Article Text
PDF
Abstract
The use of social media is pervasive throughout society and serves many purposes. Traditional forms of advertising are being upended as vendors recognize the unique abilities of social media platforms to target their messages to specific customers. Peer reviewed medical and professional journals are beginning to develop their own initiatives using social media to advertize unique content. We present the nascent Journal of NeuroInterventional Surgery experience.
- Economics
- Technology
Statistics from Altmetric.com
Introduction
Over the past decade, online social media websites have become extraordinarily popular both for private use and as a platform for marketing. Twitter and Facebook currently have more than 300 million active tweeters and 1 billion members worldwide, respectively. In response to the increasing numbers of online users, a variety of peer reviewed medical journals are expanding their social media presence in an attempt to better engage their readers.1–3 However, many have yet to join this endeavor: one recent study demonstrated that only 28% of internal medicine journals had Twitter accounts.3
The Journal of NeuroInterventional Surgery (JNIS), published by the BMJ Publishing Group, was founded in 2009 and is the official journal of the Society of Neurointerventional Surgery (SNIS). Over the past few years, JNIS manuscripts, podcasts, and important articles have occasionally been advertised on the SNIS Facebook page, JNIS Facebook page, or on the JNIS Twitter feed. Early on, this was done haphazardly and lacked a specific organizational approach. In 2014, the editorial board of JNIS decided to create an assistant editor of social media position (AESM) to expand the online social media presence of JNIS. This report details the efforts and results of these endeavors to date.
Methods
Interventions
This study does not require institutional review board approval and there are no HIPAA (Health Insurance Portability and Accountability Act) or privacy implications.
In February 2015, the JNIS AESM began sending text tweets on the JNIS Twitter account to its followers (approximately 700 people) regarding all online first publications. All articles accepted to JNIS are published online first (e-published ahead of print). Within days of each new article being published online, the AESM would tweet advertizing the new article (figure 1). Each tweet would reference the first author and briefly discuss the manuscript's content. Further, each tweet was accompanied by a link generated by the Bitly website that directs the user to the JNIS website page containing the abstract. Bitly therefore serves two purposes: first, to shorten the link considerably to allow for inclusion in the tweet; and second, provides the AESM with the number and time course of user clicks on the link provided to allow for future analysis. After each tweet was generated, the tweet was also posted on the SNIS Facebook page. Text tweets and SNIS Facebook postings were generated in this manner for all online first articles from February 2015 to the end of July 2015.
Example of a text tweet for an online first article.
Starting August 1, 2015, three additional interventions were initiated by the AESM. First, all online first articles containing figures were tweeted with one of its figures attached. If articles were published with more than one figure, the AESM would choose what they believed to be the most interesting image to include with the tweet (figure 2). Second, all tweets were additionally posted on the JNIS Facebook page as well as the SNIS Facebook page. Third, after each tweet was generated, the AESM sent an email to the corresponding author of the article describing the JNIS social media postings, with an included image of the tweet that was sent, to advertise the published article. Therefore, between August 1, 2015 and December 1, 2015, all new JNIS publications were tweeted with images (if applicable), tweets were posted on SNIS and JNIS Facebook pages, and emails were sent to corresponding authors alerting them to the online postings.
Example of an online first article tweet containing an attached image.
Data
Data were gathered for the December 2015 editor's meeting to identify the impact of social media on journal readership by reviewing the number of clicks on Bitly links per article. The primary outcome measure was defined as the number of link clicks, which directly represents the number of readers accessing the article abstract web pages via the AESM postings (readers accessing the articles through non-AESM initiated links would not have arrived at the article via the Bitly link but instead by alternative means). Articles were categorized based on subject matter (aneurysm, stroke, arteriovenous shunt, neuroradiology, carotid disease, socioeconomic, or other), article methodology (randomized controlled trial, prospective/retrospective series, case report or series (≤5 patients), animal study, editorial/commentary, or meta-analysis/systematic review), and by whether they described first experience with a new endovascular device. Articles were further characterized by those with and without associated images, and by the presence of an email to the corresponding author. As both the addition of images and emails occurred at the same time (but emails have been sent to all corresponding authors while not all manuscripts contained figures), the presence of an email best represents the factor demonstrating the impact from the interventions performed after August 2015.
Twitter analytics
The Twitter analytics webpage for the JNIS profile and number of tweets, tweet impressions, profile visits, mentions, and new followers for each month during 2015 were collected.
Statistical analysis
To assess the association between number of Bitly clicks with the social media interventions and other covariates, linear models were constructed. To determine what covariates were associated with the number of Bitly clicks, independent of the social media interventions, the main effect of each covariate was tested. All covariates significantly associated with number of Bitly clicks, independent of the presence of an email, were then entered into a full model where a backward stepwise approach was taken to construct a final model, identifying all covariates independently associated with number of Bitly clicks and formally testing the effect of social media interventions.
A level of ≤0.05 was considered to be statistically significant. To ensure adequate model fit, the response, number of Bitly clicks, was log transformed. Model assumptions were checked for each model to ensure an adequate model fit. All testing was performed using SAS V.9.4.
Results
Effect of interventions
Between February and December 1, 2015, a total of 191 online first article tweets were generated by the AESM. Overall, these 191 postings resulted in a total of 1521 article accessions via Bitly link clicks with a mean of 8.0 Bitly clicks ±7.6 per tweet. Moreover, prior to the initiation of social media interventions (February to 31 July), there were 120 tweets with a total Bitly click count of 852, resulting in a mean of 7.1 Bitly clicks ±7.1 per tweet; after the initiation of three social media interventions (August 1, to December 1; the intervention included inclusion of images, emails to corresponding authors, and posting on JNIS Facebook page), there were 71 tweets with a total Bitly click count of 669, resulting in a mean of 9.4 Bitly clicks ±8.4 per tweet. In the final model, containing day of the week (0.014), there was a significant difference before and after the initiation of social media interventions (p=0.025), where there was an average of 1.3 Bitly clicks per tweet more after the initiation of social media interventions (figure 3).
