CNS Drugs 2009;

CNS Drugs 2009; 23 (3): 225-240 1172-7047/09/0003-0225/$49.95/0

REVIEW ARTICLE

ª 2009 Adis Data Information BV. All rights reserved.

Lithium: Updated Human Knowledge Using an Evidence-Based Approach Part I: Clinical Efficacy in Bipolar Disorder Etienne Marc Grandjean1 and Jean-Michel Aubry2 1 Phidalsa Institute for Clinical Investigation, Geneva, Switzerland 2 Bipolar Program, Department of Psychiatry, Division of Adult Psychiatry, University Hospitals of Geneva, Geneva, Switzerland

Contents Abstract. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Efficacy in Acute Mania . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Critical Assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Major Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Supportive Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Use in Combination: Critical Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Efficacy in the Prophylaxis of Bipolar Disorders . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Critical Assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Major Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Data from Meta-Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Concentration-Effect Controlled Trials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Controlled Trials versus ‘Active’ Comparator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.1 Critical Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.2 Major Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Supportive Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.7 Use in Combination for Maintenance Therapy: Critical Assessment. . . . . . . . . . . . . . . . . . . . . . . 2.8 Use in Rapid Cycling Bipolar Disorder: Critical Assessment and Major Findings. . . . . . . . . . . . . . 2.9 Use in Mixed States: Critical Assessment and Major Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.10 Use in Bipolar II Disorder: Critical Assessment and Major Findings . . . . . . . . . . . . . . . . . . . . . . . . . 3. Efficacy in Acute Bipolar Depression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Use as Single Therapy: Critical Assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Antidepressant Augmentation of Lithium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Efficacy in Relation to Suicide and Mortality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Critical Assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Major Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Abstract

225 226 226 227 228 228 229 229 229 231 231 232 232 232 233 233 234 235 235 235 235 236 236 236 236 237

Although there has been a decrease in lithium use over several years, it is still recommended as a first-line mood stabilizer in all recent guidelines. It has been argued that many studies of lithium were conducted at a time when study design, assessment standards and the diagnostic criteria for patient selection were not as established as they presently are. However, recent placebo-controlled data from three-arm trials have demonstrated a definite

Grandjean & Aubry

226

efficacy of lithium in bipolar disorder. Regarding mania, recent trials of novel antimanic treatments (such as second-generation antipsychotics) that have included both placebo and lithium control groups have confirmed that lithium is effective in the treatment of moderate to severe manic episodes. The efficacy of lithium as monotherapy for acute bipolar depression is still controversial, but this therapy is recognized as a therapeutic option. For maintenance therapy, lithium is superior to placebo for the prevention of relapse or recurrence of mood episodes in bipolar I disorder patients with recent manic or hypomanic episodes. Lithium is more effective in preventing episodes of the manic/hypomanic type, including mixed episodes, than preventing depressive episodes. In rapid cycling patients, lithium improves clinical symptoms as efficiently as in nonrapid cycling persons, but is not likely to prevent recurrences. Finally, data from a number of studies suggest that lithium reduces the high suicide rates associated with mood disorders. A well designed cohort study and two independent meta-analyses are in agreement with this finding. In conclusion, most experts, and the most recent guidelines, continue to consider lithium as a keystone therapy of bipolar disorders.

Lithium is one of the oldest psychotropic agents prescribed today. Its major place remains in bipolar disorders, where it is recommended as a first-line mood stabilizer in recent guidelines.[1-3] In everyday practice, use of lithium is associated with a number of difficulties and risks, mainly because of its narrow therapeutic range. Its effectiveness is strongly dependent on patient medication adherence and on the quality of therapeutic support, follow-up and the patient’s environment. For these reasons, it has been somewhat less used in maintenance treatment, especially in the US, with preference sometimes being given instead to therapies that are not evidence-based or approved for this specific indication (valproate, carbamazepine). In the present review, we will present a critical overview of the clinical efficacy of lithium. The adverse effects of lithium, which affect several organs, and the short- and long-term risks of lithium therapy[4] are reviewed extensively in a companion paper. For our review, a computerized search of MEDLINE was made for publications, using the keywords lithium and humans, the languages English, French, German, Italian and Spanish, starting from January 1997 through to January 2008. From their title and abstract, papers were subdivided into various groups such as acute mania, maintenance therapy, rapid cycling, acute ª 2009 Adis Data Information BV. All rights reserved.

