For people with mental illness that are violent, a range of interventions have been adopted with the aim of reducing violence outcomes. Many of these interventions have been borrowed from other (offender) populations and their evidence base in a Serious Mental Illness (SMI) population is uncertain.

To aggregate the evidence base for non-pharmacological interventions in reducing violence amongst adults with SMI and PD (Personality Disorder), and to assess the efficacy of these interventions. We chose to focus on distinct interventions rather than on holistic service models where any element responsible for therapeutic change would be difficult to isolate.

We performed a systematic review and narrative synthesis of non-pharmacological interventions intended to reduce violence in a SMI population and in patients with a primary diagnosis of PD. Five online databases were searched alongside a manual search of seven relevant journals, and expert opinion was sourced. Eligibility of all returned articles was independently assessed by two authors, and quality of studies was appraised via the Cochrane Collaboration Tool for Assessing Risk of Bias.

We included 23 studies of diverse psychological and practical interventions, with a range of experimental and quasi-experimental study designs that included 7 Randomised Controlled Trials (RCTs). The majority were studies of Mentally Disordered Offenders. The stronger evidence existed for patients with a SMI diagnosis receiving Cognitive Behavioural Therapy or modified Reasoning & Rehabilitation (R&R). For patients with a primary diagnosis of PD, a modified version of R&R appeared tolerable and Enhanced Thinking Skills showed some promise in improving attitudes over the short-term, but studies of Dialectical Behaviour Therapy in this population were compromised by high risk of experimental bias. Little evidence could be found for non-pharmacological, non-psychological interventions.

The evidence for non-pharmacological interventions for reducing violence in this population is not conclusive. Long-term outcomes are lacking and good quality RCTs are required to develop a stronger evidence base.