Definition and magnitude of the problem

There are numerous ways of conceptualising the definition of violence, although at present there is no consensus as to which of these is the most appropriate. The WHO has defined violence as ‘the intentional use of physical force or power, threatened or actual, against oneself, another person, or against a group or community, that either results in or has a high likelihood of resulting in injury, death, psychological harm, mal-development or deprivation’.5 This definition includes threats, intimidation, neglect and abuse (whether physical, sexual or psychological), as well as acts of self-harm and suicidal behaviour. Although expansive and all-encompassing, it defines violence in terms of its outcomes on health and well-being rather than its characteristics as a construct that is socially or culturally determined.

Studies investigating the prevalence of violence in psychiatric patients show a wide variability, in accordance with the treatment setting in which they were conducted. The lowest prevalence rates of violence have been seen in outpatient settings (2.3–13%), and the highest in acute care settings (10–36%) and involuntarily committed patients (20–44%).6 Around 10% of the patients with schizophrenia or other psychotic disorders behave violently, compared with less than 2% of the general public.7 Although this suggests that mental illness does contribute to the risk of violence, it is important to note that the 1-year population-attributable risk (PAR) of violence associated with serious mental illness alone was found to be only 4% in the ECA (Epidemiologic Catchment Area) survey.8 This implies that even if the elevated risk of violence in people with mental illness is reduced to the average risk in those without mental illness, an estimated 96% of the violence that currently occurs in the general population would continue to occur. Although a statistical relationship with violence has been demonstrated in certain severe mental disorders such as schizophrenia, however, only a small proportion of the societal violence can be attributed to persons suffering from mental disorders.9

The dynamic interaction of social and contextual factors with the clinical variables plays an important role as a determinant of violence. However, these issues have not generated sufficient interest and the emphasis continues to be on the psychiatric diagnosis or clinical variables of the patient, while looking for causal factors of violence.

Violent victimisation of the mentally ill

Patients with severe mental illness constitute a high-risk group vulnerable to fall victims to violence in the community. Symptoms associated with severe mental illness, such as impaired reality testing, disorganised thought processes, impulsivity and poor planning and problem solving, can compromise one's ability to perceive risks and protect oneself and make them vulnerable to physical assault.10 ,11

Violent victimisation of persons with severe mental illness presents obvious dangers of physical trauma and impairs the quality of patients’ lives. Past traumatic and victimisation experiences have been found to be significantly associated with patients’ symptom severity and illness course.12 However, this issue has attracted much less attention than violent behaviour by the patients, in spite of the fact that violent victimisation of patients occurs more frequently than violent offending by the patient.6 ,13 ,14

A recent review reported that the prevalence of violent victimisation ranges between 7.1% and 56%, although the issue of comparability among the studies exists.15 Young age, comorbid substance use and homelessness were found to be the risk factors for victimisation.15 A relationship between victimisation and violent behaviour by patients with severe mental illness has also been suggested in numerous studies.16 However, it is not clear whether past victimisation predicts future violence, or past violence predicts future victimisation, or both.

Predictors of violent behaviour

The relationship between mental illness and violence has been shown to be more complex than initially suspected. From viewing mental illness as a causative agent, researchers after reanalysing the NESARC (National Epidemiologic Survey on Alcohol and Related Conditions) data have confirmed that mental illness and violence are related primarily through the accumulation of risk factors of various kinds, for example, historical (past violence, juvenile detention, physical abuse, parental arrest record), clinical (substance abuse, perceived threats), dispositional (age, sex, etc) and contextual (recent divorce, unemployment, victimisation) among the mentally ill.17 In fact, for those with mental illness without substance use, the relationship with violence was modest at best.7

With the growing repertoire of risk assessment tools, mental health professionals are often expected to predict and manage violent behaviour, especially in an acute care setting. Diagnostically, aggressive behaviour has been linked to schizophrenia, mania, alcohol abuse, organic brain syndrome, seizure disorder and personality disorders.18 Among patients in acute psychiatric settings, young age, male sex, history of psychiatric illness, comorbid substance abuse and positive symptoms have been shown as consistent predictors of violent behaviour. Among these, the history of violence is often emphasised as the most significant predictor of future violence.19 However, overall, the identified risk indicators of violent behaviour have poor predictive validity, in the short-term and the long-term. Large epidemiological studies like the ECA study also found a substantially increased risk of violent behaviour specifically within particular demographic subgroups of participants: younger individuals, males, those of lower socioeconomic status and those having problems involving alcohol or illicit drug use.8 These risk factors were statistically predictive of violence in people with or without mental illness.

