November 20, 1998

Contrary to common wisdom-both inside and outside of the field-psychiatrists are often quite successful in predicting which patients will become violent, at least in the short term. It appears, however, that they are better at predicting violence in patients with certain symptoms than in others. In her presidential address at the annual meeting of the American Academy of Psychiatry and the Law (AAPL) in October, Renée Binder, M.D., emphasized that the next time someone suggests that psychiatrists are incapable of predicting patient violence, an appropriate and accurate response is, "In some situations, we can be confident and fairly accurate in our predictions."

Where many attempts at predicting violence have gone astray, she explained, is in linking a particular psychiatric diagnosis to the likelihood of future violence. Research she and others have conducted shows that "specific clusters of symptoms are more predictive of violence than is diagnosis," Binder said. "Thus, when we are asked, 'Are schizophrenic or manic or depressed patients most likely to be violent?,' the answer is that it depends more on their acute symptoms than simply on their diagnosis."

Speculation about and fear of a connection between mental illness and violence is not a recent development-Plato, Aristotle, and Plutarch all ventured opinions on the issue. In 1751 Benjamin Franklin insisted that the establishment of a mental hospital in Pennsylvania was crucial because, he wrote, "Some [mentally ill people] going at large are a terror to their neighbors, who are daily apprehensive of the violence they may commit."

The relationship was the subject of many studies since the 1920s, most concluding that mental patients were no more violent than the general population, Binder noted.

More recent evaluations, such as those of the NIMH Epidemiologic Catchment Area (ECA) Study and by researcher Bruce Link, did, however, find higher rates of violence in samples of mental patients than in matched samples of residents of the same communities.

Binder has devoted considerable effort over the last 15 years to solving the mystery of whether there is a link between violence and mental illness, because of the major social ramifications of the controversy-persistent stigmatization of the mentally ill, for example-and legal implications that are played out in civil commitment policies and duty-to-warn statutes.

In addition, she noted, "It is crucial to understand as much as we can about which patients are likely to become violent in order to make informed admission and discharge decisions and to protect clinicians, patients, families, and caretakers from violence."

One study she and colleague Dale McNiel conducted in 1987 at the University of California, San Francisco, assessed the level of accuracy that clinicians achieve in judging patient dangerousness on a short-term basis. They chose to compare patients hospitalized for a 72-hour civil commitment because they were believed to be a danger to others or themselves or were gravely disabled and patients admitted under other criteria. They chronicled physical and verbal assaults and use of seclusion and restraint to prevent patient violence.

Their findings countered the oft-professed "pessimistic view" that clinicians are incapable of such judgments.

"What we found," she noted, "was that patients hospitalized on the basis of a judgment that they represented a danger to others engaged in more violent behavior in the first 72 hours of hospitalization than involuntary patients not judged to be dangerous."

She cautioned, however, that this prediction success held for the first 48 hours after commitment but faded by the third day, by which time the number of violent acts did not differ between the two groups. "This is likely because psychotropic medications and ward structure were able to decrease the likelihood of violence."

In a companion study to "fine-tune" their findings about predicting dangerousness, they uncovered patterns of errors that led to inaccurate assessments.

"Clinicians overpredicted violence in nonwhite patients and underpredicted violence in women," Binder said. That is, "we often think nonwhite patients will be violent, but they are not, and we think women will not be violent, but they are, and we need to be aware of this and take safety precautions."

She and McNiel have also studied factors associated with preadmission and inpatient violence. Binder noted that the ECA study found a link between being young, male, and socioeconomically disadvantaged and the likelihood to commit violence. It also found a relationship between violence and diagnoses of schizophrenia and that substance abuse comorbid with a mental illness increased the likelihood of violence.

To explore the link further, Binder looked at violent behavior in the two weeks preceding a psychiatric hospital admission and on the ward. "Men were more likely than women to engage in fear-inducing behavior," including threats and physical attacks in the community, she noted. Once hospitalized, "men continued to be overrepresented in the group of patients exhibiting fear-inducing behavior, whereas women were overrepresented in the physically assaultive group." When diagnosis was evaluated, "both schizophrenic and manic patients were overrepresented in the group who were physically assaultive in the community, but manic patients were most likely to be physically assaultive in the hospital."

Beginning about a decade ago, Binder and McNiel began to investigate whether a relationship between acute psychiatric symptoms and hospital assaults existed. Using scores on the Brief Psychiatric Rating Scale, which was administered at admission, they discovered that certain symptoms were in fact predictive of violence, Binder said, "especially high levels of thinking disturbance, hostility-suspiciousness, and agitation-excitement . . .These symptoms were especially predictive of violence in nonschizophrenic patients." She hypothesized that the reason for this finding might be that psychiatrists "intervene more quickly to decrease violence potential in hostile and paranoid schizophrenic patients."

She cautioned the AAPL audience that she was not concluding that violence risk is independent of diagnosis. "A patient with paranoid schizophrenia may be more likely to commit violence than a patient with panic disorder. However, it seems that symptoms are a better marker for violence potential than diagnoses per se," Binder stressed.

Treatment noncompliance is also a critical factor in assessing violence risk, she added, as is a patient's abuse of drugs or alcohol. Both raise the potential.

She went on to describe her research findings concerning the victims of violence by psychiatric patients. The most frequent targets are patients' family members. In one study, 54 percent of patients who assaulted someone within two weeks of admission attacked family members. She also found that "the caretaking role rather than the type of kinship relationship with the patient" was the crucial factor in whom the patient chose to assault.

Binder and her colleagues also gathered data on assault victims during inpatient stays. The finding that women were targets far more often was negated when they controlled for staff discipline; proportionately more women were nurses than physicians. "Staff discipline, however, was strongly associated with risk of assault," she pointed out. "Nurses. . .were significantly more likely to be assaulted than were doctors. This probably relates to their more frequent contact with the patient and their role relationship in which nurses are expected to set limits with patients."

She hoped that the growing body of data on violence and mental illness will go a long way toward dispelling the myth the public should fear psychiatric patients. "Our finding did not support stereotypes about psychotic patients wandering through the neighborhood randomly attacking strangers . . . . The truth is that psychotic patients most likely attack family members with whom they are living and who are setting limits on them."

While a violence risk does exist for some patients, especially if they are not following treatment recommendations or are using alcohol or drugs, it is critical when formulating social and legal policies, Binder said, "not to confound stigma with the issue of concern about dangerousness."

The Web address for the American Academy of Psychiatry and the Law is www.emory.edu/AAPL/.