Open studies, however, show that approximately 50% of patients who do not respond to their first treatment are likely to respond to the second antidepressant irrespective of whether it comes from the same class or a different one (Thase & Rush, 1997).**

Studies considered184

Altogether, six studies met inclusion criteria for the update, three of which were included in the previous guideline (two in other reviews) (Ferreri2001; Poirier1999; Thase2002a). Data were available to compare various switching strategies, including continuing with antidepressant treatment versus switching, comparison of switches to other single antidepressants, and comparison of switches to a single antidepressant versus switching to combinations of drugs. Data were available to compare continuing antidepressant treatment versus switching to olanzapine, but the GDG did not consider this relevant to clinical practice so the data are not reported (but are included in the forest plots for completeness). Summary study characteristics of the included studies are presented in Table 106, with full details in Appendix 17c, which also includes details of excluded studies.

Table 106. Summary study characteristics of included studies for continuing antidepressant treatment versus switching or switching treatment(s).

Table 106

Summary study characteristics of included studies for continuing antidepressant treatment versus switching or switching treatment(s).

Clinical evidence

Continuing with antidepressant treatment versus switching

Data were available to compare continuing nortriptyline with switching to fluoxetine, continuing fluoxetine with switching to mianserin, and continuing venlafaxine with switching to fluoxetine. There was no evidence that either strategy was more effective, or more acceptable and tolerable. Evidence from the important outcomes and overall quality of evidence are presented in Table 107. The full profile and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 107. Summary evidence profile for continuing antidepressant treatment versus switching following inadequate response to treatment.

Table 107

Summary evidence profile for continuing antidepressant treatment versus switching following inadequate response to treatment.

Switching antidepressant treatment (comparison of strategies)

Data were available to compare the following switching strategies: switching to venlafaxine versus switching to an SSRI (citalopram or paroxetine) and switching to fluoxetine plus olanzapine versus switching to fluoxetine. This part of the review updates the review of venlafaxine for treatment-resistant depression included in the previous guideline.

There was no difference between the switching strategies for which data were available on any measure, other than on the number of people leaving treatment early because of side effects, which favoured fluoxetine over fluoxetine plus olanzapine. Combining the two RCTs in which non-responders were randomised to venlafaxine or an SSRI did not show a significant advantage to venlafaxine (LENOXSMITH2008; Poirier1999). The earlier study (in severely ill patients) did suggest an advantage to venlafaxine in some outcomes as reported in the previous guideline but the later study did not. A secondary analysis of the later study did however report an advantage to venlafaxine in a secondary analysis of severely ill patients. Whether venlafaxine has an advantage in severely depressed patients is therefore undetermined. Evidence from the important outcomes and overall quality of evidence are presented in Table 108. The full profile and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 108. Summary evidence profile for switching antidepressant treatment (comparison of strategies) following inadequate antidepressant response.

Table 108

Summary evidence profile for switching antidepressant treatment (comparison of strategies) following inadequate antidepressant response.

One study randomised to both initial treatment and switching strategy, and this was analysed separately (Thase2002a). It showed no statistically significant advantage for either strategy (sertraline to imipramine or imipramine to sertraline), although there was an advantage for those starting on imipramine and switching to sertraline following inadequate response (see Appendix 16c).

In addition to the blinded RCTs that were included in the meta-analyses, the search yielded two large open randomised studies. In the first (Baldomero et al., 2005), non-responders to a single antidepressant were randomised to receive venlafaxine or another antidepressant; in the second (STAR*D, Rush et al., 2003)185, non-responders to citalopram were randomised to switch to another antidepressant or receive an augmenting drug; those who did not remit were further randomised. Both these studies were excluded from the main analyses because they were open-label, but are described narrataively here because of their importance in the field.

The first large 24-week open-label study (Baldomero et al., 2005) comprised 3502 outpatients with major depressive disorder, subthreshold depressive symptoms (8.7%) and dysthymia (16%) whose depressive symptoms (HRSD scores above 17) had not responded to treatment with an antidepressant (most commonly an SSRI) for at least 4 weeks; 1830 of the participants were randomised to venlafaxine-XR (mean dose 164 mg) and 1672 to other antidepressants different from those used in earlier treatment and including fluoxetine (17%), paroxetine (21.3%), citalopram (20.1%), sertraline (19.1%) and mirtazapine (7.9%). There was little difference in mean endpoint depression scores between the venlafaxine group and the other antidepressant group: venlafaxine 7.89 (SD 6.5) and other antidepressants 8.84 (6.7). However, 967 people (52% of the number randomised) taking venlafaxine achieved remission (HRSD <= 7) as did 755 (45% of the number randomised) taking other antidepressants. The response rate (50% reduction in baseline HRSD scores) was 1262 (69%) in the venlafaxine group and 1034 (62%) in the other antidepressants group. Figures are calculated from the number randomised rather than the ‘intention to treat’ population used by the study authors.

As the STAR*D (Rush et al., 2003) study contained both switching and augmentation arms, the data from these studies are summarised in the augmentation section below.

Clinical summary

Given the paucity of evidence from switching studies, evidence from primary efficacy studies in which antidepressants were directly compared were also considered. Caution is required in extrapolating from these studies to those whose illness has not responded to sequential trials of antidepressant drugs.

Data from switching studies and head-to-head studies suggest that there may be a very small efficacy advantage for venlafaxine and escitalopram over other antidepressants. This advantage is too small to be clinically meaningful when all people with depression are considered together, but may be large enough to be clinically worthwhile in those who have not benefited from treatment with a first or second antidepressant. However, the current evidence is not sufficiently robust to form the basis of a recommendation.

12.3.3. Combining an antidepressant with another antidepressant

Introduction

**Combining antidepressant drugs with different modes of action is increasingly used in clinical practice. Combinations of serotonergic and noradrenergic drugs may result in a ‘dual action’ combination, while combinations of serotonergic drugs with different modes of action may be expected to increase serotonergic neurotransmission more than either drug alone.

While the efficacy of these combinations may be additive (this is not proven for the majority of combinations), so too may the toxicity. Both pharmacokinetic and pharmacodynamic interactions must be considered. Fluoxetine, fluvoxamine and paroxetine may increase TCA serum levels substantially and unpredictably, thereby increasing the risk of adverse effects (Taylor, 1995). Combinations of serotonergic antidepressants increase the risk of developing serotonin syndrome, which can be fatal. Features include confusion, delirium, shivering, sweating, changes in blood pressure and myoclonus.**

Studies considered186

No new studies of combination with a second antidepressant were found after inadequate response to the first, but so that the data could be analysed together the studies are presented in the style of the update in this section. There were data for a range of strategies, including adding mianserin, desipramine (not available in the UK), mirtazapine, moclobemide and atomoxetine to an antidepressant. Summary study characteristics of the included studies are presented in Table 109, with full details in Appendix 17c, which also includes details of excluded studies.

Table 109. Summary study characteristics of included studies of combining antidepressants in people whose depression had not responded adequately to treatment.

Table 109

Summary study characteristics of included studies of combining antidepressants in people whose depression had not responded adequately to treatment.

Clinical evidence

Evidence from the important outcomes and overall quality of evidence are presented in Table 110. The full profile and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 110. Summary evidence profile for combining an antidepressant versus antidepressant with/without placebo.

Table 110

Summary evidence profile for combining an antidepressant versus antidepressant with/without placebo.

Results showed that combination treatment tended to reduce symptoms of depression more than continuing with the existing single antidepressant at ‘standard’ dose. However, the data are not strong, and participants taking combination treatment reported more side effects than those taking a single antidepressant.

In a mixed population of patients there is some evidence that combining one anti-depressant with another leads to better outcomes on response, remission and mean endpoint scores compared with a single antidepressant at ‘standard’ dose. There is insufficient evidence to determine whether this is the case when compared with a single antidepressant at high dose.

Since the majority of studies used mianserin as the second antidepressant, the analyses are weighted towards this drug. Importantly, there are no RCTs of combinations of a TCA and irreversible MAOI or any two from a choice of venlafaxine, mirtazapine and reboxetine.

Clinical summary

There is some evidence that combinations of antidepressants are associated with a higher burden of side effects than a single antidepressant at either standard or high dose, but there is insufficient evidence to comment on the number of patients leaving treatment early.

12.3.4. Augmenting an antidepressant with an antipsychotic

Studies considered187

A total of five nine studies found in the update search met inclusion criteria for the review of antipsychotic augmentation (BERMAN2007, CORYA2006, KEITNER2009, MAHMOUD2007, MARCUS2008, MCINTRYRE2007B, SONG2007, THASE2007 [two studies reported in the same paper]). The previous guideline included only one study (Shelton2001). Summary study characteristics of the included studies are presented in Table 111, with full details in Appendix 17c, which also includes details of excluded studies.

Table 111. Summary study characteristics for antipsychotic augmentation.

Table 111

Summary study characteristics for antipsychotic augmentation.

Clinical evidence

There were data for augmentation with aripiprazole, olanzapine, risperidone and quetiapine. Evidence from the important outcomes and overall quality of evidence are presented in Table 112. The full profile and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 112. Summary evidence profile for augmentation with an antipsychotic versus antidepressant with/without placebo.

Table 112

Summary evidence profile for augmentation with an antipsychotic versus antidepressant with/without placebo.

Overall, there was a moderate, clinical important effect on symptoms of depression favouring antipsychotic augmentation, which was mirrored in small effects on remission and response. Results for individual antipsychotics were similar, but tended not to be statistically significant because of the small number of studies for each drug. There were no head-to-head trials. Participants taking antipsychotics were more likely to leave treatment early for any reason and specifically because of side effects. There were also more likely to report side effects.

Clinical summary

The previous guideline found little evidence on which to make an evidence-based recommendation regarding antipsychotic augmentation of antidepressants for people whose depression had not responded to treatment with an antidepressant alone. A number of studies have been published since, which when considered together, show a statistically significant, but clinically modest advantage for antipsychotic augmentation of an antidepressant over an antidepressant alone. Patients whose antidepressant is augmented by an antipsychotic are much more likely to leave treatment early because of side effects. This was most marked for quetiapine.

12.3.5. Augmenting an antidepressant with lithium

Introduction

Lithium is an established mood stabilising drug that is used in the treatment of mania and the prophylaxis of bipolar affective disorder. It is also widely used to augment antidepressant response in depression that has not responded adequately to initial treatment with an antidepressant.

Lithium is primarily excreted renally and can cause hypothyroidism, renal damage and a number of other adverse effects. Baseline biochemical tests and ongoing monitoring are essential. For example, serum lithium levels must be monitored to achieve a stable therapeutic level (see below). This should include monitoring 1 week after initiation (and 1 week after any dose change) until stable and then every 3 months; more details can be found in the NICE guideline on bipolar disorder (NICE, 2006c).

Lithium is a potentially toxic drug. Plasma levels of 0.5 to 1.0 mmol/L are usually considered to be therapeutic. Above 1.5 mmol/L toxicity invariably develops and death may occur at levels as low as 2.0 mmol/L. Many commonly prescribed drugs can interact with lithium to precipitate lithium toxicity (British Medical Association and the Royal Pharmaceutical Society of Great Britain, 2009; Taylor et al., 2007).

