The Representation of Mental Illness in Bermudian Print Media, 1991–2011

All articles relating to mental illness published between January 1 and December 31 in 1991, 2001, and 2011 were identified. Four newspapers were in print during that time—the Royal Gazette and the Bermuda Sun in all three years, the Mid-Ocean News in 2001 and 1991, and the Bermuda Times in 1991. For years in which newspapers had online electronic search facilities, articles were identified by using the search terms mental*, schizo*, and psych*, where * related to a Boolean truncation wild card. All other articles were identified by a manual search of microfilm at the Bermuda National Library. Data were collected during January 2012.

Articles were excluded from analysis if there was a mention of psychiatric or psychological evaluation without manifest link to mental illness, if drug or alcohol misuse was mentioned with no associated link to mental illness, if the term was used in the title of books or films, or if an article identified by the electronic search was not relevant to mental health.

A total of 277 articles met inclusion criteria: 123 from 2011, 79 from 2001, and 75 from 1991. All of them were coded for year of publication, newspaper source, publication on the front page or inside the paper, presence of photographs, and presence of a photograph of a named service user.

Decisions on which features of an article to code were amalgamated from previous studies (1,4). Overall tone was categorized as positive, negative, neutral, or mixed (3). Themes were categorized in four areas—blame; dangerousness and crime; education, information, and recovery; and advocacy, actions, and concerns. The categories were chosen on the basis of previous focus group deliberations of sample articles and adapted by the research team following internal discussion. [Definitions of the themes and a coding assistance sheet containing examples of stigmatizing terms, metaphorical language, and equating descriptors are available online as a data supplement to this article.]

Five articles were randomly selected and discussed to facilitate coding training. Afterward two researchers coded a random sample of 45 articles (>10% of total) for remaining data: use of negative or stigmatizing terminology, identification of the person referred to by the terminology, metaphorical use of mental health terminology, use of equating descriptors, whether a mental health professional or service user was consulted or quoted, and overall tone and primary theme.

The average interrater reliability (κ=.9) across the 45 articles suggested almost perfect agreement. The remainder of the articles were randomly distributed to the two researchers and coded.

The data were analyzed by using SPSS, version 20, and the study design was approved by the Bermuda Health Board Ethics Committee.

There was no difference in the frequency of front-page articles across all three years.

There was a significant difference in the presence of photographs across all three years (χ²=30.91, df=2, p<.001), with significantly more photographs in 2011 (N=63, 51%) than in 2001 (N=15, 19%; χ²=21.08, df=1, p<.001) and 1991 (N=15, 20%; χ²=19.02, df=1, p<.001).

There was no difference in the frequency of photographs of a named service user across all three years; however, in 2011, 27% (N=6) of all articles in which the primary theme was violent crime contained a photo of a named service user.

There was no difference in the overall tone (Table 1), the use of negative or stigmatizing terms, the metaphorical use of mental health terms, or the use of equating descriptors across all three years. Articles with a negative overall tone constituted the largest percentage of articles in 2001 (43%, N=34) and 2011 (42%, N=51).

A significant difference in the number of mental health professionals who were consulted or quoted was found across all three years (χ²=17.58, df=2, p<.001) and in 2011 (N=62, 50%) versus 2001 (N=18, 23%; χ²=15.34, df=1, p<.001) and 1991 (N=23, 31%; χ²=7.41, df=1, p<.001).

There was no difference across all three years in the number of mental health service users who were consulted or quoted (N=17, 1991; N=16, 2001; and N=22, 2011).

The results showed that 62% (N=34) of all articles in which a service user was consulted or quoted were related to the theme of dangerousness and crime; this percentage increased to 73% (N=16) among articles from 2011. Within the dangerousness and crime category, 36% (N=20) of articles in which a service user was consulted or quoted were related to violent crime; this percentage rose to 59% (N=13) among articles from 2011. A total of 69% (N=38) of articles in which a service user was consulted or quoted were negative in tone; this percentage rose to 82% (N=18) among articles from 2011. The coders noted anecdotally that most service user quotations were from court reporting pieces in which a quote from a trial was documented.

There was a significant difference across all three years in the primary theme of the articles (χ²=15.36, df=3, p=.018). Articles were significantly more likely to favor dangerousness and crime in 2011 (N=48, 39%) and in 2001 (N=36, 45%) than in 1991, when education, information, and recovery themes were more common (N=19, 25%; χ²=8.06, df=3, p=.045, and χ²=12.17, df=3, p=.007, respectively).

There was a significant difference in primary theme between 2011 and 2001 (χ²=34.82, df=21, p=.03) and 1991 (χ²=41.64, df=20, p=.003). Violent crime was favored in 18% (N=22) of articles in 2011, compared with 4% (N=4) in 2001 and 12% (N=9) in 1991. Psychological treatment was the primary theme in 9% (N=11) of articles in 2011 and 1% (N=1) of articles in 2001. Education and information were favored in 18% (N=22) of articles in 2011 and 40% (N=31) of articles in 1991.

There was also a significant difference in primary theme between 2001 and 1991 (χ²=33.88, df=21, p=.037). Articles favored education and information in 40% (N=31) of articles in 1991, compared with 19% (N=15) in 2001.

