The association between Twitter content and suicide

Does suicide contagion (Werther effect) take place in response to social media? A systematic review

INTRODUCTION

The Werther, Copycat or contagion effect of suicidal behaviour is a complex phenomenon that can arise due to exposure to media stories in which identifiable people take their lives. On the contrary, the Papageno effect prevents people from suicide by promoting positives examples of suicidal crisis management. Impact of both effects has been widely studied in different types of situations, but its existence in social media is a source of much debate.

METHODS

A systematic search following the PRISMA guidelines of PubMed, Scopus, Embase, PsycInfo, Web of Science and the references of prior reviews yielded 25 eligible studies.

RESULTS

Most of the studies found were observational, with very different methodologies and generally with low risk of bias. In these, the results suggest the existence of the Werther effect in response to social media stories about suicide. This is mediated by multiple factors, including the characteristic of the users, the type of interaction and the content of the publications. At the same time, the Papageno effect is also described. Evidence found by type of social media and future implications are discussed.

CONCLUSION

Suicidal content on social media can be both contagious and protective. It is confirmed that the Werther and Papageno effects may occur in response to social media, so they could be an interesting target for preventive interventions.

The impact of media reporting of suicides on subsequent suicides in Asia: A systematic review

Introduction

This systematic review is aimed at (1) evaluating the association between media portrayals of suicides and subsequent copycat suicides or attempts among the general public in Asia, (2) understanding the factors associated with copycat suicides and (3) determining the positive impacts of the media reporting of suicides (e.g. increased help-seeking, coping).

Method

A systematic review and narrative synthesis of English and Chinese articles from 8 electronic databases (i.e. PsycINFO, MEDLINE, Embase, CINAHL, Web of Science, Ariti, China National Knowledge Infrastructure and OpenGrey) from January 2000 to May 2023 was conducted. Observational studies were included, and the data were analysed through narrative synthesis. The protocol was registered with PROSPERO (CRD42021281535).

Results

Among the 32 studies included (n=29 for evidence synthesis) in the review, there is good-quality evidence to show that copycat suicides and suicide attempts increase after media reports of a suicide, regardless of country, celebrity status, study design, type of media, mode of suicide or follow-up period. Females, younger age groups and those sharing similar characteristics as the deceased in publicised suicides (age, gender) were more susceptible to negative impact. Reporting of the mode of death of the deceased increased suicides by the same method among the public.

Conclusion

Media portrayals of suicide appear to have a negative impact on copycat suicides at the population level in Asia. Thus, in addition to tighter media control, healthcare systems, professional medical bodies and community outreach services should work collaboratively to promote early help-seeking in those with psychological distress.

Association of 7 million+ tweets featuring suicide-related content with daily calls to the Suicide Prevention Lifeline and with suicides, United States, 2016-2018

OBJECTIVE

The aim of this study was to assess associations of various content areas of Twitter posts with help-seeking from the US National Suicide Prevention Lifeline (Lifeline) and with suicides.

METHODS

We retrieved 7,150,610 suicide-related tweets geolocated to the United States and posted between 1 January 2016 and 31 December 2018. Using a specially devised machine-learning approach, we categorized posts into content about prevention, suicide awareness, personal suicidal ideation without coping, personal coping and recovery, suicide cases and other. We then applied seasonal autoregressive integrated moving average analyses to assess associations of tweet categories with daily calls to the US National Suicide Prevention Lifeline (Lifeline) and suicides on the same day. We hypothesized that coping-related and prevention-related tweets are associated with greater help-seeking and potentially fewer suicides.

RESULTS

The percentage of posts per category was 15.4% (standard deviation: 7.6%) for awareness, 13.8% (standard deviation: 9.4%) for prevention, 12.3% (standard deviation: 9.1%) for suicide cases, 2.4% (standard deviation: 2.1%) for suicidal ideation without coping and 0.8% (standard deviation: 1.7%) for coping posts. Tweets about prevention were positively associated with Lifeline calls (B = 1.94, SE = 0.73, p = 0.008) and negatively associated with suicides (B = -0.11, standard error = 0.05, p = 0.038). Total number of tweets were negatively associated with calls (B = -0.01, standard error  = 0.0003, p = 0.007) and positively associated with suicide, (B = 6.4 × 10-5, standard error  = 2.6 × 10-5, p = 0.015).

CONCLUSION

This is the first large-scale study to suggest that daily volume of specific suicide-prevention-related social media content on Twitter corresponds to higher daily levels of help-seeking behaviour and lower daily number of suicide deaths.

