Violence to Others, Violent Self-Victimization, and Violent Victimization by Others Among Persons With a Mental Illness

Violence to others by persons with a mental illness affects not only those who are victimized. It also fuels stigmatizing public perceptions of mental illness and is often invoked to justify fear-based mental health policies, such as loosening standards for public reporting of protected health information (1). Given these profound sequelae, it is not surprising that over the past several decades, an enormous research literature has focused on the assessment and management of the risk that persons with a mental illness will commit violence to others in the community (2).

Violence directed at others, however, is not the only form of violence that may be associated with mental illness. In fact, the “lived experience” (3) of violence among people with a mental illness includes two additional forms of violence. By simultaneously taking into account all three forms of violence, a more scientifically inclusive and clinically credible account of violence among persons with mental illness might emerge.

The first additional association between mental illness and violence reflects violence directed at oneself. Violent self-victimization—violence in which a patient is both the perpetrator and the victim, such as attempted suicide or other physically self-harming behavior—has long been recognized as a serious clinical concern, with most mental disorders associated with increased rates of suicide and self-injury (4,5).

The second additional association between mental illness and violence reflects violence directed by others at persons with a mental illness. Although this type of violence is studied much less frequently than other forms of violence associated with mental illness, patients with a mental illness are violently victimized by others in the community at a higher rate than the general population, a phenomenon that has attracted increased empirical attention in recent years (6–8).

To our knowledge, this is the first study to use a single data set, the MacArthur Violence Risk Assessment Study, to examine simultaneously the prevalence and characteristics of all three forms of violence involving people with a mental illness. The possible co-occurrence of various forms of violence could have important implications for clinical practice, including efforts to conduct routine screening, tailor treatment and preventive interventions, and pursue legal options for mandated inpatient or outpatient treatment.

Methods

The MacArthur Violence Risk Assessment Study (9–12) involved a follow-up sample of 951 patients who had been discharged from acute civil inpatient facilities at three U.S. sites. English-speaking male and female patients who were between the ages of 18 and 40; who were of white, African-American, or Hispanic race-ethnicity; and who had chart diagnoses of bipolar disorders, depressive disorders, schizophrenia or other psychotic disorders, substance use disorders, or personality disorders were included in this research.

Violence toward others by patients—the focus of the MacArthur Study—was measured by using three sources of information, including interviews with patients; interviews with persons named by patients as knowledgeable about their lives, known as collateral sources; and official sources of information (arrest and hospital records). Information from the three sources was coded and reconciled. Patients and collateral sources were interviewed in person every ten weeks for approximately one year after hospital discharge. Violence to others was defined as acts of battery that resulted in physical injury, sexual assaults, assaultive acts that involved a weapon, or threats made with a weapon in hand.

Violent self-victimization by patients was measured by means of a structured interview administered at each follow-up. Interviewers asked patients whether they had attempted to hurt themselves since the previous follow-up and, if so, how they had done so. Patients who took physical action to harm themselves were coded as having engaged in violent self-victimization (also referred to as “suicide-related behavior”) (13).

Violent victimization of patients by others was also measured by means of a structured interview at each follow-up. Violent victimization of patients by others was defined as patients’ self-reports of having been targets of acts of battery, sexual assaults, assaultive acts that involved a weapon, or threats made with a weapon in hand. Information from collateral sources or from official records regarding violent victimization and violent self-victimization was not obtained (14).

Results

As previously reported, one year after discharge, the prevalence rate for violence to others was 28% (N=262) (9); violent self-victimization, 23% (N=217) (13); and violent victimization by others, 43% (N=405) (14). One-year postdischarge prevalence was 58% (N=555) for involvement in at least one form of violence, 28% (N=262) for involvement in at least two forms of violence, and 7% (N=67) for involvement in all three. The 262 patients involved in at least two forms of violence consisted of 14 (5%) patients who were involved only in violence to others and in violent self-victimization, 124 (47%) who were involved only in violence to others and in violent victimization by others, and 57 (22%) who were involved only in violent self-victimization and in violent victimization by others, as well as the 67 (25%) patients who were involved in all three forms of violence.

Table 1 presents the characteristics of patients involved in at least one of the three forms of violence and patients involved in all three forms of violence and compares them with the characteristics of patients from the MacArthur Study who had no involvement in violence over the one-year follow-up period. Neither the patients involved in at least one form of violence nor those involved in all three forms differed significantly in gender or race-ethnicity from the patients who were uninvolved in violence. However, both of the groups that had been involved in violence were significantly more likely to have a history of prior hospitalization and of prior arrest compared with the group uninvolved in violence.

