The Spread of Infectious Diseases Conversation on Twitter – Networks, Communities and Social Mediators

Theoretical Background and research questions/hypothesis:  Public health agencies are increasingly utilizing social media as part of their communications strategy to provide guidance and manage risk perceptions during infectious disease outbreaks (IDO) like Zika. The use of social media can be strengthened by having real or near real-time information regarding what others are posting and disseminating via social media.  This social media activity reflects hidden structures of information flow and isolation, which restrict health organizations’ communication efforts. This study proposes analyzing information flow on Twitter, giving health agencies tools to identify, target and to reach out to new and existing social media communities. 

Methods:  Data. Twitter usernames, profile descriptions, retweets, and reply-to relationships were collected about users who participated in tweets and retweets about Zika from 1/1/2016 to 3/1/2016.  Zika-related topics were chosen in consultation with experts from the US Centers for Disease Control and Prevention. Network analyses. For each topic, social networks, formed by retweets and replies of Zika-related tweets, were mapped and analyzed. The clusters in the topic-networks were identified using an algorithm that assign users to groups, based on their tightness of social interactions. Social mediators were identified as users who measured high in betweenness centrality (i.e., information bridges), and high in-degree centrality (i.e., attention triggered by one’s posts). Content posted by social mediators reaches a wide audience across social media communities. Classification of social mediators.  We iteratively developed a classification system based on preliminary analysis of all users in a single day of data collection, including health organizations, media organizations, celebrities and grassroots. Initial

Results: A total of 2,596,995 Zika-related tweets were captured. Of these, 1,635,781 about virus’ transmission, 420,734 on effects on pregnancy, 206,062 were travel-related, 127,089 on social issues (e.g., abortion), 97,740 on testing and diagnosis, 89,750 about treatment and 70,773 related to conspiracy theories.   In current ongoing analysis, we are mapping and analyzing patterns of information exchange (retweets and replies) within each of the Zika-related conversations, identifying communities and social mediators. Initial analysis of key users, suggests the vital roles played by media and health institutions (e.g., WHO, UN, CDC), alongside with bloggers, grassroots and advocacy groups.

Conclusions:  Mapping patterns of social networks of the Zika Twitter discourse allows us to identify key communities discussing issues, their key information sources and the passages of information flow across communities. These networks and communities were not consistent across Zika-related topics. This study also identified how information flowed via different social mediators and a range of communities, across a variety of Zika-related topics.  

Implications for research and/or practice:  This study illustrates how Twitter data can be analyzed and used to guide public health messaging and social media efforts related to an infectious disease outbreak. Identifying patterns of information flow, the separated communities that exchange messages, and the social mediators that bridge across communities, public health communicators can evaluate and make changes to their social media strategies and messages.  While the study is contextualized around the Zika outbreak, the implications apply to other IDOs and public health emergencies.