Article Text
PDF
Abstract
Purpose To describe the stroke-related Twitter network through analysis of the #Stroke hashtag.
Materials and methods 621 653 tweets containing the #Stroke hashtag were analyzed from 20 March 2012 to 31 January 2018. Twitter activity metrics, engagement, user characteristics, content analysis, and network analysis were obtained using the healthcare social media analytics platform, Symplur Signals.
Results The number of users, the number of impressions, and the number of tweets containing the #Stroke hashtag increased by an annual average of 64.9%, 87.7%, and 89.2% over the past 6 years from 20 March 2012 to 31January 2018. 69 371 tweets (11.2%) contained novel content and 48 568 tweets (7.8%) related to patient care. 181 120 (29.1%) tweets contained at least one image and 436 132 tweets (70.2%) contained links to outside resources. Stroke prevention, diabetes, and atrial fibrillation were commonly discussed topics. With regard to engagement, there were 259 438 retweets (41.7%), 366 561 mentions (59.0%), and 8549 replies (1.4%). Physicians and patients authored 52 197 (8.4%) and 41 822 (6.7%) tweets, respectively. Advocate organizations, patients, and non-healthcare individuals most frequently used the #Stroke hashtag on Twitter.
Conclusion The use of the #Stroke hashtag on Twitter has grown significantly over the 6-year study period. The majority of the discussions were held between stroke support groups and non-healthcare-related individuals, with discussion content centering around stroke prevention, stroke symptoms, associated medical conditions, and treatment options.
- stroke
- economics
- political
Statistics from Altmetric.com
Introduction
The rapid expansion and utilization of social media has altered the way in which patients, physicians, and other healthcare stakeholders interact.1–5 This phenomenon has allowed for easy dissemination of educational resources, photographs, ideas, and medical management strategies.6 Twitter (Twitter Inc, San Francisco, California, USA) is a well-established microblogging tool which has led to the development of disease-specific communities that can categorize and aggregate their interactions using ‘hashtags’.1 7 These Twitter communities serve as readily accessible no-cost platforms which provide significant educational and professional benefits.8 9
Stroke medicine is a rapidly changing field, constantly adapting to incorporate new therapies and guidelines.10 11 Within stroke medicine, social media has been shown to benefit patients, stroke organizations, and medical education.12 Studying the intersection of social media and medicine can provide important insights into the current status of particular medical domains and how patients can best seek medical information.13
This study was designed to describe the stroke-related Twitter network by analyzing the #Stroke hashtag through assessment of tweet content, activity metrics, engagement, and user characteristics.
Methods
Hashtag search
This study did not require institutional review board approval. A retrospective search from 20 March 2012 to 31 January 2018 using the Hashtag Finder application (Symplur, Upland, California, USA) was used to identify hashtags pertaining to stroke on Twitter. Manual searching for stroke-related content on Twitter was also used. The most commonly used stroke-specific hashtag was found to be the #Stroke hashtag with 1 098 580 tweets composed by 317 698 users over a period of 5 years, 10 months, and 11 days (20 March 2012 to 31 January 2018).
Dataset exclusion criteria
Prior to analysis, Twitter accounts flagged as spam by other users as well as Twitter accounts not categorized as the following by Symplur (472 255 tweets) were excluded: physicians, healthcare professionals, patients, caregiver/advocates, researchers/academics, journalist/media, individual other health, individual non-health, and providers, research, academic, government, advocate/support, pharmaceutical, media, other healthcare, or non-healthcare organizations. Tweets that applied the #Stroke hashtag in an inappropriate manner (4672 tweets) were also excluded. Thus, the final dataset to be analyzed contained 621 653 tweets from 72 453 users.
Twitter metrics
Impressions were calculated by multiplying the number of followers for every participant by the sum of the number of tweets; thus, the impressions count represents the number of total views for all #Stroke tweets. Tweet content was determined by analyzing the Twitter transcripts for images, keywords, and links. User biography was used to identify user characteristics, language and location. Symplur Signals was used to conduct network analysis. Other hashtags included in the tweet along with #Stroke were recorded to assess concurrent topics. Engagement markers, as defined by Twitter, include follows, retweets, replies, mentions, hashtag use, and likes. Follows, retweets, replies, mentions, hashtag use, and likes were then calculated and analyzed over time.
Statistical analysis
Symplur Signals was used to conduct Twitter analytics. Symplur is a publicly available platform which performs analysis on social media content. Calculations were performed using Excel software (Microsoft, Redmond, Washington, USA). GraphPad QuickCalcs (GraphPad Software, La Jolla, California, USA) was used to perform linear regresssion analysis.
