Former APA President Calls for New Paradigm for Studying Violence, Mental Illness

In the aftermath of the most recent mass gun violence—in Dayton, Ohio; El Paso, Texas; and Odessa, Texas—the Trump administration has focused on the role of mental illness, reflecting a common public perception that serious mental illness is causally related to violence.

David Hathcox

Yet much research on the subject highlights the difficulty or impossibility of deriving predictive models of violence based on correlations with mental illness. In a recent editorial in the American Journal of Psychiatry, past APA President Paul Appelbaum, M.D., suggests that the predominant method of investigating the correlation—secondhand analyses of existing datasets—may have reached a dead end and that an entirely new paradigm is required.

“[T]he opportunistic nature of most research on violence and schizophrenia (and other mental disorders) has inhibited progress in identifying the roots of violent behavior and potential preventive measures,” Appelbaum wrote in the editorial “In Search of a New Paradigm for Research on Violence and Schizophrenia.” He is the Elizabeth K. Dollard Professor of Psychiatry, Medicine, and Law and director of the Center for Law, Ethics, and Psychiatry in the Columbia University Department of Psychiatry.

(The editorial appeared as a response to a study by Alec Buchanan, M.D., Ph.D., of Yale University and colleagues titled “Correlates of Future Violence Among People Being Treated for Schizophrenia.” That study found, again, a low rate of violence among people with schizophrenia; among patients who did commit injurious violence, previous violence and recent victimization—possibly occurring simultaneously in the context of violent altercations—were among the strongest predictors.)

In comments to Psychiatric News, Appelbaum said that notwithstanding frequent handwringing by political leaders over horrific acts of violence, there has been little interest, especially at the federal level, in funding studies dedicated to improved prediction and management of violence—whether among people with mental illness or any other group. “For this reason, researchers have turned to secondary analyses of data collected for other purposes, which often means idiosyncratic predictor variables, inadequate outcome measures, and oddly selected samples,” he said.

He said existing studies differ in sampling strategies, diagnostic inclusion criteria, research design, independent variables, methods for ascertaining a history of violence, and outcome measures. “Under these circumstances, it can hardly be surprising that—with the exception of variables of large effect such as previous violent history—studies routinely identify different predictors of violence, typically with small effects on overall risk,” he wrote.

What’s needed, he said, are prospective studies with violence as the primary outcome, with large enough study samples to account for the many variables that are likely to make some (usually modest) contribution to violence risk. The most widely cited such study—the MacArthur Violence Risk Assessment Study in the late 1990s, funded primarily by the MacArthur Foundation with supplemental support from the National Institute of Mental Health (NIMH)—found that people with mental illness but without substance use disorder are at no greater risk of being violent than the general population.

“It’s no accident that the largest primary, prospective study of violence by people with mental illness was funded principally by a private foundation, albeit with supplemental support from NIMH,” Appelbaum told Psychiatric News.

He would like to see a consortium of stakeholders—something similar to the Psychiatric Genomics Consortium that has helped to advance research on the genetics of psychiatric illness—that can formulate common standards for conducting studies on violence and mental illness and pool high-quality research. “My colleagues and I [at New York State Psychiatric Institute and Columbia University] are exploring the possibility of pulling together a group of the major researchers on violence and mental illness to see if agreement can be achieved on a set of common measures and methods that would allow aggregation of datasets across studies and enhance our abilities to identify people at very high risk of violence,” Appelbaum said.

As for predicting violence on the individual level on the basis of a history of mental illness, Appelbaum said that a large dose of reality may be in order—particularly when it comes to the kind of gun violence that captures public attention.

“Focusing on people with mental illness to solve the country’s gun violence problem is a misguided strategy for several reasons,” he said. “The amount of violence in the U.S. that’s attributable to mental illness is very small, with the best estimate around 4%—and most of that does not involve guns. Second, predicting who is going to commit a violent act is a hopeless task, given that most people with mental illnesses will never perpetrate violence, the predictors of violence are common and only weakly related to the outcome, and violent acts are rare—a combination that inevitably leads to inaccuracy, especially overprediction.

“Finally, and perhaps most importantly, focusing on mental illness ignores the most important factor that contributes to high rates of gun violence in the U.S.—the astounding number of guns in our country, especially weapons designed primarily for the purpose of killing large numbers of people.”

Yet quite apart from gun violence prevention, Appelbaum said that a better understanding of violence among people with mental illness—and a new research paradigm to elucidate correlations between the two—would still be valuable when it comes to informing decisions about involuntary commitment and the protection of third parties.

But it’s time for a new strategy, he said. “If the field keeps on its current path, with each study using different predictor and outcome measures, recruiting samples drawn from different populations, and analyzing the data differently, progress is unlikely,” Appelbaum said. ■