The Most Influential Publications on Endoscopic Submucosal Dissection: A Bibliometric Analysis
Article information
Abstract
Background/Aims
Endoscopic submucosal dissection (ESD) is the first-line treatment for superficial gastrointestinal neoplasms with negligible lymph node metastasis. It has evolved through improvements in expertise and equipment, increased understanding of indications and short- and long-term outcomes, and better management of complications. This study aimed to assess and characterize the most influential publications in ESD research.
Materials and Methods
We searched the top 50 most cited articles using Web of Science Core Collection (WoSCC) and Google Scholar (GS) from the inception of these services to January 2019. The top 50 Altmetric Attention Score (AAS) articles based on online media mentions were also searched. Each article was evaluated for the number of citations, title, journal, and publication year.
Results
The number of citations for the top 50 WoSCC articles on ESD ranged from 37 to 199; Endoscopy published the most articles (20%). Among the top 50 GS articles, Gastrointestinal Endoscopy published the most ESD articles (34%) and the most shared AAS articles (42.6%). PubMed Central article citations in WoSCC or GS showed significant correlation with those from each metric, unlike AAS. The words with the highest relevance scores were “submucosal tunnel dissection,” “guideline,” “novel submucosal gel,” “adhesive material,” “cell sheet,” “esophageal ulcer,” “hemospray,” and “endoscopic closure,” while the following words were influential: “meta-analysis,” “esophageal stricture,” “perforation,” “bleeding,” “fibrin glue,” “artificial ulcer,” “porcine model” and “esophageal squamous cell neoplasia,” excluding “ESD.”
Conclusions
This study presents a detailed list of influential articles, journals, and topic words.
INTRODUCTION
Academic impact of publication is frequently reflected by the number of times the publication is included as a citation [1]. As the amount of newly published literature that researchers need to be aware of increases and the importance of selective searching and summarization grows, the meaning of citation counts as a bibliometric index is growing. Many milestones of literature have dramatically influenced the promotion of specific research fields, and there are many cases in which the impact of older publications does not diminish. However, there has been little work in the identification of influential literature in the field of gastroenterology.
Citation counts have been the most widely adopted and useful method for measuring the impact of research activity. However, it has been criticized because the citation only occurs for a certain amount of time after publication, leaving influential literature uncited in many cases. It is also impossible to recognize the reasons the literature is cited, whether to express support or to criticize the content of the text. Moreover, solely examining citation frequency is inadequate to measure the impact beyond the particular academic field.
Social media influence in medicine is growing [2]. The current production of knowledge and distribution of data is frequently mediated by social media in the form of unstructured text data, rather than the traditional journal form [3]. Medical knowledge is no longer only limited to experts and is accessible for patients to share and disseminate. With the widespread use of social media, a new concept of social web-based metrics (‘altmetrics’) has been implemented as an indicator that more quickly reflects influence and attention than traditional indicators [4]. Altmetric Attention Score (AAS) accounts for the scientific and social impact of an article after publication and is based on the weighted count of ‘mentions’ of an article across various online sources while reflecting the relative reach of each type of source [4,5].
Endoscopic submucosal dissection (ESD) is the established treatment for superficial gastrointestinal neoplasms with minimal risk of lymph node metastasis. The most important advantage of this procedure was in achieving curative resection with minimal invasiveness, avoiding invasive surgery that inevitably accompanies substantial morbidity. It has emerged and evolved through an improvement of skills, expertise, and equipment, to arrive at a better understanding of indications, short- and long-term outcomes, and management of complications. This study aimed to assess and characterize the most influential publications in ESD research.
MATERIALS AND METHODS
1. Selection of bibliometric tools
To conduct a citation analysis in the field of ESD research, we selected the following three representative bibliometric tools: Web of Science Core Collection (WoSCC; Clarivate Analytics, Philadelphia, PA, USA), Google Scholar (GS; Google LLC., Mountain View, CA, USA), and Altmetric Explorer (Altmetric LLP, London, UK).
