Use of Twitter to monitor attitudes toward depression and schizophrenia: an exploratory study

Introduction

Mental illness affects as many as one in five people in any 12-month period (Kessler et al., 2007). Many people with a mental illness experience social and economic hardship as a direct result of their illness. They must cope with their symptoms and also with stigma and discrimination that result from misconceptions about these illnesses (McNair et al., 2002; Corrigan et al., 2003). These issues are a key concern as stigma may deter people with symptoms from seeking help due to embarrassment or the belief that others will think badly of them (Gulliver, Griffiths & Christensen, 2012; Yap, Reavley & Jorm, 2013). Stigma may also compound the experience of psychological distress and can adversely affect personal relationships and the ability to achieve educational and vocational goals (Wells et al., 1994; Link et al., 1997; Corrigan, 2004).

In an effort to better understand the factors predicting stigmatising attitudes to mental illness, a number of research studies have focused on media coverage of mental disorders (Morgan & Jorm, 2009; McGinty et al., 2014). A study of newspaper reporting on mental illness in the USA found that 39% of reports concerned dangerousness and violence (Corrigan et al., 2005). There is evidence that such reports can increase stigmatising attitudes, including beliefs about dangerousness. Data from surveys conducted in Germany showed an increase in beliefs about dangerousness after two widely-publicised attacks on politicians by people with schizophrenia (Angermeyer & Matschinger, 1996). In another German study, reading a newspaper article about mental illness and violent crime increased stigmatising attitudes (Dietrich et al., 2006). However, the majority of studies assessing the role of media coverage of mental illness has focused on traditional media, with relatively little attention paid to newer social media platforms such as Facebook and Twitter (Egan & Moreno, 2011; Moreno et al., 2011).

Twitter is a popular social networking website that allows users to post brief messages up to 140 characters in length, referred to as tweets. Twitter enables social communication and networking between users through retweeting (reposting other users’ tweets to one’s own feed), responding to tweets, and following other users’ Twitter feeds. Hashtags are included in tweets to identify a theme and enable other users to view related posts with the same hashtag. It is estimated that 18% of online US adults use Twitter (Pew Internet & American Life Project, 2011), with similar numbers reported in other developed countries (Office for National Statistics, 2013). Unless Twitter users mark their tweets as private, these are public, making Twitter a potentially valuable source of information about the views of people on a range of topics. A number of studies have analysed Twitter content on health-related topics, including an influenza outbreak (Chew & Eysenbach, 2010), problem drinking (West et al., 2012), dental pain (Heaivilin et al., 2011), palliative medicine (Nwosu et al., 2014), physical activity (Zhang et al., 2013), dementia (Robillard et al., 2013), vaccination (Love et al., 2013), and breast cancer (Thackeray et al., 2013). Martinez-Perez et al. (2014) analysed the purposes and functions of Facebook and Twitter groups for different mental illnesses, including depression. The results showed that individuals with mental illness utilise online social networks to obtain information, access support, and raise awareness. However, to our knowledge, the current study is the first to conduct a detailed analysis of Twitter content as it relates to attitudes to mental illness. The aim of the study is to provide insights into how Twitter users share information about depression and schizophrenia, the type of information shared, and the relative proportions of supportive or stigmatising attitudes in that content. The study offers the opportunity for unobtrusive attitude measurement (Lee, 2000), something which is currently lacking in much conventional stigma research, which uses self-report attitude scales that run the risk of social desirability bias because the responses are given to a mental health researcher.

Method

A qualitative analysis was conducted to explore supportive or stigmatising attitudes towards depression and schizophrenia on Twitter.

Results

A total of 7,295 depression tweets and 500 schizophrenia tweets were retrieved from Twitter over the 7-day period. After applying the inclusion and exclusion criteria, 5,907 depression tweets (81%) and 451 schizophrenia tweets (90%) were included in the analyses. NCapture identified between 24 and 29% of the included tweets as retweets. The majority of tweets included links to websites (75%–78%).