bipolar depressions, etc. The search was completed by a search of the related articles in the database, and older references from bibliographic data. Articles were selected with emphasis on original data, controlled design, then meta-analysis and systematic review, and finally guidelines and consensus statements. For all sections, an appraisal of the quality of design, conduct and analysis of the studies available was performed, and is reported, whenever possible, before the presentation of the major findings. For each issue, only publications with the highest standards of quality available were selected. Consequently, uncontrolled data are only mentioned when data of better quality are not available or are scarce. 1. Efficacy in Acute Mania 1.1 Critical Assessment

The initial clinical trials of lithium in acute mania were performed at a time when current standards for studies were not applicable; hence, only a small number of these early trials meet the stringent requirements applied today. The major drawback of those studies was the use of a crossover design, which is nowadays recognized as being inappropriate for the assessment of therapy outcome in bipolar disorders. In addiCNS Drugs 2009; 23 (3)

Efficacy of Lithium in Bipolar Disorder

tion, most of the early trials included a rather limited number of patients, such that assessment of data from those randomized controlled studies requires the modern approach of systematic review and meta-analysis. The relative deficiencies, from the modern regulatory viewpoint, of the initial clinical trials were emphasized in the systematic review of lithium efficacy in acute mania published in 2000 by the Centre for Evidence-Based Pharmacotherapy in Birmingham, UK.[5] Strict requirements were set down for inclusion of a clinical trial in this analysis. First, only randomized, double-blind clinical trials, using either placebo or an active comparator as control, could be considered. Thirty-seven published trials of this type were identified in the period from 1966 to 30 June 1999. Further requirements were that the efficacy data be expressed based on currently recognized scoring systems (either the Brief Psychiatric Rating Scale [BPRS] or the Clinical Global Impression) or in terms of response rate. Another criterion was that the data also provided the standard deviation or standard error of the mean for the various treatment groups. On this basis, 25 of the 37 initially identified trials were excluded, i.e. only 12 studies reported data suitable for pooling. Of those, only one trial[6] had data transformable to response rate. However, that particular trial was biased in disfavour of lithium, since 50% of the patients included in the trial had previously been nonresponsive to lithium. Interestingly, some recent trials of novel antimanic treatments have included a placebo group as well as a lithium control group, so that the evaluation of lithium versus placebo was also possible. Such trials currently represent the most solid evidence of lithium efficacy, in terms of an evidence-based approach. 1.2 Major Findings

Two placebo-controlled clinical trials of lithium in acute mania strictly meet current scientific requirements, including intent-to-treat analysis, and can therefore be considered pivotal trials. Both were part of three-arm comparative, prospective trials versus divalproex sodium ª 2009 Adis Data Information BV. All rights reserved.

227

(valproate semisodium)[6] and quetiapine,[7] respectively. While the first trial exhibited some lack of homogeneity in patient groups, the second trial fully met that requirement as well. In the first trial, efficacy was defined as an improvement by 50% or more after 3 weeks’ treatment.[6] The target serum lithium concentration was relatively high (1.5 mmol/L) compared with the usual recommendation of 0.6–0.8 mmol/L. Lithium proved more effective than placebo, with a positive rate ratio (defined as the proportion of improved patients in the treatment group divided by the proportion of improved patients in the control group) of 1.95 (CI 95% 1.17, 3.23). The mean number of patients needed to treat (NNT) to achieve a 50% reduction in the Mania Rating Scale score was 5. As mentioned in section 1.1, the efficacy of lithium in this trial may have been underestimated because of a patient selection bias. A post hoc analysis of this trial using a different statistical concept (effect size) detected a significant difference between lithium and placebo as early as 5 days after commencement of therapy.[8] The second trial[7] was conducted in patients with bipolar disorder type I exclusively, during a manic episode; mixed episodes and rapid cycling were exclusion criteria. The primary endpoint was clinical score improvement at 21 and 84 days, based on the Young Mania Rating Scale (YMRS). The response rate (percentage of patients showing a ‡50% reduction in YMRS score) was a secondary endpoint. The target serum lithium concentration was 0.6–1.4 mmol/L. Both the primary and secondary endpoints were significantly improved at 3 and 8 weeks by lithium treatment compared with placebo. Placebo-controlled data are also available for lithium from two three-arm controlled clinical trials conducted with topiramate as the investigational drug.[9] Although topiramate failed to demonstrate any superiority to placebo, the lithium versus placebo comparative data were used to assess the efficacy of lithium itself. Mean YMRS score reductions in lithium-treated bipolar I patients were significantly greater than in the placebo and topiramate group (p £ 0.001). A similar pattern was observed after 12 weeks of CNS Drugs 2009; 23 (3)