In terms of diagnosis (10), neither the group involved in at least one form of violence nor the group involved in all three forms of violence was more likely than patients who were uninvolved in violence to have a diagnosis of depression. The group involved in at least one form of violence was significantly less likely than the group that was uninvolved in violence to have a diagnosis of schizophrenia or bipolar disorder. Both the group that was involved in at least one form of violence and the group that was involved in all three forms of violence were significantly more likely than the group that was uninvolved in violence to have a diagnosis of alcohol or drug use disorder. Regarding co-occurring diagnoses, the proportion of patients diagnosed as having a major disorder (depression, schizophrenia, or bipolar disorder) without a diagnosis of a substance use disorder was significantly higher in the group that was uninvolved in violence compared with the groups that were involved in at least one form of violence. At the same time, the proportion of patients diagnosed as having a substance use disorder without a diagnosis of a major disorder was significantly higher among the patients involved in at least one form of violence compared with patients who were uninvolved in violence. Patients diagnosed as having a major disorder and a substance use disorder constituted the largest proportion of both the group involved in at least one form of violence and the group involved in all three forms of violence, but the proportions of patients who had both a major disorder and a substance use disorder was not significantly different between either of the groups involved in at least one form of violence and the group that was uninvolved in violence.

In terms of social history, both the patients involved in at least one form of violence and the patients involved in all three forms of violence were more likely than the patients who were not involved in violence to report having been physically abused as a child. The patients involved in at least one form of violence were more likely to report having been sexually abused as a child compared with those who were uninvolved in violence, but this comparison failed to reach statistical significance for patients involved in all three forms of violence. Both the patients involved in at least one form of violence and the patients involved in all three forms of violence were more likely than the group that was uninvolved in violence to report that their fathers had been arrested two or more times during their childhood and that they had been homeless and unemployed during at least one of the ten-week postdischarge follow-up periods.

Discussion and Conclusions

Most patients (58%) experienced at least one form of violence—whether as perpetrator, as victim, or in the case of self-harm, as both perpetrator and victim. Over one-quarter (28%) of patients experienced at least two forms of violence, and 7% of patients experienced all three forms of violence.

Although all of the groups of patients we studied were similar in gender and racial-ethnic composition, the group that engaged in at least one form of violence and, particularly, the group that engaged in all three forms of violence led much less stable and more traumatic lives than the group that was uninvolved in violence. These two groups were much more likely to have histories of prior hospitalization and arrest and to have been homeless and unemployed for some portion of the one-year follow-up. These features extended back to the patients’ childhoods: patients involved in at least one form of violence and, particularly, patients involved in all three forms of violence were much more likely to report having been physically abused as children and to report that their fathers had been arrested two or more times during their childhoods.

Although there were no differences among the groups in rates of depression, the group involved in at least one form of violence was much less likely than the group that was uninvolved in violence to be diagnosed as having schizophrenia or bipolar disorder. In terms of co-occurring diagnoses, both groups that were involved in violence were much more likely than patients who were uninvolved in violence to be diagnosed as having a substance use disorder.

We believe that the high prevalence rate of experiencing at least one form of violence and the nontrivial prevalence rate of experiencing all three forms of violence among patients who had been discharged from acute civil inpatient facilities, along with the characteristics associated with each of these groups, may have at least three implications for clinical practice.

First, given the substantial overlap among the three forms of violence studied here, clinicians should routinely screen patients who report one form for the occurrence of the other two forms of violence (7).

Second, for patients who screen positive for one or more forms of violence, clinicians should consider both treatment and preventive implications (8). For example, patients who have been violently victimized by others might benefit from trauma-informed treatment and, for those who are homeless, from efforts to obtain adequate housing, given that being domiciled reduces rates of victimization (15). The co-occurrence of several forms of violence involvement may require a package of interventions with components geared to each.

Third, risk of violence to others and of violent self-victimization are independent grounds for invoking inpatient or outpatient civil commitment, and the risk of violent victimization by others may be encompassed under state provisions that allow the commitment of a person with a mental illness who is likely to “suffer serious harm due to his lack of capacity to protect himself from harm” (Code of Virginia §37.2–815). Findings reported here indicate that these three justifications for commitment, while legally independent, often may be fulfilled simultaneously. Recognition of the co-occurrence of various forms of violence may increase the perceived benefit of inpatient or outpatient commitment and may be more persuasive to legal decision makers.