Objective Valproate is currently licensed for use in epilepsy and, in mental health practice, for the treatment of acute mania. It is also commonly used for the maintenance treatment of bipolar disorder, though on an unlicensed basis. Though few guidelines exist to support its wider “off-label” use, clinicians have extrapolated from licensed indications to a range of others that encompass mood instability or problems with impulse control at their core. We aimed to assess available evidence on the efficacy and risks of valproate in adult psychiatric practice, either as monotherapy or adjunctive treatment, in unlicensed indications, aside from bipolar disorder.

Method A systematic review was conducted on 193 published paper (31 randomised control trials) using the MEDLINE, EMBASE and Cochrane databases to identify the various unlicensed uses of valproate in mental health care, and analyse extracted data on valproate's efficacy and safety. A meta-analysis was undertaken on extracted data where appropriate.

Results Relevant literature was identified considering the unlicensed uses of valproate in aggression, alcohol-associated disorders, borderline personality disorder, dependence disorders (including cannabis, cocaine and nicotine), depressive disorders and schizophrenia.

Conclusion Analysis of results depicts a limited efficacy of valproate in schizophrenia, acute alcohol withdrawal, depressive disorders, pathological gambling, suicidal behaviour, and benzodiazepine, cannabis and cocaine dependence. Nevertheless, evidence exists to suggest efficacy in hostility amongst patients with acute alcohol-associated hallucinosis or schizophrenia, and in aggressive behaviour, either alone, or in the context of comorbid bipolar disorder or personality disorder. Common side effects of valproate include elevated liver enzymes, thrombocytopenia and weight gain. The documentation of adverse events and side effects considering valproate is inconsistent. It would improve our understanding of the risk-benefit balance of valproate if future studies reported “common” and treatment-emergent side effects in a more standardised manner.