Evaluating a patient's risk of committing violence against another person is one of the more common duties of the physician in the psychiatric emergency service. In the psychiatric emergency service at the University of Cincinnati Medical Center, with which the first author previously was affiliated, some 10 percent of patient visits are occasioned by concern about homicidal thoughts. Others have reported that up to 17 percent of patients seen in the psychiatric emergency service are homicidal, and that up to 5 percent are both homicidal and suicidal (1).
This paper presents guidelines to help clinicians in the psychiatric emergency service evaluate patients who present with homicidal ideation or violent behavior. The guidelines are styled after a similar presentation on the assessment of suicide risk published in this column last year (2).
As with suicide, no clinician can predict what is going to happen. The goal is risk assessment. The treatment decision must be related to the outcome of that risk assessment. In other words, the emergency clinician is not in the business of predicting an outcome, but of making a logical assessment of a prevailing risk of violence in order to develop a reasonable treatment plan (3).
A limited database on risk factors for violence exists, but researchers have concluded that, in general, the only factor associated with future violent behavior is a history of violence (4). In any case, epidemiological databases are largely irrelevant to the clinician in the psychiatric emergency service. Epidemiology focuses on long-term prevalence rates—for example, three-year rates—or lifetime prevalence rates. The data are related to the overall probability that a person will eventually attempt or commit a violent act. In the psychiatric emergency service, the clinician is concerned with short-term risk, not the risk over future months or years.
The definition of short-term risk is controversial. Does a homicide committed by someone who was evaluated two weeks earlier suggest faulty risk assessment? As we suggested in our discussion of suicide risk assessment (2), we consider a period of 24 hours as the short term. This period represents the maximum amount of time most emergency facilities can retain patients without admitting them as inpatients.
As in the assessment of the suicidal patient, rating scales and instruments, such as the Overt Aggression Scale, have their place for the purpose of prompting inquiries for a complete assessment. However, they cannot and must not replace individual clinical judgment about a particular patient's current risk of violence.
To formulate an adequate assessment of current homicidal risk, the clinician must combine his or her experience, knowledge base, judgment, common sense, and empathic understanding in the clinical examination of the patient. No shortcut, laboratory test, or psychometric instrument can substitute for engaging the patient in a thorough interview.
In structuring the evaluation of a patient presenting with homicidal ideation in the psychiatric emergency service, the following areas of inquiry have proved useful. We present them without implying that this listing is necessarily complete.
Make your safety the priority.
You, the evaluating clinician, must be safe. At a minimum, patients must have been searched and disarmed before meeting with the evaluating clinician. A clear route of rapid egress from the examination room must be ensured, and security personnel must be available, ideally through a panic button or other immediate means. Safety considerations may require that the patient be in restraints or that a physical barrier be present between patient and clinician. The clinician's own experience and anxiety level ought to be the determining factor in deciding the extent of safety precautions in place during a particular evaluation.
Evaluate the patient's situational context.
Many acutely homicidal patients have the perception that their situation can have no alternative resolution other than violence. The perception of rejection or humiliation is particularly important. Understanding the prevailing stressors in terms of the patient's view—not the external observer's—is most relevant to the assessment.
Expand your database to include a detailed history of past acts of violence.
The most relevant data for assessment of current risk of violence are past behaviors. Obtain old medical records, and, when possible, obtain police reports, family reports, and information from any intended victim. Identify where the patient is on the continuum of violence. What is labeled "homicidal" on the intake worker's assessment sheet may encompass a broad variety of ideation and behaviors. They may range from globally aggressive thoughts, such as "I'm going to get them," to a specific lethal plan with available means to carry it out. If a patient has identified an intended victim or purchased a firearm, the expressed violent ideation should be assumed to be part of a finalized plan. Clinicians must inquire about any ideation of harm, such as assault, rape, or destruction of property, as well as thoughts of homicide.
Assess accompanying psychopathology.
A particular concern is the loss of reality testing. A patient who is homicidal because of a delusional belief or a command hallucination is at high risk, and that risk is unlikely to be contained within the time available in the psychiatric emergency service. Symptoms such as feeling controlled by an outside force or the patient's believing that others wish him or her harm confer even higher risk of violence (5). Mental status changes associated with a reduction in impulse control, such as mania, delirium, or intoxication, weigh in as additional risk factors.
