The severely mentally ill make up only a minor share of violent crime perpetrators (and are often victims themselves), so treatment of the severely mentally ill isn't a major driver of violent crime, but they do, overall, have an elevated risk of committing violent crimes living in a world that deals with several mental illness the way that our society does. Also, they do represent a population for whom there is a fairly clear path to reducing violent crime commission risk. Moreover, some of the most sensational and shocking crimes in our society involve instances where severely mentally ill individuals run amok, even though these crimes are very rare and are difficult to predict or prevent.
[S]everely mentally ill people account for only 3 to 5 percent of violent crimes in the general population. . . . we refer only to severe mental illness—meaning schizophrenia, bipolar disorder or psychotic depression. . . .
In a 2009 meta-analysis, or quantitative review, of 204 studies exploring this connection, psychologist Kevin S. Douglas of Simon Fraser University and his associates found a slightly greater likelihood of aggressive behaviors among those with severe mental illnesses. Yet this connection is much weaker than the public seems to believe it is and does not necessarily mean that these serious disorders cause violence. The causation could be in the reverse direction: engaging in chronic aggression (stemming from some other source) may create stress that triggers the illness in those predisposed to it. Alternatively, a third factor could spawn both a psychiatric condition and violence. . . .
[T]he MacArthur Violence Risk Assessment Study in 1998, sociologist Henry J. Steadman of Policy Research Associates and his colleagues reported that almost a third of severely mentally ill patients with substance abuse problems engaged in one or more violent acts in the year after they left the hospital. For discharged patients who did not abuse drugs, the corresponding figure was only 18 percent. . . .
In its meta-analysis, Douglas’s team also flagged drug abuse as one of several factors that contributed to the connection between mental illness and violence. In addition, it found the link was even stronger for patients who suffered from delusions, hallucinations or disorganized thinking. Thus, a mentally ill person is more at risk of committing an act of aggression when that individual is also abusing a drug and shows particular symptoms. . . .
In the MacArthur study, Steadman’s team found no difference in the prevalence of violence between the severely mentally ill who were on their medications and mentally healthy people, whereas unmedicated patients lashed out at significantly higher rates. . . . improving adherence to treatment may lessen the chances that severely ill people will behave violently.
From Scientific American.
I am not quite as optimist as Hal Arkowitz and Scott O. Lilienfeld, who are the authors of this Scientific American article. For instance, individuals who have been hospitalized for mental health conditions (generally, because someone determines that they pose a risk of bodily harm to themselves or others as a result of one or more prior specific incidents), clearly pose a much greater risk to the public than individuals who have received merely outpatient mental health treatment. An 18% risk of engaging in violent acts is on the same order of magnitude as the risk that a felon about two years out of prison without incident will do so and is fantastically elevated relative to someone who has neither a criminal record nor a record of commitment to an institution for a mental health condition. A 33% risk of engaging in violent acts is on the same order of magnitude as the risk that a felon just release from prison will commit a new crime.
The claim that "severely mentally ill people account for only 3 to 5 percent of violent crimes" also fails to reflect that a much larger percentage of people whom the criminal justice system deems it necessary to incarcerate have moderate to several mental health issues. The number of mentally ill people in jail and prison vastly outnumbers the number of mentally ill people who are in civil institutional settings.
This account also fails to dissect the overall problem of mental health issues adequately. There are multiple very different paths linking mental health issues and crime.
One common path links traits like impulsivity, novelty seeking, and weak anger management capacity with "heat of passion" crimes, often under the influence of alcohol and aggravated by the possession of weapons. These traits do not mitigate guilt in the criminal justice system but often are considered in mitigation of a sentence for a crime.
A second common path links the concept of psychopathic absence of conscience, with pre-meditated crimes like first degree murder, serial rapes and fraud. The violent crimes seem to usually involve a plus factor of impulsivity with pscyhopathy. The white collar crimes and immoral but not illegal business practices tend to fit psychopaths who are at the opposite end of the conscientious to impulsive dimension of personality. Psychopathic traits, rather than being mitigating circumstances, are generally aggravating circumstances in the criminal justice system and closely overlap with circumstances justifying the death penalty.
The third common path is the one focused on in the article - the delusional, and also manic or depressed individual who goes amok, a path where illegal substance abuse greatly elevates the risk and taking prescribed meds
Fourth, a significant share of people who commit serious crimes are not mentally ill, per se, but are developmentally disabled, i.e. they have IQs so low that they are considered mentally retarded. In the extreme, a low IQ can affect guilt in the criminal justice system, but more often it mitigates a sentence and heightens scrutiny of the veracity of the individual's confessions to law enforcement officers, their capacity to carry out the crime, and their vulnerability to being framed or participating ineffectively in the criminal justice system.
Fifth, a certain share of people who commit serious crimes do so in response to traumatic circumstances, extreme stress, abuse and manipulation, in a form of extended self-defense, in the form of poor judgment in connection with the mental breakdown of a not congenitally mentally ill individual, or in the form of an inability to resist the pressure from a dominant person in their life to commit a crime. Criminal justice responses to these circumstances are all over the map, sometimes producing leniency from one district attorney or judge or jury, while producing draconian responses from another just a few miles away.