PREREGISTRATION

As Predicted, #66922, 26 May 2021.

Is the narrative the message? The relationship between suicide-related narratives in media reports and subsequent suicides

OBJECTIVES

When journalists report on the details of a suicide, the way that they contextualize the meaning of the event (i.e. the 'narrative') can have significant consequences for readers. The 'Werther' and 'Papageno' narrative effects refer to increases and decreases in suicides across populations following media reports on suicidal acts or mastery of crises, respectively. The goal of this study was to investigate the impact of these different narrative constructs on subsequent suicides.

METHODS

This study examined the change in suicide counts over time in Toronto, Canada. It used latent difference score analysis, examining suicide-related print media reports in the Toronto media market (2011-2014). Articles (N = 6367) were coded as having a potentially harmful narrative if they described suicide in a celebrity or described a suicide death in a non-celebrity and included the suicide method. Articles were coded as having potentially protective narratives if they included at least one element of protective content (e.g. alternatives to suicide) without including any information about suicidal behaviour (i.e. suicide attempts or death).

RESULTS

Latent difference score longitudinal multigroup analyses identified a dose-response relationship in which the trajectory of suicides following harmful 'Werther' narrative reports increased over time, while protective 'Papageno' narrative reports declined. The latent difference score model demonstrated significant goodness of fit and parameter estimates, with each group demonstrating different trajectories of change in reported suicides over time: (χ2[6], N = 6367) = 13.16; χ2/df = 2.19; Akaike information criterion = 97.16, comparative fit index = 0.96, root mean square error of approximation = 0.03.

CONCLUSION

Our findings support the notion that the 'narrative' matters when reporting on suicide. Specifically, 'Werther' narratives of suicides in celebrities and suicides in non-celebrities where the methods were described were associated with more subsequent suicides while 'Papageno' narratives of survival and crisis mastery without depictions of suicidal behaviours were associated with fewer subsequent suicides. These results may inform efforts to prevent imitation suicides.

Selfie Videos to Reduce Stigma and Increase Treatment Seeking Among Youths: Two Noninferiority Randomized Controlled Trials

OBJECTIVE

Confronting stigma early in life could enhance treatment seeking. In two randomized controlled trials (RCTs), one focused on psychosis and the other on adolescent depression, the efficacy and equivalence of brief social contact-based videos were evaluated and compared with a control condition. The outcomes of interest were changes in illness-related stigma and treatment-seeking intention. The hypotheses were that the intervention videos would show greater efficacy than control conditions and that traditional and selfie videos would demonstrate similar efficacy.

METHODS

Young adults (study 1, N=895) and adolescents (study 2, N=637) were randomly assigned to view intervention videos (in traditional or selfie styles) or to a control condition. In short videos (58-102 seconds), young presenters humanized their illness by emotionally describing their struggles and discussing themes of recovery and hope.

RESULTS

Repeated-measures analyses of variance and paired t tests showed significant differences in stigma and treatment seeking between the intervention and control groups and similar efficacy of the traditional and selfie videos. Cohen's d effect sizes ranged from 0.31 to 0.76 for changes in stigma from baseline to 30-day follow-up in study 1 and from 0.13 to 0.47 for changes from baseline to postintervention in study 2.

CONCLUSIONS

The RCTs demonstrated the efficacy of brief videos, both traditional and selfie, in reducing illness-related stigma among young adults and adolescents and in increasing treatment-seeking intention among adolescents. Future studies should explore the effects of brief videos presented by social media influencers on mental health stigma and treatment engagement.

Associations of suicide-related media reporting characteristics with help-seeking and suicide in Oregon and Washington

OBJECTIVE

Specific content characteristics of suicide media reporting might differentially impact suicides in the population, but studies have not considered the overarching theme of the respective media stories and other relevant outcomes besides suicide, such as help-seeking behaviours.

METHODS

We obtained 5652 media reports related to suicide from 6 print, 44 broadcast and 251 online sources in Oregon and Washington states, published between April 2019 and March 2020. We conducted a content analysis of stories regarding their overarching focus and specific content characteristics based on media recommendations for suicide reporting. We applied logistic regression analyses to assess how focus and content characteristics were associated with subsequent calls to the US National Suicide Prevention Lifeline (Lifeline) and suicides in these two states in the week after publication compared to a control time period.