On the other hand, some patients who present with violent behavior or thoughts of killing someone do not meet criteria for any axis I disorder but nevertheless present high risk. Individuals with antisocial personality disorder may need psychiatric help, but they also belong in a restrictive setting where they can be safeguarded. A psychiatric emergency service should discharge them quickly into the custody of law enforcement authorities.
Patients who present with homicidal ideation are also at a greater risk of killing themselves. The assessment of any individual who harbors thoughts of violence directed at others must include a suicide risk assessment.
A religious belief or fear of the legal consequences are possible factors mitigating against the probability of homicidal behavior. Any evidence that a person has the capacity for empathy likewise mitigates against such behavior.
Imagine the situation awaiting the discharged patient.
It is important to know if the patient returns to a situation that is essentially unchanged from the one he or she left behind. The evaluator should try to see this situation from the patient's perspective and, in particular, try to envision the range of options available to the patient. Is the range as narrow—from the patient's point of view—as before, or has the evaluation interview opened up new viable alternatives? Are drugs, alcohol, or weapons readily available to the discharged patient?
Consider the effect of the emergency service visit, but avoid a "no-homicide" contract with the patient.
"No-homicide" contracts seem to be less popular than "no-suicide" contracts, and they have similar problems (2). Some of the issues mentioned above, such as the perceived lack of alternative solutions, can improve as a result of the clinician's intervention or other factors associated with the visit to the emergency service, and the assessment of risk may change accordingly. However, we caution against the use of contracts. They obviously fail to meet defined standards of legal contracts. They falsely reassure the clinician that a risk has disappeared when, in fact, the homicidal patient may agree to the contract so he or she can leave the emergency service. Contracts in themselves do not decrease the risk of violence.
This strategy is not just defensive medicine. Certainly, another clinician's concurring opinion can be useful evidence in supporting a claim of having used sound professional judgment if litigation arises. But a second opinion can also help the clinician get a perspective on his or her own countertranferential attitude toward the patient, correct distortions, and raise or alleviate concern correspondingly. We have found that a second opinion provided by a colleague of the opposite sex may sometimes be helpful in identifying distortions resulting from countertransference.
Get a urine drug screen, and do not discharge an intoxicated patient.
Patients who are under the influence of alcohol or other substances are at an increased risk of causing harm to themselves or others, often accidentally or impulsively. A patient should generally not be discharged from the psychiatric emergency service into a less restrictive setting unless he or she has attained clinical sobriety, which is not to be equated with a particular blood concentration of the intoxicant. If certain drugs, such as amphetamine and phencyclidine, are detected in the urine, the patient should be retained for additional observation. The observation should focus on erratic behaviors. The duration of observation will be a function of the pharmacokinetics of the circulating drug.
Document the dispositional decision and its rationale.
The documentation can help the clinician who receives the emergency service's disposition assess the patient's continuing homicidal risk and response to treatment. Documentation is also obviously essential for forensic reasons. If a patient is discharged from the psychiatric emergency service rather than admitted to a hospital bed, the exit discussion should include clear directions about what to do if the patient experiences homicidal thoughts again.
Fulfill your legal duties.
The 1976 Tarasoff decision specifies that clinicians have a duty to protect a patient's intended victims. This duty may include notifying the police and hospitalizing the patient (5). States vary in their application of the decision. We recommend that emergency service psychiatrists familiarize themselves with their state's case law before they are faced with a Tarasoff -type case. In addition, violence towards others is a tort, and the patient should be informed that the consequences of his or her violent behavior would include criminal sanctions.
Physicians in the psychiatric emergency service are frequently called on to assess a patient's current risk of violent behavior. The guidelines presented in this paper can help clinicians conceptualize the task of risk assessment and can serve as a reminder about important clinical areas and legal duties to cover during an actual assessment of a potentially violent patient.
Dr. Thienhaus, editor of this column, is professor and chairman of the department of psychiatry at the University of Nevada School of Medicine, Mail Stop 354, Reno, Nevada 89557-0046. Dr. Piasecki is assistant professor of psychiatry at the school.