Key points

  • A total of 20,520 Twitter accounts which shared 93,776 tweets related to 23,686 articles from 91 journals were analysed.

  • The British Dental Journal had the highest number of Twitter mentions followed by the Journal of Dental Research and the Journal of the American Dental Association.

  • Children, dental caries, and periodontal disease were the hottest topics among the highly tweeted dental articles.

Introduction

In recent years, altmetrics have been growing fast.1,2 These non-traditional bibliometrics discover both the volume and nature of online attention surrounding research findings3 and act as a complement to the traditional citation-based matrices.4 The term altmetrics, short for alternative metrics, was introduced in 2010 by scientists such as Jason Priem, DarioTaraborelli, Paul Groth and Cameron Neylon, who were in charge of writing the Altmetrics manifesto.5

Altmetrics involved broad and diverse data resources, including, but not limited to, social media (Twitter, Facebook, Sina Weibo, YouTube, Vimeo, Reddit, Pinterest, Wikipedia and Google+), post-publication peer-reviews (F1000 Prime, PubPeer), policy documents, patents, mainstream news outlets, scientific blogs, and social bookmarking sites (Mendeley, CiteULike).6,7 The traditional citation-based measurements can only be provided several years after publication.8 Yet, altmetric analysis is very fast and the majority of altmetric data resources are updated on a real-time feed or daily basis.

Recently, research funders such as the Wellcome Trust and the John Templeton Foundation have been attracted to altmetrics.9 Steve Fitzmier, the director of the planning and evaluation team at the John Templeton Foundation, says:

'At the core of the Foundation's mission is a desire to both fund high-quality research and to generate greater public engagement with the research we support; while analysing metrics such as citations can be helpful to assess impact, these methods provide an incomplete picture.'10

Let us analyse a paper published in Scientific Reports in January 2017, 'Promotion of natural tooth repair by small molecule GSK3 antagonists'.11 Traditional Web of Science analysis revealed that the journal impact factor was 4.12 and the number of article citations was 14. However, this is not the whole story. The article had the highest Altmetric score (1,399) in the field of dentistry. It was discussed in 157 mainstream news outlets, five scientific blogs, 177 tweets (with an upper bound of 902,642 followers), 40 Facebook pages, ten Google+ users, one review on Publons, one recommendation in F1000 Prime, and was cited in one Wikipedia article.

Nevertheless, one of the most important features in altmetrics is analysis of Twitter mentions.12,13 Twitter is a popular worldwide news and social networking service, with more than 330 million active users. Healthcare professionals use Twitter to create virtual communities and share information with other health professionals and the public, which will have a potentially positive impact on public health.14,15 For example, the Mayo Clinic (@MayoClinic) regularly tweeted on several medical problems to communicate with its 1.9 million followers.16

Twitter is a potential tool to bridge the gap between biomedical research and health policymaking.17 It has been used effectively in medical conferences.18,19 Prestigious high impact journals are quite active in the Twittersphere. A number of tweets are associated with article citation count in medical and biological sciences.20 The number of Twitter followers is also significantly correlated with the citations and impact factor of general and internal medicine journals.21 The number of tweets an article gets in the first three days of publication is a useful tool to forecast the highly cited articles.22 In contrast, a survey in dental science showed tweets were sometimes generated by bots with no evidence of conversation and original thought.23 However, with respect to the rapidly growing pressure on academics to disseminate their knowledge and research findings on the internet, a new generation of scholars developed the so-called 'Twitter science stars'.24 To measure the over/under activity of scientists on Twitter, the Kardashian index was introduced.25

Twitter is becoming a ground breaking issue in modern medicine. A total of 2,000 health care providers are active on Twitter, with at least 300 followers and at least one tweet per day.26 Twitter is the most common social media and altmetric resource in dental science.27,28,29 Hence, in the present study, we aimed to rank dental journals according to the number of Twitter mentions related to their articles. We set out to analyse and visualise the knowledge structure and bibliometric characteristics of highly tweeted dental articles. This science mapping approach, based on network theory, allowed us to summarise the complex network of bibliometric data to reveal the structure and dynamics of scientific knowledge of the most popular dental articles on Twitter. We also tried to find the most active Twitter accounts related to the dental articles.

Methods

The data of Twitter mentions were extracted from the Altmetric database (Altmetric LLP, London, UK) via the ISSN of journals on 7 July 2018. Articles with more than 50 tweets were extracted from the Altmetric database. Bibliometric data of these articles were obtained from the Web of Science and analysed by VOSviewer 1.6.6 (Leiden University, Centre for Science and Technology Studies).30,31 This study realised the science mapping of highly tweeted articles at two levels: keyword co-occurrence and bibliographic coupling.

To find hot topics among highly tweeted dental articles, keyword co-occurrence network analysis was used.32,33 It is well-known that keywords signify the core of a research article. By way of definition, the keywords A, B, C, and D may be defined to 'co-occur' if they all appear in a specific article. One more article may encompass the keywords C, D, E, and F. Connecting these keywords creates a co-occurrence network of the six keywords (Appendix 1). The number of co-occurrences of two keywords is the number of articles in which both keywords co-occur.