Studies considered188

No new studies were found that met inclusion criteria, but so that the data could be analysed together the studies are presented in the style of the update in this section. The data from the ten remaining studies were reanalysed without dividing the dataset by antidepressant-response history. Summary study characteristics of the included studies are presented in Table 113, with full details in Appendix 17c, which also includes details of excluded studies.

Table 113. Summary study characteristics for lithium augmentation.

Table 113

Summary study characteristics for lithium augmentation.

Clinical evidence

There was some evidence that lithium augmentation was effective in reducing symptoms of depression. Evidence from the important outcomes and overall quality of evidence are presented in Table 114. The full profile and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 114. Summary evidence profile for augmentation with lithium versus antidepressant with/without placebo.

Table 114

Summary evidence profile for augmentation with lithium versus antidepressant with/without placebo.

Clinical summary

There is some evidence of a clinically important advantage of adding lithium to an antidepressant over adding placebo, although this effect was not found for mean endpoint scores on all outcome measures. Adding lithium to an antidepressant appears to be less acceptable to patients, with just over 30% leaving treatment early compared with 17.4% taking placebo. There is insufficient evidence to determine whether this is due to side effects.

12.3.6. Augmenting an antidepressant with anticonvulsants

The following sections on augmenting an antidepressant with anticonvulsants marked by asterisks (**_**) are from the previous guideline and have not been updated except for style and minor clarification.

Introduction

Anticonvulsants are increasingly being prescribed for people with bipolar disorder; there is growing data related to their efficacy in the treatment of depression and mania and in the prophylaxis of bipolar disorder. No new data were found for augmentation of an antidepressant with carbamazepine or valproate.

Carbamazepine

**Carbamazepine has attracted the most interest because it was the first anticonvulsant to be shown to have efficacy in bipolar disorder and because it shares some neurochemical properties with tricyclic antidepressants. However, no RCTs met the inclusion criteria set by the GDG. There are some open studies (Dietrich & Emrich, 1998), and one RCT in major depression (Zhang et al., 2008), and some open studies in treatment-resistant depression (Cullen et al., 1991; Ketter et al., 1995) that show some benefit. It is noteworthy that in Cullen’s study a high percentage of the older patients who responded had to discontinue carbamazepine because of adverse effects.

Carbamazepine has a wide range of side effects, contraindications and interactions with other drugs. In the context of depression, it is noteworthy that co-administration of carbamazepine reduces TCA levels by up to 50% (Dietrich & Emrich, 1998) and SSRIs may interfere with carbamazepine metabolism leading to intoxication.

There is a lack of controlled data and a high likelihood of adverse effects or clinically important interactions and, therefore, carbamazepine cannot be recommended as a routine next-step treatment for poorly responsive depression.**

Valproate

There are no RCTs of valproate in unipolar major depression. Evidence to date suggests that valproate is more effective in preventing hypomania rather than depression in people with bipolar disorder.

One open study enrolled 33 patients with major depressive disorder in an 8-week study of valproate as monotherapy (Davis et al., 1996). Approximately 50% of the patients achieved remission. Valproate is associated with a number of side effects including significant weight gain. It can also increase plasma levels of other commonly prescribed drugs such as TCAs, quetiapine and warfarin. Fluoxetine may elevate valproate levels by interfering with its metabolism. Valproate is also a major human teratogen.

There are a lack of controlled data and a high likelihood of adverse effects or clinically important interactions and, therefore, valproate cannot be recommended in the routine management of depression that has not responded adequately to other treatments.

Lamotrigine

Lamotrigine is used in the treatment of partial and generalised seizures. In clinical trials in epilepsy it was noted that those who received lamotrigine reported improvements in mood, alertness and social interaction.

Studies have shown evidence of efficacy for lamotrigine in bipolar depression (Geddes et al., 2009). However, in a study of 437 patients with major depressive disorder randomised to lamotrigine, desipramine or placebo, ‘last observation carried forward’, ratings demonstrated no difference between groups (Hurley, 2002). In a further RCT (Normann et al., 2002), 40 patients with depression (30 unipolar, 10 bipolar) were given lamotrigine (200 mg) or placebo added to paroxetine (40 mg) for 9 weeks. There was no benefit for lamotrigine over placebo in HRSD scores at endpoint. There was a high frequency of adverse effects and dropouts in both groups. Barbosa and colleagues (2003) reported on 23 patients with depression (65% major depressive disorder) who had failed at least one trial of an antidepressant, and were randomised to receive either placebo or 25 mg to 100 mg of lamotrigine in addition to fluoxetine 20 mg/day. There was no statistical difference in HRSD or MADRS ratings between the two groups at 6 weeks, although there was a benefit in a secondary outcome measure of responders based on the CGI. A further small study (Santos et al., 2008; N = 34) of outpatients whose depression had not responded to at least two antidepressants of different classes for at least 6 weeks at the highest tolerated dose, compared augmentation with lamotrigine in doses up to 200 mg with augmentation with placebo for 8 weeks. Participants continued with their existing antidepressant. There was no advantage for lamotrigine augmentation when endpoint depression scores were compared.

Finally, in an 8-week randomised open-label study of antidepressant augmentation with either lamotrigine (150 mg) or lithium (serum level 0.6 to 0.8 mmol/L) in 34 inpatients with a diagnosis of major depressive disorder whose depression had not responded to two trials of different antidepressants, Schindler and Anghelescu (2007) reported no significant difference between the treatment groups at endpoint based on HRSD scores, remission or response.

**In view of the lack of positive data, lamotrigine cannot be recommended for use in unipolar disorder. Although it is generally well tolerated and free of major interactions, it can cause a severe rash that can be life-threatening in a small minority of cases. Its profile in epilepsy and bipolar disorder suggests that further trials of lamotrigine in treatment-resistant depression are worthwhile.

There are no data that indicate that other anticonvulsants – for example, gabapentin or topiramate – can be recommended in the treatment of depression.**

12.3.7. Augmenting an antidepressant with pindolol

The following sections on augmenting an antidepressant with pindolol marked by asterisks (**_**) are from the previous guideline and have not been updated except for style and minor clarification.

Introduction

**Serotonergic antidepressants inhibit the reuptake of serotonin into the presynaptic neurone thus increasing serotonergic neurotransmission. The immediate effect of this increase is to stimulate serotonin 1a autoreceptors, which results in a decrease in serotonin release. In time, these autoreceptors become desensitised and serotonin release returns to normal. This, in combination with the inhibition of serotonin reuptake, is thought to lead to the onset of the antidepressant effect.

Pindolol is primarily an adrenergic b-blocking drug, which also blocks serotonin 1a autoreceptors. The co-administration of pindolol with a serotonergic antidepressant could be expected to result in an immediate increase in serotonin neurotransmission, thus eliminating the delay in onset of antidepressant response.

As well as being used to speed the onset of antidepressant response, pindolol has also been used to augment the efficacy of antidepressant drugs in acute-phase non-responders and treatment-resistant depression.**

Studies considered189,190

**Twenty-four studies were found in a search of electronic databases, six of which met the inclusion criteria set by the GDG (Bordet1998, Maes1999, Perez1997, Perez1999, Tome1997, Zanardi1997) and 18 of which did not.** No new studies were found in the update search.

**Only studies comparing pindolol plus an antidepressant with pindolol plus placebo were included in the analyses. Apart from one study (Perez1999), which included clomipramine as well as a range of SSRIs, all studies used a single SSRI as the antidepressant. Efficacy data were available from up to 282 participants and tolerability data from up to 333 participants.

All included studies were published between 1997 and 1999 with participants being randomised to an experimental treatment phase of between 10 days and 6 weeks (mean = 4.25 weeks).

In two studies participants were described as inpatients (Maes1999, Zanardi1997), in a further two as outpatients (Perez1999, Tome1997), in one as primary care (Perez1997) and in the remaining trial participants were from mixed sources (Bordet1998). In no trial were participants exclusively older or experiencing atypical depression. The mean dose of pindolol was 9.23 mg, ranging from 7.5 mg to 15 mg.

No trial was classified acute-phase non-responder, and only one study included patients who had not responded to previous antidepressant treatment (Perez1999). Here patients were randomised to receive augmentation for 10 days with either pindolol (7.5 mg) or placebo after receiving fluoxetine (40 mg), fluvoxamine (200 mg), paroxetine (40 mg) or clomipramine (150 mg) for at least 6 weeks beforehand. In addition the participants’ depression had already failed to respond to between one and four courses of antidepressants (median two). Most patients were outpatients aged 18 to 65. Results from a separate analysis of this trial are presented below.

Outcomes are classified according to when assessment measures were taken. Up to 14 days after treatment was begun was categorised ‘early assessment point’ and more than 20 days was categorised ‘late assessment point’. Three studies (Bordet1998, Tome1997, Zanardi1997) gave outcomes at both assessment points.

Clinical evidence statements: effect of treatment on efficacy191

Early assessment point

There is evidence suggesting that there is no clinically important difference between SSRIs plus pindolol and SSRIs plus placebo on increasing the likelihood of achieving a 50% reduction in symptoms of depression by the tenth day of treatment (N = 2; n = 160; RR = 0.95; 95% CI, 0.82 to 1.11).

There is insufficient evidence to determine whether there is a clinically important difference between SSRIs plus pindolol and SSRIs plus placebo on:

  • increasing the likelihood of achieving remission by the 10th or 14th day of treatment (K = 3; N = 222; Random effects RR = 0.73; 95% CI, 0.44 to 1.20)
  • reducing symptoms of depression by the 10th or 14th day of treatment (K = 3; N =237; Random effects SMD = –0.30; 95% CI, –0.88 to 0.28).
Late assessment point

There is insufficient evidence to determine whether there is a clinically important difference between SSRIs plus pindolol and SSRIs plus placebo on increasing the likelihood of achieving a 50% reduction in symptoms of depression by the 35th or 42nd day of treatment (K = 3; N = 214; RR = 0.75; 95% CI, 0.54 to 1.03).

There is some evidence suggesting that there is a clinically important difference favouring SSRIs plus pindolol over SSRIs plus placebo on increasing the likelihood of achieving remission by the 21st, 28th or 42nd day of treatment (K = 3; N = 253; RR = 0.73; 95% CI, 0.55 to 0.98).

There is evidence suggesting that there is a clinically important difference favouring SSRIs plus pindolol over SSRIs plus placebo on reducing symptoms of depression by the 21st, 35th or 42nd day of treatment, but the size of this difference is unlikely to be of clinical importance (K = 4; N = 282; SMD = −0.26; 95% CI, −0.49 to −0.02).

Acceptability of treatment

There is insufficient evidence to determine whether there is a clinically important difference between SSRIs plus pindolol and SSRIs plus placebo on any measure of tolerability.

Clinical evidence statements: effect of treatment on efficacy for people whose depression is treatment resistant192

Early assessment point

For people whose depression is treatment resistant there is evidence suggesting that there is no clinically important difference when assessment is made between days 10 and 14 between pindolol augmentation and antidepressant monotherapy on:

  • increasing the likelihood of achieving a 50% reduction in symptoms of depression (K = 1; N = 80; RR = 1; 95% CI, 0.85 to 1.18)
  • increasing the likelihood of achieving remission (K = 1; N = 80; RR = 1.03; 95% CI, 0.88 to 1.2).

