Communicating Anxiety and Depression in Twitter: Content and Network

1. Introduction

When communicating anxiety and depression, stigma and discrimination are the issues concerned.

Computer-medicated communication (CMC) provides a more desirable context for people to discuss this topic.

2. Literature review

Computer-mediated communication (CMC) provides a range of impetus for people to discuss mental health problems on the internet (Fage-Butler & Jensen 2015; Wright & Bell, 2003); for example, reduction of prejudice due to the lack of social markers (e.g. age, gender and race), acceptance of divergent opinions and personal anonymity.

Cavaxos-Rehg et al. (2016) conducted a content assessment of tweets related to depression with the hashtags #depression and #depressed. The study revealed that 40% of the tweets convey supportive or helpful tweets about depression and 32% of those disclose depressive symptoms. Relatively small number of tweets contains the messages of contemplating suicide or desires death and self-harm.

Park et al.’s research (2012) suggests that some individuals are willing to tweet about their conditions with depression and constantly update their treatments on Twitter. Also, the participants tended to follow the same celebrities of popular culture.

2.1 Research Objectives

This pilot study aims to explore the way the users communicate anxiety and depression in Twitter through answering two research questions:

  1. What is the content the users deliver?
  2. How is the network of the communication related to anxiety and depression structured?

3. Methodology

Tweets (n=18000) with the hashtags depression or anxiety were retrieved through TAGv 6.1. No particular period for data collection was set due to the independence of the topic. 

Two predominant approaches were employed to analyze the data and address the research questions respectively, namely data visualizations (Tableau, Voyant and Gephi) and content analysis (300 tweets coded).

4. Results

4.1 Cirrus of the corpus


The theme, anxiety and depression, are the most predominant words in the corpus.

The word i’m indicates that the users talk about themselves in terms of their depressive conditions. The content tends to be individual.

The swear words fuck and wtf mark that the messages posted might be expressive.

The reason for having anxiety or depression can also be associated with the words living, ugly and stress; for example, life experiences, appearance and stress.

4.2 Content analysis

Themes Frequency Percentage
Supportive or helpful tweets 22 7%
Self-disclosure of anxiety or depression 122 40%
Related Information 106 35%
Protest 16 5%
Appeal 10 3%
Advice 11 4%
Questions 16 5%

Most of the themes of the tweets are self-disclosure of having anxiety or depression. For example,

RT @chellyjae: who tf stole my xanax prescription out of my gym bag? i have anxiety, you crackhead ass bitch wtf

Related information about the two mental disorders is the second most salient theme. The content of this theme vary. It can be any information related to anxiety and depression; for example, scientific studies or news about the mental disorders and stories of people having anxiety and depression. For instance,

Depression linked to higher risks of a heart rhythm disorder #healthnews

RT @AntiStigma: Five Anxiety Facts that Help You Understand Yourself – HealthyPlace

The rest of the themes are relatively infrequent in the tweets.

4.3 The network of retweeting activities

Whole network
Graph A

Graph A exhibits that the vast majority of the clusters in the network are isolated without connections with other clusters. This implies that most of the tweets with #depression or #anxiety are in fact neglected.

Graph B

Graph B focuses on the network of the retweeting activities at the middle. It shows that the largest cluster is public figure A called KatHeartwell, who is a writer. The relevant tweets posted by this account are re-tweeted the most frequently.

The account has currently 35.8K followers and the tweets involved in the account tend to be personal, psychological and emotional. It has the most social ties and receives and disseminates information with more accesses.  It possesses the highest closeness with other nodes and hence can spread information faster.

The second largest cluster is specified account A called depressionnote with currently 38.2K followers, and the third largest cluster is personal account A called chellyjae with 4.5K followers. The themes of the accounts KatHeartwell and chellyjae are similar.

In general, personal accounts appear to be more salient in the communication of anxiety or depression due to their salience in the retweeting network.

4.4 Comparison of the top 10 retweets and user visibility


The creators of the top 6 retweets, including Katheatwell, depressionnote, chellyiae, leaklestfacucet, kallitgurl and mandosie8, are at the same time the most visible accounts. Digging deeper into the content of the most retweeted texts, it is found that anxiety and depression in the mostly tweeted messages are taken as usual emotions of an individual rather than a serious mental illness. For instance,

RT @KatHeartwell: Cheers to all the women who living with anxiety, depression, ptsd, BPD etc. Mental health can make you feel ugly but I’m here to remind you that you cute as fuck

RT @chellyjae: who tf stole my xanax prescription out of my gym bag? i have anxiety, you crackhead ass bitch wtf

This also corresponds to the content analysis conducted in which disclosure of anxiety and depression is the most prevalent theme. As to the most retweeted posting of the specified account, it tends to be medical information related the mental illness.

RT @depressionnote: depression symptoms ⚠
unable to imagine a happy future, lack of organization, procrastination, social withdrawal, tiredness, anxiety, insomnia, feeling “numb” or “empty”
can you RT for depression awareness?