Bibliographic coupling measures subject similarity relationships between articles. It is an effective tool to cluster research papers and allows us to map the intellectual structure of research fields. This phenomenon occurs when the articles A and B cite article C in their bibliographies. The 'coupling strength' is higher when more citations, the two referring article have in common.34 Considering articles as nodes and connections in coupling strength allows us to create a bibliometric coupling network (Appendix 2).35 Knowledge network analysis, based on bibliographic coupling, directly maps the recent publications based on how they are cited. In the same way, two journals, organisations, and countries are bibliographically coupled if the cumulative citations of their respective research output contain a citation to a communal article. Correspondingly, we analysed all tweets related to dental articles to find the most active Twitter accounts.

Results

A total of 20,520 Twitter accounts which shared 93,776 tweets related to 23,686 articles from 91 journals were analysed in this cross-sectional survey. The British Dental Journal (39,087) had the highest number of Twitter mentions related to dental articles, followed by the Journal of Dental Research (4,503) and the Journal of the American Dental Association (3,960) (Fig. 1). Among the tweeted articles, each article had 3.9 Twitter mentions on average. The majority of tweets were from the UK (30.4%) and the US (18.8%). The most highly tweeted dental article was 'Tooth damage in captive orcas', which had 2,268 Twitter mentions (Table 1).

Fig. 1
figure 1

Top ten journals with the highest number of Twitter mentions

Full size image
Table 1 Top ten highly tweeted dental articles
Full size table

A total of 155 articles met the inclusion threshold of more than 50 tweets. Bibliometric data of 134 highly tweeted dental articles were found on Web of Science and employed for science mapping through keyword co-occurrence and bibliometric coupling network analysis. The h-index of these articles was 21 (average citations per item:12.96). The minimum number of occurrences of a keyword was set at two to include in the study. Among 755 keywords, 124 met the threshold. Children, dental caries, and periodontal disease were the hottest topics among the highly tweeted dental articles (Fig. 2). To better show the research topics, three massively repeated inappropriate keywords, including association, health, and risk were removed from the study.

Fig. 2
figure 2

Hot topics among 134 highly tweeted dental articles. Red-yellow-green-blue colour scheme was used; red corresponds to the highest keyword co-occurrence in density and blue corresponds to the lowest density. The distance-based approach was used to create this map; which means the smaller the distance between two terms, the higher their relatedness

Full size image

Bibliometric coupling network analysis of resources showed the British Dental Journal had the highest influence on the network, followed by the Journal of the American Dental Association and Clinical Periodontology (Fig. 3). At organisation level, King's College London had the highest influence on the network, followed by the University of Leeds and University College London (Fig. 4). At country level, the UK had the highest influence on the network, followed by the US and Germany (Fig. 5). Nevertheless, among Twitter accounts, @The_BDJ had the highest number of tweets related to dental articles, followed by@Dddent2 and @gary_takacs. @TheBDA had the highest number of followers, followed by @Dddent2 and @The_BDJ (Fig. 6).

Fig. 3
figure 3

Network visualisation of bibliometric coupling among resources of highly tweeted dental articles. The network involves 15 nodes and five clusters, in which journals in the same cluster with the same colour are being commonly bibliographically coupled together

Full size image
Fig. 4
figure 4

Network visualisation of bibliometric coupling among organisations of highly tweeted dental articles. The network involves 32 nodes and nine clusters, in which organisations in the same cluster with the same colour are being commonly bibliographically coupled together

Full size image
Fig. 5
figure 5

Network visualisation of bibliometric coupling among countries of highly tweeted dental articles. The network involves 22 nodes and five clusters, in which countries in the same cluster with the same colour are being commonly bibliographically coupled together

Full size image
Fig. 6
figure 6

Top ten twitter accounts, according to the number of tweets related to dental articles

Full size image

Discussion

Today, Twitter acts as a cornerstone in spreading scholarly information and disseminating cross-disciplinary knowledge.36 To our knowledge, this is the first attempt to rank dental journals according to the number of Twitter mentions related to their articles. Notably, this exercise showed that the British Dental Journal and its Twitter account, @The_BDJ, and British universities, such as King's College London, were the most active accounts. Interestingly, @The_BDJ and its sister account @TheBDA, with 49,700 followers and 13,595 tweets, had a great social impact on dissemination of dental research findings.