There is insufficient evidence to determine if there is a clinically important difference between pindolol augmentation and antidepressant monotherapy on reducing symptoms of depression in people whose depression is treatment resistant (K = 1; N = 80; WMD = 1.6; 95% CI, −0.96 to 4.16).

Acceptability of treatment

There are no data on the acceptability of treatment for people whose depression is treatment resistant.

Clinical summary

While there is some evidence of a modest advantage at 21 to 42 days favouring the addition of pindolol to antidepressants over adding placebo on achieving remission, this effect is not evident for response or mean endpoint scores. There is no evidence of any effect on outcomes in people whose depression is treatment resistant at early assessment point. No data were available for late assessment points.

There is insufficient evidence to comment on the tolerability of adding pindolol to antidepressants.

It should be noted that there is uncertainty regarding optimum dose and duration of treatment.**

12.3.8. Augmenting an antidepressant with triiodothyronine

The following sections on augmenting an antidepressant with triiodothyronine (T3) marked by asterisks (**_**) are from the previous guideline and have not been updated except for style and minor clarification.

Introduction

**Consistent with the observations that the prevalence of depression is increased in hypothyroidism (Loosen, 1987), and subclinical hypothyroidism is more prevalent in people who are clinically depressed (Maes et al., 1993), T3 has been used as an anti-depressant augmenting agent both to increase the speed of onset of antidepressant response and to increase the magnitude of response.

Increase the speed of onset of antidepressant response

T3, at a dose of 25 mcg per day, may hasten response to TCAs and this effect may be more robust in women (Altshuler et al., 2001). The optimal duration of treatment is unknown although there is a suggestion in the literature that T3 may be safely withdrawn once response has been achieved (Altshuler et al., 2001). There are no studies with SSRIs or any of the newer antidepressants.**

Increase the magnitude of antidepressant response

Although the RCT that satisfied the inclusion criteria set by the GDG found T3 and lithium to be equally effective and superior to placebo (see below), several ‘negative’ non-RCTs also exist (Steiner et al., 1978; Gitlin et al., 1987; Thase et al., 1989). The response rate has been variable across studies (Aronson et al., 1996). All studies used TCAs. There are no studies with SSRIs or any of the newer antidepressants apart from STAR*D (Rush et al., 2003), which used an open-label design. T4 has been shown to be inferior to T3 in one study (Joffe & Singer, 1990). Most studies used a dose of 37.5 mcg T3 per day. The optimum duration of treatment is unknown.

Studies considered193,194

One study was found in a search of electronic databases (Joffe1993a), and this met the inclusion criteria set by the GDG. It compared a range of antidepressants augmented with T3 (37.5 mcg) with antidepressants augmented with placebo. Participants were outpatients who had not achieved remission after 5 weeks’ treatment with either desipramine or imipramine. No new double-blind studies were found in the update search, although the STAR*D (Rush et al., 2003) trial includes a T3 augmentation arm (described elsewhere in this chapter).

Clinical evidence statements195

Effect of treatment on efficacy outcomes

**There is some evidence suggesting that there is a clinically important difference favouring T3 augmentation over antidepressant plus placebo on increasing the likelihood of achieving a 50% reduction in symptoms of depression (K = 1; N = 33; RR = 0.51; 95% CI, 0.27 to 0.94).

There is insufficient evidence to determine if there is a clinically important difference between T3 augmentation and antidepressant plus placebo on reducing symptoms of depression (K = 1; N = 33; WMD = −3.9; 95% CI, −8.86 to 1.06).

Acceptability of treatment

There was no evidence on which to assess the acceptability of treatment.

Clinical summary

There is little evidence on which to make an evidence-based recommendation of augmentation of antidepressants with T3 for the treatment of treatment-resistant depression. The prevalence of cardiovascular disease is increased in people with depression (Glassman & Shapiro, 1998) and T3 should be used with caution in cardiovascular disease. Potential adverse effects include tachycardia, anginal pain and arrhythmias. TCAs also have cardiac side effects including arrhythmias, tachycardia and postural hypotension. Caution is advised in combining TCAs and T3.**

12.3.9. Augmenting an antidepressant with a benzodiazepine

The following sections on augmenting an antidepressant with a benzodiazepine marked by asterisks (**_**) are from the previous guideline and have not been updated except for style and minor clarification.

Introduction

**Depression and anxiety commonly co-exist and insomnia is a common symptom of depression. Antidepressants usually take 2 to 4 weeks to have a clinically important effect.

Benzodiazepines are effective anxiolytic and hypnotic drugs with an immediate onset of action and therefore could be expected to produce early improvement in some symptoms of depression. They do not have a specific antidepressant effect.

Benzodiazepines are associated with tolerance and dependence and withdrawal symptoms can occur after 4 to 6 weeks of continuous use. To avoid these problems, it is recommended that they should not routinely be prescribed for their hypnotic or anxiolytic effects for longer than 4 weeks (Royal College of Psychiatrists, 1997; British Medical Association and the Royal Pharmaceutical Society of Great Britain, 2009).

The National Service Framework for Mental Health (Department of Health, 1999) discourages the use of benzodiazepines and many primary care prescribing incentive schemes include low prescribing rates for benzodiazepines as a marker of good practice. A Cochrane review, however, concludes that early time-limited use of benzodiazepines in combination with an antidepressant drug may accelerate treatment response (Furukawa et al., 2002b).

Studies considered196,197

The GDG used an existing review (Furukawa et al., 2002b) as the basis for this section. The original review included nine studies of which four met the inclusion criteria set by the GDG (Feet1985, Nolen1993, Scharf1986, Smith1998). New searches of electronic databases found an additional study (Smith2002), which was included in the review. Together these studies provided tolerability data from up to 196 participants and efficacy data from up to 186 participants.**

No new studies were found in the update search.

**All included studies were published between 1985 and 2002 and were between 3 and 12 weeks’ long (mean = 7 weeks). One study was of inpatients (Nolen1993), three of outpatients (Feet1985, Smith1998, Smith2002) and in the remaining study (Scharf1986) participants were from mixed sources. No study was undertaken in primary care and none was exclusively of older participants or people with atypical depression. Other than in Feet1985, where participants had been ‘treated in general practice without success’, study participants were not described as having failed previous courses of antidepressants.

All studies compared an antidepressant plus benzodiazepine with an antidepressant plus placebo. The included trials used the following antidepressant/benzodiazepine combinations:

The mean dose of TCAs was between 122.5 mg and 200 mg, and fluoxetine was given at between 20 mg and 40 mg.

Clinical evidence statements198

Effect of treatment on efficacy

There is insufficient evidence to determine whether there is a clinically important difference between antidepressants plus a benzodiazepine and antidepressants plus placebo on any efficacy measure.

Acceptability of treatment

There is insufficient evidence to determine whether there is a clinically important difference between antidepressants plus a benzodiazepine and antidepressants plus placebo on any tolerability measure.

Clinical summary

There is insufficient evidence to determine whether there is any effect of adding a benzodiazepine to antidepressant treatment in terms of both efficacy and tolerability.**

12.3.10. Augmenting an antidepressant with buspirone

The sections on augmenting an antidepressant with buspirone marked by asterisks (**_**) are from the previous guideline and have not been updated except for style and minor clarification.

Introduction

There are no extractable efficacy data from double-blind RCTS of buspirone augmentation. Buspirone was used in the STAR*D study (Rush et al., 2003), which had an open-label randomised design in which buspirone augmentation of citalopram did not differ significantly in efficacy from bupropion addition in terms of response to treatment, but there was a greater reduction in self-rated depression scores in people taking bupropion.

Clinical evidence statements199

Acceptability of treatment

There is insufficient evidence to determine if there is a clinically important difference between buspirone augmentation and SSRI monotherapy on any tolerability measure. In the STAR*D study (Rush et al., 2003), dropout because of side effects was greater with buspirone augmentation than bupropion addition to citalopram.

Clinical summary

**There is no double-blind placebo-controlled evidence on which to make an evidence-based recommendation of augmentation of antidepressants with buspirone for the treatment of treatment-resistant depression.**

12.3.11. Augmenting an antidepressant with atomoxetine

The following section on augmenting an antidepressant with atomoxetine is new for this guideline update.

Studies considered200

One study was found in the update search of augmentation with atomoxetine. Summary study characteristics of the included studies are presented in Table 115, with full details in Appendix 17c, which also includes details of excluded studies.

Table 115. Summary study characteristics for augmentation with atomoxetine.

Table 115

Summary study characteristics for augmentation with atomoxetine.

Clinical evidence

Evidence from the important outcomes and overall quality of evidence are presented in Table 116. The full evidence profiles and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 116. Summary evidence profile for atomoxetine augmentation.

Table 116

Summary evidence profile for atomoxetine augmentation.

Clinical summary

Augmenting an antidepressant with atomoxetine showed no significant effect on symptoms of depression, and increased the number of people leaving treatment early for any reason because of side effects compared with those taking an antidepressant alone.

12.3.12. Sequenced Treatment Alternatives to Relieve Depression (Rush et al., 2003)

Sequenced Treatment Alternatives to Relieve Depression (STAR*D) (Rush et al., 2003) is a four-level study designed to assess treatments in patients who had not responded to previous treatment. At each level patients who had not responded to treatment at the previous level were randomised to different treatment options. At the first level, all patients received citalopram. Those not responding (QIDS-SR >5) moved to level 2 where they were randomised to switch to another antidepressant (bupropion, sertraline or venlafaxine-ER) or to receive an augmentation treatment (bupropion, buspirone or CBT). Those not responding to treatment in level 2 moved to level 3 where they were randomised again to switch to mirtazapine or nortriptyline or to receive an augmentation agent (lithium or T3 for those on bupropion, sertraline or venlafaxine-ER). In addition, those who had not responded to CBT at level 2 were randomised to bupropion or venlafaxine-ER to ensure that all those in level 3 had failed two courses of antidepressants. Those not responding moved to level 3. Those not responding to level 3 treatment moved to level 4 and were re-randomised to tranylcypromine or mirtazapine plus venlafaxine-ER.

The study was designed to be as analogous as possible to real clinical practice. In order to achieve this, patients were allowed to opt out of being randomised to drug switching, augmentation treatments and, in level 2, to CBT. They were not allowed to opt out of randomisation to a particular agent within the drug switching or drug augmentation arms. Also all treatments were given open label. Medication was free to trial participants but they had to pay for CBT treatment (Weissman, 2007). The patient preference aspect of the trial meant that there were 12 permutations of randomisation preferences at level 2, which greatly adds to the complexity of the trial. For example, only data from patients accepting randomisation to an augmenting or switching option including CBT can be used in comparisons with CBT (either as a switching option or as an augmenting treatment).

It is difficult to draw conclusions about suitable sequencing options since there are so many permutations of treatments possible within the trial. Patients who reach level 4 (that is, have failed three drug trials or three drugs plus a course of CBT) will have taken a variety of routes through the study. They may have taken citalopram continuously (augmented with two separate agents), or may have tried three different single antidepressants, or switched from single to combination drugs and back again. The percentage remission achieved by each treatment strategy is shown in Table 117.

Table 117. Percentage remission by treatment strategy in STAR*D (Rush et al., 2003).

Table 117

Percentage remission by treatment strategy in STAR*D (Rush et al., 2003).