Number of clicks per article prior to (left) and after (right) initiation of the three social media interventions.
Article subject
There was no significant difference between number of Bitly clicks and article subject (p=0.47). Of article subjects, ischemic stroke had the highest click rate (47 articles; mean 10.3, range 0–47, SD 10.3), while neuroradiology had the lowest (17 articles; mean 5.1, range 0–9, SD 2.6).
Study type
There was no significant difference between number of Bitly clicks and study type (p=0.92). Of study types, animal studies had the highest click rate (9 articles; mean 9.1, range 1–25, SD 8.6) and randomized trials had the lowest (2 articles; mean 6.5, range 6–7, SD 0.7).
New device
There was no significant difference between number of Bitly clicks on those articles featuring a new device versus those that did not (p=0.12). Articles featuring new devices had a non-significant but slightly higher number of clicks (18 articles; mean 9.7, range 0–32, SD 8.3) compared with those articles that did not (173 articles; mean 7.8, range 0–47, SD 8.3).
Number of tweets on day of post
There was no significant association between number of Bitly clicks and the number of tweets on the day of posting (p=0.78).
Day of the week of post
The day of the week the tweet was generated was significantly associated with number of Bitly clicks (p=0.014) (table 1). Saturday tweets resulted in the highest number of clicks (15 articles; mean 10.1, range 2–23, SD 6.9), while Monday tweets resulted in the lowest number of clicks (36 articles; mean 5.4, range 3–15, SD 3.5).
Number of link clicks based on day of the week posted. Saturday and Wednesday posts resulted in the highest numbers of clicks, while Monday posts resulted in the lowest
Discussion
Over a 10 month period, the AESM initiatives to increase the social media presence for JNIS resulted in over 1500 additional article website accessions. Neurointerventional surgery is a small specialty. Its major organization's (SNIS) total membership is just over 800 physicians; 1500 additional accessions represent a substantial increase in online traffic to JNIS scientific content. These numbers strongly support the use of regular social media posting to expand journal readership. Further, additional interventions to advertise for newly published articles, including email notifications for corresponding authors, the use of imaging with tweets, and expanding posts to all available major social media platforms, significantly increased the number of article accessions. These interventions are further supported by the Twitter analytics data, which suggests a substantial increase in profile visits, mentions, and new followers with expanded social media efforts. We have separately analyzed the impact of social media on the JNIS podcast platform (accepted to JNIS for publication: neurintsurg-2015-012170.R1-JNIS Podcasts: The early part of our journey).
Nearly two-thirds of physicians use some form of social media for professional purposes, but the vast majority (∼90%) participate on these venues for private or personal reasons.4 ,5 Twitter is one service that is particularly suited to the peer reviewed medical journal, as it requires messaging to be brief: tweets are limited to 140 characters, or 116 characters if an image is attached. Twitter, therefore, allows for concise messaging that mandates a given post be shortened to one or two major principles, allowing physicians to rapidly peruse topics of interest. Although the number of journals with Twitter accounts remains low (less than one-third of general medical journals in one study3), journals actively expanding their social media presence have reported significantly increased online website traffic1 as was seen with JNIS in the present study. Additionally, one ecological study demonstrated a positive correlation between the number of Twitter followers for a given journal and its impact factor and article citations.3 In fact, highly tweeted articles at the time of publication can predict highly cited articles.6 These reports suggest that journals may not only increase their online traffic and followers by expanding their social media presence, but may additionally increase article citations and impact factor through these platforms.
Aside from the obvious benefits of increased exposure to scientific content, social media sites allow for relatively simple and easy quantification of its value. The number of followers, ‘likes’, ‘retweets’, ‘mentions’, and comments are readily accessible via Twitter. Further, other metrics, such as with the aid of Bitly or other link shortening services, may allow for direct quantification of the impact of social media on article accessions. Given the ease through which quantifiable data are obtained, journal editors may be able to study the impact of subject matter, prose, images, or other factors on its readers’ responses to its posts. For instance, we noted that tweets posted on Mondays had drastically lower click rates than those posted on other days. In response to this phenomenon, we are now avoiding Mondays and preferentially tweeting on Wednesday and Saturdays (the 2 days with the highest click rate). Such information allows journals to continue to optimize their social media efforts in the complex and ever changing realm of online possibilities.
There are limitations to the present study. We preferentially analyzed the link click rate as a means of understanding the number of article accessions from AESM efforts, as opposed to other metrics. The data collected were only over a 10 month time period, and represents only 191 tweets. The three interventions were initiated simultaneously and therefore could not be individually studied to determine effect. Further analyses of these interventions will be performed as we gain greater experience using social media platforms to promote JNIS content.
Conclusions
Over a 10 month period, journal initiatives to increase the social media presence for JNIS resulted in over 1500 additional article website accessions. Additional interventions to advertise for new published articles, including email notifications for corresponding authors, the use of imaging with tweets, and expanding posts to all available major social media platforms, significantly increased the number of article accessions. These findings were further substantiated by the dramatic increase in Twitter analytics metrics, such as profile views. Expanding the social media presence with Twitter and Facebook appears to be an effective way to increase online traffic to newly published peer reviewed journal articles.
Footnotes
Contributors KMF composed the original draft. All of the other authors (AFD, MH, JAH, and RWT) had the opportunity to review and provide meaningful feedback, and helped to construct the final manuscript.
Competing interests None declared.
Provenance and peer review Not commissioned; internally peer reviewed.