The public perception of mental health and violent behaviour has often been influenced by poor media coverage and a focus on specific crimes committed by people with psychosis. There’s a nice summary of work in this area on the Mind website, which includes a number of enlightening stats, including this one:

The fear of random unprovoked attacks on strangers by people with mental health problems is unjustified. This has been highlighted by a US finding that patients with psychosis who are living in the community are 14 times more likely to be the victims of a violent crime than to be arrested for such a crime.

Of course, interpersonal violence does sometimes occur in mental health and criminal justice populations. This new health technology appraisal conducted by researchers from the University of Liverpool set out to examine the effectiveness of pharmacological, psychosocial and organisational interventions that have been developed to help prevent and manage violent behaviour.

The authors conducted a systematic search to identify studies that looked at interventions aimed at reducing violence against other people. Participants in the studies had to be aged ≥ 17 years and either have a mental disorder, be offenders or have committed indictable offences. Studies also had to report an outcome measure of violence either directly (e.g. reconviction for a violent offence) or indirectly through a proxy measure (e.g. a validated anger measure).

They found 198 studies, including 51 randomised controlled trials. The literature was extremely diverse and covered 94 different types of intervention and 55 different outcomes. This heterogeneity made it very difficult to conduct a robust meta-analysis and pool the results.

However, three interventions were highlighted as effective in the results:

• Atypical antipsychotic drugs [OR 0.21, 95% CI 0.16 to 0.27, fixed effects; OR 0.24, 95% CI 0.14 to 0.43, random effects; 10 studies, I2 = 72.2, Q = 32.4 (df = 9), p < 0.0001]
• Psychological interventions [OR 0.63, 95% CI 0.48 to 0.83, fixed effects; OR 0.53, 95% CI 0.31 to 0.93, random effects; nine studies, I2 = 62.1, Q = 21.1 (df = 8), p = 0.007]
• Cognitive behavioural therapy (CBT) as a primary intervention [OR 0.61, 95% CI 0.42 to 0.88, fixed effects; OR 0.61, 95% CI 0.37 to 0.99, random effects; seven studies, I2 = 21.6, Q = 7.65 (df = 6), p = 0.26].

The authors concluded:

Whilst some interventions targeted at mental health populations to reduce violence are well supported by the evidence a scattergun approach in the research literature provides little firm evidence for the majority of interventions.

Evidence shows small-to-moderate effects for cognitive behavioural therapy for psychological interventions, and larger effects for atypical antipsychotic drugs.

The report also made the following recommendations for future research:

• Improvements are needed in the design quality of future research studies. Of particular note is the relative dearth of RCTs, especially in the evaluation of non-pharmacological interventions. Furthermore, RCTs themselves should be improved by extending the study follow-up period wherever possible. The quality and rigour of research in the field could be improved by more consistent attention to the protocols that have been published with respect to the reporting of both randomised and quasi-experimental designs. Researchers should identify a single primary outcome variable against which effectiveness is judged.
• Any approach that could increase the homogeneity of research in this field will be welcomed. Greater homogeneity in study design, the interventions applied and outcome measures used, would all be beneficial, especially if actual aggression or violence rather than some proxy for these were to be adopted as the primary outcome measure. If the best-validated measures were to be more widely used it would strengthen internal validity and also facilitate comparability across studies for review purposes.
• A programme of research funded and co-ordinated at a national or international level should be developed, as this would improve the capacity to conduct robust MAs and increase confidence in their results. The review has revealed the extensive literature that has been produced in just the past few years but this is coupled with relatively low design quality. Much of the research is conducted opportunistically by practitioners on the basis of what is possible within their clinical setting. Although this is laudable as a contribution to the principle of evidence-based practice, without adequate resources to improve study design the cumulative evidence base will never produce knowledge that is generalisable beyond specific local settings.
• Some treatment approaches are particularly lacking in evidence-based interventions, such as psychosocial interventions other than CBT. A greater focus on improving the quantity and quality of research here is likely to prove very beneficial.
• Psychosocial and other non-pharmacological interventions should be defined more clearly so that the theoretical elements they are testing is made explicit. In this way, the key components that make up a broad intervention, such as CBT, will be identified and examined for effectiveness.

Hockenhull J, Whittington R, Leitner M, Barr W, McGuire J, Cherry G, et al. A systematic review of prevention and intervention strategies for populations at high risk of engaging in violent behaviour update 2002-8 (PDF). Health Technol Assess 2012;16(3).