Results
Stroke-related hashtags
The most utilized stroke-related hashtags are shown in table 1. From 20 March 2012 to 31 January 2018 the #Stroke hashtag was used in 1 098 580 tweets for an average of 13 504 tweets per month.
Top five most frequently used stroke-related hashtags on Twitter
Content
Content analysis for the #Stroke hashtag is shown in table 2. 436 132 tweets (70.2%) contained links to outside websites or scholarly journals, 181 120 tweets (29.1%) possessed at least one image, 69 371 tweets (11.2%) included novel topics, and 48 568 tweets (7.8%) were related directly with patient care. Topics which were commonly discussed throughout the #Stroke network on Twitter are also presented in table 2, with prevention, diabetes, and atrial fibrillation being commonly tweeted topics.
User identity, tweet content, and topic
Engagement
A total of 366 561 (59.0%) tweets made mentions of other users, 259 438 (41.7%) were subsequently retweeted, and 8549 (1.4%) had a response (table 3). Throughout the study period, 72 453 Twitter users composed a tweet containing the hashtag #Stroke in at least one tweet, averaging 8.6 tweets per user (table 3). The distribution of user engagement on Twitter is shown in table 3, with 57 607 users (79.5%) composing <5 tweets and 1620 (2.2%) users composing >50 tweets during the study period.
The #Stroke hashtag experienced an annual increase in users from 20 March 2012 to 31 January 2018 of 136.9%, 76.4%, 53.3%, 39.4%, and 18.5% (figure 1A). The number of tweets containing the #Stroke hashtag increased annually over the same interval by 176.0%, 108.8%, 73.2%, 57.7%, and 30.2% (figure 1B). Furthermore, the impressions increased annually by 149.2%, 121.4%, 69.6%, 57.2%, and 41.0% (figure 1C).
#Stroke hashtag engagement metrics
Quarterly cumulative users (A), tweets (B), and impressions (C) bearing the #Stroke hashtag from 20 March 2012 to 31 January 2018.
The #Stroke hashtag has experienced continued monthly growth since its inception (figure 2). Through the use of a time series linear regression, a statistically significant positive trend was found in the number of tweets bearing the #Stroke hashtag from March 2012 through January 2018 (R2=0.7646, P<0.0001). The #Stroke hashtag experienced an increased use at an average rate of 214.6 additional tweets per month (95% CI 200 to 229). Of note, tweet volume peaks corresponded to National Stroke Awareness Month during May.
Volume of tweets per month bearing the #Stroke hashtag from 20 March 2012 to 31 January 2018.
User characteristics
Tweets were mainly composed by advocacy/support organizations (21.5%; table 2). During the study period, 4351 individual advocacy/support organizations composed 133 368 tweets, averaging 30.7 tweets per organization (table 2). In comparison, physicians contributed to a smaller proportion of the total #Stroke tweets with 52 197 (8.4%) tweets composed by 9727 physicians, averaging 5.4 tweets per physician (table 2). Furthermore, 87 108 tweets (14.0%) originated from individuals not known to be directly working in in the healthcare industry, averaging 6.4 tweets per individual.
Network analysis
Network analyis of the #Stroke hashtag is shown in figure 3. The #Stroke network mainly comprised individuals not known to be directly working in the healthcare industry interacting with patients directly, particularly with Stroke Hope and Peter Coghlan, a stroke survivor and advocate. There were also many interactions between patients and advocacy/support organizations, predominately the Stroke Association and Sign Against Stroke. Additionally, there were many advocacy/support organizations which communicated directly on Twitter.
#Stroke Twitter network labeled by healthcare stakeholder. Node size is related to user influence which is directly correlated to the amount a user is mentioned in a tweet. Arrows between nodes signify communication between users and arrow size denotes the total number of tweets between users.
Associated hashtags
The hashtags included alongside tweets also containing the #Stroke hashtag were documented (figure 4). Associated hashtags served as a representation of the content themes found among the tweets bearing the #Stroke hashtag. The hashtags most used alongside the #Stroke hashtag included #health (25 441 tweets), #Afib (19 019 tweets), #heartdisease (18 963 tweets), #diabetes (15 311 tweets), and #FourOurHearts (14 859 tweets). #WorldStrokeDay and #StrokeMonth were other noteworthy hashtags related to stroke awareness campaigns.
Top 20 hashtags associated with #Stroke from 20 March 2012 to 31 January 2018.
Discussion
Twitter is classified as a microblog, arguably the most dynamic and concise form of social media.14 Most forms of microblogging social media possess character restrictions that can be supplemented with shortened hyperlinks to other digital media such as videos or websites. Specifically, Twitter limits tweets to 280 characters, six or fewer images, and videos of limited duration.