The trend in themes can be grossly explained by a decreasing number of articles regarding education and information, from 40% in 1991 to 19% in 2001 and 18% in 2011, and an increasing number relating to violent crime (12%, 4%, and 18%, respectively). [A table summarizing numbers of articles by theme is available in the online supplement.]

The articles that were located provided a comprehensive data set comprising three time points, enabling us to comment on longitudinal trends.

Despite increased consultation of mental health professionals—evidence of positive change in print media coverage of mental illness—over time mental illness was increasingly connected with dangerousness and crime, and publication of positive articles was less likely than publication of negative ones.

In contrast to data from the United Kingdom and the United States (1,2), our data have shown an increase in reporting of violent crime in articles related to mental health over the last 20 years. An important point of consideration is whether this may be a function of a general increase in crime in Bermuda. Studies showed a two- to sixfold increase in the likelihood that people with mental illness will commit violent crime compared with matched controls (12). A recent Swedish study demonstrated that the fraction of attributable risk associated with patients with mental illness was 5%, suggesting that persons with severe mental illness commit one in 20 violent crimes (13). It also found that severe mental illness was consistently a poorer predictor of violent crime than age, ethnicity, and gender. Although most people with mental illness are neither violent nor criminal, the overrepresentation of stories relating to these traits is likely to cause the Bermudian lay public to disbelieve this assertion.

Worrisome numbers of photographs of named service users had been printed over the last 20 years. In this small, formal community, this practice could severely limit a service user’s anonymity and place in jeopardy prospective community rehabilitation and recovery. Service users may be discouraged from seeking treatment for fear of having their names, photographs, and purported diagnoses appear in the national press.

Although there were many positive aspects to our study, including inclusion of all published copies of national newspapers in sampled years, we must consider potential limitations.

First, our electronic search strategy failed to identify some stories, particularly articles focusing on specific psychiatric diagnoses. Because of human error, the manual microfilm search may have missed articles, particularly articles containing terms beginning with schizo*, that an electronic search would have located. Metaphorical terms in nonobvious mental health–based articles may have been harder to identify by searching by hand.

Second, the study included only national newspapers and not imported international newspapers, local television, radio, or electronic media sources. The pure national newspaper sample remains highly relevant in Bermudian culture. In 2011 alone, the Royal Gazette, a cheap ($1), daily source of local information, was the newspaper of choice for 80% of the island’s population (10).

Third, the results may have been skewed by high-profile events. Although multiple articles concerning the same events occurred in all three years, in 2011 two cases contributed significantly to the volume of articles reporting violent crime. One could argue that without these events, fewer articles with a negative tone and a primary theme of violent crime would have emerged. However, given that these events were chosen as topics to be repeatedly reported, they formed part of the article set to which the Bermudian lay public were consistently exposed and remain relevant.

Last, the coding was completed by two mental health professionals, which may have incurred bias.

In order to develop the study, other media sources—particularly electronic sources—would be important to take into consideration. For example, the newly established bermy.net, a media source with much less regulation, is pervasive among younger generations.

None of these limitations was sufficient to overcome the trends demonstrated, given that our data represent the most extensive sampling of Bermudian newspaper media regarding mental illness ever undertaken.

This study was timely given the founding, in February 2011, of the Media Council of Bermuda, a nongovernmental regulatory body established to deal with media complaints by the public. The council’s media code of conduct contains the following assertions: “The media shall avoid prejudicial or pejorative reference to an individual’s . . . disability, illness,” “Details of an individual’s . . . disability, illness, shall be avoided unless genuinely relevant to the story,” and “Media reports shall not place gratuitous emphasis on . . . disability, illness. Journalists should also avoid stereotyping based on the above.”

These statements clarify that the national media has several obligations to the island’s service users. The Mid-Atlantic Wellness Institute, as the island’s only mental health service provider, has a duty to its service users to correct potentially misleading metaphors and stereotyping in the press. Despite many articles breaching these assertions in the year since its inception, the council has upheld only one complaint and has received no complaints regarding coverage of mental health (e-mail communication, Ebbin M, 2012). Strategies to combat this disappointing result may include resurrection of service user–led mental health advocacy groups or judicious engagement of the media council by mental health professionals or service users with the encouragement and advocacy of mental health services.

Another way in which media may be challenged regarding their portrayal of mental health includes use of targeted information, educational programs, and destigmatization campaigns aimed at local journalists. The World Psychiatric Association has identified journalists as a specific target of their international program to fight stigma faced by persons with schizophrenia (14).

We set out to discover changes in reporting of mental health issues and discussed how the findings may relate to the level of social stigma experienced by persons with mental illness in Bermuda.

Although positive changes were found in terms of increased consultation with mental health professionals, the predominant negativity of newspaper article tone did not change. In addition, the primary theme of coverage shifted away from education and information toward violent crime. These findings suggest that the time is ripe for action by the island’s advocacy groups. Given the close-knit nature of the community and the nation’s high newspaper readership, antistigma interventions in Bermuda may have an enormous impact.

Ongoing analysis of Bermudian newspapers is warranted to observe whether these trends continue and whether they can be altered by intervention by island mental health services. Interventions designed to improve media representation, for example, consideration of specific mental health training for the island’s press, judicious use of the Media Council to highlight stigma, or promotion of service user advocacy, may be required to halt the patterns described above.

The authors report no competing interests.