Grandjean & Aubry

228

double-blind treatment in studies with doubleblind extensions.[9] 1.3 Supportive Data

Further studies can be considered as providing supportive data. This is the case with two older placebo-controlled, crossover trials, which were analysed in the UK systematic review.[5] The first controlled clinical trial by Schou et al.[10] reported results in terms of qualitative assessment, classified as positive, possible or negative effect. 84% of patients showed ‘positive effect’ or ‘possible effect’ when treated with lithium. A further early controlled crossover trial[11] evaluated clinical symptoms using the Wittenborn scale. Severity of mania was reduced significantly during lithium treatment compared with placebo. The mean scores on cluster III (manic states) of this scale before and after 2 weeks of lithium treatment were 12.0 and 4.33, respectively. The corresponding values for placebo were 16.11 and 11.22. A meta-analysis of data from comparative controlled trials of lithium and the antipsychotic agent chlorpromazine included a total of 143 patients from four trials.[5] Lithium was more effective than chlorpromazine, with a mean NNT of 4 (CI 95% 3, 9). A number of controlled trials of antiepileptics versus lithium were also included in that meta-analysis.[5] There was no difference in efficacy between lithium and valproate, or lithium and carbamazepine, even after pooling of the individual studies as much as possible. Another independent statistical meta-analysis[12] found no difference in efficacy among lithium, valproate and carbamazepine, although that analysis included at least one preventive trial and had other methodological deficiencies.[13] Finally, in a recent meta-analysis of randomized, placebo-controlled studies in which lithium was used as the third study arm, Storosum et al.[14] concluded that lithium is effective in the treatment of moderate to severe manic episodes. Over the last few years, lithium has also been compared with second-generation antipsychotics. In the direct comparative trial of lithium versus quetiapine mentioned in section 1.2, ª 2009 Adis Data Information BV. All rights reserved.

similar efficacy was demonstrated for both drugs in acute mania, with exclusion of mixed states and rapid cycling patients.[7] Similar results were obtained in comparisons of lithium with risperidone[15] and olanzapine,[16] with the limitation that only small groups of patients were included in the trials. In these cases, as in most recent investigational studies, lithium was taken as the reference compound. More specifically, a randomized, controlled trial was performed using both lithium and haloperidol as active controls and risperidone as the investigational drug, with no placebo group.[15] Forty-five patients with bipolar disorder, manic phase, according to DSM-IV criteria,[17] were randomly assigned to one of the three drugs for 28 days, on a twice-daily schedule. The target serum lithium concentration range was 0.6–1.2 mmol/L. The primary endpoint was a reduction in the YMRS score. Six patients withdrew (three in the haloperidol group, two in the risperidone group and one in the lithium group). Data were analysed on a last observation carried forward basis. The actual average serum lithium concentrations were 0.53, 0.62 and 0.72 mmol/L at 7, 21 and 28 days, respectively. Results showed a highly significant improvement on the YMRS in all groups, with no significant difference among groups. Similar improvements were obtained on the BPRS for all three groups, with no difference among groups. The sample size was small, however, and more subtle differences among groups may have been overlooked. The same may be true with respect to extrapyramidal adverse effects, the frequency of which was not significantly different among all three groups. 1.4 Use in Combination: Critical Assessment

A relevant number of patients with acute mania do not respond to a single mood-stabilizing agent, including lithium. Therefore, several recent trials have investigated combination treatment consisting of an antipsychotic (particularly a newer agent such as risperidone, olanzapine and quetiapine) and a mood stabilizer in the treatment of acute manic or mixed episodes of bipolar disease. Unfortunately, with respect to assessment of CNS Drugs 2009; 23 (3)


lithium itself, the designs of these trials have been such that no distinction was made between mood stabilizers, especially lithium and valproate, which were evaluated as a single group, and placebo or the antipsychotic agent. Three well designed, placebo-controlled trials of this type were published between 2002 and 2004 (one each involving olanzapine, risperidone and quetiapine).[18-20] A further trial[21] with a similar control group compared the effect of risperidone and haloperidol. A further therapeutic trend is initial use of triple therapy, adding lithium and valproate to an antipsychotic. No evidence-based data for this approach are available. An uncontrolled study of 12 patients with severe mania reported a high remission rate, with a mean remission time of 30 days.[22] 2. Efficacy in the Prophylaxis of Bipolar Disorders Lithium is still considered the gold standard for prophylaxis of bipolar disorders. This is the case despite increasing use of valproate (notwithstanding its lack of approval in this particular indication) and lamotrigine (which has recently been demonstrated to be effective and approved for prophylactic use, predominantly for depressive episodes). 2.1 Critical Assessment

Use of lithium as a maintenance treatment has long been based on the results of a number of controlled, but small, studies conducted in the 1960s and early 1970s. Their design, inclusion criteria and endpoints do not meet current requirements for clinical research. However, since then, placebo-controlled data have become available from recently conducted, well designed, three-arm trials of investigational drugs using lithium as a reference treatment.[23-25] The first of these, published in 2000,[23] was presented as the first placebo-controlled trial of any drug for maintenance therapy for 25 years. Although these studies were recently designed, all have some kind of deficiency. In all of these trials, a problematic issue was the pre-study treatment ª 2009 Adis Data Information BV. All rights reserved.

