A common question posed to mental health researchers is whether people with mental illnesses are more violent than those in the general population. For years, the clarion call from advocacy groups was that the answer to this question was a flat "no". However, recent research is beginning to challenge that rather dogmatic view, and in so doing, has enveloped into a controversy. In particular, a recent study from the landmark National Institute of Health CATIE study suggests that for some people with mental illness the answer is yes. Of course, when examining the complex phenomena of mental illness and violent behavior a lot of caveats are in order.
The first among these is the important disclaimer that while some people with mental illnesses do seem to have elevated risks for violence, most do not. Similarly, many types of mental illness are not associated with increased risks for violence. Advocacy groups are quite right in jealously guarding against the many stereotypes that have engendered people with mental illnesses as discrimination against the mentally ill is an enduring problem.
Notwithstanding, posts such as this and this are indicative of the advocacy culture that has gone too far. But first some background. The CATIE study was the largest ever multi-center, non-industry funded study conducted to assess various aspects of the multiple antipsychotic drugs. The main purpose of this study was to test head to head these drugs to assess their efficacy and cost. The results, which are still being published in numerous journals, have been quite surprising and controversial. The main conclusion being that newer antipsychotic drugs don't appear more effective than the much cheaper older ones. One of the advantages of doing such a large-scale, multi-center trial, is the ability to collect other valuable data, including reports of violence.
In this seminal article, Dr. Swanson and colleagues present data demonstrating that during a 6 month period, the rate of violence among participants in the CATIE study averaged 19.1%. This is a rather high number and demonstrated contrary to prevailing wisdom that people with psychotic disorders, on average, are more prone to violence than the general population. Of course, the "devil is in the details" like so many other things in life. Upon closer examination of the study results, a few notable factors stand out:
First, participants who were symptomatic were more likely to be violent than those whose symptoms were well managed. But it wasn't just any symptoms. In psychiatric research, we talk about positive and negative symptoms of psychosis. Positive symptoms are the ones that get the public's attention: hallucinations (seeing, hearing, or smelling things not really there), delusions (belief in ideas not based on reality), disorganized speech (loose association between ideas, derailment of sentences, incoherence, illogical statements, excessive detail, and rhyming of words), or bizarre behavior. Negative symptoms tend to be more subtle: social withdrawal, apathy (decreased motivation), poverty of speech (brief replies), inability to experience pleasure (anhedonia), limited emotional expression, or defects in attention control. In the CATIE study, positive psychotic symptoms were associated with violence.
Second, substance abuse was significantly associated with risk for violence. In fact, a number of studies have shown substance abuse to account for a large variance of the risk for both violence and incarceration among the mentally ill. In this sense, people with mental illness are like everyone else: using drugs and alcohol greatly increases one's risk of violence and incarceration generally.
Third, the mean time of treatment for CATIE participants who were violent was 16 years. This likely reflects the fact that schizophrenia is a chronic illness that usually follows a downward course (although some have argued against this).
Fourth, prior criminal arrests were predictive of violence during the CATIE study. This is hardly earth shattering news, but taken with the other factors described above, the overall picture is one suggesting that risk for violence among those with psychotic illnesses mirrors that of the general population plus untreated psychotic symptoms.
The CATIE study, as mentioned, was truly groundbreaking. But it wasn't perfect. Like so many studies that report on criminal justice involvement, it relied mainly upon self-report. And there are other weaknesses of the study. However, posts such as this one are large on rhetoric and short on intellectual candor.
While it is true that the CATIE study had a long list of exclusion criteria, the author fails to acknowledge that this is quite the norm for research studies and for a good reason. In examining complex phenomena researchers are always struggling with including as many folks as possible to get a "real world" picture and guarding against contamination of the variables of interest. Simply put, if you put everything and the kitchen sink into your study you can't generalize your results to much of anything. The result is that you have a study that is essentially worthless.
The author next states that baseline data for violence was collected on only 42% of the sample. Why? Because of limited resources, only some of the CATIE participants were given the lengthy assessment for violence. But that's not unusual in large, multi-center trails. As mentioned previously, CATIE's main goal was not the assessment of violence but antipsychotic drug effectiveness. As long as the selection was random (and it was) then there are not statistical issues here to complain about.
Update: Dr. Swanson has provided a correction. Baseline data on violence was 92% for the CATIE sample.
The author criticizes Swanson et al.'s reconstruction of the MacArthur Community Violence Interview variables. But Swanson et al. admits this fact in their paper and give a good reasons for doing so:
Published reports from the MacArthur Violence Risk Assessment study, which developed the MacArthur Community Violence Interview, have labeled these 2 constructs as “violence” and “other aggressive acts." In this article, however, to be consistent with standard definitions of violent crime used in reports by the Bureau of Justice Statistics, we use the common term violence, qualified by level of severity, to refer to these assaultive behaviors.
Thus, instead of "violence" and "other aggressive acts" Swanson et al. used "severe violence" and "minor violence". Contrary to the critics contentions, this really is just semantics as commonsense dictates. Whether an assault leads to physical injury or not it is still violence. Moreover, the purpose of the Swanson et al. paper was to provide an overview of the total violence risk of the CATIE participants. Anyone who has ever been slapped, punched or shoved knows that such behavior is violent albeit not on the order of a felonious crime.
Astoundingly, the author's next point is that because CATIE relied upon self-reports of violence the actual prevalence might, in fact, have been lower. Anyone who has worked in the criminal justice system knows that if anything the number reported probably underestimated the prevalence of violence as Swanson et al. conclude in their paper. Putting aside the obvious fact that folks generally don't want to own up to their violent behaviors, memory impairment is noted problem in schizophrenia and it is very possible that CATIE participants simply didn't recall every instance of violence that had occurred. While Swanson et al. were interested primarily in violence broadly defined and self-reported, it is a shame that NIH decided not to collect arrest data as a comparison (although most researchers suspect that much violence by the mentally ill goes unreported because family members are often the target as the Swanson et al. article notes).
So what can we conclude from all of this? CATIE is just one study, but it is an important one. It was methodologically sound, but imperfect just like all experimental science is. Groups like the Treatment Advocacy Center, albeit very controversial, exist because so many families, professionals, policy-makers, and patients themselves understand that violence among the mentally ill is largely a preventable tragedy. Of course, there are very important liberty questions inherent in how we decide to handle individuals and groups with elevated risk for violence, but we should at least acknowledge that groups like TAC make some thoughtful and powerful arguments as to why the status quo isn't good for society and the mentally ill alike.