Data from RCTs (see Table 118) suggest that switching from one antidepressant to another may be clinically worthwhile, with increased remission rates of around 20% but with some drugs reporting higher remission rates; within-class switches are associated with remission rates of approximately 20%. Open switching studies report higher remission rates when SSRI non-responders are switched to venlafaxine (BALDOMERO2005 [Baldermero et al., 2005]; Rush et al., 2003). This advantage holds in blinded studies, but the magnitude of the benefit is considerably more modest (Rush et al., 2003).

Table 118. Raw remission rates following switch to another antidepressant (data from RCTs).

Table 118

Raw remission rates following switch to another antidepressant (data from RCTs).

12.3.13. Clinical summary for ‘next-step’ treatments

The evidence for effective strategies in people whose depression has not responded adequately to treatment is not strong. A common first-line strategy, increasing the dose, is also not supported by convincing evidence of effectiveness, although this strategy may well be effective in some people, particularly if they have been able to tolerate the drug at the initial dose.

The evidence for switching to another antidepressant is stronger, but data for switching between classes of antidepressant is not. Overall, however, switching is likely to be a worthwhile strategy, and data from primary efficacy head-to-head studies suggest that venlafaxine and escitalopram may offer marginal benefits over other antidepressants in this regard. Augmenting with lithium, a second antidepressant or an antipsychotic is also worthwhile, but the effect size is modest clinically and the side effect burden increased. The main message from the STAR*D study (Rush et al., 2003) is that some patients will achieve remission with each successive treatment strategy although the proportion doing so falls each time. The lack of good objective data to clearly demonstrate the superior efficacy of one strategy over another probably reflects the fact that the overall difference in effect size between strategies is likely to be small. As was seen in the STAR*D study (Rush et al., 2003), some patients have clear preferences for one treatment over another based, at least in part, on perceived acceptability of the treatment and on degree of response to the current treatment.

12.3.14. Health economic evidence and considerations

No evidence on the cost effectiveness of ‘next-step’ treatments was identified by the systematic search of the economic literature. Details on the methods used for the systematic search of the economic literature are described in Chapter 3, Section 3.6.1.

12.3.15. From evidence to recommendations

Since the evidence for sequencing pharmacological strategies for people whose depression has not responded adequately to initial treatment is weak, the recommendations in the previous guideline are largely unchanged, although they have been updated to reflect new NICE style. Choice of new medication should be guided by similar principles to those guiding choice of initial medication, for example, a drug’s potential for side effects. Since it is possible that poor response to initial treatment may be because the treatment was not properly initiated or adhered to, these factors should be reviewed first and increased frequency of follow-up considered.

12.3.16. Clinical practice recommendations

12.3.16.1.

When reviewing drug treatment for a person with depression whose symptoms have not adequately responded to initial pharmacological interventions:

  • check adherence to, and side effects from, initial treatment
  • increase the frequency of appointments using outcome monitoring with a validated outcome measure
  • be aware that using a single antidepressant rather than combination medication or augmentation (see 12.3.16.9 to 12.3.16.13) is usually associated with a lower side-effect burden
  • consider reintroducing previous treatments that have been inadequately delivered or adhered to, including increasing the dose
  • consider switching to an alternative antidepressant.

The evidence for an advantage of switching to another antidepressant over continuing treatment with the existing antidepressant is not strong. In addition, there is insufficient robust evidence about which antidepressant to switch to. Choice should therefore be guided by side effects and possible interactions during the period of the switch.

12.3.16.2.

When switching to another antidepressant, be aware that the evidence for the relative advantage of switching either within or between classes is weak. Consider switching to:

  • initially a different SSRI or a better tolerated newer-generation antidepressant
  • subsequently an antidepressant of a different pharmacological class that may be less well tolerated, for example venlafaxine, a TCA or an MAOI.
12.3.16.3.

Do not switch to, or start, dosulepin because evidence supporting its tolerability relative to other antidepressants is outweighed by the increased cardiac risk and toxicity in overdose.

12.3.16.4.

When switching to another antidepressant, which can normally be achieved within 1 week when switching from drugs with a short half-life, consider the potential for interactions in determining the choice of new drug and the nature and duration of the transition. Exercise particular caution when switching:

  • from fluoxetine to other antidepressants, because fluoxetine has a long half-life (approximately 1 week)
  • from fluoxetine or paroxetine to a TCA, because both of these drugs inhibit the metabolism of TCAs; a lower starting dose of the TCA will be required, particularly if switching from fluoxetine because of its long half-life
  • to a new serotonergic antidepressant or MAOI, because of the risk of serotonin syndrome201
  • from a non-reversible MAOI: a 2-week washout period is required (other antidepressants should not be prescribed routinely during this period).

Following several courses of treatment it may be appropriate to refer someone with depression to a specialist (for example, someone with a special interest in treating depression or a specialist service). Before deciding the next course of action, there should be a thorough assessment of factors affecting treatment choice, including suicide risk and associated comorbidities. It may be appropriate to re-introduce previous treatments, if these were not adequately delivered or adhered to.

12.3.16.5.

For a person whose depression has failed to respond to various strategies for augmentation and combination treatments, consider referral to a practitioner with a specialist interest in treating depression, or to a specialist service202.

12.3.16.6.

The assessment of a person with depression referred to specialist mental health services should include:

  • their symptom profile, suicide risk and, where appropriate, previous treatment history
  • associated psychosocial stressors, personality factors and significant relationship difficulties, particularly where the depression is chronic or recurrent
  • associated comorbidities including alcohol and substance misuse, and personality disorders203.
12.3.16.7.

In specialist mental health services, after thoroughly reviewing previous treatments for depression, consider reintroducing previous treatments that have been inadequately delivered or adhered to204.

12.3.16.8.

Medication in secondary care mental health services should be started under the supervision of a consultant psychiatrist.

Given the higher side-effect burden of taking two drugs rather than one, combining medication would not normally be an initial next-step option. However, there is some evidence of efficacy. Most of the data published since the previous guideline are for augmentation of an antidepressant with an antipsychotic, and this shows some benefit. However, antipsychotics do not have UK marketing authorisation for use in depression. There is still limited evidence for combinations of antidepressants. The recommendations are largely unchanged, but the one for augmentation with a benzodiazepine has been amended since this strategy is recommended elsewhere in the guideline for the short-term management of agitation.

12.3.16.9.

When using combinations of medications (which should only normally be started in primary care in consultation with a consultant psychiatrist):

  • select medications that are known to be safe when used together
  • be aware of the increased side-effect burden this usually causes
  • discuss the rationale for any combination with the person with depression, follow GMC guidance if off-label medication is prescribed, and monitor carefully for adverse effects
  • be familiar with primary evidence and consider obtaining a second opinion when using unusual combinations, the evidence for the efficacy of a chosen strategy is limited or the risk–benefit ratio is unclear
  • document the rationale for the chosen combination.
12.3.16.10.

If a person with depression is informed about, and prepared to tolerate, the increased side-effect burden, consider combining or augmenting an antidepressant with:

12.3.16.11.

When prescribing lithium:

  • monitor renal and thyroid function before treatment and every 6 months during treatment (more often if there is evidence of renal impairment).
  • consider ECG monitoring in people with depression who are at high risk of cardiovascular disease
  • monitor serum lithium levels 1 week after initiation and each dose change until stable, and every 3 months thereafter.
12.3.16.12.

When prescribing an antipsychotic, monitor weight, lipid and glucose levels, and side effects (for example, extrapyramidal side effects and prolactin-related side effects with risperidone).

12.3.16.13.

The following strategies should not be used routinely:

  • augmentation of an antidepressant with a benzodiazepine for more than 2 weeks as there is a risk of dependence
  • augmentation of an antidepressant with buspirone, carbamazepine, lamotrigine or valproate as there is insufficient evidence for their use
  • augmentation of an antidepressant with pindolol or thyroid hormones as there is inconsistent evidence of effectiveness206.

12.4. ELECTROCONVULSIVE THERAPY

12.4.1. Introduction

Electroconvulsive therapy (ECT) has been used as a treatment for depression since the 1930s. In its modern form ECT is perceived by many healthcare professionals to be a safe and effective treatment for severe depression that has not responded to other standard treatments (Geddes et al., 2003b). But many others, including many patient groups, consider it to be an outdated and potentially damaging treatment (Rose et al., 2003). During ECT, an electric current is passed briefly through the brain, via electrodes applied to the scalp, to induce generalised seizure activity. The individual receiving treatment is placed under general anaesthetic and muscle relaxants are given to prevent body spasms. The ECT electrodes can be placed on both sides of the head (bilateral placement) or on one side of the head (unilateral placement). Unilateral placement is usually to the non-dominant side of the brain, with the aim of reducing cognitive side effects. The standard bilateral placement is bitemporal/temporofrontal but some studies have used bifrontal placement in the hope of reducing cognitive side effects associated with the standard placement. The number of sessions undertaken during a course of ECT usually ranges from six to twelve, although a substantial minority of patients responds to fewer than six sessions. ECT is usually given twice a week in the UK; less commonly it is given once a fortnight or once a month as continuation or maintenance therapy to prevent the relapse of symptoms. It can be given on either an inpatient or day patient basis.

ECT causes short-term disorientation immediately after treatment and may cause short- or long-term memory impairment for past events (retrograde amnesia) and current events (anterograde amnesia). These effects appear to be dose related and depend on electrode placement, possibly the type of electrical stimulus and patient characteristics (Ingram et al., 2008). However the persistence, severity and precise characterisation of such impairments are still a subject of debate. There is preliminary evidence that prolonged short-term disorientation immediately after treatment predicts retrograde amnesia after the end of a course of treatment (Sobin et al., 1995) but not 2 months after the course. Cognitive impairments have been highlighted as a particular concern by many patients, especially retrograde amnesia for autobiographical events (Rose et al., 2003). There is no simple relationship between subjective cognitive impairment and cognitive test measures, which has contributed to polarising views about the relative risks and benefits of ECT.

At present there is a lack of consensus as to the best method of assessing cognitive function during a course of ECT. The benefit of using only a global measure such as the MMSE in its original or modified form (3MSE) is uncertain given the inconsistent effects of ECT on these measures in trials. And given the evidence that the ability to learn new material (anterograde memory) recovers after the end of ECT treatment, a main concern is in the early detection and minimisation of persistent retrograde memory loss, particularly for important autobiographical memories. Detecting cognitive impairments only at the end of treatment does not give the practitioner the opportunity to alter treatment to attempt to minimise this, although it may lead the practitioner to consider cognitive remediation; there is no evidence, however, to show that this is effective. A battery consisting of a formal mood rating scale (MADRS), the 3MSE, an autobiographical memory task, a word learning task, and tests of digit span forward and backward has been suggested (Porter et al., 2008), but it takes an hour to administer.

In line with NICE policy regarding the relationship of technology appraisals to clinical practice guidelines, this guideline updates the NICE technology appraisal on ECT (TA59) only for depression in adults (the TA covered the use of ECT in the treatment of mania and schizophrenia as well as depression in children and adolescents; NICE, 2003).

