Elsevier

Psychiatry Research

Volume 136, Issues 2–3, 15 September 2005, Pages 153-162
Psychiatry Research

Findings from the National Comorbidity Survey on the frequency of violent behavior in individuals with psychiatric disorders

https://doi.org/10.1016/j.psychres.2005.06.005Get rights and content

Abstract

Previous studies using probability samples have found a noticeable, but small association between violence and psychiatric disorder. In this article, we analyze data from the National Comorbidity Survey (NCS) to further examine this question. Psychiatric diagnosis of survey responses was based on a modified version of the Composite International Diagnostic Interview. The NCS study also included items that permitted self-report of violent behaviors in the past year. People with 12-month diagnoses of anxiety disorders, dysthymia and major depression were three to four times more likely to admit violent behaviors than those with no disorders. People with bipolar disorder or drug and alcohol abuse were eight times more likely to report violent behaviors. People with co-occurring non-substance and substance abuse disorders were more likely to report violence than those with only non-abuse disorders. Adjusting violence rates by population base rates shows demographics including ethnicity and gender to be a better predictor of violent behavior than psychiatric diagnosis. The NCS findings approximate those in other probability studies and echo the conclusions of the 1996 Consensus Statement by Advocates and Researchers on violence and mental illness; namely, mental illness is only a weak predictor of violent behavior.

Introduction

Regardless of where people stand on the issue of psychiatric disorder and violence, most agree that it is a major source of prejudice and discrimination against people with mental illness. Results of a nationwide probability survey showed that as much as 75% of the public view people with mental illness as violent (Link et al., 1999, Pescosolido et al., 1999). A separate analysis of these data found that twice as many Americans in 1996 viewed people with mental illness as violent than did so 40 years ago (Phelan et al., 2000). Why do so many members of the general public think mental illness is strongly linked to a potential for violence, and why is this perception on the rise? Some view rising public concern in this area as an accurate reflection of increased violence by people with mental illness that has resulted from diminished institutional care (Torrey, 1994). Others argue that media portrayals of people with mental illness as violent, especially in the entertainment industry, contribute to the widespread misperception that mental illness and violence go hand in hand (Wahl, 1995). Several studies with varying methodological rigor have addressed this question. The National Comorbidity Survey (NCS) (Kessler, 1994) provides a rigorously collected set of data from a probability sample that provides additional perspective on this issue.

Members of the Treatment Advocacy Center (TAC) have cited statistics that suggest mental illness is strongly associated with a potential for violence. The TAC, for example, estimates that persons with serious mental illness commit approximately 1000 homicides per year in the United States (Treatment Advocacy Center, 2002). The TAC findings are echoed by other research. A 1979 review of research comparing offenses of people with mental illness to other community residents found ratios varied from as low as 1.16 to as high as 15 (Rabkin, 1979). Rabkin's summary was mirrored by subsequent studies in which people with mental illness were consistently more likely to be arrested for violent crimes compared with non-psychiatric samples (Harry and Steadman, 1988, Holcomb and Ahr, 1988, McFarland et al., 1989, Shore et al., 1990). Finally, a body of practice-based evidence seems to provide empirical support for the perception of a strong link between mental illness and violence. For example, reviews of records of admissions to inpatient units have found that between 18% and 41% of patients had been violent (e.g., struck others) before admission (Johnstone et al., 1986, Humphreys et al., 1992).

Several research groups have criticized the studies on which these conclusions are based (Link et al., 1992, Link et al., 1999, Wahl, 1995). In summarizing the state of research on mental illness and violent behavior, Monahan (2002) said answers to this issue can only be advanced by studying probability samples that were unselected for treatment status in the community at large. Several studies have been completed using probability samples; that is, groups of people with mental illness that were randomly selected from geographic areas in stratified formats so that key demographics relevant to crime and violence are appropriately distributed. Comparison groups were also drawn from the general population and are representative of all key variables. This line of research suggests that people with mental illness are more likely to commit violent crimes than are comparable samples of people without mental illness in the general population. Research completed in the United States and Britain show a two- to six-fold increase in the rate of violence in samples of people with mental illness compared with samples of people without mental illness drawn from the general population (Link et al., 1992, Stueve and Link, 1997, Swanson et al., 1997, Steadman et al., 1998). The prototypic study of this kind was the analysis conducted on the data set from the Epidemiological Catchment Area (ECA) survey (Robins and Regier, 1991). Conducted between 1980 and 1983, the ECA survey was the largest study of psychiatric disorders conducted in the United States to that date. ECA analyses completed by Swanson et al. (1990) showed a four-fold increase in violent behavior for some mental illnesses (e.g., schizophrenia) compared with no disorder, and a 10-fold increase for substance abuse disorders.

We seek to extend and advance ECA findings using the NCS data set. The NCS was a congressionally mandated survey designed to study the comorbidity of substance use and non-substance use disorders in the United States. It was designed to take the next step beyond the ECA study. Two main advances are noteworthy. First, NCS diagnoses are based on DSM-II-R (American Psychiatric Association, 1987) rather than DSM-III (American Psychiatric Association, 1980) criteria. Questions are also included in the interview that allow some comparisons with DSM-IV (American Psychiatric Association, 1994) and with the International Classification of Diseases (ICD-10) Diagnostic Criteria for Research (World Health Organization, 1991). Second, the NCS yielded a national probability sample and was the first nationally representative data set that can be used in the current debate about health care policy in the United States.

