When a patient threatens violence to another person, the responsible mental health care professional faces a decision with potential clinical, ethical and legal consequences. The clinician must first decide whether there is a realistic risk of violence or whether the patient is expressing fantasies or just blowing off steam.
Clinical assessments of dangerousness should be regarded as assessments of risk, rather than as predictions of violence. As we are not able to predict with certainty or near certainty whether or not a person will be violent, assessment should seek to describe the individual with a full clinical and behavioral description. Whenever possible, the individual should be assessed in an ongoing manner, rather than just once. The probability of future dangerous behavior can be related to the similarity of the person's current situation to situations in which they were violent in the past.
Prior history of violent behavior is the single best predictor of future violence. It is important for clinicians to fully characterize prior acts of violence so that they can examine the patient's history for patterns of violent behavior and current situational similarities. Who did what to whom? When, where and under what circumstance? Who was the victim and what was their relationship to the patient? Were there perceived precipitants? Was there a motive or provocation? Did the violence appear to be impulsive or spontaneous, or was it premeditated?
Other historical details may provide information that would increase the patient's risk category. Was the subject under the influence of drugs or alcohol? Were weapons used? Was the injury serious? What was the most violent thing the individual has done? If there is no history of actual violence, what was the closest they ever came to being violent?
Demographic variables such as younger age and lower socioeconomic status may increase risk. Race is not a significant factor when socioeconomic status and other demographic variables are controlled. Males in the general population do commit more violent offenses. However, among male and female psychiatric patients, the rates are similar. Mental status can influence the likelihood of violence. The strongest correlation has been the presence of active psychotic symptoms, especially command auditory hallucinations of a familiar voice or of influence and control. One study, however, found no relationship between delusions and violence in the absence of substance abuse one year after discharge from hospitalization (Appelbaum et al., 2000).
Is the individual using substances? Is the person angry with anyone for either realistic or delusional reasons? Do they say that someone has disrespected them? Do they make frequent and open threats or describe a concrete plan? If the answers to these questions are affirmative, the patient's risk category is higher.
An assessment of clinical factors, including Axis I mental disorder, character pathology and possible medical conditions, is also important. There is a modest relationship between violence and mental disorder; the risk for violence increases with a greater number of comorbid diagnoses. Active psychotic symptoms and comorbid substance abuse or dependence present much more of a risk than any diagnosable mental disorder alone.
Patients may be violent based on character as well as psychosis, and it is wise to identify traits such as sociopathy, narcissism, risk-taking, low frustration tolerance, inability to tolerate criticism and difficulty managing temper. In specifically considering potential for impulsive violence, it is important for clinicians to evaluate general impulsiveness. Is this a person who makes quick decisions and acts on them? Does heavy drinking or other behavior increase impulsiveness?
Risk of violence depends on the person and their situation. Risk for violence in one situation may differ from risk in another. Stress, whether based on family, peer group or employment stressors, can increase risk.
After a careful assessment of potential for violence, the clinician can make an informed estimate on whether or not a threat exists. If violence is unlikely, then the legal and ethical duties are simple: Maintain the patient's confidentiality and continue the evaluation or treatment as clinically indicated. New legal duties may come into play when the clinician concludes that potential violence is an issue. Potential conflict arises between the duty to protect and the duty to maintain confidentiality. The duty to protect may involve warning a third party or initiating involuntary hospitalization of the patient, both of which breach confidentiality.
The fundamental principle in this situation is to treat the potential violence as a clinical issue. Once the clinician has decided that violence is a real risk, then they should discuss this assessment with the patient and treat the risk of violence as a serious problem for the two of them to work through together. In our experience, this approach usually leads to an agreement between the clinician and the patient on how to proceed--whether to warn the potential victim, modify the current treatment or hospitalize the patient.
Note that when agreement is reached, the ethical dilemma vanishes. Confidentiality is only breached when the clinician acts without the patient's consent. When the patient consents to share information either with significant others or clinicians, there is no breach of confidentiality. If such an agreement cannot be worked out, the clinician has a clear obligation to break confidentiality. The imminent threat of violence to a specific individual trumps the issue of protecting confidentiality, and the courts have generally supported this position.