Key points to emerge from the reviews underpinning the NICE TA on ECT (NICE, 2003), which concluded that ECT is an effective treatment, include:

  • real ECT had greater short-term benefit than sham ECT
  • ECT had greater benefit than the use of certain antidepressants
  • bilateral ECT was reported to be more effective than unilateral ECT
  • the combination of ECT with pharmacotherapy was not shown to have greater short-term benefit than ECT alone
  • cognitive impairment does occur but may only be short term
  • compared with placebo, continuation pharmacotherapy with tricyclic antidepressants and/or lithium reduced the rate of relapses in people who had responded to ECT
  • preliminary studies indicate that ECT is more effective than repetitive transcranial magnetic stimulation.

12.4.2. Databases searched and the inclusion/exclusion criteria

For the updated review double-blind RCTs were sought that compared ECT either with sham ECT or another active treatment in the treatment of people experiencing an acute depressive episode or in relapse prevention following successful treatment (either with ECT or another treatment). Information about the databases searched and the inclusion/exclusion criteria used are presented in Table 119. Details of the search strings used are in Appendix 8.

Table 119. Databases searched and inclusion/exclusion criteria for clinical effectiveness of ECT.

Table 119

Databases searched and inclusion/exclusion criteria for clinical effectiveness of ECT.

12.4.3. Studies considered207

In total, 21 new trials were found from searches of electronic databases. These included: ten trials comparing ECT with transcranial magnetic stimulation (TMS), which the GDG did not review since NICE has produced guidance on TMS (NICE, 2007d); four trials of continuation treatment following successful treatment with ECT (two of which included continuation ECT), which are considered in the section on relapse prevention, and eight comparing bilateral with unilateral ECT, which are considered in the section on next-step treatments. Several studies included populations with a relatively high proportion of participants with bipolar disorder (up to 30%). These were included since ECT is not known to cause switching to mania (and, indeed, is used as a treatment for mania).

Summary study characteristics of the included studies are presented in Table 120, with full details in Appendix 17c, which also includes details of excluded studies.

Table 120. Summary study characteristics of studies of ECT or of treatment following successful ECT published since the systematic reviews underpinning the NICE TA were undertaken.

Table 120

Summary study characteristics of studies of ECT or of treatment following successful ECT published since the systematic reviews underpinning the NICE TA were undertaken.

Two older trials on relapse prevention following response to ECT were also discussed narratively (Lauritzen1996, Sackheim2001); see Section 12.4.5.

12.4.4. Clinical evidence for ECT as a next-step treatment

The TA reviews of ECT compared with sham ECT and with pharmacological interventions were not updated because no new studies were found. However, the review comparing bilateral ECT with unilateral ECT, including a sub-analysis by dose, was updated. In addition a narrative review of cognitive impairment related to electrode placement and dose was undertaken.

Bilateral ECT versus unilateral ECT

A review by Geddes and colleagues (2003b) was used as the basis of this review. The effect sizes reported in the published paper were input into CMA (Comprehensive Meta-Analysis) and combined with effect sizes from the eight new studies found (see Table 120 for a summary of these studies). The overall SMD calculated by Geddes and colleagues (2003b) from 22 studies and 1,137 particpiants was −0.322 (Random effects) (−0.458 to −0.186). With the addition of the relevant new data the SMD effect size was reduced slightly to −0.23 (Random effects) (−0.37, −0.09) (31208 studies, 1,693 participants; I2 = 39%), thus confirming an overall small to medium effect favouring bilateral ECT (see Figure 10).

Figure 10. Bilateral ECT versus unilateral ECT: updated forest plot.

Figure 10

Bilateral ECT versus unilateral ECT: updated forest plot.

Bilateral ECT versus unilateral ECT – the effect of dose and electrode placement on efficacy

For this guideline update, a sub-analysis by dose was also undertaken on efficacy related to electrode placement. This topic was also included in the review by Geddes and colleagues (2003b), which included seven studies comparing different doses of unilateral ECT and different doses of bilateral ECT, as well as five that specifically compared bilateral ECT with unilateral ECT at doses related to seizure threshold. These five studies were included in the sub-analysis (SACKHEIM1993, SACKHEIM2000; Malitz et al., 1986; Sackeim et al., 1987; Letemendia et al., 1993).

Dose was classified based on percentage above seizure threshold (one new study described doses as ‘high’ [STOPPE2006]). Doses described as ‘just above seizure threshold’ were classified 0%. The doses given in the studies available for the sub-analysis are in Table 121.

Table 121. Doses (% above seizure threshold) of bilateral ECT and unilateral ECT given in the available studies.

Table 121

Doses (% above seizure threshold) of bilateral ECT and unilateral ECT given in the available studies.

Low-dose unilateral ECT was defined as doses up to 150% above seizure threshold (that is, including low and standard doses used clinically) and high-dose unilateral ECT was defined as doses over 150% above seizure threshold. There was insufficient evidence to show a difference between low-dose bilateral ECT and low-dose unilateral ECT from the available studies in this subset, although the direction of effect was similar to that in the full set (see Table 122). On one outcome measure (non-remission) high-dose unilateral ECT tended to be more effective than low-dose bilateral ECT but this was not clinically important and no differential benefit was suggested with the other outcome measures. Evidence from the important outcomes and overall quality of evidence are presented in Table 122. The full evidence profiles and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively.

Table 122. Summary evidence profile for acute-phase ECT: bilateral ECT versus unilateral ECT.

Table 122

Summary evidence profile for acute-phase ECT: bilateral ECT versus unilateral ECT.

A visual inspection of the forest plots indicated that there appears to be neither no consistent effect for different bilateral electrode placement (bifrontal or bitemporal) nor a consistent relationship between electrode placement and dose, although there are insufficient studies to allow these factors to be explored systematically.

Cognitive side effects related to electrode placement and dose

Geddes and colleagues (2003b) reported that patients who received bilateral ECT seemed to take longer to recover orientation than those treated with unilateral ECT (based on six trials that reported this), and that they showed greater impairment in retrograde memory (based on four trials that reported this) and anterograde memory (seven trials reported this). Geddes and colleagues (2003b) also report that they found only two trials reporting long-term data, which were both small and underpowered, and which found no long-term differences between bilateral and unilateral ECT on cognitive functioning.

In the studies considered the GDG has taken bifronto-temporal placement as bitemporal. Combining the new studies with relevant studies from Geddes and colleagues (2003b) there was comparison between different doses of bitemporal ECT and unilateral ECT in six studies, between bifrontal ECT and unilateral ECT in four studies and between bifrontal ECT and bitemporal ECT in one study (see Table 123). In SACKHEIM1993 and SACKEIM2008 approximately 30% of patients had bipolar disorder and in SIENAERT2008 20% of patients had bipolar disorder; both were included in this review of cognitive effects.

Table 123. Studies comparing bilateral and unilateral ECT: reported differences in cognitive functioning and efficacy.

Table 123

Studies comparing bilateral and unilateral ECT: reported differences in cognitive functioning and efficacy.

The new studies had differences in bilateral electrode placement (bifrontal compared with the standard bitemporal placement) and in stimulus pulse width (ultra brief pulse compared with standard brief pulse). There was variation in the lower/‘standard’ dose of bitemporal ECT with 150% above seizure threshold often used in key US studies compared with lower UK recommendations from the Royal College of Psychiatrists (50 to 100% above seizure threshold) (Royal College of Psychiatrists, 2005). As explored quantitatively below (see Table 123), high dose (≥400% above seizure threshold) unilateral ECT generally appeared as effective as low/standard dose (0 to 150% above seizure threshold) bilateral ECT, whether bitemporal or bifrontal. One study including low dose unilateral ECT arms found them to be less effective than standard dose bilateral and high dose unilateral ECT. Another study found that threshold dose unilateral ECT was less effective than low/standard dose bilateral ECT.

The range of cognitive side-effects assessments varied between studies and were not consistent with regard to global scores (MMSE/3MS), but more consistent memory effects (including autobiographical memory impairment) were seen.

Previous studies have suggested that bifrontal ECT may cause fewer cognitive effects than bitemporal ECT but with similar efficacy (Lawson et al., 1990; Letemendia et al., 1993; Bailine et al., 2000) so the two types of bilateral ECT were considered separately.

In the five studies in which bitemporal low/standard dose ECT was compared with unilateral high dose ECT, two found no difference in cognitive effects, two found that bitemporal ECT caused a greater global decrease and one found that bitemporal ECT caused greater impairment of autobiographical memory but not other measures of retrograde and anterograde memory. In one study a global decrease in cognitive function with high dose bitemporal ECT compared with high dose unilateral ECT was seen. The studies in which bitemporal ECT worsened cognitive function compared with unilateral ECT mostly used high standard doses (150% above seizure threshold).

In the three studies where bifrontal low/standard dose ECT was compared with high dose unilateral ECT, two studies found no difference in global cognitive effects and one found less impairment. A study where both doses were low found no difference in most cognitive effects except less non-verbal anterograde amnesia with bifrontal ECT. A study of low and standard doses of bitemporal and unilateral ECT found effects of both dose, electrode placement and their interaction depending on the test used, which had recovered to above baseline 2 months after ECT. In two studies there was faster onset of improvement with high dose unilateral ECT.

Ultra-brief pulse (0.3 msec) high dose ECT caused no cognitive impairment in two studies and cognitive impairment was significantly less than standard brief pulse (1.5 msec) treatment in one study.

A soon-to-be reported large study comparing bitemporal (50% above seizure threshold), bifrontal (50% above seizure threshold) and right unilateral (400% above seizure threshold) with a 1msec pulse width, similar to treatment practice in the UK, has found few differences in cognitive effects and efficacy between placements (Charles Kellner, personal communication, 2009).

The NICE TA on ECT (NICE, 2003) concluded that cognitive impairment is greater in individuals who have had electrodes applied bilaterally than in those who have had them placed unilaterally, and that unilateral placement to the dominant hemisphere causes more impairment than placement to the non-dominant hemisphere. They also found that raising the stimulus threshold above the individual’s seizure threshold increased the efficacy of unilateral ECT at the expense of increased cognitive impairment. Overall the conclusion was that reduction in the risk of cognitive impairment is mirrored by a reduction in efficacy.

The new studies provide insufficient evidence to determine whether efficacy and cognitive side effects can be dissociated by manipulating electrode placement and stimulus dose or parameters. Results with high dose ultra-brief unilateral ECT need to be replicated.

Effect of ethnicity

The data from the acute phase of the KELLNER2006 trial included in the analyses above were also analysed by race, looking at data for black and white participants separately (Williams, M. D., et al., 2008). Of 515 participants, 483 were white and 32 black. Of these, 63.4% of white participants and 71.9% of black participants achieved remission. The difference was not statistically significant, although may indicate a trend towards ECT being more effective in black participants. It should be noted that the study was undertaken in the US where the ethnic populations are different from those in England and Wales so the results of this study are unlikely to be generalisable.

12.4.5. Relapse prevention following successful treatment with ECT in relapse prevention

Four studies were found of continuation treatment after successful treatment with ECT, two of which included maintenance ECT (see Table 124; the full evidence profiles and associated forest plots can be found in Appendix 16c and Appendix 19c, respectively). In these studies, there was little difference after 6 months between adding ECT to an antidepressant and maintaining the antidepressant alone, or between ECT alone compared with a combination of nortriptyline and lithium. However, at 12 months, fewer participants experienced relapse if they had received ECT plus nortriptyline compared with those continuing treatment with nortriptyline alone. Similar data were not available for the other study.

