U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Forum on Global Violence Prevention; Board on Global Health; Institute of Medicine; National Research Council. Contagion of Violence: Workshop Summary. Washington (DC): National Academies Press (US); 2013 Feb 6.

Cover of Contagion of Violence

Contagion of Violence: Workshop Summary.

Show details


Madelyn S. Gould, Ph.D., M.P.H.

Columbia University and New York State Psychiatric Institute


Alison M. Lake, M.A.

New York State Psychiatric Institute


Evidence has accumulated to support the idea that suicidal behavior is “contagious” in that it can be transmitted, directly or indirectly, from one person to another (Gould, 1990). This evidence is derived from three bodies of research: studies of the impact of media reporting on suicide, studies of suicide clusters, and studies of the impact on adolescents of exposure to a suicidal peer. In each case, suicide contagion can be viewed within the larger context of behavioral contagion or social learning theory. While research has also addressed the distinct but related topic of the contagion of nonsuicidal self-injurious behavior (Jacobson and Gould, 2009; Hawton et al., 2010; Whitlock, 2010), the current review focuses specifically on attempted and completed suicide.

Impact of Media Reporting on Suicide

Research into the impact of media stories about suicide has demonstrated an increase in suicide rates after both nonfictional and fictional stories about suicide. Most research in this area has addressed nonfictional reporting, which has been shown to have a more powerful effect (Stack, 2003). More than 50 studies on nonfictional stories reported in newspapers, on television, and more recently on the Internet, have yielded consistent findings. Suicide rates go up following an increase in the frequency of stories about suicide (e.g., Hagihara et al., 2007). Moreover, suicide rates go down following a decrease in the frequency of stories about suicide (e.g., Motto, 1970). A dose-response relationship between the quantity of reporting on completed suicide and subsequent suicide rates has consistently been demonstrated (e.g., Phillips, 1974; Phillips and Carstensen, 1986; Pirkis et al., 2006). Changes in suicide rates following media reports are more pronounced in regions where a higher proportion of the population is exposed (Etzersdorfer et al., 2004). The prevalence of Internet users, with access to Internet stories about suicide, has been associated with general population suicide rates in males, but not females (Hagihara et al., 2007; Shah, 2010).

The way suicide is reported is a significant factor in media-related suicide contagion, with more dramatic headlines and more prominently placed (i.e., front page) stories associated with greater increases in subsequent suicide rates (Phillips, 1974, 1979; Kuess and Hatzinger, 1986; Michel et al., 1995). Repetitive reporting on the same suicide and definitive labeling of the death as a suicide have also been associated with greater increases in subsequent suicide rates (Niederkrotenthaler et al., 2009, 2010). Content analyses of suicide newspaper reports from six countries with different suicide rates (Austria, Finland, Germany, Hungary, Japan, and the United States) found that attitudes toward suicide in newspaper reports varied by country, and that national suicide rates were higher in countries where media attitudes toward suicide were more accepting (Hungary) and suicide completers were more positively portrayed (Japan) (Fekete et al., 2001). Conversely, national suicide rates were lower in countries (Finland, Germany, and the United States) where reporting tended to portray the suicide victim and act of suicide in terms of psychopathology and abnormality, and to describe the negative consequences of the suicide. Moreover, media stories about individuals with suicidal ideation who used adaptive coping strategies to handle adverse events and did not attempt suicide have been negatively associated with subsequent suicide rates (Niederkrotenthaler et al., 2010).

The impact of media reporting on subsequent suicides is not monolithic, but interacts with characteristics of the reported suicide and characteristics of the media audience, as well as with characteristics of the media portrayal, as noted above. For example, celebrity suicides are more likely and the suicides of criminals are less likely to be followed by increased suicide rates (Stack, 2003; Niederkrotenthaler et al., 2009); individuals with a recent history of suicide attempt and/or a concurrent severe depression are more likely to attempt suicide in the wake of a media report (Cheng et al., 2007a,b).

Ecological studies of the impact of media on suicide rates, like those described above, meet four of Hill's five criteria for demonstrating causality (namely, consistency, temporality, strength of association, and coherence), but provide less convincing evidence of specificity (Hill, 1965; Gould, 1990; Insel and Gould, 2008). A handful of extant individual-level studies, however, have examined whether individuals who attempted suicide following a media story were exposed to and influenced by the media report, and have contributed evidence to support the specificity of the media effect. Hawton and colleagues (1999) conducted a study in emergency departments in the United Kingdom, examining the pattern of suicide attempts before and after a fictional Royal Air Force pilot took an overdose of paracetamol (i.e., acetaminophen) in an episode of a popular weekly TV drama. Presentations for self-poisoning increased by 17 percent in the week after the broadcast and 9 percent in the second week. Increases in overdoses using the specific drug used by the model were more marked than increases in other types of overdoses. The most compelling evidence of modeling from this study was that use of the specific drug for overdose among overdose patients who were viewers of the drama doubled after the episode in question, compared with overdose patients who were viewers of the drama prior to that episode. Twenty percent of the interviewed patients reported that the model had influenced their behavior. In a more recent study, 63 individuals who attempted suicide in Taipei, Taiwan, following the suicide of a young female pop singer were assessed for exposure to media reporting about her death. Forty-three (68 percent) respondents had been exposed to the media reporting, of whom 37 percent reported that the media stories influenced their suicide attempts (Chen et al., 2010). This study also demonstrated a positive modeling effect on the chosen method of suicide (burning charcoal inside a closed car), with an adjusted odds ratio of 7:3 (for additional evidence of a modeling effect based on choice of suicide method, see also Etzersdorfer et al., 2004; Cheng et al., 2007b; Chen et al., 2012).