Content

The content of the tweets by mental illness is presented in Fig. 1.

Depression: Tweets on depression mostly related to resources for consumers (34%; 2,011 tweets) or advertised services or products for individuals with depression (20%; 1,201 tweets). The remaining tweets described personal experiences with depression (16%; 977 tweets), promoted awareness of depression (15%; 917 tweets), linked to news media articles on depression (6%; 354 tweets), reported research findings (5%; 311 tweets), or described an individual’s personal opinion or view (2%; 136 tweets). The hashtags which most commonly accompanied the hashtag #depression were #anxiety, #mentalhealth, #bipolar, #stress, and #suicide.

Schizophrenia: The majority of schizophrenia tweets aimed to increase awareness of schizophrenia (29%; 132 tweets) or reported on research findings (22%; 100 tweets). Tweets providing resources, advice, or support for people with schizophrenia accounted for 17% of the tweets (76 tweets). Twelve percent of tweets described personal opinions (54 tweets), while 10% described personal experiences with schizophrenia (47 tweets). A small number of tweets consisted of advertising (7%; 34 tweets) and news media articles (2% 8 tweets). The hashtags which most commonly accompanied the hashtag #schizophrenia were #mental, #health, #mentalhealth, #bcss, #psych. The hashtag #bcss refers to the British Columbia Schizophrenia Society.

Attitudes towards depression or schizophrenia

The numbers of stigmatising attitudes by mental illness is shown in Fig. 2. Figure 3 presents the types of stigmatising attitudes that were evident.

Depression: Tweets on depression were largely supportive (65%; 3,839 tweets) or neutral (27%; 1,600 tweets). A number of tweets were specifically anti-stigma (7%; 421 tweets). Less than 1% of tweets reflected stigmatising attitudes (0.7%; 43 tweets) or personal experience of stigma (<0.1%; 4 tweets). More than one third of the tweets which reflected stigmatising attitudes were mocking or trivialising towards individuals with depression (37%; 16 tweets). The attitude that individuals with depression should “snap out of it” was evident in 30% (13 tweets) of the stigmatising tweets. The belief that people with depression are weak was displayed in 21% (9 tweets) of tweets. A further 12% (5 tweets) of the tweets suggested self-stigma.

Schizophrenia: The majority of tweets relating to schizophrenia were categorised as neutral (43%; 193 tweets) or supportive (42%; 191 tweets). Almost 10% of tweets were explicitly anti-stigma (44 tweets). The percentage of stigmatising attitudes was 5% (21 tweets), while less than 1% of tweets described personal experiences of stigmatising attitudes towards individuals with schizophrenia (2 tweets). Of the tweets that indicated stigmatising attitudes, most reflected inaccurate beliefs about schizophrenia being multiple personality disorder (52%; 11 tweets) or mocked or trivialised individuals with schizophrenia (33%; 7 tweets). The remaining tweets indicated an unwillingness to be in social contact with people with schizophrenia (5%; 1 tweet), a belief that individuals with schizophrenia are dangerous (5%; 1 tweet), or were suggestive of self-stigma (5%; 1 tweet).

Discussion

Content analysis of Twitter information about depression and schizophrenia can provide insights into the type of information shared, who is sharing the information, and the types of attitudes displayed, whether supportive or stigmatising. An analysis of tweets with the hashtags #depression or #schizophrenia posted on Twitter during a 7-day period revealed that most tweets for both illnesses were supportive or neutral, although stigmatising attitudes were more prevalent in the schizophrenia tweets. More than one third of the depression tweets which reflected stigmatising attitudes were mocking or trivialising towards individuals with depression and the belief that individuals with depression should “snap out of it” was evident in almost one third. To some extent, these findings reflect surveys assessing stigmatising attitudes towards people with depression, which have shown the beliefs that people are weak rather than sick are relatively common (Reavley & Jorm, 2011), although views about the dangerousness of people with depression, which are relatively common in surveys were not reflected in the tweets. Of the schizophrenia tweets that indicated stigmatising attitudes, just over half reflected inaccurate beliefs about schizophrenia being multiple personality disorder while one third mocked or trivialised individuals with schizophrenia.

