Our study showed that social media promotion of gastroenterology journal publications on Twitter independently predicted a greater number of Google Scholar citations in 3 years after controlling for journal impact factor, type of article, and time since publication.

We also found that tweeted articles that were retweeted had significantly increased citations, perhaps a reflection that public engagement with these tweets might predict further academic interest. Our study also demonstrated that guidelines/technical reviews and meta-analyses/systematic reviews are among the most heavily promoted article types by official journal accounts on Twitter. Articles involving motility/neurogastroenterology, cancer, education, and quality improvement were independently associated with increased social media exposure, while basic science manuscripts were less promoted. This may reflect the perceived areas of public interest in gastroenterology and influence perception of gastroenterology research, resulting in potential secondary impacts on factors, such as funding. Further studies are needed to evaluate why these promotional differences exist and how this might affect research funding availability, public interest, and clinical and epidemiological outcomes.

As done in studies from other disciplines, Twitter was chosen given that it is the predominant mainstream social media platform where journal publications are regularly promoted, where academic discussion exists, and where analytics are most readily available. The evidence regarding the association between Twitter promotion and publication citations has varied across different disciplines. One study of 20 ecological journals showed a similar positive correlation between social media exposure and citations, independent of time since publication and impact factor [7]. In the American Journal of Psychiatry, a study of 438 published articles suggested that a greater frequency of Twitter mentions was associated with more citations [8]. One study of 286 articles from the Journal of Medical Internet Research showed that highly tweeted articles were 11 times more likely to be highly cited and that Twitter mentions in the first 3 days could predict highly cited articles [6].

However, this observation was not consistent across all disciplines. In a randomized trial of social media impact on 243 publications from Circulation, there was no significant association between social media exposure and 30-day article website views according to Google analytic data [9]. There were also no differences noted by article subtype (referring to the general categories of clinical, population, or basic science articles). One study of all 1.4 million biomedical publications between 2010 and 2012 found a weak correlation between Twitter mentions of journal articles and traditional bibliometric indicators, although the authors concluded that Twitter mentions did not reflect traditional research impact [10].

The number of annual citations by a journal is used to calculate a journal’s impact factor [11]. Historically, journal impact factor has been directly correlated with citations, as impact factor is the ratio of the number of citations to the number of publications by a journal in a given year. Google Scholar was chosen as the source of citations given its inclusivity of citations among all sources, unlike Web of Science and Scopus, which only consider citations from within journals listed in their libraries. However, some researchers have raised concerns that certain tactics can be used to boost the journal impact factor, such as publishing more review articles, strategic publication timing, and internal citations. Citations also take time to accumulate. Social media may therefore play another role in measuring the societal impact of an individual article. As a result, other metrics collectively known as “altmetrics,” which include not only number of citations, but also social media views and engagement, references by databases, and news media, are emerging [12].

In our study, guidelines and reviews published in major gastroenterology journals were the most heavily promoted types of manuscripts on social media. This promotional pattern may be a result of known interest in these articles, given existing knowledge that guidelines and reviews accrue more citations than other publication types, and they may be more relevant to a broader audience that includes general clinicians and practitioners [13]. For gastrointestinal subtopics, our finding of functional bowel disease and cancer articles as the most tweeted is also not surprising. Communities for these conditions are very active on social media, and journals may be preferentially promoting these articles knowing the discussion and engagement they tend to generate. Moreover, the high general prevalence of functional bowel conditions and the increased interest cancer-related topics tend to generate in the lay media may also play roles in the higher social media activities of these articles.

Social media engagement by the gastroenterology community is lagging behind other medical subspecialties. Adoption of Twitter by journals still varies widely by specialty, with 70% of the 20 leading radiology journals on Twitter, but only 28.1% of general medical journals and 39% of urological journals on Twitter [2,14]. Latest data suggest that the proportion of overall tweeted gastroenterology publications (7.8%) falls below the average rate of Twitter promotion (9.4%) of all medical journal articles [10]. A survey of 265 gastroenterologists in 2015 found that 82.1% of respondents did not access social media for journals or other educational purposes, and 47.7% reported never having used any form of social media. This is in contrast to patient acceptance of social media, where 84.4% (n=112) of inflammatory bowel disease patients and 72.9% (n=68) of chronic viral hepatitis patients favored interaction with health care professionals on social media [15]. This dynamic may be improving in recent years, as there has been a growing presence of gastroenterologists and young physicians on social media.

One limitation of this study is the inability to establish causality and elucidate the potential mechanisms of social media impact on citations. Articles addressing subjects of higher interest or “more popular” topics (therefore, more highly cited) may be selectively tweeted at a higher frequency. For instance, social media promotion has been associated with greater downloads of journal articles in clinical pain sciences, which could suggest greater readership as a catalyst for future citations [16]. The radiology community has observed a greater distribution of web links to published articles on Twitter [14]. In fact, the effect of social media on journal article readership has been previously demonstrated in a study of the Journal of the American College of Radiology, which showed that a planned Twitter-based discussion increased monthly website journal article views by 31.4%, unique visitors by 20.0%, and website visits by 25.5% [17]. In a randomized controlled study of the same journal, Twitter promotion of publications was independently associated with significantly greater weekly webpage visits (18.2 vs 7.6 page visits) [17]. Another study showed that 7 out of 11 radiology journals new to Twitter experienced increases in impact factor after 1 year [14].

Moreover, it is unknown if social media alone can be credited for greater citations or if social media is simply one arm of a larger promotional effort that includes other avenues, such as press releases for traditional media coverage. It is worth noting, however, that because the initial promotional tweet precedes article downloads and subsequent citations, it is not possible for the number of citations to cause a reactionary increase in social media presence for that particular article. Additionally, various journals may adopt different methods to selectively promote certain subtopics or manuscript types (such as guidelines) on social media. To best account for these biases, we controlled for subtopics and manuscript types in the multivariate analysis. Another potential limitation was the potential overestimation of baseline academic impact using Google Scholar citations. Though Google Scholar may include duplicate citations or citations of a paper in a nonpeer-reviewed publication without scholarly relevance, the Google algorithm is standard and therefore objectively compares citation volume.

In conclusion, social media promotion on Twitter of gastroenterology publications independently predicted a greater number of citations in 3 years in this exploratory study. Publications and researchers should consider wider adoption of social media to increase reach and measure uptake of published research. Social media promotion of publications can not only potentially boost journal citations, but also help define the societal impact of an individual article and thus influence academic promotion. However, beyond citations and academic uptake, journals and physicians should be aware of other benefits of social media for professionals, patients, and family members. In addition to academic productivity, it is important to recognize how social media could have other important roles in public health. For instance, social media may help propel academic medicine by informing other professionals. Moreover, it is our public health responsibility as a medical community to serve as primary sources of accurate up-to-date medical information online in order to preserve the integrity of what readers consume. Internally, social media can provide an open forum for discussion, boost professional and institutional recognition, attract referrals for trial enrollment and research purposes, and perhaps encourage funding to sustain academic research.