RESULTS

Compared to a focus on suicide death, a focus on suicidal ideation, suicide prevention, healing stories, community suicide crises/suicide clusters and homicide suicide was associated with more calls. As compared to a focus on suicide death, stories on suicide prevention and stories on community suicide crises/suicide clusters were also associated with no increase in suicides. Regarding specific content characteristics, there were associations that were largely consistent with previous work in the area, for example, an association of celebrity suicide reporting with increases in suicide.

CONCLUSION

The overall focus of a media story may influence help-seeking and suicides, and several story characteristics appear to be related to both outcomes. More research is needed to investigate possible causal effects and pathways.

Detecting Potentially Harmful and Protective Suicide-Related Content on Twitter: Machine Learning Approach

Background

Research has repeatedly shown that exposure to suicide-related news media content is associated with suicide rates, with some content characteristics likely having harmful and others potentially protective effects. Although good evidence exists for a few selected characteristics, systematic and large-scale investigations are lacking. Moreover, the growing importance of social media, particularly among young adults, calls for studies on the effects of the content posted on these platforms.

Objective

This study applies natural language processing and machine learning methods to classify large quantities of social media data according to characteristics identified as potentially harmful or beneficial in media effects research on suicide and prevention.

Methods

We manually labeled 3202 English tweets using a novel annotation scheme that classifies suicide-related tweets into 12 categories. Based on these categories, we trained a benchmark of machine learning models for a multiclass and a binary classification task. As models, we included a majority classifier, an approach based on word frequency (term frequency-inverse document frequency with a linear support vector machine) and 2 state-of-the-art deep learning models (Bidirectional Encoder Representations from Transformers [BERT] and XLNet). The first task classified posts into 6 main content categories, which are particularly relevant for suicide prevention based on previous evidence. These included personal stories of either suicidal ideation and attempts or coping and recovery, calls for action intending to spread either problem awareness or prevention-related information, reporting of suicide cases, and other tweets irrelevant to these 5 categories. The second classification task was binary and separated posts in the 11 categories referring to actual suicide from posts in the off-topic category, which use suicide-related terms in another meaning or context.

Results

In both tasks, the performance of the 2 deep learning models was very similar and better than that of the majority or the word frequency classifier. BERT and XLNet reached accuracy scores above 73% on average across the 6 main categories in the test set and F₁-scores between 0.69 and 0.85 for all but the suicidal ideation and attempts category (F₁=0.55). In the binary classification task, they correctly labeled around 88% of the tweets as about suicide versus off-topic, with BERT achieving F₁-scores of 0.93 and 0.74, respectively. These classification performances were similar to human performance in most cases and were comparable with state-of-the-art models on similar tasks.

Conclusions

The achieved performance scores highlight machine learning as a useful tool for media effects research on suicide. The clear advantage of BERT and XLNet suggests that there is crucial information about meaning in the context of words beyond mere word frequencies in tweets about suicide. By making data labeling more efficient, this work has enabled large-scale investigations on harmful and protective associations of social media content with suicide rates and help-seeking behavior.

Differences in Suicide-Related Twitter Content According to User Influence

Background: The content of suicide-specific social media posts may impact suicide rates, and putatively harmful and/or protective content may vary by the author's influence. Aims: This study sought to characterize how suicide-related Twitter content differs according to user influence. Method: Suicide-related tweets from July 1, 2015, to June 1, 2016, geolocated to Toronto, Canada, were collected and randomly selected for coding (n = 2,250) across low, medium, or high user influence levels (based on the number of followers, tweets, retweets, and posting frequency). Logistic regression was used to identify differences by user influence for various content variables. Results: Low- and medium-influence users typically tweeted about personal experiences with suicide and associations with mental health and shared morbid humor/flippant tweets. High-influence users tended to tweet about suicide clusters, suicide in youth, older adults, indigenous people, suicide attempts, and specific methods. Tweets across influence levels predominantly focused on suicide deaths, and few described suicidal ideation or included helpful content. Limitations: Social media data were from a single location and epoch. Conclusion: This study demonstrated more problematic content vis-à-vis safe suicide messaging in tweets by high-influence users and a paucity of protective content across all users. These results highlight the need for further research and potential intervention.