Table 124. Summary evidence profile for relapse prevention with ECT.

Table 124

Summary evidence profile for relapse prevention with ECT.

In studies of pharmacological maintenance strategies (see Table 125), only nortriptyline plus lithium was effective (compared with placebo), although there was a trend towards nortriptyline plus lithium compared with nortriptyline alone being more effective. The data are weak since there is only one study comparing each strategy, with relatively low numbers. However, the data suggest that combination treatment with nortriptyline and lithium may be effective in reducing the likelihood of relapse following successful treatment with ECT.

Table 125. Summary evidence profile for studies of pharmacological strategies for relapse prevention following successful ECT.

Table 125

Summary evidence profile for studies of pharmacological strategies for relapse prevention following successful ECT.

A further small study randomised 74 patients following response to ECT to paroxetine or placebo in those with cardiovascular disease and paroxetine or imipramine in those without (Lauritzen et al., 1996). Using survival analysis there was a significant benefit for paroxetine over placebo although this was only at trend level at the end of 6 months, and for paroxetine over imipramine.

12.4.6. Continuation/maintenance ECT and cognitive function

A particular concern in the NICE TA on ECT (NICE, 2003) about continuation or maintenance ECT was the lack of evidence about potential long-term cognitive effects. Since then there have been further data published although the numbers of patients studied remains relatively small.

In the only prospective RCT of continuation ECT compared with continuation antidepressants after acute ECT treatment (Kellner et al., 2007), the MMSE improved in both groups over the 6 months after the end of acute-phase treatment with no difference between those who had not relapsed or dropped out. At 3 months, however, the continuation ECT group had improved less than the antidepressant group and one of the 15 who stopped treatment early in the ECT group did so because of memory loss. Russell and colleagues (2003) reported a retrospective evaluation of 43 patients who had received maintenance ECT for at least a year. They had an improved clinical status and slight improvement in their MMSE scores compared with before starting ECT. Adverse effects included falls, delirium and cardiac dysrhythmia, each in about 10% of patients but none causing significant morbidity. Rami-Gonzalez and colleagues (2003) undertook a cross sectional study of 11 patients on maintenance ECT compared with a matched group not receiving ECT. The patients receiving ECT had impaired encoding of new information and frontal lobe test results compared with the control group but no difference in delayed recall. Vothknecht and colleagues (2003) undertook a prospective study (mean 61 weeks) of 11 patients receiving maintenance ECT compared with 13 patients receiving only antidepressants. There was no difference between groups on a test battery including attention and concentration, anterograde memory and frontal lobe function. An equal number in each group had subjective memory complaints. Rami and colleagues (2004) reported results on a prospective assessment of 26 patients of whom 20 carried on with maintenance ECT over 1 year in comparison with 10 controls. There were no differences found between groups or significant changes over 1 year in attention and concentration, anterograde memory and frontal lobe function. There have also been a few case reports showing no effects on cognitive function with maintenance ECT (Wijkstra & Nolen, 2005; Zisselman et al., 2007).

12.4.7. Health economic evidence and considerations

The systematic literature search identified only one economic evaluation on ECT by Greenhalgh and colleagues (2005) as part of the HTA on ECT. The economic evaluation was undertaken to determine the cost effectiveness of ECT for depressive illness as well as schizophrenia, catatonia and mania. The authors developed an economic model based on how ECT is used in the UK for people with major depressive disorder who require hospitalisation. The analysis compared inpatient administered ECT with other pharmacological treatments (TCAs, SSRIs, SNRIs and lithium augmentation). These therapies were sequenced in several ways so as to form eight scenarios in which ECT featured as a first-, second- and third-line therapy. Expert opinion and data from the clinical effectiveness evidence review and other relevant studies were used to develop the model. Resource use patterns and costs were sourced from published literature. Health utility scores were adapted from a study by Bennett and colleagues (2000) and incorporated in the model. The evaluation failed to demonstrate, however, that any of the scenarios had a clear economic benefit over any of the others. This was due to high levels of uncertainty around the effectiveness data and the utility estimates.

The Greenhalgh and colleagues’ (2005) study was one of the first attempts to evaluate the cost effectiveness of ECT and although many of the model inputs were based on published literature many assumptions underlay the results due to the lack of available data. The authors pointed out that one of the main drawbacks in terms of cost effectiveness of prescribing ECT was the associated high resource use. They also mention a higher rate of relapse with ECT than pharmacological therapies. This statement points to one of the limitations of this evaluation. Studies with very dissimilar populations were combined to compute model inputs such as relapse and response rates, while medication trials with patient populations that were less depressed or not treatment resistant were combined with populations who were treatment resistant or referred specifically for ECT. Underlying patient characteristics do play a vital role in determining the outcomes of studies and using data in this way makes the accuracy of the effectiveness estimates used in the model questionable. However, the authors did acknowledge the lack of data and conducted many sensitivity analyses, which further emphasised the uncertainty of the results. The authors of the HTA pointed to the clear need for RCTs that directly compare the efficacy of treating severely depressed patients with ECT versus pharmacological treatments.

For the effectiveness update, reviews of ECT with pharmacological interventions were not updated since no new studies were found. As a result, the cost effective analysis was not updated. However, the review comparing bilateral ECT with unilateral ECT, including a sub-analysis by dose, was updated. The HTA explored these differences by varying the efficacy, outcomes and cost in the sensitivity analysis to incorporate the different approaches used in providing ECT with no effect on results. There should be no resource use differences between bilateral versus unilateral treatment. The clinical evidence review shows little difference in effect between bilateral and unilateral ECT with a slight advantage for bilateral ECT. These results are in keeping with previous effectiveness evidence.

The authors also mentioned uncertainty around the utility estimates used from the study by Bennett and colleagues (2000). In this study the depression-specific McSad health state classification system was utilised; NICE recommends using a generic tool (NICE, 2004a). The health state descriptions used referred to untreated depression. The population of the study consisted of patients who had experienced at least one episode of major unipolar depression in the previous 2 years but who were currently in remission. This is not typical of the patients who are usually prescribed ECT. This study therefore, may underestimate quality of life gains from the treatment and also potentially overestimate benefit if cognitive impairment following ECT is taken into account. However, utility data for mental health related conditions are very sparse and at the time this study was one of a very small number of studies available for patients with depression. The utility values were also subject to sensitivity analysis, with no effect on the results. To date no studies have been found describing health-related quality of life in which the health states have been determined in a group of patients with chronic or severe depression requiring or having received ECT.

ECT is resource intensive, however, patients who require such treatment usually have a chronic form of the illness or undergo several treatment options before being referred on for ECT. This group of people usually makes up a small proportion of the entire depressive population in a health system and the costs they incur to health systems can be quite significant. The clinical evidence points to ECT having a higher success rate for certain groups of people with severe depression, and providing this high cost intervention may prove to be cost effective as it may reduce subsequent resource use and potentially improve quality of life if prescribed as recommended.

12.4.8. From evidence to recommendations

The review of ECT for the updated guideline found relatively little additional data to update the reviews undertaken for the original NICE TA (NICE, 2003). There were no new data comparing ECT with sham ECT, antidepressants, or combination treatment in the acute phase and limited new data in the continuation phase after acute treatment.

Integrating the evidence for ECT with that for other treatments for depression it is evident that many people with depression have a poor response to treatment. In addition the definition of the severity of depression has altered between the previous guideline and this guideline update so that many patients previously defined as severely depressed would now be included in the moderate severity category. For this reason, while ECT is still not recommended as a routine treatment for moderately severe depression, it is presented as an option in those with moderate depression who have repeatedly not responded to both drug and psychological treatment.

The new data comparing bilateral ECT with unilateral ECT did not change the conclusion that bilateral ECT is more effective than unilateral for people with depression, although the effect size is small and complicated by variations in dosing and electrode placement. A sub-analysis by dose suggests that high dose unilateral ECT (doses over 150% above seizure threshold) may be at least as effective as low/standard dose bilateral ECT but there are relatively few data and it was not possible to explore this quantitatively.

For cognitive impairment, it is still not clear to what degree the trade-off between efficacy and cognitive side effects can be avoided by manipulating dose and electrode placement. There is, however, evidence that bilateral ECT causes more cognitive impairment than unilateral ECT and that the cognitive impairment and efficacy from unilateral ECT are dose related. This has now been included in the guidance together with more detailed advice on how and when to measure cognitive side effects and on the principles of choice of electrode placement and dose in relation to efficacy and cognitive side effects.

There are some data on continuation/maintenance ECT that support at least equal efficacy in preventing relapse compared with pharmacotherapy but the evidence is limited. Systematic, prospective assessment of longer-term cognitive effects of continuation/maintenance ECT are also limited although those available do not suggest cumulative cognitive adverse effects. Given the relative lack of data, no treatment recommendation is made with regard to continuation/maintenance ECT.

However, in recognition that continuation/maintenance ECT will continue to be used in exceptional circumstances, and that conclusive RCT data are unlikely to be available in the short-to- medium term, a research recommendation on collecting data for national audit when continuation/maintenance ECT is used has been added (see Section 12.4.10).

Relapse prevention using pharmacological strategies has also been examined, and the data suggest that continuation antidepressants particularly with lithium augmentation of antidepressants is effective.

12.4.9. Recommendations

12.4.9.1.

Consider ECT for acute treatment of severe depression that is life-threatening and when a rapid response is required, or when other treatments have failed.

12.4.9.2.

Do not use ECT routinely for people with moderate depression but consider it if their depression has not responded to multiple drug treatments and psychological treatment.

12.4.9.3.

For people whose depression has not responded well to a previous course of ECT, consider a repeat trial of ECT only after:

  • reviewing the adequacy of the previous treatment course and
  • considering all other options and
  • discussing the risks and benefits with the person and/or, where appropriate, their advocate or carer.
12.4.9.4.

When considering ECT as a treatment choice, ensure that the person with depression is fully informed of the risks associated with ECT, and with the risks and benefits specific to them. Document the assessment and consider:

  • the risks associated with a general anaesthetic
  • current medical comorbidities
  • potential adverse events, notably cognitive impairment
  • the risks associated with not receiving ECT.

The risks associated with ECT may be greater in older people; exercise particular caution when considering ECT treatment in this group.

12.4.9.5.

A decision to use ECT should be made jointly with the person with depression as far as possible, taking into account, where applicable, the requirements of the Mental Health Act 2007. Also be aware that:

  • valid informed consent should be obtained (if the person has the capacity to grant or refuse consent) without the pressure or coercion that might occur as a result of the circumstances and clinical setting
  • the person should be reminded of their right to withdraw consent at any time
  • there should be strict adherence to recognised guidelines about consent, and advocates or carers should be involved to facilitate informed discussions
  • if informed consent is not possible, ECT should only be given if it does not conflict with a valid advance decision and the person’s advocate or carer should be consulted.
12.4.9.6.

The choice of electrode placement and stimulus dose related to seizure threshold should balance efficacy against the risk of cognitive impairment. Take into account that:

  • bilateral ECT is more effective than unilateral ECT but may cause more cognitive impairment
  • with unilateral ECT, a higher stimulus dose is associated with greater efficacy, but also increased cognitive impairment compared with a lower stimulus dose.
12.4.9.7.