229

(run-in or ‘stabilization’) phase, during which a wide variety of treatments, dosages, durations of treatment and treatment discontinuation rates were reported, resulting in a lack of homogeneity in the treatment groups. Moreover, the second trial was closed prematurely by the sponsor, apparently for nonmedical reasons, resulting in treatment discontinuation in 20% of patients.[24] Finally, in the third study[25] of bipolar patients with a depressive episode as the index episode, a high percentage of patients withdrew (31–37% in the various groups). These considerations illustrate the difficulties of designing and performing maintenance controlled trials in patients with bipolar disorder, especially if a placebo arm is included. A well conducted meta-analysis of long-term lithium therapy for bipolar disorder, covering five recent trials, is also available.[26] Furthermore, concentration-effect comparisons, based on therapeutic drug monitoring as usually recommended for lithium, have been published (see section 2.4). Limited data are available regarding specific subgroups of patients, such as those with rapid cycling illness, mixed episodes or bipolar II disorder. 2.2 Major Findings

Three major, well designed, prospective, placebo-controlled, parallel group, three-arm trials were performed recently, one with lithium and divalproex sodium, and two with lithium and lamotrigine.[23-25] In one of the latter studies,[25] patients had recently suffered a depressive episode, whereas in the other two trials,[23,24] patients had recently presented with a manic or hypomanic episode. The study of lamotrigine in patients with recent manic or hypomanic episodes was conducted over 18 months in 175 patients.[24] Lithium was titrated to serum concentrations of 0.8–1.1 mmol/L. During the acute phase, all patients had received lamotrigine, either as monotherapy, or as adjunctive therapy to valproate or carbamazepine. Use of chloral hydrate, temazepam, oxazepam or lorazepam was allowed during the trial. The primary endpoint was the time to intervention for any CNS Drugs 2009; 23 (3)

230

mood episode; changes in various clinical scores were included as secondary endpoints. The primary endpoint was significantly improved compared with placebo by both lithium (p = 0.003) and lamotrigine (p = 0.02). Lithium and lamotrigine did not differ from each other on this measure. In terms of overall survival in the study (considering all causes of early discontinuation), lamotrigine was superior to placebo (p = 0.03) and there was a statistical trend in favour of lithium compared with placebo (p = 0.07). In conclusion, lithium was superior to placebo for the prevention of relapse or recurrence of mood episodes in bipolar I disorder patients with recent manic or hypomanic episodes. Lithium was more effective at preventing episodes of the manic/hypomanic type, including mixed episodes, than depressive episodes. In the second trial of lamotrigine, 463 bipolar I patients with a depressive episode as the most recent episode were treated for 18 months with lithium, lamotrigine or placebo.[25] Lithium was titrated to serum concentrations of 0.8–1.1 mmol/L. Rapid cycling patients were excluded. The primary endpoint was the time to a therapeutic intervention for any mood episode. Use of chloral hydrate, temazepam, oxazepam, lorazepam or midazolam was allowed. Thirtyfour percent of patients terminated the study early. Other than a trend (p = 0.076) toward a higher rate of withdrawal because of adverse events with lithium compared with lamotrigine, rates of discontinuation classified by reason were similar across treatment groups. The final median dosage for lithium carbonate was 900 mg/day (range 450–1800 mg/day). The measured steadystate serum lithium concentration was 0.8 – 0.3 mmol/L. Both lithium and lamotrigine were significantly superior to placebo on the primary endpoint (p = 0.029 for both). Median times to treatment intervention for any mood episode were 93 days (95% CI 58, 180) for placebo, 170 days (95% CI 105 to ‘not evaluable’) for lithium and 200 days (95% CI 146, 399) for lamotrigine. For overall survival in the study (considering all early discontinuation causes), both lamotrigine and lithium were superior to placebo (p = 0.003 and p = 0.022, respectively). In other words, withdrawal was more frequent in the placebo ª 2009 Adis Data Information BV. All rights reserved.

