Longitudinal relationship between social media activity and article citations in the journal Gastrointestinal Endoscopy.

BACKGROUND AND AIMS: Social media activity in the area of scientific journal publications has increased significantly over the last decade. Several studies have suggested that journal article citations can be positively impacted by an increase in social media activity. Preliminary data have also suggested that published articles in the field of gastroenterology and hepatology that were discussed on Twitter were independently associated with higher citation rates compared with those that were not. Gastrointestinal Endoscopy (GIE) is the most widely cited endoscopy-focused journal in the world. We aimed to assess the association of social media exposure of published articles in GIE and its impact on article citations. METHODS: Data regarding journal article publication status, number of citations per article, and social media exposure per article using Altmetric data were collected from the publisher. All original articles published in GIE from 2000 to 2016 were reviewed. Editorials, case reports, and other nonresearch correspondence were excluded. Multivariable linear regression was used to assess for independent Altmetric predictors of higher citation rates over time. Logistic regression was used to assess the impact of an article mention on various social media platforms and whether it was cited during the study period. RESULTS: A total of 2361 original research articles were evaluated; 2050 articles (86.8%) were cited at least once during the follow-up period. Mean number of citations per article was 16.8 ± 23. The average Altmetric Attention score was 2.0 ± 13. The most profound independent predictor of article citation was whether an article was tweeted (odds ratio [OR], 14.2; 95% confidence interval [CI], 8.93-22.45). Other predictors were Facebook posts (OR, 1.08; 95% CI, 1.03-1.51) and number of Mendeley readers (OR, 1.359; 95% CI, 1.28-1.45). On multivariable linear regression, number of tweeters (ß = 2.3, P = .022), F1000 reviews (ß = 5.87, P < .001), policy documents (ß = 7.6, P < .001), and number of Mendeley readers (ß = 14.21, P < .001) were significantly associated with higher citation rates of published articles. CONCLUSIONS: For original articles published in GIE, there was a strong association between social media exposure on Twitter and rates of journal article citations. To a lesser degree, Facebook posts and Mendeley readers also were associated with a higher rate of article citations. This could represent both cause and effect and may be representative of higher quality articles being more commonly mentioned on social media by third-party users. A randomized controlled study evaluating different degrees of social media exposure on individual articles could be considered to further assess for causality.