*Center for Healthcare Outcomes and Policy, University of Michigan, Ann Arbor, MI†Department of Surgery, Massachusetts General Hospital, Boston, MA‡Department of Surgery, Washington University School of Medicine, St. Louis, MO.
Checking for direct PDF access through Ovid
Nearly all major academic research journals have adopted social media platforms, such as Twitter, to disseminate their publications and make them more accessible to readers.1 One recent study suggested that articles featured on Twitter may be 3 times more likely to be read versus those that were not.2 Despite the widespread adoption of Twitter by academic journals, the extent to which the social media platforms and strategies can influence practical outcomes, such as number of article reads, remain understudied.In July of 2016, Annals of Surgery adopted the use of “visual abstracts” as a novel strategy to improve dissemination of the journal's publications. A visual abstract is simply a visual representation of the key findings typically found in the abstract portion of an article. They are produced by the journal after an article is accepted. Examples can be found in Figure 1. As of March 2017, more than 15 journals have utilized visual abstracts in their social media dissemination strategy,3 yet no data exist describing how their use impacts dissemination of publications.In this context, a case-control crossover study was conducted to compare tweets that included only the title of the article versus tweets that contain the title and a visual abstract. Such information would be valuable to help journals and authors understand the impact of different dissemination strategies for their publications.METHODSBetween July 2016 and December of 2016, a prospective case-control crossover study was performed using 44 original research articles published that same year in the Annals of Surgery. Each article was tweeted from the Annals of Surgery Twitter account in 2 formats; as the title of the article only and as the title with a visual abstract. Half (n = 22) of the articles were tweeted as title alone, then after a 4-week “washout” period, the same article was also tweeted as a visual abstract. The other half of the articles (n = 22) were tweeted on the same protocol, but in the opposite order. Thus, by the end of the study period, all 44 articles were tweeted using the both formats allowing for a matched-pair t test analysis (Fig. 2).The primary outcomes of this study were (i) number of times the tweet was seen (impressions), (ii) the number of times the tweet was shared (retweets), and (iii) the number of times the article link was clicked on (article visits) that were tracked prospectively using Twitter Analytics.As a secondary outcome, we wanted to assess if nonvisual abstract tweets for articles besides the 44 original contributions benefited from a “spill over” effect after visual abstracts were implemented. To do so, we performed an interrupted time series in the six months before and after visual abstracts were implemented and compared average number of impressions per tweet between the 2 groups. All analyses were 2 tailed, using 0.05 as the threshold for significance.RESULTSWe found a strong correlation between the use of visual abstract tweets and increased dissemination on social media (Table 1, Fig. 3). When article titles were tweeted, each tweet averaged 3073.3 impressions and 11.0 retweets. However, when the same articles were tweeted as a visual abstract, each tweet averaged 23,611 impressions (7.7-fold increase; P < 0.001) and 92 retweets (8.4-fold increase; P < 0.001). Similarly, tweets with title only resulted in average of 65.