Grandjean & Aubry

group. Only lithium was superior to placebo in the prevention of manic symptoms, whereas only lamotrigine was superior to placebo for the prevention of a depressive episode. In conclusion, both lamotrigine and lithium were superior to placebo for the prevention of mood episodes in bipolar I disorder patients with a recent major depressive episode. Lithium was predominantly effective against mania, whereas lamotrigine was predominantly effective against depression. In the divalproex sodium trial,[23] which was performed in the late 1990s, 372 patients with a recent manic episode received a 12-month treatment of divalproex sodium, lithium or placebo. The inclusion criteria of bipolar disorder were still those of DSM-III-R,[27] with no distinction among the atypical presentations of the disease. The randomization ratio was 2 : 1 : 1 for divalproex sodium, lithium and placebo, respectively. The target therapeutic serum concentration range for lithium was 0.8–1.2 mmol/L. A number of patients were already receiving divalproex sodium or lithium on the day of randomization. The primary endpoint was the time to intervention for any mood episode. Changes in various clinical scores were included in the secondary endpoints. Two courses of lorazepam and a 1-week course of haloperidol were permitted. Analysis of the data revealed no significant difference among the three treatment groups with respect to the primary endpoint. The negative outcome of this trial for the two major mood stabilizers in clinical use in the US came as a troublesome and worrying surprise. The success rate in the placebo group exceeded expectations. A number of possible biases were suggested by the investigators, particularly in the selection of patients, which may explain these negative findings. In our opinion, the number and lack of homogeneity of previous and concomitant treatments permitted may have also played a role, as well as dechallenge and rechallenge with some of the study drugs after randomization and the unbalanced randomization scheme. The nondiscriminant outcome of the trial may also have reflected the relative inexperience with randomized clinical trials in the field (according to the investigators, no placebo-controlled trial of maintenance therapy CNS Drugs 2009; 23 (3)


231

for bipolar disorder had been conducted for the previous 25 years) and the difficulty some investigators had complying with the placebocontrolled approach (12 of 37 investigation centres did not randomize at least one patient to each treatment group). 2.3 Data from Meta-Analyses

statistical significance, presumably because of the small sample size. Overall, withdrawal from the trials was significantly lower with lithium than with placebo. Somnolence, nausea and diarrhoea were statistically more common with lithium than with placebo. 2.4 Concentration-Effect Controlled Trials

A systematic review and meta-analysis of randomized controlled trials recently compared the effect of lithium and placebo as maintenance therapy for bipolar disorder.[26] The investigators reviewed five published trials, including the three discussed in section 2.2,[23-25] and two earlier well designed trials.[28,29] Such an analysis proves particularly useful when apparently contradictory results, or data that are not statistically significant, are obtained from various trials. From more than 300 articles screened, only five (corresponding to 770 patients) met the selection criteria for the review and meta-analysis. Four of these trials were three-arm, lithium-test, drugplacebo, comparative trials,[23-25,29] from which only the lithium versus placebo comparison was included. Test drugs were imipramine, divalproex sodium and lamotrigine (twice). As shown in table I, lithium proved to be more effective than placebo for the prevention of any episode of mood disturbance and of manic episodes. For the prevention of depressive episodes, the effect was smaller and just failed to reach

A small number of studies have compared the prophylactic efficacy of different dosages of lithium based on the therapeutic blood concentrations obtained by therapeutic drug monitoring. A key study is that conducted by Gelenberg et al.[30] This study was well designed, with clear inclusion criteria and patient stratification according to three clinical criteria. The study was a 3-year, prospective, randomized, double-blind, controlled, parallel-group trial of two different dose ranges of lithium. The two target serum lithium concentrations were ‘standard’ (0.8–1.0 mmol/L) and ‘low’ (0.4–0.6 mmol/L). The primary endpoint was clinical relapse, according to well defined scores and criteria. Ninety-four patients were included in the trial. Data were assessed according to intent-to-treat analysis and withdrawals were clearly defined and analysed. Thirteen percent of patients in the ‘standard range’ group had relapses while receiving protocol treatment, compared with 38% in the ‘low range’ group. The risk of relapse was 2.6-fold higher (CI 95% 1.3, 5.2; p = 0.004) for patients in the ‘low range’ group.

Table I. Meta-analysis of various endpoints in randomized controlled trials of lithium vs placebo evaluated in a systematic review and metaanalysis[26] Effectiveness

Relative riska

95% CI

p-Value

Prevention of any new episode of mood disturbance

0.66

0.57, 0.77