Using Twitter to Track Unplanned School Closures: Georgia Public Schools, 2015-17.

OBJECTIVES: To aid emergency response, Centers for Disease Control and Prevention (CDC) researchers monitor unplanned school closures (USCs) by conducting online systematic searches (OSS) to identify relevant publicly available reports. We examined the added utility of analyzing Twitter data to improve USC monitoring. METHODS: Georgia public school data were obtained from the National Center for Education Statistics. We identified school and district Twitter accounts with 1 or more tweets ever posted ("active"), and their USC-related tweets in the 2015-16 and 2016-17 school years. CDC researchers provided OSS-identified USC reports. Descriptive statistics, univariate, and multivariable logistic regression were computed. RESULTS: A majority (1,864/2,299) of Georgia public schools had, or were in a district with, active Twitter accounts in 2017. Among these schools, 638 were identified with USCs in 2015-16 (Twitter only, 222; OSS only, 2015; both, 201) and 981 in 2016-17 (Twitter only, 178; OSS only, 107; both, 696). The marginal benefit of adding Twitter as a data source was an increase in the number of schools identified with USCs by 53% (222/416) in 2015-16 and 22% (178/803) in 2016-17. CONCLUSIONS: Policy-makers may wish to consider the potential value of incorporating Twitter into existing USC monitoring systems.

Public Health Implications of Image-Based Social Media: A Systematic Review of Instagram, Pinterest, Tumblr, and Flickr.

INTRODUCTION: Image-based social media Instagram, Pinterest, Tumblr, and Flickr have become sources of health-related information and tools for health communication. No known systematic review exists that summarizes the existing research and its health implications. METHODS: We searched EBSCOhost Academic Search Complete, PubMed, and Web of Science in January 2016, April 2017, and November 2017, with the following keywords: Pinterest, Instagram, Tumblr, or Flickr. Only public health-related, peer-reviewed research articles were included (excluding surveys of self-reported social media use). RESULTS: Sixty-six research articles were included. All 12 experimental studies used Instagram. Of the 54 observational studies, 38 studied Instagram; 12, Pinterest; 7, Tumblr; and 6, Flickr; some studies investigated more than 1 platform. More than half of the included articles were related to mental health and substance abuse (n = 39, 59%). Other topics included chronic diseases, infectious diseases, surgery, injury prevention, pharmacovigilance, sexual and reproductive health, and adolescent psychology with functional magnetic resonance imaging. DISCUSSION: Public health-oriented research on Instagram, Pinterest, Tumblr, and Flickr is increasing. Most observational studies investigated questions ranging from public perception of diseases or interventions (vaccination) and undesirable media exposure ("echo chamber," distorted body image, underage substance use, and pro-suicide messages) to information dissemination and online engagement (likes and comments). A few studies attempted to use image-based social media as intervention tools, but the results were mainly exploratory. CONCLUSION: More research is needed to study how social media users discuss health through sharing images online, and how public health professionals can communicate effectively using image-sharing social media.

Twitter Conversations and English News Media Reports on Poliomyelitis in Five Different Countries, January 2014 to April 2015.

INTRODUCTION: Twitter and media coverage on poliomyelitis help maintain global support for its eradication. OBJECTIVE: To test our hypothesis that themes of polio-related tweets and media articles would differ by location of interest (hashtag of country name mentioned in the tweet; country name mentioned in media articles) but would be similar to each other (tweets and media articles) for each location of interest. METHODS: We retrospectively examined a 40% random sample of Twitter data containing the hashtag #polio from January 1, 2014, to April 30, 2015 (N = 79,333), from which we extracted 5 subcorpora each with a co-occurring hashtag #India (n = 5027), #Iraq (n = 1238), #Nigeria (n = 1364), #Pakistan (n = 11,427), and #Syria (n = 2952). We also retrieved and categorized 73 polio-related English-language news stories from within the same timeframe. We assessed the association between polio-related English news themes and the Twitter content. Descriptive analyses and unsupervised machine learning (latent Dirichlet allocation modeling) were conducted on the 5 Twitter subcorpora. RESULTS: The results of the latent Dirichlet allocation modeling on the specific subcorpora with country co-occurring hashtags showed significant differences between the 5 countries in terms of content. English mass media content focused largely on violence/conflicts and cases of polio, whereas social media focused on eradication and vaccination efforts along with celebrations. DISCUSSION: Contrary to our hypothesis, our evidence suggests Twitter content differs significantly from English mass media content. Evidence from our study helps inform media monitoring and communications surveillance during global public health crises, such as infectious disease outbreaks, as well as reactions to health promotion campaigns.

