There’s no way to identify dangerous psychiatric patients

If we can’t predict violent acts, then we can’t prevent them.

Powered by automated translation

Aaron Alexis, the former US navy reservist who used a shotgun and two handguns to kill 12 people at the Washington Navy Yard last week, has been described as experiencing serious mental health problems. He reportedly heard voices, and experienced paranoia.

Such cases invariably raise questions about the care of people with severe mental health problems, and the risk they pose to society.

In 2011, a man named Jared Lee Loughner shot and injured Gabrielle Giffords, a US congresswoman, and her supporters, killing six. Similar questions were raised. Loughner's mental competence to stand trial was discussed and his legal team acknowledged that their client had "serious mental afflictions".

New York Times columnist David Brooks wrote that one key question raised in the case was: “Do we need to make involuntary treatment easier for authorities to invoke?”

The reality however is that the trend in psychiatric care over the past 50 years has emphasised getting people out of institutions, to receive care in the community. In 1955 the US had more than 500,000 people in hospitals for mental health problems. Today the number is under 40,000.

However, high-profile violent crimes committed by people experiencing mental health problems have prompted some hard questions about this policy.

Sometimes people with such problems do commit extreme acts of violence. But so do the mentally healthy. A study published in the British Journal of Psychiatry in 2006 suggested that around five per cent of people incarcerated for murder in England and Wales were suffering from severe mental problems; that works out to about 20 homicides per year.

But even this five per cent figure is highly misleading. People with severe mental health problems (psychosis), are also far more likely than others to have problems with alcohol and narcotics. When this is taken into account, any semblance of a relationship between psychosis and violent crime vanishes.

Similarly, a Swedish study spanning three decades looked at the risk for homicide and other violent crimes among 3,743 people with severe mental illness, comparing them with a control group of 37,429 healthy people from the general population. Among the first group, 8 per cent were convicted of violent crime during the study, while only 4 per cent of the healthy controls received comparable convictions.

However, as in the UK study, once researchers controlled for drug and alcohol use, the apparent difference vanished – in other words: drug and alcohol use predicts violent crimes, psychosis doesn’t. In short, you are not more likely to be killed by a person with mental health problems than by a “normal” person.

The real problem, then, is how to predict which psychiatric patients are most likely to commit violent crimes. If we can predict, we can better prevent, by means of detention in a suitable hospital.

However, it has proven terribly difficult to predict which patients are most prone to violence, and almost impossible to identify in advance those capable of extreme violence.

George Szmukler, a professor of psychiatry and society at King’s College, London, has explored sophisticated predictive models in this field. His conclusion is that in a population of 100 patients, where 20 are prone to any degree of violence, the best the predictive models can do is to correctly identify 14 (70 per cent) of the 20.

Unfortunately, even this modest performance is achieved at the cost of misidentifying 24 (30 per cent) of the non-aggressive patients as aggressive. And when we raise the bar and use such models to try to detect patients likely to commit extreme acts of violence, the best we can do is get it right three times out of every 100; in other words we would be wrong 97 times out of 100.

How many patients is it acceptable to detain unnecessarily, to prevent one violent incident?

If we can’t predict violent acts, then we can’t prevent them, at least not without trampling all over the civil liberties and human rights of psychiatric patients, vulnerable individuals who are all too often already maltreated.

Lowering the threshold for detention and increasing the size of inpatient psychiatric populations would be a regressive move, and highly unlikely to make the world safer.

Justin Thomas is an associate psychology professor at Zayed University in Abu Dhabi