Since the Tarasoff (1976) decision, clinicians have reasonably been concerned that when patients are violent, legal action against the therapist may follow. These concerns can at times frighten clinicians to the point that their clinical judgment is affected. It is necessary to keep careful records, including ongoing notes outlining the clinician's thinking (pro and con) for each of their decisions. Equally important is an emphasis for appropriate consultation in questionable cases.
We believe that clinicians can best deal with these difficult situations when they know what has actually happened in prior cases that led to suits. In our clinical experience, impulsive violence poses the greatest dilemma for clinicians. Because impulsive patients are, by definition, unpredictable, they naturally pose a greater risk of harm when they are delusional or angry. In what follows, we review the cases in which impulsive violence by patients has led to Tarasoff-like suits against the clinician.
Case 1a. White v United States (1986) involved a man who killed a police officer. He was found not guilty by reason of insanity and then committed. After some years in the hospital, he had progressed to off-grounds privileges, and he had met and married another patient who was subsequently discharged. While he was visiting her, she showed him pictures of herself and another man, both of them in bathing suits. The husband impulsively grabbed a pair of scissors and superficially stabbed his wife 55 times in the back. The wife sued the patient's therapist alleging a failure of duty to protect. The court found for the therapist, noting, "Dr. Brown gave a detailed and persuasive explanation of why the single fantasy did not represent a danger [to the wife]."
In this case, there was a prior threat to a clearly named victim and a credible motive. The court found the violence to be foreseeable, but ruled that the clinician's clinical judgment had been made with sufficient care so that she was not held responsible for the harm to the patient's wife. This case illustrates that careful clinical assessment of potential for future violence protects the clinician from a finding of malpractice, even when future violence occurs.
Case 1b. Nine weeks after being discharged from the state hospital, a mentally ill man became annoyed at some neighbors and asked them to keep the noise down. When they failed to respond to his satisfaction, he fired a shotgun, killing one and wounding the other. The survivor sued the hospital and the psychiatrist who had discharged the shooter (Sharpe v South Carolina, 1987). The Court found for the defense, holding that, as a matter of law, there was no duty to the victims, because they were not identifiable.
Case 1c. Moye v United States (1990) concerned a veteran with a history of alcohol and marijuana use who was in outpatient treatment for paranoia for several years. After refusing to comply with treatment, the patient was taken home by his parents. Subsequently, he impulsively shot and killed both of them. The administrator of the parents' estate filed a civil action claiming the defendants were negligent in failing to hospitalize the patient. The court held that there was no affirmative duty to seek involuntary commitment of any person or to control a voluntary patient. Therefore, as a matter of law, the U.S. Department of Veterans' Affairs (VA) and the physicians could not be liable. The United States' motion for summary judgment was granted.
Appelbaum PS, Robbins PC, Monahan J (2000), Violence and delusions: data from the MacArthur Violence Risk Assessment Study. Am J Psychiatry 157(4):5665-72.
Bardoni v Kim, 151 Mich. App. 169, 390 NW.2d 218 (Mich. 1986).
Boulanger v Pol, 258 Kan. 289, 295, 900 P.2d 823 (1995).
Eckhardt v Kirts, 179 Ill. App. 3d 863, 534 N.E.2d 1339, 1344, 128 Ill. Dec. 734 (Ill. 2d Dist. 1989).
Hamman v County of Maricopa, 775 P.2d 1122 (Ariz. 1989).
Jablonski v United States, 712 F. 2d 391 (9th Cir. 1983).
Leedy v Hartnett, 510 F. Supp. 1125 (MD Pa. 1981).
Moye v United States, 735 F. Supp. 179 (E.D.N.C. 1990).
Sellers v United States, 870 F.2d 1098, 1102 (6th Cir. 1989).
Sharpe v South Carolina, 354 S.E.2d 778 (S.C. App. Ct. 1987).
Tarasoff v Regents, 17 Cal. 3d 425, 551 P.2d 334, 131 Cal. Rptr. 14 (Cal. 1976).
Webb v Jarvis, 575 NE.2d 992, 994 (Ind. 1991).
White v United States, 780 F.2d 97 (DC Cir 1986).