Bibliometric analysis of research on the role of intestinal microbiota in obesity

Search strategy

Bibliometric data can be acquired through various search engines. In this study, the Scopus database was selected to perform the literature search for all published articles on enteric microbiota and obesity. It was justifiable to use Scopus as our data source to retrieve abstracts, citations and other bibliometric data given that it has wider resources and is consistently more accurate than other alternatives such as PubMed, Web of Science and Google Scholar (Choudhri et al., 2015; Falagas et al., 2008; Kulkarni et al., 2009).

The Scopus database was searched from its inception to December 31, 2017 with no language limitation. The synonyms for gut microbiota and obesity were included in the search strategy. The keywords pertaining to gut microbiota were: gastrointestinal microbiomes, gut microflora, gut microbiota, gastrointestinal flora, gut flora, gastrointestinal microbiota, gut microbiome, gastrointestinal microflora, intestinal microbiome, intestinal microbiota, intestinal microflora, and intestinal flora. The keywords regarding obesity were: obesity, corpulence, fatness, and overweight. The two sets of keywords were searched with the AND logic in the Article title/ Abstract/ Keywords fields. The final search query was built like this:

((TITLE-ABS-KEY ( gastrointestinal AND microbiomes ) OR TITLE-ABS-KEY ( gut AND microflora ) OR TITLE-ABS-KEY ( gut AND microbiota ) OR TITLE-ABS-KEY ( gastrointestinal AND flora ) OR TITLE-ABS-KEY ( gut AND flora ) OR TITLE-ABS-KEY ( gastrointestinal AND microbiota ) OR TITLE-ABS-KEY ( gut AND microbiome ) OR TITLE-ABS-KEY ( gastrointestinal AND microflora ) OR TITLE-ABS-KEY ( intestinal AND microbiome ) OR TITLE-ABS-KEY ( intestinal AND microbiota ) OR TITLE-ABS-KEY ( intestinal AND microflora ) OR TITLE-ABS-KEY ( intestinal AND flora ) ) AND PUBYEAR <2018 ) AND ( ( TITLE-ABS-KEY ( obesity ) OR TITLE-ABS-KEY ( corpulence ) OR TITLE-ABS-KEY ( fatness ) OR TITLE-ABS-KEY ( overweight ) ) AND PUBYEAR <2018 ).

All of the citation information, bibliographical information, abstract and keywords, funding details and other information of the retrieved publications was exported with CSV format through the University of Chicago Library’s access, for the further data processing.

Data analysis

All bibliometric information was exported into CSV format from the Scopus database. Microsoft Excel was applied to for sorting and to perform statistical procedures. Top prolific authors, countries, journals, institutions and most cited papers were ranked according to the standard competition ranking (SCR). Data visualization was conducted using the VOSviewer technique to create scientific landscapes and networks based on the citation frequency, countries, journals, authors and other information (Van Eck & Waltman, 2010). GunnMap 2 (http://lert.co.nz/map/) was used to generate the world map to show the publication distribution.

Some bibliometric indicators were applied in the analysis. The strength of publications included was assessed by impact factor (IF) obtained from the latest Journal Citation Report (2016) published by Thomson Reuters. Bradford’s Law was used as a scattering index to reveal the distribution of the scientific literature in a particular discipline. Bradford proposed a model of concentric zones of productivity, termed as Bradford zones, with decreasing densities of literature. Each of these zones would contain a similar number of articles, but the number of journals in which these articles are published would increase on passing from one zone to another (Bradford, 1948). This model permits the identification of journals that are most widely used or have the greatest weightage in a given field of scientific production.

GraphPad Prism version 6.0c (San Diego, CA, USA) was used to conduct statistical analysis. Pearson’s correlation analysis was used to investigate relationships between the publication numbers of different countries/regions and some related variables, such as gross domestic product (GDP), GDP per capita (purchasing power parity, PPP), etc. A P value of <0.05 was considered statistically significant.