Media coverage of Canadian Veterans, with a focus on post traumatic stress disorder and suicide

BACKGROUND

A large corpus of research indicates that the media plays a key role in shaping public beliefs, opinions and attitudes towards social groups. Some research from the United States indicates that military Veterans are sometimes framed in a stereotypical and stigmatizing manner, however there is a lack of research on Canadian media coverage of Veterans. As such, the overarching aim of this study is to assess the tone and content of Canadian media coverage of military Veterans, with a focus on PTSD and suicide. The first objective is to document and analyze common themes, content and temporal patterns in Canadian media coverage of Veterans per se. The second objective is to examine common themes and content in the sub-set of articles having PTSD as a theme. The third objective is to assess adherence to responsible reporting of suicide guidelines in the sub-set of articles having suicide as a theme.

METHODS

We used validated and systematic methods including use of key words, retrieval software and inter-rater reliability tests to collect and code news articles (N = 915) about Veterans from over 50 media sources during a 12-month period, with specific coding of articles about PTSD (N = 93) and suicide (N = 61).

RESULTS

Analysis revealed that the most common theme is 'honour or commemoration of Veterans' which occurred in over half of the articles. In contrast 14% of articles focused on danger, violence or criminality. In the sub-set of articles with PTSD as a theme, over 60% focused on danger, violence or criminality, while only around 1 in 3 focused on recovery, rehabilitation, or health/social service intervention. In the sub-set of articles about suicide, there was generally strong adherence to responsible reporting guidelines, though less than 5% gave help-seeking information. Moreover, most reporting on PTSD and suicide focused on a single anomalous murder-suicide incident, with few articles about suicide prevention, helpful resources and modifiable risk factors.

CONCLUSIONS

The results reveal some encouraging findings as well as a need to diversify media coverage of Canadian Veterans. This could be achieved through targeted educational outreach to help Canadian journalists responsibly report on Veterans and their mental health issues.

A Pilot Case-Control Study of the Social Media Activity Following Cluster and Non-Cluster Suicides in Australia

Social media may play a role in the "contagion" mechanism thought to underpin suicide clusters. Our pilot case-control study presented a novel methodological approach to examining whether Facebook activity following cluster and non-cluster suicides differed. We used a scan statistic to identify suicide cluster cases occurring in spatiotemporal clusters and matched each case to 10 non-cluster control suicides. We identified the Facebook accounts of 3/48 cluster cases and 20/480 non-cluster controls and their respective friends-lists and retrieved 48 posthumous posts and replies (text segments) referring to the deceased for the former and 606 for the latter. We examined text segments for "putatively harmful" and "putatively protective" content (e.g., discussion of the suicide method vs. messages discouraging suicidal acts). We also used concept mapping, word-emotion association, and sentiment analysis and gauged user reactions to posts using the reactions-to-posts ratio. We found no "putatively harmful" or "putatively protective" content following any suicides. However, "family" and "son" concepts were more common for cluster cases and "xx", "sorry" and "loss" concepts were more common for non-cluster controls, and there were twice as many surprise- and disgust-associated words for cluster cases. Posts pertaining to non-cluster controls were four times as receptive as those about cluster cases. We hope that the approach we have presented may help to guide future research to explain suicide clusters and social-media contagion.

Suicide-related Twitter Content in Response to a National Mental Health Awareness Campaign and the Association between the Campaign and Suicide Rates in Ontario

OBJECTIVE

Mental health awareness (MHA) campaigns have been shown to be successful in improving mental health literacy, decreasing stigma, and generating public discussion. However, there is a dearth of evidence regarding the effects of these campaigns on behavioral outcomes such as suicides. Therefore, the objective of this article is to characterize the association between the event and suicide in Canada's most populous province and the content of suicide-related tweets referencing a Canadian MHA campaign (Bell Let's Talk Day [BLTD]).

METHODS

Suicide counts during the week of BTLD were compared to a control window (2011 to 2016) to test for associations between the BLTD event and suicide. Suicide tweets geolocated to Ontario, posted in 2016 with the BLTD hashtag were coded for specific putatively harmful and protective content.

RESULTS

There was no associated change in suicide counts. Tweets (n = 3,763) mainly included content related to general comments about suicide death (68%) and suicide being a problem (42.8%) with little putatively helpful content such as stories of resilience (0.6%) and messages of hope (2.2%).

CONCLUSIONS

In Ontario, this national mental health media campaign was associated with a high volume of suicide-related tweets but not necessarily including content expected to diminish suicide rates. Campaigns like BLTD should strongly consider greater attention to suicide-related messaging that promotes help-seeking and resilience. This may help to further decrease stigmatization, and potentially, reduce suicide rates.