"Using a decision-tree approach, each question is asked based on the clinician’s previous response. At the end of 20 minutes, the clinician knows what the patient’s risk of committing violence is in the next 20 weeks," said John Monahan, Ph.D.
Monahan, a professor of law, psychology, and legal medicine at the University of Virginia in Charlottesville, presented the Manfred S. Guttmacher Award lecture at APA’s 2002 annual meeting last month in Philadelphia.
The National Institute of Mental Health is funding the development of the software program, which is expected to be released late next year.
The tool is based on 106 variables culled from the MacArthur Violence Risk Assessment Study. Monahan and his colleagues won the Manfred S. Guttmacher Award for their book Rethinking Risk Assessment: The MacArthur Study of Mental Disorder and Violence in which they described the development of the study and the assessment tool. The Guttmacher Award is given annually by APA and the American Academy of Psychiatry and the Law for outstanding contributions to forensic psychiatry literature.
Patients were recruited from the Western Psychiatric Institute and Clinic in Pittsburgh, Western Missouri Mental Health Center, Worcester (Mass.) State Hospital, and the University of Massachusetts Medical Center.
The baseline sample size was 1,136 patients, with a final sample size of 951 patients and a dropout rate of 29 percent.
Patients were assessed at two 10-week intervals in their communities using a large number of demographic, historical, and psychosocial variables, said Monahan. Other people, such as relatives and friends, were also interviewed at these intervals to corroborate patients’ self-reports. Hospital and police records were reviewed as well.
Fifty-nine percent of the subjects were men. Caucasians made up 69 percent of the subjects; the remainder were Hispanic and African American.
The patients’ primary diagnoses were major mood disorders, personality disorders, and substance abuse. Seventy-two percent of the patients had been hospitalized previously, and 42 percent had been involuntarily committed. Seventeen percent had committed violent acts in the two months prior to hospitalization.
Aggressive behaviors including biting, choking, pushing, and hitting were classified as violent if the patient caused physical injury. Other violent behaviors were sexual assault, assault with a potentially lethal weapon such as a knife or gun, or threats made while holding a weapon, said Monahan.
The perception that people with serious mental illnesses may be violent persists among the public despite educational campaigns to the contrary.
"Public perceptions drive mental health policies. The criterion of dangerousness to self or others is the basis of involuntary commitment laws and more recently mandatory community treatment laws," said Monahan.
Another reason to study risk assessment is that the public and the courts rely on psychiatrists and mental health professionals to assess the risk of violence accurately, said Monahan. "However, research on the accuracy of their predictions shows that it is only slightly better than chance."
In addition, previous studies have been limited to a small number of variables, white men, or one site, added Monahan.
The one-year follow-up results showed that about 25 percent of the subjects had committed one or more violent acts. Most of these acts were committed within 20 weeks after discharge from the hospital.
"This led to the 20-week timeframe in our risk-assessment tool because we found the accuracy of our predictions declined dramatically after 20 weeks," said Monahan.
The most common violent behaviors were kicking, biting, choking, and beating up someone, followed by weapon use or threats with a weapon, according to the authors.
The researchers compared the patient sample in Pittsburgh with a control group of 519 people who reside in the same neighborhoods. The subjects were randomly selected and matched by similar demographic variables. The comparison assessment period was limited to 10 weeks due to funding limitations, said Monahan.
The results showed that the patients and the controls had similar rates of violence when the subjects with substance abuse were excluded (5 percent for the patients versus 3 percent for the control group). Discharged psychiatric patients who had more than one symptom of alcohol or drug abuse at the first 10-week follow-up, however, were twice as likely as the control subjects with substance abuse to be violent (22 percent versus 11 percent), said Monahan.
In addition, the psychiatric patients had nearly double the rate of substance abuse than the control subjects, said Monahan.
The results dispelled the stereotype that people with mental illness are likely to victimize strangers. In fact, control subjects were twice as likely to victimize strangers as were patients (22 percent versus 11 percent). Discharged patients were more likely to victimize family members, followed by friends and acquaintances, according to Monahan.
"Although our study wasn’t on managing the risk of violence, our results showed a link between the number of treatment sessions and rate of violence," said Monahan.
Fourteen percent of the group that had no treatment during the first 10-week follow-up period committed one or more violent acts in the second follow-up period. By contrast, the group that had approximately weekly treatment sessions in the first follow-up period had a 3 percent rate of committing violence in the next follow-up, according to Monahan.
"These rates held up when we controlled for substance abuse, prior violence, age, sex, education, and diagnosis," he said.
The MacArthur Violence Risk Assessment Study is posted on the Web at http://macarthur.virginia.edu/read_me_file.html. The study’s data can be downloaded at the end of the file at by clicking on "Go to Download Area." ▪