Assess clinical status after each ECT treatment using a formal valid outcome measure, and stop treatment when remission has been achieved, or sooner if side effects outweigh the potential benefits.

12.4.9.8.

Assess cognitive function before the first ECT treatment and monitor at least every three to four treatments, and at the end of a course of treatment.

12.4.9.9.

Assessment of cognitive function should include:

  • orientation and time to reorientation after each treatment
  • measures of new learning, retrograde amnesia and subjective memory impairment carried out at least 24 hours after a treatment.

If there is evidence of significant cognitive impairment at any stage consider, in discussion with the person with depression, changing from bilateral to unilateral electrode placement, reducing the stimulus dose or stopping treatment depending on the balance of risks and benefits.

12.4.9.10.

If a person’s depression has responded to a course of ECT, antidepressant medication should be started or continued to prevent relapse. Consider lithium augmentation of antidepressants.

12.4.10. Research recommendations

12.4.10.1.

The effectiveness of maintenance ECT for relapse prevention in people with severe and recurring depression that does not respond to pharmacological or psychological interventions

Is maintenance ECT effective for relapse prevention in people with severe and recurring depression that does not respond to pharmacological or psychological interventions?

Why this is important

A small number of people do not benefit in any significant way from pharmacological or psychological interventions but do respond to ECT. However, many of these people relapse and need repeated treatment with ECT. This results in considerable suffering to them and it is also costly, because ECT often necessitates inpatient care. A small number of studies suggest possible benefits from maintenance ECT but it is used little in the NHS. The outcome of the audit and clinical trial should supply information on patient characteristics, outcomes, feasibility and acceptability in relation to the use of maintenance ECT and potentially inform its wider use in the NHS. The results therefore may have important implications for the provision of ECT in the NHS.

This question should be addressed through first establishing a national audit for the collection of data on all people receiving maintenance ECT. The characteristics of the people who are likely to be considered for maintenance ECT make a randomised controlled trial unfeasible, but a clinical trial using alternative methods (for example, mirror image or a carefully characterised non-randomised study) should be undertaken depending on the outcome of the audit.

The number of people receiving maintenance ECT is small, and considerable uncertainty surrounds its use, such as its long-term efficacy and acceptability and possible side effects, which include cognitive impairment. The outcomes chosen for the audit and clinical trial should reflect both observer and patient-rated assessments of improvement, the impact on cognitive function and an assessment of the acceptability of ECT as a maintenance treatment.

12.5. OTHER NON-PHARMACOLOGICAL PHYSICAL TREATMENTS

12.5.1. Transcranial magnetic stimulation

Transcranial magnetic stimulation (TMS) involves focal stimulation of the superficial layers of the cerebral cortex using a rapidly changing magnetic field applied using an external coil. It does not require anaesthesia and can be performed on an outpatient basis. Treatment with TMS usually involves daily sessions lasting about 30 minutes for 2 to 4 weeks and possibly longer. Its use in the treatment of depression has recently been the subject of NICE Interventional Procedures Guidance (IPG 242; NICE, 2007d).

The main points highlighted in the review and guidance were:

  • Uncertainty about the procedure’s clinical efficacy, which may depend on higher intensity, greater frequency, bilateral application and/or longer treatment durations than have appeared in the evidence to date.
  • No major safety concerns associated with TMS.

Included in the review was consideration of a meta-analysis of 33 short-term RCTs in depression (Herrmann & Ebmeier, 2006), which found a large significant effect size of 0.71 against sham treatment. However, the studies were small, heterogeneous in methodology and effect size and it was not possible to identify any significant predictors of outcome. A more recent meta-analysis for patients with treatment-resistant depression, which included 24 studies (1,092 patients) meeting their inclusion criteria (Lam et al., 2008), found that active repetitive transcranial magnetic stimulation (rTMS) was significantly superior to sham conditions in producing clinical response, with a risk difference of 17%. However the pooled response and remission rates were only 25% and 17%, and 9% and 6% for active rTMS and sham conditions respectively. They concluded that further studies are required before adopting rTMS as a first-line treatment for treatment-resistant depression.

12.5.2. From evidence to recommendations

The guideline uses the recommendation from the current NICE Interventional Procedure Guidance on TMS (IPG 242, NICE, 2007d).

12.5.3. Recommendation

12.5.3.1.

Current evidence suggests that there are no major safety concerns associated with transcranial magnetic stimulation (TMS) for severe depression. There is uncertainty about the procedure’s clinical efficacy, which may depend on higher intensity, greater frequency, bilateral application and/or longer treatment durations than have appeared in the evidence to date. TMS should therefore be performed only in research studies designed to investigate these factors.

12.5.4. Vagus nerve stimulation

Vagus nerve stimulation (VNS) therapy is a type of treatment where a small electrical pulse is administered through an implanted neurostimulator to a bipolar lead attached to the left vagus nerve. A battery-powered pulse-generating device is implanted under the skin of the upper left chest. A wire is tunnelled under the skin and connected to the left vagus nerve in the neck.

The stimulation parameters (pulse width and frequency, current intensity, and on/off cycles) are programmed into the pulse generator via a programming wand. The battery lasts 8 to 10 years and can be replaced under local anaesthesia. A typical treatment regimen might comprise intermittent stimulation for 30 seconds every 5 minutes throughout the day and night. This procedure has been studied in patients with treatment-resistant epilepsy and it is indicated for use as an adjunctive therapy in reducing the frequency of seizures in patients who are refractory to anti-epileptic medication. NICE guidance on VNS for refractory epilepsy in children concluded that current evidence appears adequate to support the use of this procedure ‘provided that the normal arrangements are in place for consent, audit and clinical governance’ (IPG 50, NICE, 2004c). In addition antidepressant effects of VNS in epilepsy patients have been described, independent of reduction of seizure frequency (for example, Harden et al., 2000).

The efficacy and safety of VNS for treatment-resistant depression is currently under consideration by the NICE Interventional Procedures Advisory Programme. Readers concerned with the efficacy and safety of VNS, and recommendations about its use to treat depression, should refer to this document which is expected to be published in 2010.

12.6. THE PHARMACOLOGICAL MANAGEMENT OF RELAPSE PREVENTION

The following sections on the pharmacological management of relapse prevention marked by asterisks (**_**) are from the previous guideline and have not been updated except for style and minor clarification.

12.6.1. Introduction

**Major depressive disorder is among the most important causes of death and disability worldwide in both developing and developed countries (Murray & Lopez, 1997a). Because of the long-term nature of depression, with many patients at substantial risk of later recurrence, there is a considerable need to establish how long such patients should stay on antidepressants. Existing clinical guidelines recommend that treatment should be continued for 4 to 6 months after the acute episode (Anderson et al., 2000; APA, 2000b; Bauer et al., 2002a). There is considerable variation in practice, suggesting that many patients do not receive optimum treatment. Geddes and colleagues (2003a) reviewed all published and unpublished trials available for review by August 2000 in which continued antidepressant drug therapy was compared with placebo in patients who had responded to acute treatment with antidepressants. It was found that antidepressants reduced the risk of relapse in depression and continued treatment with antidepressants appeared to benefit many patients with recurrent depression. The treatment benefit for an individual patient depended on their absolute risk of relapse with greater absolute benefits in those at higher risk. It was estimated that for patients who were still at appreciable risk of recurrence after 4 to 6 months of treatment with antidepressants, another year of continuation treatment would approximately halve their risk. The authors found no evidence to support the contention that the risk of relapse after withdrawal from active treatment in the placebo group was due to a direct pharmacological effect (for example, ‘withdrawal’ or ‘rebound’) since there was not an excess of cases within a month of drug discontinuation.**

12.6.2. Studies considered209,210

The GDG used the review by Geddes and colleagues (2003a) as the basis for the review in the previous guideline. This included 37 studies of which 20 met the inclusion criteria set by the GDG. An additional five studies were identified in searches for the previous guideline, one of which was excluded. Another study was identified through searching journal tables of contents and a further study was identified from searches undertaken for the review of lithium augmentation elsewhere in this guideline. Both of these were included. Therefore, 26 studies formed the basis of this review in the previous guideline (Alexopoulous2000, Bauer2000, Cook1986, Doogan1992, Feiger1999, Frank1990, Georgotas1989, Gilaberte2001, Hochstrasser2001, Keller1998, Kishimoto1994, Klysner2002, Kupfer1992, Montgomery1988, Montgomery1992, Montgomery1993, Prien1984, Reimherr1998, Robert1995, Robinson1991, Sackheim2001, Schmidt2000, Terra1998, Thase2001, Versiani1999, Wilson2003) and 18 were excluded.

A further nine studies were identified in update searches and added to the review (GORWOOD2007 [escitalopram versus placebo]; KORNSTEIN2006A [escitalopram versus placebo]; MCGRATH2006 [fluoxetine versus placebo]; PERAHIA2006 [duloxetine versus placebo]; PREVENT STUDY [studyA and study B: venlafaxine-ER versus placebo]; RAPAPORT2004 [escitalopram versus placebo]; RAPAPORT2006A [risperidone + citalopram versus placebo + citalopram]; VAN den BROEK2006 [imipramine versus placebo]).

**Studies included a pre-maintenance phase during which participants continued to receive medication after they had achieved remission. This was followed by a maintenance phase in which participants who had achieved remission were randomised either to pharmacological treatment or to placebo. Studies were included provided participants were classified as remitted, only if they no longer met diagnosis for major depression or had achieved an HRSD or MADRS score below the cutoff for mild depression. Similarly, studies were included only if participants had been assessed as having relapsed using some kind of formal criteria such as exceeding a specific HRSD or MADRS score or meeting formal diagnostic criteria for depression rather than clinical judgement alone.

A single outcome (number of study participants experiencing relapse) was extracted. Since the length of both the pre-maintenance and the maintenance phase varied between studies, sub-analyses were undertaken splitting the dataset as follows:

  • by length of continuation treatment (that is, length of time continued with medication after remission but before randomisation) – less than or more than 6 months
  • by length of maintenance treatment – less than or more than 12 months.

The longest maintenance phase was 2 years. Further sub-analyses were undertaken combining these factors – for example, studies with pre-maintenance treatment of less than 6 months and maintenance treatment of less than 12 months.**

Fifteen studies used an SSRI as the maintenance treatment, eight studies used a TCA, and seven studies used other antidepressants. Three studies compared lithium (with and without an antidepressant) with an antidepressant or placebo211. One study compared SSRIs augmented with other agents with the SSRI alone. Twenty-seven studies used the same treatment in both acute and maintenance phases and four did not.

All included studies were published between 1984 and 2008. In 21 studies participants were described as outpatients, one was from primary care and in the others it was either not clear from where participants were sourced or they were from mixed sources. There were no studies of inpatients. Five studies were classified elderly, and none was of atypical depression.