Suicide Clusters

A suicide cluster is an excessive number of suicides occurring in close temporal and/or geographical proximity (Gould et al., 1989). Clusters occur primarily among teenagers and young adults, with between 1 percent and 5 percent of teen suicides occurring in clusters (Gould, 1990; Gould et al., 1990; Hazell, 1993). A case-control study of two teen suicide clusters in Texas indicated that the clusters included teens who had close personal relationships with others in the cluster, as well as teens from the same community who were not directly acquainted with one another (Davidson et al., 1989). When compared with matched living controls, suicide completers were more likely to have preexisting vulnerabilities (e.g., emotional illness, substance abuse problems, frequent changes of residence, recent or anticipated relationship break-up) that may have increased their susceptibility to suicide contagion.

It has been suggested that teen suicide clusters may result from the combination of assortative relating, the tendency for similar individuals (in this case, teens at high risk of suicide) to preferentially associate with one another, with shared life stress (Joiner, 2003). According to this argument, which should apply only to those teens within a suicide cluster who were directly acquainted with one another, teen suicides may cluster within a peer group because of high levels of preexisting vulnerability across the peer group, not because of suicide contagion. A recent study used agent-based computer simulation modeling to test this hypothesis and to explore the possible mechanisms behind suicide clustering (Mesoudi, 2009). As programmed in the simulation model, social learning was sufficient to generate suicide clusters localized both in time and space. The simulation model further found that assortative relating, also known as homophily, was likely to generate spatially localized suicide clusters among high-risk peer groups, but less likely to generate spatiotemporal suicide clusters and unlikely to generate purely temporal clustering of suicides. As the study's author notes, homophily seems to provide no reason why suicides should be clustered in time. Finally, the model confirmed that media effects, in combination with the effects of prestige and similarity biases, were capable of generating suicide clusters localized in time, but not space.

Even within spatiotemporal suicide clusters, where decedents are more likely to have direct contact with one another, media reporting on suicide can play a role. A recent analysis of the Foxconn suicides in China found support for a temporal clustering effect (Cheng et al., 2011). National (but not local) newspaper reporting on the suicides and the occurrence of a Foxconn suicide or suicide attempt were each associated with elevated chances of a subsequent suicide 3 days later, demonstrating the impact of both media-related contagion and direct contagion within the Foxconn company.

Impact on Adolescents of Exposure to a Suicidal Peer

Of 16 studies reviewed by Insel and Gould (2008) on the impact on adolescents of exposure to a suicidal peer, the majority found a significant association between exposure to the suicidal behavior of an adolescent peer and a subsequent adolescent suicide attempt. Odds ratios ranged from 2.8 to 11.0 for attempted suicide. Analysis of data on a nationally representative sample of U.S. high school students from the National Longitudinal Study of Adolescent Health (ADD Health) found that “teens who know friends or family members who have attempted suicide are about three times more likely to attempt suicide than are teens who do not know someone who attempted suicide” (Cutler et al., 2001). Girls were more likely to attempt suicide if they knew someone who had survived a suicide attempt, while boys were more likely to attempt suicide if they knew someone who had died by suicide. Teens who had not made a suicide attempt in wave one of the study were more likely to have attempted suicide in wave two if they knew someone who had attempted suicide in the interim; this temporal sequencing lends support for the role of contagion alongside the possible effect of assortative relationships among high-risk teens. In the context of exposure to the suicidal behavior of an intimate, contagion may operate via the impact on a vulnerable teen of stress or grief at the loss of a loved one, as well as via social learning about suicide.

Strategies to Prevent Suicide Contagion

A number of evidence-based interventions capable of combating suicide contagion have been developed. Studies have shown that it is possible to intervene to mitigate media-driven suicide contagion by implementing media guidelines for suicide reporting (Gould, 2001; Pirkis and Nordentoft, 2011). Media guidelines can interrupt the transmission of suicidality by identifying the types of media reporting through which suicidality is likely to be transmitted, and by modifying the volume and content of media reporting, with resultant decreases in suicide rates. For example, suicides in the Vienna subway system decreased by approximately 75 percent in 1987 following implementation of media guidelines for reporting on subway system suicides (Etzersdorfer et al., 1992). Applying media guidelines to new electronic media, including social networking websites, presents a new challenge to the suicide prevention community (Pirkis and Nordentoft, 2011; Robertson et al., 2012).

Screening for suicide risk can also interrupt the transmission of suicidality by identifying in advance individuals who may be susceptible to suicide contagion (Gould et al., 2009). In addition, suicide screening works to alleviate that susceptibility by enabling services to be directed to at-risk individuals identified by the screen. Key settings for suicide screening include schools and primary care practices. A range of school- and community-based psychosocial programs may also work to alleviate susceptibility to suicide contagion by, for example, changing adolescent peer norms through positive messaging (Wyman et al., 2010), or educating and empowering parents to communicate with teens (Toumbourou and Gregg, 2002). Finally, research suggests that coordinated postvention/crisis intervention efforts following a death by suicide may minimize and contain the effects of suicide contagion (Poijula et al., 2001; Hacker et al., 2008).


While the complex etiology of suicidal behavior is recognized (Gould et al., 2003), it has become increasingly apparent that suicide contagion exists and contributes to suicide risk along with psychopathology, biological vulnerability, family characteristics, and stressful life events. Strategies to prevent suicide contagion are essential and require ongoing evaluation.

Copyright 2013 by the National Academy of Sciences. All rights reserved.
Bookshelf ID: NBK207262


  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (2.2M)

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...