#CDCGrandRounds and #VitalSigns: A Twitter Analysis.

BACKGROUND: The CDC hosts monthly panel presentations titled 'Public Health Grand Rounds' and publishes monthly reports known as Vital Signs. Hashtags #CDCGrandRounds and #VitalSigns were used to promote them on Twitter. Objectives: This study quantified the effect of hashtag count, mention count, and URL count and attaching visual cues to #CDCGrandRounds or #VitalSigns tweets on their retweet frequency. METHODS: Through Twitter Search Application Programming Interface, original tweets containing the hashtag #CDCGrandRounds (n = 6,966; April 21, 2011-October 25, 2016) and the hashtag #VitalSigns (n = 15,015; March 19, 2013-October 31, 2016) were retrieved respectively. Negative binomial regression models were applied to each corpus to estimate the associations between retweet frequency and three predictors (hashtag count, mention count, and URL link count). Each corpus was sub-set into cycles (#CDCGrandRounds: n = 58, #VitalSigns: n = 42). We manually coded the 30 tweets with the highest number of retweets for each cycle, whether it contained visual cues (images or videos). Univariable negative binomial regression models were applied to compute the prevalence ratio (PR) of retweet frequency for each cycle, between tweets with and without visual cues. FINDINGS: URL links increased retweet frequency in both corpora; effects of hashtag count and mention count differed between the two corpora. Of the 58 #CDCGrandRounds cycles, 29 were found to have statistically significantly different retweet frequencies between tweets with and without visual cues. Of these 29 cycles, one had a PR estimate < 1; twenty-four, PR > 1 but < 3; and four, PR > 3. Of the 42 #VitalSigns cycles, 19 were statistically significant. Of these 19 cycles, six were PR > 1 and < 3; and thirteen, PR > 3. Conclusions: The increase of retweet frequency through attaching visual cues varied across cycles for original tweets with #CDCGrandRounds and #VitalSigns. Future research is needed to determine the optimal choice of visual cues to maximize the influence of public health tweets.

#Globalhealth Twitter Conversations on #Malaria, #HIV, #TB, #NCDS, and #NTDS: a Cross-Sectional Analysis.

BACKGROUND: Advocates use the hashtag #GlobalHealth on Twitter to draw users' attention to prominent themes on global health, to harness their support, and to advocate for change. OBJECTIVES: We aimed to describe #GlobalHealth tweets pertinent to given major health issues. METHODS: Tweets containing the hashtag #GlobalHealth (N = 157,951) from January 1, 2014, to April 30, 2015, were purchased from GNIP Inc. We extracted 5 subcorpora of tweets, each with 1 of 5 co-occurring disease-specific hashtags (#Malaria, #HIV, #TB, #NCDS, and #NTDS) for further analysis. Unsupervised machine learning was applied to each subcorpus to categorize the tweets by their underlying topics and obtain the representative tweets of each topic. The topics were grouped into 1 of 4 themes (advocacy; epidemiological information; prevention, control, and treatment; societal impact) or miscellaneous. Manual categorization of most frequent users was performed. Time zones of users were analyzed. FINDINGS: In the entire #GlobalHealth corpus (N = 157,951), there were 40,266 unique users, 85,168 retweets, and 13,107 unique co-occurring hashtags. Of the 13,087 tweets across the 5 subcorpora with co-occurring hashtag #malaria (n = 3640), #HIV (n = 3557), #NCDS (noncommunicable diseases; n = 2373), #TB (tuberculosis; n = 1781), and #NTDS (neglected tropical diseases; n = 1736), the most prevalent theme was prevention, control, and treatment (4339, 33.16%), followed by advocacy (3706, 28.32%), epidemiological information (1803, 13.78%), and societal impact (1617, 12.36%). Among the top 10 users who tweeted the highest number of tweets in the #GlobalHealth corpus, 5 were individual professionals, 3 were news media, and 2 were organizations advocating for global health. The most common users' time zone was Eastern Time (United States and Canada). CONCLUSIONS: This study highlighted the specific #GlobalHealth Twitter conversations pertinent to malaria, HIV, tuberculosis, noncommunicable diseases, and neglected tropical diseases. These conversations reflect the priorities of advocates, funders, policymakers, and practitioners of global health on these high-burden diseases as they presented their views and information on Twitter to their followers.