The WoSCC is an online scientific indexing platform that provides a comprehensive citation searching mechanism for their collection of high-quality scholarly content, providing access to multiple databases for research discovery, covering multidisciplinary content [6], which allows for an in-depth exploration of specialized academic fields. The WoSCC search of the top 50 most-cited articles was able to retrieve articles published from 2013 to 2017 and included relatively recent citations from high-quality journals that have been integrated annually from Journal Citation Reports (JCR) (timespan option of searching was set for all years).
GS is an online search engine that indexes publications from academic and scientific web domains. It provides bibliographic references and citations from most peer-reviewed online academic journals, books, conference papers, theses, dissertations, preprints, and court opinions and patents [7]. GS has indexed approximately 389 million documents, making it the world's largest academic search engine as of January 2018 [8]. Researchers can also obtain information from non-JCR journals [9]. Harzing’s Publish or Perish software is a third party tool for calculating numbers of citations that could retrieve articles published from 1999 to 2015, pulling citations from various sources, not limited to JCR journals and recent research [10].
Altmetric.com provides real-time online mentions of published articles from public policy documents, mainstream media, online reference managers, post-publication peer-review platforms, Wikipedia, Open Syllabus Project, patents, blogs, research highlights from F1000Prime, and social media platforms, including Facebook, Twitter, and YouTube, from 2011 [4]. It provides AAS to measure the overall level of online impact using Altmetric Explorer [4].
PubMed is the most widely used tool for searching medical literature. Since it does not provide a tool for analysing the citation counts by bibliometric index, it was used to validate the citation counts of WoSCC and GS.
2. Identification of the top 50 most-cited articles and the top 50 most-mentioned AAS articles in online media
We searched the top 50 most-cited articles using the WoSCC and GS from the inception of these services to January 2019 with two independent evaluators (YJY and CSB). The search term was ‘endoscopic submucosal dissection’. The site was accessed on 1 February 2019 to avoid changes in the online activity of publications. Articles with the same number of citation counts were included in the analysis even if the number of enrolled articles exceeded 50. Then, we performed a comparison of the citation counts for the top 50 articles from WoSCC and GS with those of PubMed Central (PMC), using the PubMed platform on the same day. Since the publication dates of the GS top 50 most-cited articles ranged from 1999 to 2015, it was difficult to reflect recent citations. Therefore, the top 50 articles published after 2015 among the top 1,000 most-cited articles according to GS bibliometrics were extracted and analysed separately. AAS was also accessed and retrieved on 1 February 2019 using the ‘advanced search’ function in Altmetric Explorer.
All articles were retrieved and ranked in descending order based on the citation counts obtained by WoSCC, GS, and AAS. The titles and abstracts of all identified articles were reviewed to exclude irrelevant publications. Publications relevant to ESD, such as endoscopic mucosal resection (EMR), peroral endoscopic myotomy (POEM), or submucosal tunnel dissection, were included. Full-text reviews were conducted for cases where it was impossible to determine the relevance of ESD research. Disagreements between the evaluators were resolved by discussion.
3. Data extraction and bibliometric analyses
Each of the top 50 most-cited or -mentioned articles was examined for 1) citation counts from WoSCC, GS, and from PMC articles; 2) AAS; 3) title; 4) journal name with its impact factor and EigenfactorTM score [11]; 5) published year; 6) region (stomach vs. colon vs. esophagus vs. duodenum vs. all regions) of ESD; and 7) article type (original research vs. guideline/consensus report vs. review vs. case report).
A network-analysis was conducted to determine the most influential words in titles, abstracts, and of both titles and abstracts among the top 500 most-cited articles in WoSCC. Visualization was impossible when analysing the top 50 most-cited articles due to paucity of data; therefore, the analysis scheme was increased to the top 500 most-cited articles [12]. The units of analysis were words in a title or abstract, and a full counting method was adopted-all occurrences of a term in a title or abstract were counted.
