Mentally ill individuals who go through a specialized court system instead of the criminal justice system are less likely to be arrested again, new research hints.

The study suggests that these so-called mental health courts can achieve the public safety outcomes that funders and the public want, Dr. Henry J. Steadman of Policy Research Associates in Delmar, New York, and his colleagues write. But the jury's still out, they say, on who will benefit the most from going through this system, and which approaches will produce the best results.

There are currently about 250 mental health courts in the US, up from just a couple in the late 1990s, Steadman and his team note in the Archives of General Psychiatry. These courts are intended to direct people with serious mental illness who commit crimes into treatment, rather than sending them to jail, where they may wind up in a cycle of release and re-incarceration while getting little or no treatment for their psychiatric issues.

According to a 2009 report from the Council of State Governments Justice Center, about 17 percent of people who enter the criminal justice system have serious mental illness.

A 2007 study by the RAND Corporation, the first to look at the cost-effectiveness of mental health courts, found that a Pennsylvania mental health court saved taxpayers $3.5 million over a two-year period.

But because implementation of mental health courts widely varies by jurisdiction, by judge, and across time, Steadman and his team point out, single-court studies of their effectiveness offer only a limited perspective.

To address this issue, the researchers looked at four different mental health courts -- two in California, one in Minnesota, and one in Indiana. They compared 447 people assigned to the mental health court system to 600 people who were also mentally ill, but went through the regular criminal justice system.

In the 18 months before entering the criminal justice or mental health court system, 93 percent of the individuals in the mental health court group and 95 percent of the control group were arrested at least one additional time, the researchers found. But in the 18 months following their entry into the system, 49 percent of participants in the mental health group were re-arrested, compared to 58 percent of the control group.

The number of days people spent incarcerated in the 18 months before entering the mental health court system was 73, compared to 82 days for the following 18 months. But for those who entered the regular criminal justice system, the number of days incarcerated increased much more sharply, from 74 days to 152 days.

People with schizophrenia or depression who went through the mental health court system were significantly more likely to be arrested again than people with bipolar disorder or people who used illegal drugs in the previous 30 days.

Our data do not comprehensively address the key questions of who the courts are most effective for or what mechanisms produce positive outcomes, the researchers caution in their report. These important questions await further data from this and other studies.