key: cord-0689989-blrp0gdn authors: Hamamsy, T. C.; Bonneau, R. title: Twitter activity about treatments during the COVID-19 pandemic: case studies of remdesivir, hydroxychloroquine, and convalescent plasma. date: 2020-06-20 journal: nan DOI: 10.1101/2020.06.18.20134668 sha: 945c07fab35dd9e7ca3462d92ea84977f4f3add5 doc_id: 689989 cord_uid: blrp0gdn Since the COVID-19 pandemic started, the public has been eager for news about promising treatments, and social media has played a large role in information dissemination. In this paper, our objectives are to characterize the public discussion of treatments on Twitter, and demonstrate the utility of these discussions for public health surveillance. We pulled tweets related to three promising COVID-19 treatments (hydroxychloroquine, remdevisir and convalescent plasma), between the dates of February 28th and May 22nd using the Twitter public API. We characterize treatment tweet trends over this time period. Most major tweet/retweet/sentiment trends correlated to public announcement made by the white house and/or to new clinical trial evidence about treatments. Most of the websites people shared in treatment-related tweets were non-scientific media sources that leaned conservative. Hydroxychloroquine was the most discussed treatment on Twitter, and over 10% of hydroxychloroquine tweets mentioned an adverse drug reaction. There is a gap between the public attention/discussion around COVID-19 treatments and their evidence. Twitter data can and should be used public health surveillance during this pandemic, as it is informative for monitoring adverse drug reactions, especially as many people avoid going to hospitals/doctors. Since the COVID-19 pandemic started, the public has been eager for news about promising treatments. In mid-March, treatments became a focus of the White House Coronavirus briefings, with hydroxychloroquine taking center stage. In reaction to these communications, people panicbought hydroxychloroquine, and in March over 40,000 health care professionals became firsttime prescribers of the drug, with the prescribing rate in New York increasing over 40 times. [1] In March, the success of most treatments was speculative, but as of May 5th, there were 1209 COVID-19 clinical trials underway. Social media discussions about treatments during a pandemic are relatively new territory, [2] [3] and with new clinical evidence constantly coming out, scientific information gets amplified to the public. While Twitter trends during the COVID-19 pandemic have been well characterized by several studies, [4] [5] [6] the extent of social media activity related to COVID-19 treatments has not been well described. In this study, we characterize trends on Twitter (San Francisco, CA), that reflect discussion of COVID-19 treatments, we show how they temporally relate to White House communications, new evidence about treatments, and also illustrate their utility for public health surveillance. We used Twitter's public API, and the R package 'rtweet' to pull tweets mentioning several promising treatments with clinical trials underway for COVID-19, focusing on hydroxychloroquine, remdevisir and convalescent plasma. [7] We retrospectively analyzed posts made between February 28th and May 22nd, 2020, on Twitter. We calculated the number of tweets and retweets made per day mentioning each treatment. We performed sentiment analysis on these tweets using the AFFIN lexicon, and calculated the average sentiment per day for each . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134668 doi: medRxiv preprint treatment. We analyzed the urls mentioned in these tweets, counting how often different websites were shared over this time period. We searched these urls for mentions of journals/preprints, and calculated the number of retweets for each treatment that linked to research articles. We screened all hydroxychloroquine tweets for mentions of adverse drug reactions (ADRs), using the MEDRA ontology for low-level ADRs, and the spaCy NLP library for term matching. We visualized all of these data alongside major treatment/clinical trial/government announcements. Among 1209 clinical trials, 133 interventions included hydroxychloroquine, 45 convalescent plasma, and 13 remdesivir. Between February 28th and May 22nd, we found 441,052 tweets containing "hydroxychloroquine" or "hcq", 60,050 containing "remdevisir", and 63,958 containing "convalescent" or "plasma" along with a treatment-related phrase. The three hydroxychloroquine peaks appeared after Trump promoted hydroxychloroquine on March 21 st , April 5 th , and May 18 th , and 31% of "hydroxychloroquine" tweets mentioned Trump. Our analysis of the links included in tweets mentioning hydroxychloroquine, showed that the top sites were mostly conservative media websites. We found 72 tweets made by congress members about treatments, and their sentiments about hydroxychloroquine were divided along party lines: average sentiment was negative for Democrats (-.41) and positive for Republicans (.88). We found 68,934 mentions of ADRs in treatment tweets, including 61,144 in hydroxychloroquine tweets (covering 10.4% of hydroxychloroquine tweets). . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134668 doi: medRxiv preprint Health misinformation during this pandemic about treatments, vaccines, and prevention strategies, have been described as an infodemic. [8] Social media is playing a crucial role in information dissemination, but it has been a double-edged sword. As Figure 2 illustrates, most information about treatments reaching the public on social media comes from non-scientific sources, and frequently these sources spin scientific results. As an example, before evidence of the efficacy of hydroxychloroquine for treating COVID-19 came out, the white house advertised it, doctors over prescribed it, and governments stockpiled it. We've seen from the rapid pace of preprint articles, and premature publishing of the Surgisphere papers,[9] that science during a pandemic is messy. As evidence about treatments for COVID-19 comes out, scientific communication of the evidence, not extrapolation or speculation, will be essential. No ethical approval was needed for this cohort study that uses publicly accessible Twitter data. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 20, 2020. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134668 doi: medRxiv preprint websites are right-wing websites, including "thegatewaypundit", "foxnews", and "breitbart". There were 620 different ADRs mentioned in hydroxychloroquine -related tweets. Given the mass promotion of this drug, and a public that is fearful of visiting hospitals during a pandemic, social media mentions of ADRs can inform public health surveillance. . CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134668 doi: medRxiv preprint Hydroxychloroquine Prescription Volume Changes During COVID-19 How did Ebola information spread on twitter: broadcasting or viral spreading? Pandemics in the age of Twitter: content analysis of Tweets during the 2009 H1N1 outbreak Conversations and Medical News Frames on Twitter: Infodemiological Study on COVID-19 in South Korea Top Concerns of Tweeters During the COVID-19 Pandemic: Infoveillance Study Tracking Social Media Discourse About the COVID-19 Pandemic: Development of a Public Coronavirus Twitter Data Set. JMIR Public Health and Surveillance Collecting and analyzing Twitter data How to fight an infodemic. The Lancet The authors have no conflicts of interest to report. RB and TH acknowledge support from the following sources: NIH R01DK103358, Simons Foundation, NSF-IOS-1546218, R35GM122515, NSF CBET-1728858, NIH R01AI130945.. CC-BY-NC 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 20, 2020. . https://doi.org/10.1101/2020.06.18.20134668 doi: medRxiv preprint