Using the NCS data set, we have three goals in this article: (1) We seek to contrast findings from the NCS sample with those reported by Swanson et al. (1990) on the ECA sample in terms of violent behavior and diagnosis. Central to the ECA article was the finding that frequency of violent behaviors increased significantly in cases of dual diagnosis, especially where one of the co-occurring disorders was substance abuse. (2) Previous studies of violent behavior and diagnosis, including the ECA study, have limited diagnosis to the past year. A second goal of this study is to contrast rates of violent behavior in people who meet diagnostic criteria during the past year versus those who meet lifetime criteria but not past year criteria. This analysis helps to address the question whether recent symptoms and associated disabilities are important for the correlation with violent behavior or if some other characteristic of the diagnosis is associated. Corrigan and Cooper (2005) attempted to make sense of the effect size of the ECA findings on violent behavior and mental illness diagnosis by juxtaposing frequency of violent behaviors by diagnostic groups and by demographic subgroups of the general sample against base rates of those groups in the population. Results showed that demographic variables including gender and age are significantly stronger correlates of violence than psychiatric diagnosis. This kind of base-rate analysis will be repeated in our NCS data analysis.

Section snippets

Methods

All aspects of the NCS have been discussed in detail elsewhere (Kessler, 1994, Kessler, 1995, Kessler et al., 1994). Briefly, the NCS is based on a stratified, multistage area probability sample of persons aged 15–54 years in the non-institutionalized civilian population in the 48 coterminous states. The survey was administered by the Survey Research Center at the University of Michigan between September 14, 1990, and February 6, 1992. The response rate was 82.6%. A total of 8098 respondents

One-year prevalence of violent behavior

Of the survey participants responding to item 1, 158 (2.7%) people reported trouble with the police in the past year. For item 2, 88 (1.5%) reported one physical fight that led to injury in the past year, 46 (0.8%) had such fights two or three times in the past year, nine (0.1%) reported four to five incidents, and 10 (0.2%) reported more than five. This yields a total of 157 (2.6%) people reporting fights in the past year. Interestingly, combining the two items together to yield a single index

Discussion

This article used the NCS data to address the question about violent behavior in people diagnosed with single and multiple psychiatric disorders. In many ways, findings from this study mirrored the results found in Swanson et al. (1990) on the ECA data. People with 12-month single diagnoses in the anxiety or depressive spectrum will have a three to almost four times greater likelihood of reporting violent behaviors than those with no 12-month disorder. People with bipolar disorder will have a

Acknowledgments

This research was supported by a research infrastructure support program grant from the National Institute of Mental Health (5 R24 MH62198) to the first author. We are grateful to John Monahan and Jeffrey Swanson for helpful comments on an earlier version of this manuscript.

References (33)

  • American Psychiatric Association

    DSM-III: Diagnostic and Statistical Manual of Mental Disorders

    (1980)
  • American Psychiatric Association

    DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders

    (1987)
  • American Psychiatric Association

    DSM-IV: Diagnostic and Statistical Manual of Mental Disorders

    (1994)
  • P.W. Corrigan et al.

    Mental illness and dangerousness: fact or misconception, and implications for stigma

  • B. Harry et al.

    Arrest rates of patients treated at a community mental health center

    Hospital and Community Psychiatry

    (1988)
  • W.R. Holcomb et al.

    Arrest rates among young adult psychiatric patients treated in inpatient and outpatient settings

    Hospital and Community Psychiatry

    (1988)
  • M.S. Humphreys et al.

    Dangerous behavior preceding first admissions for schizophrenia

    British Journal of Psychiatry

    (1992)
  • A. Jablensky

    Schizophrenia: the epidemiological horizon

  • E.C. Johnstone et al.

    The Northwick Park study of first episodes of schizophrenia: I. Presentation of the illness and problems relating to admission

    British Journal of Psychiatry

    (1986)
  • R.C. Kessler

    The National Comorbidity Survey of the United States

    International Review of Psychiatry

    (1994)
  • R.C. Kessler

    The National Comorbidity Survey: preliminary results and future directions

    International Journal of Methods in Psychiatric Research

    (1995)
  • R.C. Kessler et al.

    Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States: results from the National Comorbidity Study

    Archives of General Psychiatry

    (1994)
  • B.G. Link et al.

    The dangerous and illegal behavior of mental patients reconsidered

    American Sociological Review

    (1992)
  • B.G. Link et al.

    Public conceptions of mental illness: labels, causes, dangerousness, and social distance

    American Journal of Public Health

    (1999)
  • B.H. McFarland et al.

    Chronic mental illness and the criminal justice system

    Hospital and Community Psychiatry

    (1989)
  • P.O. Monahan

    The Mantel-Haenszel Procedure for DIF: Alternative Matching Scores to Control Type I Error and Improve Distributional Properties

    (2002)
  • Cited by (0)

    View full text