Of the 25 trials of antidepressant medication, 13 (Bauer2000, Cook1986, Frank1990, Gilaberte2001, Hochstrasser2001, Kishimoto1994, Kupfer1992, Montgomery1988, Montgomery1993, PERAHIA2006, Robinson1991, Terra1998, Versiani1999) included only participants who had had at least one previous depressive episode. Five studies (Alexopoulous2000, Feiger1999, Klysner2002, Thase2001, Wilson2003) were of participants with a mix of first episode and previous episode depression. For the purpose of a sub-analysis by number of episodes, two of these (Klysner2002, Wilson2003) were classified first episode since more than 70% of participants were in their first episode. In the remaining seven studies (Doogan 1992, Georgotas1989, Keller1998, Montgomery1992, Robert1995, Schmidt2000, Sackheim2001) it was not possible to assess the proportion of participants with first or subsequent episode depression. Additional sub-analyses were undertaken by number of previous episodes.

12.6.3. Clinical evidence statements212

Effect of treatment on relapse

In an analysis of all available data comparing maintenance treatment with an antidepressant with placebo, there is strong evidence suggesting that there is a clinically important difference favouring continuing antidepressant treatment over discontinuing antidepressant treatment on reducing the likelihood of relapse (K = 32; N = 4982; RR = 0.46; 95% CI, 0.4 to 0.52; RD =−0.25 [−0.29 to −0.22]).

**There was little difference in the results of sub-analyses by length of pre-randomisation treatment or by post-randomisation treatment, by a combination of these factors, or between results for SSRIs and TCAs analysed separately. Nor was any difference found for patients in their first episode or for those with previous episodes.

With regard to lithium augmentation:

There is some evidence suggesting that there is a clinically important difference on reducing the likelihood of relapse favouring continuing lithium augmentation of an antidepressant over:

  • discontinuing lithium (that is, continuing on antidepressant monotherapy) (K = 3; N = 160; RR = 0.58; 95% CI, 0.37 to 0.92).
  • discontinuing lithium and antidepressant treatment (that is, taking a placebo) (K = 2; N = 129; RR = 0.42; 95% CI, 0.28 to 0.64).

In patients who have achieved remission while taking an antidepressant plus lithium, there is some evidence suggesting that there is a clinically important difference favouring discontinuing lithium treatment (that is, continuing with the antidepressant alone) over discontinuing antidepressant treatment (that is, continuing lithium alone) on reducing the likelihood of patients experiencing a relapse in symptoms of depression (K = 1; N = 77; RR = 1.75; 95% CI, 1.03 to 2.96).

In patients who have achieved remission while taking an antidepressant plus lithium, there is insufficient evidence to determine if there is a clinically important difference between discontinuing antidepressant treatment (that is, continuing with lithium alone) and discontinuing antidepressant and lithium treatment (that is, taking a placebo) on reducing the likelihood of patients experiencing a relapse in symptoms of depression (K = 1; N = 71; RR = 0.88; 95% CI, 0.60 to 1.28).

12.6.4. Clinical summary

The majority of study participants in this review had experienced multiple depressive episodes. There is strong evidence that responders to medication, who have had multiple relapses, should stay on medication to avoid relapse, irrespective of the length of treatment pre-response (between 6 weeks and 12 months). This effect holds true beyond 12 months. From the available data, it is not possible to determine effects beyond 2 years. These effects were evident with both TCAs and SSRIs. Whether this effect is evident in those recovering from a first episode or with placebo is unknown. Since most studies randomised participants either to continue with medication or to a placebo, there is little data comparing lengths of maintenance treatment with active medication.

12.6.5. Health economic evidence and considerations

No evidence on the cost effectiveness of the pharmacological management of relapse prevention was identified by the systematic search of the economic literature. Details on the methods used for the systematic search of the economic literature are described in Chapter 3, Section 3.6.1.

12.6.6. From evidence into recommendations

The previous guideline recommended initially continuing treatment for at least 6 months after remission, and up to 2 years for patients who are high risk of relapse. There is no new evidence that suggests that these recommendations should be changed. For patients who have achieved remission while taking lithium in addition to an antidepressant it appears to be worthwhile continuing both treatments. If one or other drug is stopped, the evidence, while suggestive that lithium should be stopped in preference to the antidepressant, is based on a single small study and this was not considered sufficient to support a strong recommendation. The recommendations have been updated to match the new NICE style.

12.6.7. Recommendations

12.6.7.1.

Support and encourage a person who has benefited from taking an antidepressant to continue medication for at least 6 months after remission of an episode of depression. Discuss with the person that:

  • this greatly reduces the risk of relapse
  • antidepressants are not associated with addiction.
12.6.7.2.

Review with the person with depression the need for continued antidepressant treatment beyond 6 months after remission, taking into account:

  • the number of previous episodes of depression
  • the presence of residual symptoms
  • concurrent physical health problems and psychosocial difficulties.
12.6.7.3.

For people with depression who are at significant risk of relapse or have a history of recurrent depression, discuss with the person treatments to reduce the risk of recurrence, including continuing medication, augmentation of medication or psychological treatment (CBT). Treatment choice should be influenced by:

  • previous treatment history, including the consequences of a relapse, residual symptoms, response to previous treatment and any discontinuation symptoms
  • the person’s preference.
12.6.7.4.

Advise people with depression to continue antidepressants for at least 2 years if they are at risk of relapse. Maintain the level of medication at which acute treatment was effective (unless there is good reason to reduce the dose, such as unacceptable adverse effects) if:

  • they have had two or more episodes of depression in the recent past, during which they experienced significant functional impairment
  • they have other risk factors for relapse such as residual symptoms, multiple previous episodes, or a history of severe or prolonged episodes or of inadequate response
  • the consequences of relapse are likely to be severe (for example, suicide attempts, loss of functioning, severe life disruption, and inability to work).
12.6.7.5.

When deciding whether to continue maintenance treatment beyond 2 years, re-evaluate with the person with depression, taking into account age, comorbid conditions and other risk factors.

12.6.7.6.

People with depression on long-term maintenance treatment should be regularly re-evaluated, with frequency of contact determined by:

  • comorbid conditions
  • risk factors for relapse
  • severity and frequency of episodes of depression.
12.6.7.7.

People who have had multiple episodes of depression, and who have had a good response to treatment with an antidepressant and an augmenting agent, should remain on this combination after remission if they find the side effects tolerable and acceptable. If one medication is stopped, it should usually be the augmenting agent. Lithium should not be used as a sole agent to prevent recurrence.

12.6.8. Research recommendations

12.6.8.1.

Sequencing antidepressant treatment after inadequate initial response What is the best medication strategy for people with depression who have not had sufficient response to a first SSRI antidepressant after 6 to 8 weeks of adequate treatment?

Why this is important

Inadequate response to a first antidepressant is a frequent problem but the best way of sequencing treatments is not clear from the available evidence. There is good evidence that the likelihood of eventual response decreases with the duration of depression and number of failed treatment attempts so that maximising the response at an early stage may be an important factor in final outcome. The results of this study will be generalisable to a large number of people with depression and will inform choice of treatment.

This question should be addressed using a randomised controlled trial design and compare the effects of continuing on the same antidepressant (with dose increase if appropriate) and switching to another SSRI or to an antidepressant of another class. Built into the design should be an assessment of the effect of increased frequency of follow-up and monitoring alone on improvement. The outcomes chosen should reflect both observer and patient-rated assessments of improvement and an assessment of the acceptability of the treatment options. The study needs to be large enough to determine the presence or absence of clinically important effects using a non-inferiority design, and mediators and moderators of response should be investigated.

Footnotes

182

This section updates the NICE Technology Appraisal on ECT (for depression only).

183

Here and elsewhere in the guideline, each study considered for review is referred to by a ‘study ID’ made up of first author and publication date (unless a study is in press or only submitted for publication, when first author only is used). Study IDs in title case refer to studies included in the previous guideline and study IDs in capital letters refer to studies found and included in this guideline update. References for studies from the previous guideline are in Appendix 18 and references for studies for the update are in Appendix 17c.

184

Study IDs in title case refer to studies included in the previous guideline and study IDs in capital letters refer to studies found and included in this guideline update. References for studies from the previous guideline are in Appendix 18.

185

Many papers have been published from the STAR*D study. Those containing data used in this guideline are listed in Appendix 17, and the study is referred to with the Rush and colleagues (2003) reference which gives an overview of the study design.

186

Study IDs in title case refer to studies included in the previous guideline and study IDs in capital letters refer to studies found and included in this guideline update. References for studies from the previous guideline are in Appendix 18.

187

Study IDs in title case refer to studies included in the previous guideline and study IDs in capital letters refer to studies found and included in this guideline update. References for studies from the previous guideline are in Appendix 18.

188

Study IDs in title case refer to studies included in the previous guideline. References for studies from the previous guideline are in Appendix 18.

189

Details of standard search strings used in all searches are in Appendix 8. Information about each study along with an assessment of methodological quality is in Appendix 17c, which also contains a list of excluded studies with reasons for exclusions.

190

Study IDs in title case refer to studies included in the previous guideline. References for these studies are in Appendix 18.

191

The forest plots can be found in Appendix 19c.

192

The forest plots can be found in Appendix 19c.

193

Details of standard search strings used in all searches are in Appendix 8. Information about each study along with an assessment of methodological quality is in Appendix 17c, which also contains a list of excluded studies with reasons for exclusions.

194

Study IDs in title case refer to studies included in the previous guideline. References for these studies are in Appendix 18.

195

The forest plots can be found in Appendix 19c.

196

Details of standard search strings used in all searches are in Appendix 8. Information about each study along with an assessment of methodological quality is in Appendix 17c, which also contains a list of excluded studies with reasons for exclusions.

197

Study IDs in title case refer to studies included in the previous guideline. References for these studies are in Appendix 18.

198

The forest plots can be found in Appendix 19c.

199

Ibid.

200

Study IDs in capital letters refer to studies found and included in this guideline update.

201

Features of serotonin syndrome include confusion, delirium, shivering, sweating, changes in blood pressure and myoclonus.

202

The evidence for this recommendation has not been updated since the previous guideline. Any wording changes have been made for clarification only.

203

Ibid.

204

Ibid.

205

Aripiprazole, olanzapine, quetiapine and risperidone do not have UK marketing authorisation for the indication in question at the time of publication. Informed consent should be obtained and documented.

206

Buspirone, carbamazepine, lamotrigine, valproate, pindolol and thyroid hormones do not have UK marketing authorisation for the indication in question at the time of publication. Informed consent should be obtained and documented.

207

Study IDs in capital letters refer to studies found and included in this guideline update.

208

There are 30 studies, but SACKHEIM2008 includes four treatment groups that were used as two separate comparisons.

209

Details of standard search strings used in all searches are in Appendix 8. Information about each study along with an assessment of methodological quality is in Appendix 17c, which also contains a list of excluded studies with reasons for exclusions.

210

Study IDs in title case refer to studies included in the previous guideline and study IDs in capital letters refer to studies found and included in this guideline update. References for studies from the previous guideline are in Appendix 18.

211

One four-arm trial (Prien1984) has both antidepressant and lithium treatment groups.

212

The forest plots can be found in Appendix 19c.

Copyright © The British Psychological Society & The Royal College of Psychiatrists, 2010.

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.

Cover of Depression
Depression: The Treatment and Management of Depression in Adults (Updated Edition).
NICE Clinical Guidelines, No. 90.
National Collaborating Centre for Mental Health (UK).
Leicester (UK): British Psychological Society; 2010.

Download

NICE (National Institute for Health and Care Excellence)

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...