Twitter is the most popular social media network among dental researchers and practitioners, and tweets are generally from the UK and US.27,28,29 Well-known high impact dental journals must pay more attention to Twitter as a newly-emerging scholarly tool. As we have emphasised previously, some high impact journals have no Twitter account and refer to the account of societies and publishers.29 This situation is the same among the general and internal medicine journals. A survey published in 2015 showed 28% of these journals had a Twitter profile.21 Dental journals are strongly recommended to be proactive on Twitter, to set up their own profile, and to promote their visibility and social impact by immediately tweeting about new articles and gaining more followers.37

A good example to demonstrate the power of Twitter is 'Tooth Damage in Captive Orcas (Orcinusorca)', which achieved an upper bound of 2,106,629 followers (Table 1). A demographic breakdown showed 94% of tweets were carried out by members of the public, not by scientists, practitioners, doctors, other healthcare professionals, or science communicators (journalists, bloggers, editors). This shows that Twitter is a good tool for public dissemination of research findings. This idea may be criticised by the concept that people may be attracted to buzzwords in the title of this article. A good example for a buzzword appearing in a title would be 'Fellatio by fruit bats prolongs copulation time',38 which was mentioned in 400 tweets, with an upper bound of 1,757,117 followers (82% being members of the public). Let us analyse the second highly tweeted dental article, 'Coronally Advanced Flap with Different Designs in the Treatment of Gingival Recession: A Comparative Controlled Randomised Clinical Trial' (Table 1), which has no buzzword in the title. It reached an upper bound of 1,549,411 followers, with 79% of tweets from members of the public.

Of more interest, alongside well-known dental journals and scientific organisations, some independent scientists are also active on Twitter. For example, Gary Takacs, @gary_takacs, with 111,000 tweets and 15,100 followers; and Héctor J. Rodriguez, @flyerLPA, with 104,000 tweets and 14,200 followers. These independent tweeters are more active than some recognised dental organisations, such as International Association for Dental Research, @IADR, with 1,200 tweets and 3,500 followers. These findings confirmed the results of a recent large-scale study, which showed a considerable number of non-academics were interested enough in research to tweet academic articles.36 Since the dissemination of scholarly information to a wider range of audiences is a growing task in the academic world, it is now more and more important for scholars to be able to write tweets that are understandable by a non-professional audience.36

Hot topic analysis showed ground breaking issues such as genomic medicine, stem cells, tissue engineering, nanotechnology, and artificial intelligence were not seen among the highly tweeted dental articles. The situation was the same among dental literature in 2017.39 Unfortunately, this is a persistent and deep-rooted problem in dentistry. Pioneering scientific discoveries and cutting-edge technologies are slowly recognised by dental scholars.40,41,42

Remarkably, the highly tweeted dental articles had an acceptable citation rate (h-index = 21, average citations per item: 12.96). This finding is compatible with previous reports, demonstrating a possible association between the number of tweets and citations.20,21 A recent survey, based on 40,000 biology articles, showed the highly tweeted articles were tweeted on the day of publication. Another positive factor associated with popularity is tweeting by respective journal accounts or accounts with a high number of followers. On the other hand, the highly cited articles which are not highly tweeted experienced delayed tweeting and were not promoted by active journal accounts.37 Similarly, the number of tweets can forecast the highly cited articles within the first three days of publication.22

The limitations of our study should be noted. Twitter mentions and number of followers seem to fluctuate more than traditional metrics over time. Also, Twitter is blocked in some rapidly growing countries, such as China and Iran. Censorship of Twitter may cause bias in the results of this study. This study also only covered dental journals with an impact factor. Some popular dental articles may be published in other prestigious scientific journals such as Scientific Reports or PLoS One.

Another weakness of this study could be gaming. Tweets can be created through fake accounts and 'robot tweeting'.23,43 Also, it is well-known that there are spam companies marketing tweets, retweets, Twitter followers, and likes.27 More recently, some anti-gaming strategies are available which analyse more than one million papers using artificial intelligence to differentiate between organic and artificial patterns of attention.44,45 Gaming is an old problem, which can belong to the traditional metrics like impact factor.46,47 Readers must note, as with citations,48 it is not always clear whether Twitter mentions are negative or positive. Furthermore, this investigation only analysed popular dental articles on Twitter. It would be interesting to analyse the knowledge structure of the highly shared dental articles on Facebook, Google+, Wikipedia etc.

As a final point, in spite of the numerous benefits of Twitter as a newly-emerging scholarly tool to disseminate and discuss science, dental researchers and practitioners must be alert to its overuse in scholarly communications.24,49,50,51 The Kardashian-index would be a useful tool to measure the over/under activity of a researcher on Twittersphere.25

Appendix 1
figure 7

A simple keyword co-occurrence network. Article alpha represent keywords A, B, C, D and article beta represent keywords C, D, E, F. The numbers on the links showed number of co-occurrences of two keywords. In this network keywords C and D were hot topics

Full size image
Appendix 2
figure 8

A simple bibliographic coupling network. This phenomenon occurs when the articles alpha and beta cite article gamma in their bibliographies. Upper part of diagram showed citation pattern between five articles. Lower part showed bibliographic coupling network. The numbers on the links showed coupling strength of two articles. In this network articles D and E had the highest bibliographic coupling strength

Full size image