Skip to content Skip to navigation

Mass shootings, criminal violence: Can't be predicted, but can be mitigated

May 1, 2013
by Dennis Grantham, Editor-in-Chief
| Reprints
Osher: Additional screening of at-risk individuals, followed by treatment, may prevent future violence

Following a year in which mass violence, notably the Sandy Hook shootings, riveted the attention of lawmakers and the public on the possible linkage between mental illness and criminal violence, longtime community and criminal-justice psychiatrist Fred Osher, who heads the behavioral health division of the Council of State Governments’ Justice Center (Bethesda, Md.), asked a capacity crowd at the 2013 National Council Conference a question:  Can we predict or prevent violent behavior?

Osher began his discussion by exploring what he called the “complex relationship between mental illness and violence,” a relationship that challenges mental health advocates to walk a fine line between two powerful and conflicting perceptions.  On one hand, he said, “We must fight against media interpretations that are distorted and stigmatizing toward those who are mentally ill. But we must also own up to the association between some types of mental health and substance use disorders. Denying that such disorders play any role in violent behavior is disingenuous and ultimately counterproductive.“

So, he asked, “What can we say about the issue?”

On the positive side, he noted, “We can say that most people with mental illness aren’t violent and most violence in our country is not caused by people with mental illness.  In fact, people with mental illnesses are far more likely to be the victims of violence, often within their own homes and families.” In some cases, he said, this victimization can contribute to future violent behavior.

“Mentally ill people with no history of substance abuse or violence are statistically no more likely than anyone else to behave violently,” he added, explaining that demographic factors, such as age, gender, and socioeconomic status “are far better predictors of the likelihood of violence than is the presence or absence of a mental illness.“

On the other hand, Osher said, “Mental illness may be a consistent but modest risk factor in the likelihood of violence. “ While he said “there is no clear causal link between mental illness and violence, symptoms associated with some mental illnesses “increase the statistical risk of violence” for some individuals.  He explained that those whose mental illnesses result in psychosis, command hallucinations (hearing voices that “command” one to commit harm), paranoia, or paranoid delusions may face — in the absence of effective treatment — a higher risk of future violent behavior.

But there’s another even more important risk factor. “We must acknowledge that substance abuse, in people with or without mental illnesses, is a large driver of violent behavior in our communities. And, having said that, it’s important that we distinguish violence from criminal behavior.” He observed that violent crime is defined by statute as murder, robbery, rape, and assault and that just 3-5 percent of violent crimes are committed by individuals with a mental illness. 

It is only in arrests for non-violent crimes — possession, panhandling and other sometimes petty crimes — that the mentally ill are disproportionately represented. Osher reported that of 13 million annual arrests in the US, about one in six, some 2 million, involve individuals with a mental illness. Among men placed under arrest, he said that 15% have a mental illness, while the number of mentally ill among women arrested is over 30%.

Predicting violent behavior — Impossible

Osher introduced the next segment of the discussion with a statement:  “We do not have the capability to predict violent actions.”  Violence of the type witnessed at Sandy Hook, in Tucson, or Virginia Tech represent “extremely rare events.” And, he continued, “It is not so easy to predict the occurrence of rare and low probability events.” Staistically speaking, “it’s a base rate problem — these things just don’t happen that often.”  

Predictive mechanisms are extremely difficult to develop, he added, because each represents “a tradeoff between sensitivity and specificity.” For example, he said that any approach that sought to identify all those who might become violent would lead professionals to “overpredict” and “accept a whole lot of false positives.” On the other hand, an approach that sought to be highly specific in terms of defining criteria that predict violence would likely mean that “you’ll miss a lot of potentially violent individuals down the road.” In any event, he debunked the whole notion of prediction.

Rather than risk the high inaccuracy of trying to predict specific instances of individual violence, Osher instead suggested that “professionals can assess the risk for violence” by screening individuals and placing them into what he called “bands of risk.” He called on the field to develop greater expertise in using risk assessment tools that combine three factors:

·         static factors — age, gender, family, criminal history

·         dynamic (changeable) factors —substance abuse, antisocial thoughts or beliefs, relationships, trust or fear issues

·         protective (positive) factors —potentially stabilizing or supportive factors including family and family status, employment status, etc.    

Pages

Topics