In terms of the type of information, the results of the current study showed that tweets on depression mostly related to resources for consumers or advertised services or products for individuals with depression, whereas the majority of schizophrenia tweets aimed to increase awareness of schizophrenia or reported on research findings. These differences are likely to relate to the higher prevalence of depression than schizophrenia and therefore the potentially greater market for products and services. These findings may be compared to those of a recent study by Martinez-Perez et al. (2014), who analysed the purposes and functions of Facebook and Twitter groups for different mental illnesses, including depression. They classified groups according to whether they were support groups, self-help groups, advocacy and awareness groups and fundraising groups. They concluded that self-help groups were the most common category (64%), followed by support groups (15%), and advocacy and awareness groups (10%). They also identified the creators of the groups, listing the following categories: associations, societies, foundations, collectives, individuals, companies, caregivers, and specialists. Over 70% of creators were identified as individuals. However, they did not do further analysis on the content of the tweets.

As far as we are aware, this is the first study to assess the content of tweets on the topic of mental illness. Strengths of the study include the opportunity for unobtrusive measurement of freely volunteered views (Lee, 2000). Results of such studies are less likely to be affected by social desirability bias than responses to surveys, which are more commonly used to assess stigmatising attitudes (Angermeyer & Matschinger, 2004; Reavley & Jorm, 2011).

With the increasing prominence of social media, the results of the study may be useful in assisting mental health promotion and advocacy organisations to provide information about resources and support, raise awareness and reduce stigma. Twitter may also be an increasingly useful resource for people seeking health information on the internet as a number of tweets provided links to health information. There is evidence that the quality of online information is improving, particularly that relating to affective disorders, and it is likely that online information and interventions will continue to become more important (Reavley & Jorm, 2010). In addition, advocacy organisations may want to specifically counter common stigmatising attitudes such as beliefs about weakness in people with depression or beliefs about ‘split personality’ in people with schizophrenia. A recent campaign to raise awareness and counter the use of stigmatising terms for people with intellectual disabilities has taken the approach of directly responding to celebrities who use these terms (Sibley, 2014).

The main limitation of the study is our inability to determine what proportion of existing tweets referring to depression and anxiety were analysed. Comparisons between the Streaming API and the “firehose” suggest that the Streaming API may not sufficiently represent the activity on Twitter as a whole (Morstatter et al., 2013). Despite this, we believe that it was appropriate to access the Streaming API as the significant costs involved in accessing the “firehose” would not yield proportionate benefits. We analysed a sufficient number of tweets to reach data saturation. Furthermore, the Streaming API is available to the public, and therefore representative of how the average person experiences Twitter.

A second potential limitation is our focus on tweets using the depression and schizophrenia hashtags. This may have meant that content related to other hashtags or slang terms such as “schizo” (which people with stigmatising attitudes are potentially more likely to use) may have been missed. Moreover, the anonymity of most users makes it impossible to verify the authenticity of content or to assess stigmatising attitudes according to age, gender, or location. Finally, these findings cannot be generalised to communities outside Twitter.

Given the speed of reactions to events on social media, it would be useful for future research to compare attitudes before and after significant events, for which it is not possible to anticipate a ‘pre-test’ (e.g., a mass shooting that involves a person with a mental illness) as this may also provide useful information on stigmatising attitudes.

Conclusion

The study supports the use of analysis of Twitter content to unobtrusively measure attitudes towards mental illness, both supportive and stigmatising. The results of the study may be useful in assisting mental health promotion and advocacy organisations to provide information about resources and support, raise awareness and counter common stigmatising attitudes.

Supplemental Information