• K A Macritchie, J R Geddes, J Scott, D R Haslam, and G M Goodwin.
• Department of Psychiatry, University of Oxford, Oxford, UK, OX3 7JX. john.geddes@psychiatry.oxford.ac.uk
• Cochrane Db Syst Rev. 2001 Jan 1(3):CD003196.

BackgroundAlthough lithium has been the most commonly used maintenance treatment in bipolar disorder for several decades, valproate is being used increasingly - especially in the United States of America. There is a need to clarify whether the increasingly prominent prophylactic role of valproate in bipolar disorder is justified.ObjectivesTo review the effectiveness of valproate, relative to placebo, other mood stabilisers and antipsychotics, in the prevention and/or attenuation of acute episodes of bipolar disorder. The effectiveness of valproate was considered in terms of mood symptoms, mortality, general health, social functioning, adverse effects and overall acceptability to patients.Search StrategyThe CCDAN group search strategy was used. The following databases were searched: The Cochrane Collaboration Depression, Anxiety and Neurosis Controlled Trials Register (CCDANCTR), The Cochrane Controlled Clinical Trials Register (CCCTR), EMBASE, MEDLINE, LILACS, PsycLIT and Psyndex. Reference lists of relevant papers and major textbooks of mood disorder were examined. Authors, other experts in the field and pharmaceutical companies were contacted for knowledge of suitable published or unpublished trials.Selection CriteriaRandomised controlled trials which compared valproate with placebo, alternative mood stabilisers (including lithium and carbamazepine) or neuroleptics, where the stated intent of intervention was the maintenance treatment of bipolar disorder. Participants were males and females of all ages with a diagnosis of bipolar disorder however diagnosed, approximating to ICD 10 Code F31 and DSM IV 296, but including patients diagnosed as ICD-9 manic depressive psychosis and DSM-III and DSM-IIIR bipolar disorder.Data Collection And AnalysisData were extracted from the original reports individually by two reviewers. The main outcomes to be assessed were: 1. The effectiveness of valproate treatment in preventing or attenuating further episodes of bipolar disorder, including its effectiveness in rapid cycling disorder. 2. The acceptability of valproate treatment to patients. 3. The prevalence of side-effects. 4. Mortality on valproate treatment. Outcomes concerning relapse/recurrence were analysed excluding data from discontinuation studies, which were to be analysed separately. Sub-group analyses were to be performed to examine the effects of valproate treatment in rapid cycling bipolar disorder and previous mood stabiliser non-responders. Data were analysed using Review Manager version 4.1.Main ResultsOne trial of 12 months duration with 372 participants was identified comparing lithium, divalproex and placebo. It had several methodological limitations. The primary analysis of time to occurrence of mood episode described in the main trial report found no reliable difference between the treatments, although there was a trend for divalproex to be more effective than lithium. In the analysis in this review, patients taking divalproex who left the study because of the occurrence of an mood episode were significantly less in number than those on placebo (RRR 37%; RR 0.63; 95% CI 0.44 to 0.90). There was no significant difference in the numbers of patients in receipt of divalproex compared with those in receipt of lithium who left the study because they suffered any mood episode. (RRR 22%; RR 0.78; 95% C.I. 0.52 to 1.17). There was insufficient information to allow sub-group analyses of rapid-cycling disorder. The divalproex group had significantly more patients suffering tremor (RRI 223%; RR 3.23; 95% C.I. 1.85 to 5.62), weight gain (RRI 187%; RR 2.87; 95% C.I. 1.34 to 6.17) and alopecia (RRI 143%; RR 2.43; 95% C.I. 1.05 to 5.65) than the placebo group. In comparison with the lithium, divalproex was associated with more frequent sedation (RRI 58%; RR 1.58; 95% C.I. 1.08 to 2.32) and infection (RRI 107%; RR 2.07; 95% C.I. 1.16 to 3.68), but less suffered thirst (RRR 62%; RR 0.38; 95% C.I. 0.18 to 0.81) and polyuria (RRR 57%; RR 0.43; 95% C.I. 0.22 to 0.82).Reviewer's ConclusionsIn view of the equivocal findings of this review, conclusions about the efficacy and acceptability of valproate compared to placebo and lithium cannot be made with any degree of confidence. With current evidence, patients and clinicians would probably wish to use lithium before valproate for maintenance treatment. At present, the observed shift of prescribing practice to valproate is not based on reliable evidence of efficacy

10 keywords...

hide…