ACC.20: Impact of social media at the virtual scientific sessions during the COVID‐19 pandemic

Abstract Background The COVID‐19 pandemic led to the American College of Cardiology (ACC) Annual Scientific Session 2020 (ACC.20) being held as a virtual event. Hypothesis Social media activity around a virtual event might be quite different to that of a physical meeting. The goal of this study was to assess impact of ACC.20 through Twitter and compare it to ACC.19. Methods Data were extracted using NodeXL, with analysis in Excel. Results ACC.20‐related tweeting was demonstrated globally. However tweeting and participants fell substantially for ACC.20. Tweeting, participation and tweet views were overestimated by the most widely used social media analysis tool used at medical conferences (Symplur). Conclusion Comparing the 2019 and 2020 Scientific Sessions, the global cardiology community continued to communicate despite COVID‐19, but with reduced social media activity potentially due to the briefer format, no physical interaction and private virtual chatroom during live sessions, reducing visibility of new cardiology research findings.


| INTRODUCTION
The COVID-19 pandemic has meant that healthcare workers and with the same ACC acronym. 1 Social media activity around a virtual event might be expected to be quite different to that around a physical meeting. The analysis presented here looks at social media activity related to ACC.20/WCC and compares this with ACC.19.

| METHODS
Data were extracted using NodeXL, 2 with analysis in Excel, as described in the ACC.19 paper. 1 The hashtags used were the official hashtags for the virtual event (#ACC20, #WCCardio), and some tweeters used #ACC2020. A further 77 tweets were identified using the term "ACC.20" without hashtags, but these were not included in this analysis as the comparable data were not available for ACC.19.
ACC.19 data were re-analyzed to allow direct comparison with the ACC.20 data, focussing on the day before the events, the 3 days of the conference, and the day after the events (15-19 March 2019 and 27-31 March 2020, coordinated universal time). Data on individual retweets were used rather than the aggregate count of retweets recorded in tweets, to allow a direct comparison of activity over the 5-day period rather than retweets accumulated after this period (see Supporting Information Data S1). Preliminary data and findings and other mapping data were shared in a tweet thread, including the global spread of ACC.20 related tweets and retweets. 3 A summary of tweets posted during the 3 days of the virtual event was shared via a Wakelet summary. 4 The most widely disseminated tweets were viewed and one tweet with a large number of interactions selected 5 for further mapping to document the branching structure of replies and quoting tweets (collectively referred to as "responses" in this paper). These responses were identified by viewing and expanding replies in Twitter in an internet browser, recording the tweet URL and searching for quoting tweets by copying the tweet URL into the Twitter search box after removing the leading "https://www.", repeating until no further responses were found. 6 The 19 character tweet IDs for individual responses were then mapped using NodeXL, which also collects information about tweeters and numbers of retweets obtained for individual tweets. 7 Finally, the number of tweets was recorded using the Symplur healthcare hashtags website, as this is a tool commonly used and shared in conference tweeting to track headline statistics for single hashtags. 8 Symplur provides estimates of "tweets" (which they calculate by adding tweets and retweets), and "tweeters" (calculated by adding tweeters and retweeters) for single, registered hashtags.
Symplur also provides estimates of "impressions" (number of times a tweet has been displayed on a Twitter-enabled device) by adding the number of tweets and retweets made by a tweeter and multiplying by total number of followers for that tweeter. Looking at #ACC20 by itself, and focusing on the American College of Cardiology (@ACCinTouch) Twitter account, Symplur data were compared with NodeXL data (separate records of tweets, retweets, tweeters and retweeters) and Twitter Analytics (a record of impressions direct from Twitter), 9 and the ACCinTouch Twitter feed (providing a record of retweets). 10        there was evidence of "hashtag drift," with a greater proportion of tweeters using the incorrect hashtag (#ACC2020) in 2020 than 2019.

| RESULTS
Each of these factors reduce opportunities for effective dissemination of scientific findings from a flagship medical event. In an era when "fake news" can trump evidence-based approaches it is important to maximize sharing and discussion of research and progress in the field of cardiology.
There was no "hashtag confusion" evident, with no other events using the #ACC20 hashtag, but this reflects the impact of the COVID-19 pandemic rather than an active change in hashtag use by other events. In future years the hashtag should be adapted to include a specific cardiology term. One option would be #ACCardio21, ideally matching the name of the conference to the hashtag.
The mapping of responses to Dr Tang's tweet illustrates the very con-