The ‘follow’ feature of Twitter allows all of a particular user’s posts to gain wider notoriety among their followers by appearing in their ‘Twitter feed'. These tweets can then garner further attention when they are reposted or ‘retweeted’ by other users, once again becoming visible to all of the followers of the reposter. This form of social media makes for a fast-paced environment in which discussions can center around specific user-generated ‘hashtags’ (eg, #heart or #Cancer). Access to current conversations, news, and information is what has allowed Twitter to adapt to over 140 applications within medicine and healthcare.15
The substantial increase and subsequent steady rise in the number of tweets, impressions, and users composing tweets containing the #Stroke hashtag reflects the worldwide trend regarding increased awareness for prompt identification of stroke and evidence-based treatment guidelines. No other stroke-related hashtag has similar rates of use, indicating that #Stroke is the predominant channel for discussions between patients, advocacy groups, and physicians. #Stroke represents an easily remembered hashtag and, given its continued success, it is likely to remain the dominant hashtag for the recognition and management of stroke disease.
The community of users participating in stroke-related discussions on Twitter is continually growing. Although the annual growth in the rate of retweets and users tweeting has plateaued during the course of the study period (figure 1), the absolute number of tweets and users indicates that the community recognizes #Stroke as a well-established conduit for stroke-related discussions. Additionally, although there appears to be an overall decrease in the volume of tweets containing the #Stroke hashtag, this appears to reflect an overall decrease in the monthly number of active users engaged in Twitter during 2017 rather than a reflection of the usage of stroke-related Twitter posts.16
Furthermore, there is a group of 1620 users (2.2%) mainly composed of advocacy/support organizations (Stroke Association, American Heart Association), individuals not known to be directly working in the healthcare industry, and organizations filling roles within the healthcare industry that are not providing direct clinical care who are extremely active within the #Stroke web, composing over 50 unique tweets with the #Stroke hashtag throughout the duration of the study. Notably, there was a similar proportion of total tweets with the #Stroke hashtag generated by physicians (8.4%) and patients (6.7%) during the study period and apparent minimal network communication between physicians and patients. Following the example of other medical specialties, this area represents an opportunity for board-certified physicians to provide credible medical information to patients in a readily accessible platform.3 4 7 An opportune time to attempt further outreach to patients would be during the annual National Stroke Awareness campaign in May, as patients are more likely to become engaged in community outreach campaigns and visit stroke center web pages during that month.10
The content of the discussions on Twitter included recognition of the signs of a stroke, associated risk factors (eg, atrial fibrillation, heart disease, and diabetes), and findings of peer-reviewed journals regarding stroke treatment. The results of the recently published DAWN and DEFUSE 3 trials have caused an increase in discussions regarding stroke treatment windows and, in turn, generated more interest in the #Stroke hashtag and discussion centered around thrombectomy.17 18 This represents an excellent example of Twitter’s benefits in encouraging learner engagement, collaboration, and professional development.19
Practically speaking, increased social media engagement surrounding stroke medicine is important because it provides patients with an opportunity to learn and recognize the signs of stroke early and subsequently reach the appropriate level of care sooner. As the temporality involved in stroke is of great importance, increasing patient awareness regarding stroke signs will hopefully increase the number of patients who can benefit from stroke interventions. Additionally, increased physician engagement in social media can encourage the sharing of innovative ideas, generate networking opportunities, and provide a conduit for professional activities including advertisement and patient recruitment into clinical trials.20
There are limitations to this study. The #Stroke tweets that were recorded and analyzed represent a subset of the entire stroke medicine Twitter network. There are additional tweets containing stroke-related content which did not contain the #Stroke hashtag that were not included in this analysis. Additionally, analysis has shown that 62% of all Twitter users are less than 49 years old. While patient demographics were not collected in this analysis, it would hold that the demographics were skewed toward younger participants.21
As social media continues to grow and become more entrenched within the medical community, the American Medical Association’s guidance for physicians regarding the ethical use of social media requires more attention.22 In particular, maintaining professionalism, respecting patient confidentiality, and respectful discourse between colleagues should be upheld as the scope of content shared continues to expand.
Conclusion
The Twitter #Stroke hashtag serves as a proxy for overall interest for stroke medicine and advocacy. Overall use of the hashtag has grown significantly over the nearly 6-year study period. Most lines of communication were between stroke support groups and non-healthcare-related individuals, with discussion content centering around stroke prevention, stroke symptoms, associated medical conditions, and treatment options.
References
Footnotes
Contributors All authors contributed equally to this manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data involved in this study is available for sharing.
Correction notice Since this article was first published online, the date January 20th 2018 has been updated to January 31st 2018 in the article text.