5. Concluding discussion

This study revealed that the predominant content of tweets related to anxiety and depression is self-disclosure of mental suffering. The findings are contradictive to the results of Cavax-Rehg et al.’s study (2016), which found that supportive or helpful tweets are the dominant theme in the tweets that contain the content of depression. This could be due to the difference of hashtags studied. The hashtag #depressed tends to be used to express personal feelings and it is often associated with encouraging messages.

The users are more willing to disclose themselves probably due to the contextual merits of online social networks. Computer-mediated health communication provides more communicative freedom than face-to-face settings and a secure platform wherein people ask awkward, sensitive and detailed questions and diminish harassment at the same time (Mullany et al., 2016). Users seldom sense the feeling of being judged, disapproval or disappointment by the addressees of their messages (Wood & Smith, 2001). In addition, anonymity or pseudonymity of online communication gives the participants a sense of safety that encourages them to share their problems in the support group (Wallace, 1999).

The network analysis of this study reflects that the opinion leader in the communication of anxiety and depression seems to be the public figure whose account is in high-level degree and high closeness with other accounts in the retweeting activities. Most frequently retweets were self-disclosure of having anxiety or depression and the frequency is more than the informative tweets. The most active users’ tweets are at the same time the most retweets in the data set. This is consistent with the claim of McCandless (2009) who found approximately 5% of the tweeters create about 75% of all tweets.  This phenomenon also suits the topic of this study. Furthermore, the salience of tweets with mental health-related information can be illuminated through the fact that some online information seekers are interested in information of specific diseases, symptoms, medical procedures, living styles, drugs, therapies and health systems (Underhill & McKeown, 2008). Twitter acts as a medium which helps search and deliver the relevant information and satisfies the needs of the users.

 6. Limitations

There are several limitations of this study which manifest the needs for future research to explore. First, content analysis per se overlooks the non-verbal aspects of texts. The tone and stress of the tweets are hard to be analyzed. This might lead to inaccuracy of data analysis. Data analysis along with interviews of the users involved is applicable to solve the problem. Second, despite the independence of the topic, specific time or social affairs could have an effect on the content and communicative acts in the communication of the mental disorders. A larger amount of data composed of tweets in a period of time could be collected for data analysis.


Berry, N., Lobban, F., Belousov, M., Emsley, R., Nenadic, G., & Bucci, S. (2017). #WhyWeTweetMH: Understanding Why People Use Twitter to Discuss Mental Health Problems. Journal of Medical Internet Research19(4), e107.

Cavazos-Rehg, P. A., Krauss, M. J., Sowles, S., Conolly, S., Ros, C.,Bharadwaj, M., & Bierut, L. (2016). A content analysis of depression-related tweets. Computers in Human Behavior. 54, 351-357

McCandless, D. (2009). Information is beautiful: If twitter was 1000 people… Retrieved

Mullany, L., Smith, C., Harvey, K., & Adolphs, S. (2016). ‘Am I anorexic?’ Weight, eating and discourses of the body in online adolescent health communication. Communication & Medicine, 12(2–3), 211–223.

Park, M., Cha, C., & Cha, M. (2012). Depressive moods of users portrayed in Twitter. In Proceedings of the ACM SIGKDD Workshop on healthcare informatics (HI-KDD) (pp. 1-8).

Underhill, C., & McKeown, L. (2008). Getting a second opinion: Health information and the Internet. Retrieved from

Wallace, P. (1999). The psychology of the Internet. Cambridge. Cambridge, UK: Cambridge University.

Wood, A. F., & Smith, M. J. (2001). Online communication: Linking technology, identity, and culture. Mahwah, NJ: Lawrence Erlbaum.


Appendix 1 Coding scheme for the content analysis

Themes Description Example tweets
Supportive or helpful tweets Empathetic messages that shows understanding and support or information of possible assistance Simple strategies to overcome fear & #anxiety during a #highriskpregnancy & stay #pregnant as long as possible
Self-disclosure of anxiety or depression An admission of having anxiety or depression with narrative about person experiences I’m only now realizing the extent of the anxiety I’ve lived with my entire life. This feeling, to me, was normal life.  GAWD! #SickNotWeak
Related Information Texts related to anxiety or depression, which enhance people’s understanding about the illness Panic, chronic anxiety and burnout: doctors at breaking point
Protest Emotional protest that shows desire to overcome or distance from the mental illness I deserve everything tbh. I deserve love, money, great sex, success, a stress free life, a fire body, even skin, perfect teeth, minimum depression and anxiety. I deserve it all 😭
Appeal Appeals for others to do something related to the mental illness RT @hailee_mccraw: Anxiety and depression are haunting people that you would never guess, so let’s all just do our best to be kind and sensitive to everyone. You never know what giants others are facing.
Advice that helps improve the conditions of people with anxiety or depression You can’t just say that to me because I said it to you. Pathetic. I think you need psychotherapy, like I had. It’s scary digging around, but you find an explanation for depression & anxiety, rather than some made up diagnosis from a shrink.
Questions Asking questions about anxiety or depression What really causes depression?