The Virginia Polytechnic Institute and State University shooting renewed national attention to violent acts by an individual who had previous contact with a psychiatrist. John Hinckley had been under psychiatric care before he shot President Reagan, Columbine High School shooter Eric Harris had received an antidepressant, and the “unabomber,” Theodore Kaczynski, had visited a psychiatrist. These cases have raised concerns about the adequacy of psychiatrists’ assessment and the treatment of individuals who subsequently commit violence. Very few mentally ill persons are violent, and the pattern of mental disorder most commonly associated with violence, comorbid psychopathic personality and substance abuse, is not considered particularly treatable. However, bizarre, extremely violent acts by a person during a psychotic or mood disorder or their prodrome, although rare, are a particular challenge for our field.

Although there is no single clinical picture associated with violent behavior, a building crescendo of paranoid fear in a patient with a history of aggression should create a serious concern about violence for the evaluating psychiatrist. Such patients may strike out at their imagined persecutors in misperceived self-defense. The dilemma with respect to our expertise in assessing the risk for violence in these patients is quite analogous to the evaluation of suicidal risk. Unless a patient specifically acknowledges an intent and plan, we can only determine the extent to which that individual is within a group at greater risk for violence than the general population. Factors that determine the time and place of a violent act, such as a later chance stressor or provocation, may not be known either to doctor or patient at the time of an interview. Violent behaviors are often well planned over a long time but then executed impulsively in a brief period of high emotional arousal. Potentially violent individuals may not provide information regarding their plans or schemas because they are exceptionally guarded. Warning signals may be more frequently observed in public settings than in the psychiatrist’s office, where lower levels of stimulation allow patients to remain more circumspect.

Adding to the problem of prediction are many false positives, particularly in younger individuals. After the Columbine shooting, schools became so hypersensitive to the issue of violence that if a young child wrote the word “bomb” on his desk, he was immediately removed from the classroom for evaluation. Most of these children were not at high risk of committing violence. Violent imagery is common in children, particularly boys, and making their treatment a reflexive punitive response further distances potentially violent children from positive treatment alliances with psychiatrists. Early, sustained assessment and treatment of mental disorders, offered to all affected individuals and their families whether or not violence seems imminent, may be more effective in reducing violence than efforts aimed solely at detection of specific future acts.

For the individual whose violence derives from a severe mental disorder, the potential for violence diminishes with effective treatment of the underlying mental illness. However, this therapeutic effect may require a longer period of treatment than the patient will accept voluntarily. Options include civil commitment on an inpatient unit in order to initiate treatment and assisted outpatient treatment for patients who repeatedly fail to comply with treatment recommendations. The context of the psychiatrist’s relationship with the patient must also be considered. The appropriate intervention for a patient who reveals violent acts or intent in an already ongoing treatment may be different from what might be appropriate for a patient who is brought in for evaluation of potential violence by a third party such as a school.

Involuntary outpatient treatment is an underused strategy for violent patients. Although some fine tuning of commitment statutes may be helpful (e.g., eliminating the imminence requirement in Virginia and a small number of other states), radical changes in the mental health laws are not needed to address the risk of violence. Although the current system allows dangerous patients to be confined as long as substantial violence risk remains, in most states, the current mental health system does not do a good job of keeping those patients stable after discharge from the hospital. Broader use of mandatory treatment after release—as the newer generation of outpatient commitment statutes enables—should be helpful for involuntary patients who are unlikely to adhere to treatment and likely as a result to present a substantial violence risk. Many mentally ill individuals who become violent toward others, like many suicidal patients, do so when they appear to be resolving their acute illness and are discharged prematurely, in retrospect. Longer-term mandated treatments are generally not resisted by patients and could effectively reduce violence. They often lead to increased insight and compliance with treatment. More evidence of their effectiveness is required to gain support for their increased prescription by psychiatrists and for their appropriate funding.

We can successfully treat some conditions associated with risk for anger and violence, such as angry outbursts in mood disorders, especially major depression. Similar progress has not been made in treating violence in psychotic individuals. Families of psychotic patients often bear the brunt of violence by psychotic individuals. High tension, hostility, suicidality, verbal aggression, and physical violence happen within the family and within other care settings. There is little evidence about which interventions, both pharmacologic and psychotherapeutic, may be best for this target of treatment. We do not do enough to educate and prepare families and caretakers on how to manage violence. We have not fully evaluated the effectiveness of psychoeducational approaches and anger management techniques, as well as of efforts to remove guns from the environment.

Research on dysregulated anger pales in comparison to that on other emotional disorders. As with substance abuse in the past, the challenges of working with violent patients appear to have limited the robustness of our clinical research efforts. One constraint is ethical. Patients who are incarcerated or who are hospitalized against their will are relatively powerless, and we have an obligation to ensure that they are not coerced to participate in research projects. However, involuntary patients who are not charged with or convicted of a crime are not identified as a vulnerable group by federal regulations on protection of research subjects. Currently, institutional review board policies cover the spectrum from treating them the same as voluntary patients to severely limiting the research in which they can participate. We need to more clearly inform these review boards that the preponderance of evidence supports the conclusion that involuntary patients are able to make informed decisions about consenting to research. We need to create a new ethical framework to facilitate appropriate voluntary research with incarcerated patients as well.

Many innovative, ground-breaking investigations of violent patients have been accomplished, but research comparable in scope to that currently being conducted with other patients who have serious mental disorders is needed. Some persons committing impulsive violent acts appear to have a reduced threshold for aggression based on reduced top-down control from the prefrontal cortex, perhaps because of structural brain differences or impaired cortical neuromodulation. Reduced serotonin, enhanced catecholamines, and hypersensitivity of the amygdala and other portions of the limbic system in response to negative stimuli modulated by cholinergic and g-aminobutyric acid (GABA)-glutamate imbalance are among the array of candidate neurobiological mechanisms. Pharmacologic, imaging, and genetic paradigms developed for patients with aggressive borderline personality disorder or intermittent explosive disorder could be extended to other violent patients. Similarly, the cognitive impairment identified by research in schizotypal personality disorder should be investigated in violent, paranoid patients. An expanded research agenda is necessary to develop and validate new assessments and treatments for violent patients.

Violence is a complex, multicausal phenomenon, and psychiatrists are not experts on all of its aspects. Our focus is on the individual factors that contribute to violence, especially among people with mental illnesses. The work of other disciplines, including sociology and criminology, has taught us that many other variables predict violence—often more strongly than the factors that psychiatrists consider. Ethical concerns about inappropriately stigmatizing mentally ill patients as potentially violent and then mandating treatment reflect current limitations in the accuracy of our assessment and the effectiveness of our treatment. Society’s acceptance of inadequate funding for the treatment of such patients in an already overburdened mental health system further impedes more comprehensive efforts. The unfortunate end result includes infrequent but periodic acts of violence by mentally ill individuals whom we might have helped. A research agenda that encompasses pathophysiology and intervention, along with model treatment programs that can be widely adopted for public and even private funding, may enhance psychiatrists’ ability to advocate more effectively for their role in helping mentally ill violent patients and protecting the potential innocent victims of violence.