4. Statistical analyses
This study’s primary outcome was a narrative synthesis of data. However, a quantitative synthesis was conducted to find correlations between citation counts from WoSCC or GS and those from PMC articles. The correlation coefficient was analysed using Pearson’s correlation test; the partial correlation coefficient was analysed to control for confounding variables. To find the differences in citation counts according to the publication year, a Kruskal-Wallis test was performed. To find the differences in publication numbers of different regions of ESD according to the publication year, a Cochran-Armitage test for trend was executed. Harzing’s Publish or Perish software version 6 was used for the citation analysis [10], and VOSviewer version 1.6.9 was used for the network analysis [12]. A P-value <0.05 (two-tailed) was defined as the statistical significance threshold for all tests. All analyses were performed using SPSS version 22.0 (SPSS Inc., Chicago, IL, USA), except for the Cochran-Armitage test, which was completed using EpiTools calculators (http://epitools.ausvet.com.au/content.php?page=trend; Ausvet Pty Ltd, Bruce, Australia).
RESULTS
1. Bibliometrics of WoSCC
Table 1 lists the top 10 most-cited articles from WoSCC and Supplementary File 1 lists the top 50 most-cited articles from WoSCC including detailed information. The citation counts for the top 50 articles ranged from 37 to 199; the publication years ranged from 2013 to 2017. The top 50 articles were published in 15 different journals; Endoscopy published the largest share of articles (20%), followed by Digestive Endoscopy/Gastrointestinal Endoscopy (18%) (Table 2).
In the WoSCC bibliometrics of the top 50 most-cited articles, citations from PMC articles showed a significant correlation with those from WoSCC (r=0.93, P<0.001). The result was consistent after controlling for publication year, region of ESD, and type of article (partial correlation coefficient=0.94, P<0.001). A Kruskal-Wallis test showed no significant difference (P=0.12) in citation counts according to the publication year.
Of the top 50 most-cited articles, original research comprised 78%, review 14%, guidelines or consensus reports 6%, and case series 2%. The proportion of systematic reviews/meta-analyses was 8% (Supplementary File 1).
For the region of ESD in each study, stomach comprised 32%, colon 30%, esophagus 20%, all regions 12%, and duodenum 6%. A Cochran-Armitage test for trends showed no significant difference (P=0.48) in publication numbers on different regions of ESD from WoSCC according to the publication year.
2. Bibliometrics of GS
Table 3 lists the top 10 most-cited articles from GS, and Supplementary File 2 lists the top 50 most-cited articles from GS. The citation number for the top 50 most-cited articles found through GS ranged from 221 to 1,195 and the publication years ranged from 1999 to 2015. The top 50 most-cited articles were published in 12 different journals. Gastrointestinal Endoscopy published the largest share of articles (34%), followed by Endoscopy (22%) (Supplementary Table 1).
The GS bibliometrics of the top 50 most-cited articles indicated that the citations of PMC articles were significantly correlated with those from GS (r=0.80, P<0.001). The result was consistent after controlling for publication year, region of ESD, and type of article (partial correlation coefficient=0.81, P<0.001). A Kruskal-Wallis test showed no significant difference (P=0.38) in citation counts according to the publication year.
Original research comprised 90%, review 8%, guidelines or consensus reports 2%, and systematic reviews/meta-analyses 4% of the top-50 cited articles (Supplementary File 2).
For the region of ESD in each study, stomach comprised 50%, esophagus 24%, colon 16%, and all regions 10% of the top 50 most-cited articles of GS bibliometrics. A Cochran-Armitage test for trends showed no significant difference (P=0.05) in publication numbers on different regions of ESD from GS according to the publication year.
