This systematic review investigated the efficacy and acceptability of valproate compared to that of placebo, lithium, carbamazepine, olanzapine and haloperidol in the treatment of acute episodes of bipolar disorder. Ten randomised studies were included in the review: all examined patients with mania. Valproate was more effective than placebo in the treatment of mania. There was no significant difference between efficacy in valproate and lithium or between valproate and carbamazepine, but there were fewer data on these comparisons and so these questions remain unanswered. The evidence suggests that valproate may be less effective in reducing manic symptoms than olanzapine but may cause less sedation and weight gain. There remains a need for more trials comparing medicines in treating all affective episodes.
There is consistent, if limited, evidence that valproate is an efficacious treatment for acute mania. Valproate may be less efficacious than olanzapine. More, rigorously designed, trials over the full range of acute affective episodes are required.
Valproate has become a leading adjunctive and alternative treatment to lithium in bipolar disorder.
To determine the efficacy and acceptability of valproate in acute episodes of bipolar disorder.
Registers and databases:
Cochrane Collaboration Depression, Anxiety and Neurosis Controlled Trials Registrar (version 2-2002)
Cochrane Controlled Clinical Trials Register (3-2002)
Medline (1966-January 1999 )
PsychLit (1996-June 1999)
Embase (1980-January 1999)
Reference lists of relevant papers/books.
Trial authors, experts and pharmaceutical companies.
Handsearches (specialist journals and conference proceedings, listed below)
Randomised controlled trials comparing valproate with placebo and other medications for any acute episode. Bipolar patients, male and female, of all ages were included.
Data extraction and methodological quality assessment were each performed independently by two reviewers. For analysis, relative risk was used for binary efficacy outcomes and the weighted mean difference or standardised mean difference used for continuously distributed outcomes.
Ten trials compared valproate with other interventions in mania. None examined depression or mixed affective episodes. Data were extracted on 'failure to respond by the end of the study' (i.e.less than 50% reduction in Young Mania Rating Scale or SADS-S mania scale). Three trials (316 participants) compared valproate with placebo. Three trials (158 participants) compared valproate with lithium. Two trials (363 participants) compared valproate with olanzapine. One trial (36 participants) compared valproate with haloperidol. Two trials (59 patients) compared valproate with carbamazepine.
Treatment acceptability was estimated by the 'total number withdrawing from the study'. Three trials (321 patients) compared valproate and placebo, two (144 patients) compared valproate with lithium. One study (30 patients) compared valproate and carbamazepine. Pooled relative risks ( 95% confidence intervals) were calculated using fixed effect.
Valproate was more efficacious than placebo (RRR 38%; RR 0.62; 95% C.I. 0.51 to 0.77) in the treatment of mania. There was no significant difference between valproate and lithium (RRI 5%; RR 1.05; 95% C.I. 0.74-1.50) or between valproate and carbamazepine (RRR 34%; RR 0.66; 95% C.I. 0.38 to 1.16). Valproate was less effective than olanzapine (failure to achieve clinical response; RRI 25%; RR 1.25, 95% C.I. 1.01 to 1.54; average of 2.8 point less change on the Mania Rating Scale (95% CI 0.83 to 4.79). There were no significant differences in those withdrawing from the study.