Evidence for available treatment options

It is now well established that adequate treatment, including management of comorbid substance use, leads to better outcomes for patients with severe mental illness. This improvement lowers the risk of violence, even up to that seen in the general population. However, there is little evidence that any of the available antipsychotics have specific ‘antiaggressive’ properties, although clozapine may be superior to other drugs in this regard.29 Antiepileptics have shown benefit in reducing aggression in persons with intellectual disability and seizure disorder, but their effectiveness for this indication in severe mental disorders is unproven.30 Thus, the best possible strategy seems to be to the reduction in psychopathology and functional deficits.

Research and public health challenges

The assessment of violence-specific risk prediction in the past studies presents several limitations: unclear definition of violence, use of non-standardised scales for the evaluation of aggressive behaviour, non-homogeneous samples, absence of control groups and of prospective design in the majority of the studies.31 These limitations might explain the heterogeneity of conclusions drawn by various studies, and particularly the wide variations in risk ratios for mental illness as a contributor to the violence. An attempt to resolve this heterogeneity is important from a public health perspective as the association of violence with mental illness hampers community reintegration of people with schizophrenia.

Also, most studies have primarily examined the association between violence and severe mental illness, for example, schizophrenia, in terms of relative risk (ie, the amount of risk posed by those with schizophrenia relative to others). However, there is a dearth of literature on indices of greater public health significance, such as PAR %: the percentage of violence in the population that can be ascribed to schizophrenia and thus could be eliminated if schizophrenia was eliminated from the population.32 A shift of research focus from relative to attributable risk will help provide a more balanced picture and prevent unnecessary stigmatisation of people suffering from severe mental illness. Another major issue is that, since causality between mental illness and violent behaviour cannot be definitively determined, these indices need to take into account the various social-related, contextual-related and comorbidity-related factors which would act as confounders. Better ways are required for presenting risk magnitudes in a comprehensive manner.

The public health importance of resolving these issues is, to a certain extent, in disassociating mental illness from the concept of dangerousness. Any attempt to resolve these issues must begin with an acceptable operational definition of violence, and clear distinctions between various types (towards self/others, verbal/physical, intended/actual, etc) for more consistent and reliable reporting.

Additionally, studies of violence among people with mental illness must go beyond linking various conditions or categories with rates or severity of violence, and instead include a careful examination of contextual and comorbid factors, so that the complex patterns of confounding may be unravelled. It is only with such an understanding that the appropriate intervention(s) might be formulated, and provided to patients at an appropriate time and setting.

Evidence regarding the effectiveness of psychotropic drugs on violent behaviour as one of the treatment outcomes is not yet adequately researched. Moreover, investigating the effectiveness of specific psychotropic drugs on violent behaviour as an outcome is also riddled with numerous challenges. Although pharmacoepidemiological studies provide an opportunity to assess the effectiveness of psychotropic drugs in reducing incidence of violent behaviour, they are subject to a number of confounding factors. These studies have often failed to look into the individual, social, economic and contextual factors responsible for variability in the risk of violence in these patients. Similarly, randomised controlled trials to investigate the efficacy of drugs to reduce violence in particular are also mired with feasibility issues.

Violent patients are often difficult to recruit and the attrition rates are also high in such studies. Also, since the outcome has a lower rate of occurrence, the sample size of studies needs to be high.30 Moreover, conducting such studies will pose an ethical dilemma as violence in a psychiatric patient is considered as an acute emergency, warranting immediate intervention.