3. Bibliometrics of GS published after 2015
Among the top 1,000 most-cited articles on GS bibliometrics, there were 50 papers published between 2015 and 2017. Supplementary Table 2 lists the top 10 most-cited articles on GS published after 2015, and Supplementary File 3 lists the top 50 most-cited articles on GS published after 2015 among the top 1,000 most-cited articles. The citation counts ranged from 24 to 323; the publication years ranged from 2015 to 2017. The top 50 most-cited articles were published in 19 different journals; Endoscopy published the largest share of articles (20%), followed by Gastrointestinal Endoscopy (18%) (Supplementary Table 3).
In the GS bibliometrics of the top 50 most-cited articles published after 2015, citations of PMC articles showed a significant correlation with those from GS (r=0.96, P<0.001). The result was consistent after controlling for publication year, region of ESD, and type of article (partial correlation coefficient=0.91, P<0.001). A Kruskal-Wallis test showed no significant difference (P=0.45) in citation counts according to the publication year.
Of the top 50 most-cited articles published after 2015, original research comprised 88%, review 6%, guidelines or consensus reports 6%, and systematic reviews/meta-analyses 10% (Supplementary File 3).
For the region of ESD in each study, stomach comprised 38%, esophagus 32%, colon 22%, all regions 6%, and duodenum 2%. A Cochran-Armitage test for trends showed no significant difference (P=0.46) in publication numbers on different regions of ESD according to the publication year.
4. Bibliometrics of AAS
Supplementary Table 4 lists the top 10 AAS articles, and Supplementary File 4 lists the top 50 AAS articles. The AAS for the top 50 most-mentioned articles in online media ranged from 6 to 21. The publication years ranged from 2006 to 2019. The citation counts of PMC articles for the top 50 most-cited AAS articles ranged from 0 to 55. The top 50 most-cited AAS articles were published in 15 different journals. Gastrointestinal Endoscopy published the largest share of articles (42.6%), followed by VideoGIE (29.5%) (Supplementary Table 5).
In the top 50 most-cited AAS articles, citations of PMC articles showed no correlation with AAS (r=-0.05, P=0.68). The result was consistent after controlling for publication year, region of ESD, and type of article (partial correlation coefficient=-0.05, P=0.74). A Kruskal-Wallis test showed no significant difference (P=0.36) in citation counts according to the publication year.
Of the top 50 AAS articles, original research comprised 60.7%, case reports 31.1%, review 6.6%, guidelines or consensus reports 1.6%, and systematic reviews/meta-analyses 6.6% (Supplementary File 4).
For the region of ESD in each study, colon comprised 42.6%, stomach 26.2%, esophagus 23%, duodenum 6.6%, and all regions 1.6%. A Cochran-Armitage test for trends showed no significant difference (P=0.38) in publication numbers on different regions of ESD from AAS bibliometrics according to the publication year.
5. Network-analysis for the most influential words in titles, abstracts, and of both titles and abstracts among the top 500 most-cited articles in WoSCC
Among the 1,040 words in the titles of the most-cited articles, Supplementary Table 6 lists the words with high relevance-score. Words or terms with a high relevance-score included ‘colorectal ESD/EMR,’ ‘gastric ESD,’ ‘submucosal tunnel dissection,’ ‘guideline,’ ‘novel submucosal gel,’ and ‘esophageal squamous cell neoplasia’ [13]. Fig. 1A demonstrates the overlay visualization with no normalization layout. Fig. 1B demonstrates the density visualization of Fig. 1A [14]. The following were influential title-words: ‘early gastric cancer (EGC),’ ‘meta-analysis,’ ‘risk factor,’ ‘esophageal stricture,’ ‘perforation,’ ‘bleeding,’ ‘colorectal tumor,’ ‘feasibility,’ ‘technique,’ ‘esophagus,’ ‘fibrin glue,’ ‘artificial ulcer,’ ‘porcine model,’ etc.
