When a major new longitudinal study was greeted earlier this month with headlines like "Mental Illness Doesn't Predict Violent Behaviour," experienced news consumers could have been forgiven for rolling their eyes at yet another attempt to convince them that two plus two don't add up to four. Psychiatric discoveries are invariably presented in the media in the most destigmatizing way possible, and whenever a lunatic enjoys some unusual success in following the instructions inside his head, an army rushes forth to convince us that it is all just coincidence, and that you and I are inherently quite as likely as any paranoid schizophrenic to take up the august professions of celebrity stalker or school shooter.
What I wonder is, if mental illness is not even correlated with violence, let alone a recognizable cause of it, why do our courts formally designate the insane as not responsible for criminal actions? Surely an obvious corollary of "mental illness doesn't predict violent behaviour" is that treating mental illness more effectively will do nothing to make our communities safer, and that the psychiatric profession has no appropriate place in the courtroom. When we "offer" therapy or hospitalization to a mad person who has behaved violently, we might as well be giving them free ice cream, for all the good it will do the rest of us.
It turns out, of course, that the study itself -- which appears in the February edition of Archives of General Psychiatry -- says something very different. The data were gathered from the National Epidemiologic Survey on Alcohol and Related Conditions (NESARC), a strongly representative sample of the American populace containing more than 34,000 members. Under the NESARC, subjects were interviewed in two "waves" about three years apart. The authors of the new paper looked at the characteristics of those who had committed some violent act in the meantime; in the phrasing of their abstract, "severe mental illness did not independently predict future violent behaviour" during the period in question.
It's that word "independently" that demands some exegesis. People who were severely mentally ill were, in fact, more than twice as likely to have earned the "violent" tag between interviews. So from a social standpoint, mental illness is, in fact, an excellent predictor of violence. If your sane neighbour moves out, a demographically identical mentally ill person moves in and all you know about that person is that they are mentally ill, your known risk of having violence next door has, as you would expect, risen significantly.
(The degree of increase in the risk is quantitatively less, mind you, than you would experience from, say, merely having a female neighbour replaced with a male one. But in matters like these, we are often rightly more concerned with a 10% increase to a high background risk than we are with a doubling of a low one.)
What the authors found is that the severely mentally ill do not, in fact, display a statistically significant increase in propensity for violence if they are not substance abusers. Rates of violence exploded when drugs and alcohol were thrown into the mix with a major psychiatric syndrome. The depressed, for example, were only two-thirds as likely to be violent as the rest of the sample, including the apparently well, if they were sober; those who suffered both depression and addiction were 2.7 times more likely to become violent. Where schizophrenia was combined with substance abuse, the rate of violence was elevated by a factor of three. In the case of bipolar disorder it was more like five.
So this is the limited sense in which mental illness is not an "independent" predictor of violence; it does not seem to make violence much more likely on its own. Unfortunately, this is cold comfort, since it merely conceals the fact that mental illness is a relatively strong predictor of becoming a self-medicating drug or alcohol abuser -- which, in turn, is an excellent indicator of violent propensity.
The authors played a similar game with a history of violent behaviour as a distinct analytical factor. They write that "if a person has severe mental illness without substance abuse and history of violence, he or she has the same chances of being violent during the next three years as any other person in the general population."
This was literally true of the sample, whose median age was 43, and which therefore contained a great many people whose impulsive boiling-over-with-testosterone days are over for good. But it was also true that the mentally ill were three times more likely than the general public to have accumulated a history of violent behaviour in the first place. This, indeed, could have been the finding in the headlines quite as easily as the somewhat misleading feel-good message that was propagated email@example.com