Letter

Revisiting the MacArthur Study

To the Editor: I read the debate in the February issue ( 1 ) revisiting the MacArthur Violence Risk Assessment Study with interest. It presented opposing views of the results of the MacArthur study that was conducted between 1992 and 1995 to determine the prevalence of community violence in a cohort of patients discharged from psychiatric hospitals. On the basis of follow-up data from 951 patients, the study essentially found no significant difference in the prevalence of violence between the patients and other people living in the same neighborhoods ( 2 ).

In the February debate, Dr. Torrey and Mr. Stanley presented several points on which they disagree with the findings or the methodology of the MacArthur study. Dr. Monahan, Dr. Steadman, and the MacArthur Study Group then responded to each point. Dr. Torrey and Mr. Stanley argued at one point that the results of the MacArthur study are not applicable to all patients with mental illness. To illustrate their concern, they cited press coverage of the MacArthur study and a reference to the study in my 2007 article on criminal recidivism ( 3 ) as implying that "the results were applicable to all psychiatric patients." As noted by Dr. Monahan and Dr. Steadman in their response, the MacArthur study did not make that claim. Neither did my article.

My article discussed eight central risk factors that are predictive of criminal behavior, including violent crimes—albeit to a lesser extent because such crimes are low-frequency events. These risk factors broadly apply to individuals whether or not they are mentally ill, and they have been established through an extensive body of research ( 4 ). However, this research has received relatively little attention in the psychiatric literature. The risk factors are substance abuse, family problems, lack of healthy recreational pursuits, low levels of performance or satisfaction with school or work, history of antisocial behavior, antisocial personality pattern, antisocial cognition, and antisocial attitudes, including associating with criminal companions. The MacArthur study's negative finding is consistent with the established relationship between known risk factors and criminal behavior. In particular, the study used an appropriate control group of individuals who lived in the same neighborhoods as the study patients and were therefore likely to share many of the same risk factors.

My article also made the point, which was based on a review of recent literature, that active psychosis is an additional risk factor for violent crime. Since publication of the landmark MacArthur study, several studies have been published that provide new evidence that active psychosis is a risk factor independent of substance abuse or other known risk factors. For example, a rigorous national prospective study of 1,410 adults with schizophrenia recently found that "psychotic symptoms were strongly associated with both minor and serious violence" ( 5 ).

Given the predictive strength of identifiable risk factors and the fact that some are modifiable, the important question for the field is no longer whether persons with mental illness are more violent than others. Rather, the question is how to engage high-risk individuals in treatments that effectively target modifiable risk factors for criminal behavior. Thanks to the authors for an interesting debate and to the journal for this opportunity for clarification.