Among the 7,293 words in abstracts, Supplementary Table 7 lists the words with high relevance-score. Words or terms with a high relevance-score included ‘adhesive material,’ ‘cell sheet technology,’ ‘cultured autologous oral mucosal epithelial cell sheet,’ and ‘esophageal ulcer.’ Fig. 2A demonstrates the overlay visualization with association strength layout and Fig. 2B demonstrates the density visualization of Fig. 2A. The following were influential abstract-words: ‘en bloc resection,’ ‘recurrence,’ ‘second look endoscopy,’ ‘EGC,’ ‘endoscopic ultrasound,’ ‘neoplasia,’ ‘lymphovascular invasion,’ ‘additional surgical resection,’ ‘curative ESD,’ ‘ulcer healing,’ ‘rebamipide,’ ‘metastasis,’ ‘extragastric recurrence,’ ‘active bleeding,’ ‘proton pump inhibitor,’ ‘device,’ ‘CO2 insufflation,’ ‘sedation,’ ‘speed,’ etc.
Among the 7,691 words in the titles and abstracts, Supplementary Table 8 lists the words with high relevance-scores. Fig. 3A demonstrates the overlay visualization with linlog layout and Fig. 3B demonstrates the density visualization of Fig. 3A. The following were influential words and terms in the titles and abstracts: ‘perforation,’ ‘efficacy,’ ‘adverse event,’ ‘EGC,’ ‘curative resection,’ ‘bleeding,’ ‘esophageal stricture,’ ‘colorectal ESD,’ ‘colorectal tumor,’ ‘EMR,’ ‘tumor size,’ ‘gastric ESD,’ ‘bleeding,’ ‘risk factor,’ ‘second look endoscopy,’ ‘delayed bleeding,’ ‘post ESD bleeding,’ ‘elderly patients,’ ‘surgery,’ ‘technical difficulty,’ ‘adenoma,’ etc.
DISCUSSION
The citations of PMC articles among the top 50 WoSCC or GS articles showed a significant correlation with those from each bibliometrics; however, AAS was not correlated. This was consistent with previous studies exploring the association between altmetric indicators and citation counts [15,16]. A recent bibliometric analysis of neurointervention articles also suggested that AAS operates independently of citations and can be used as an alternative index for measuring an article’s impact [17]. Considering that attention itself does not reflect the quality of research, AAS might not represent a qualified impact in specialized academic fields; however, information will no longer be inaccessible, and AAS, as a disseminative impact, might be a complementary indicator in the evaluation of the impact of a specific article [4].
In terms of the topic of all four bibliometrics, guidelines or evidence-based technical reviews of each academic society [18-21] were common in the WoSCC or GS top 50 most-cited articles published after 2015, including recent citations. Studies with emerging issues, such as management of ESD-induced gastric perforation with an over-the-scope clip [22], ESD of early esophageal cancers [23] or superficial Barrett’s neoplasia [24,25], ESD of EGC with undifferentiated-type histology [26], therapeutic outcomes of POEM [27], or natural orifice transluminal endoscopic surgery techniques, such as laparoscopic endoscopic cooperative surgery [28] were noted in the WoSCC or GS top 50 most-cited articles published after 2015, including recent citations. Large-scale cohort studies with over 1,000 cases that evaluated the efficacy and safety of ESD were also common in the WoSCC or GS top 50 bibliometrics [29-34].
Upon increasing the analysis scheme to GS top 50 bibliometrics, many milestone papers and expert review articles that provided evidence for current practice standardization were included. A study on the introduction of POEM [35] and review publications that greatly affected the standardization of gastric ESD techniques, indications, and procedures were also included [36,37]. In addition to gastric ESD, earlier studies that influenced the widespread use of ESD in the esophagus or colorectal area were included [38,39].
The composition of the top 10 most-cited AAS articles were slightly different from other bibliometrics. Additional validation of topics that may already be considered as having sufficient evidence were noted, such as a comparison of long-term outcomes of gastric ESD with surgery for EGCs that fulfilled the expanded indication [40], or therapeutic outcomes of ESD of differentiated EGCs [41] or colorectal tumors [42] in Western countries. Studies that compared EMR with ESD for the treatment of superficial colorectal neoplasia [43-45] received attention from various audiences. Studies about the development and validation of novel ESD techniques, such as endoscopic mucosa-sparing lateral dissection for the treatment of gastric subepithelial tumours or ESD for subepithelial tumours of the colon and rectum also received attention [46,47]. Studies on emerging issues, such as the Western ESD trial on superficial Barrett’s neoplasia [48] or the transplantation of autologous esophageal mucosa to prevent stricture after circumferential ESD of esophagus [49] were included. Upon increasing the analysis scheme to the top 50 most-cited AAS articles, most of the articles were video cases published in VideoGIE.
In terms of the subject categories of published journals, ‘Gastroenterology and Hepatology’ (especially, endoscopy-related journals such as Gastrointestinal Endoscopy, Endoscopy, Digestive Endoscopy, Surgical Endoscopy and Other Interventional Techniques) was prevalent in all of the bibliometrics. However, other categories such as ‘Surgery,’ ‘Oncology,’ or ‘Medicine, General, and Internal’ were also prevalent, indicating that ESD research is not confined to one specialized academic field and journals with wider readerships are also eligible to publish research on ESD.
Original research was common in all of the top 50 bibliometrics (60.7~90%); however, video cases were also prevalent (31.1%) in top 50 AAS bibliometrics. Guidelines/consensus reports (1.6~6%) and systematic reviews/meta-analyses (4~10%) were prevalent and in the upper ranks of all bibliometrics, likely due to the consolidated and quantitative review of the large, complex, and sometimes conflicting field of articles in this era of overflowing publications.
For the region of ESD in each set of bibliometrics, stomach comprised 50% in GS bibliometrics (esophagus: 24%, colon: 16%); however, the proportion of colon and esophagus was increased in the bibliometrics of WoSCC (stomach: 32%, colon: 30%, esophagus: 20%) and GS published after 2015 (stomach: 38%, esophagus: 32%, colon: 22%), while the proportion of colon was most prevalent in the AAS bibliometrics (colon: 42.6%, stomach: 26.2%, esophagus: 23%).
A network-analysis of influential words in the titles or abstracts among the top 500 most-cited articles on WoSCC showed the recent interest of researchers relevant to ESD research. Words and terms with high relevance-scores were ‘submucosal tunnel dissection,’ ‘guideline,’ ‘novel submucosal gel,’ ‘adhesive material,’ ‘cell sheet,’ ‘esophageal ulcer,’ ‘hemospray,’ and ‘complete endoscopic closure,’ and the following words and terms were influential: ‘meta-analysis,’ ‘esophageal stricture,’ ‘perforation,’ ‘bleeding,’ ‘fibrin glue,’ ‘artificial ulcer,’ ‘porcine model,’ ‘esophageal squamous cell neoplasia,’ excluding ‘ESD.’
To the best of our knowledge, this is the first study using bibliometric analysis relevant to ESD research. We adopted various bibliometric sources with in-depth analyses of citation counts, journals, publication years, and words in titles and abstracts. Despite its strengths, this study has limitations. First, only the ‘number of citations’ was adopted as a bibliometric index, excluding various other indices (such as article influence score, cited half-life or immediacy index, etc.). Previous studies exploring the impact of each bibliometric index suggested the significance and importance of specific bibliometric indices representing influence in each academic field [50,51]. However, there has been no study stated above in the field of gastroenterology and we could adopt only ‘number of citations’ as a bibliometric index. Another limitation was that the number of top 50 most-cited articles might be insufficient in determining the most influential articles on ESD research. Visualization of network-analysis to determine the most influential title-words was impossible among the top 50 most-cited articles and this scheme should be increased to the top 500 most-cited articles.
In conclusion, this study can help researchers identify the most influential articles on ESD and provide insight into the most noteworthy scientific trends and to visualize future research needs of the topic.
Acknowledgements
This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) and funded by the Korean government, Ministry of Science and ICT (MSIT) (grant number NRF2017M3A9E8033253).
Notes
No potential conflict of interest relevant to this article was reported.