Publications analysis based on numbers and citations

The applied strategy yielded a total of 3,446 publications. The first article on gut microbiota and obesity was published in 1968, and the annual publication numbers were stable in the following nearly 40 years till 2004. A dramatic growth was observed in the last 15 years from 2003 to 2017, and the number of annual documents during this period showed an exponential growth trend (y = 0e^0.4263x, R² = 0.94). The specific numbers of annual documents and accumulated documents are shown in Fig. 1A. The highest number of annual publications was seen in 2017, totaling 702 publications. The publication numbers of obesity research (Fig. 1B) can serve as a background frame to present a better understanding of the growth trend of intestinal microbiota in obesity research. As shown in Fig. 1C, the percentage of intestinal microbiota related publications in the obesity research was increasing gradually, especially after the year of 2006.

In terms of document type, the majority of the retrieved papers were research articles (n = 1826, 53.0%), followed by review articles (n = 1110, 32.2%) and book chapters (n = 103, 3.0%) (Fig. 2A). The primary source of the publications was journals (Fig. 2B). As shown in Fig. 2C, most of the retrieved documents belong to medical subjects (n = 2, 324, 67.4%), followed by Biochemistry, Genetics and Molecular Biology (n = 1, 106, 32.1%), Nursing (n = 554, 16.1%), Immunology and Microbiology (n = 466, 13.5%), and Agricultural and Biological Sciences (n = 459, 13.3%). The retrieved papers were written in 23 different languages, mainly in English (n = 3232, 93.8%), followed by French (n = 65, 1.9%), Chinese (n = 37, 1.1%), German (n = 33, 1.0%), and Spanish (n = 24, 0.7%).

All of the retrieved 3,446 papers were cited a total of 141,918 times. A total of 2,858 (82.9%) articles had at least one citation while 588 (17.1%) articles had no citations.

Among the total 3,446 documents there were multiple source types (Fig. 2B) that were written in different languages (Fig. 2D). Considering the heterogeneity of the documents may add background noises to the study, we only included the journal articles written in English to perform the further analyses. According to this inclusion criteria, a total of 390 documents (books, book series, conference proceedings and other types) were excluded. 3,056 English papers were included for the subsequent analyses.

Publications analysis based on top cited articles

Top 20 cited papers are listed in Table 1 (Bäckhed et al., 2004; Bäckhed et al., 2005; Bäckhed et al., 2007; Cani et al., 2007; Cani et al., 2008; Cani et al., 2009; Henao-Mejia et al., 2012; Kau et al., 2011; Le Chatelier et al., 2013; Ley et al., 2005; Ley et al., 2006; O’Hara & Shanahan, 2006; Qin et al., 2010; Ridaura et al., 2013; Tremaroli & Bäckhed, 2012; Turnbaugh et al., 2008; Turnbaugh et al., 2009a; Turnbaugh et al., 2006; Turnbaugh et al., 2009b; Vijay-Kumar et al., 2010). The highest citation number was 3,961 for the article entitled “An obesity-associated gut microbiome with increased capacity for energy harvest” (Turnbaugh et al., 2006). Among the top 20 highly cited papers, eight were published in Nature, followed by Science and Proceedings of the National Academy of Sciences of the United States of America (three papers), Diabetes (two papers). As the first author, Turnbaugh PJ published four papers among the top 20 cited papers, followed by Bäckhed F. and Cani P.D. (three papers), then by Ley RE (two papers). Among the top 20 cited papers, Gordon, J.I. participated 12 papers, followed by Ley R.E. (six papers) and Bäckhed F. (five papers); they were the most productive of the highest quality authors.

Publications analysis based on authors

A total of 10,648 authors contributed to the publication of all the 3,056 papers included. The number of authors for a single document, also known as the transience index, is 8,224, accounting for 77.2% of all the authors. There are 85 authors who published more than ten articles in this field. Cani, P.D. was the most productive author in this field with 76 publications, the following were Delzenne, N.M. (52 publications) and Bäckhed, F. (48 publications). Details of the top 20 most prolific authors are presented in Fig. S1A, and the co-authorship analysis is shown in Fig. S2.

A total of 9,590 authors have been cited at least once, accounting for 91.7% of the total 10,462 authors, 6,064 authors had citation numbers at least 10 (58.0%), and 1,597 authors have been cited at least 100 times (15.3%). Gordon JI was the most influential author with the greatest citation number of 25,073, followed by Ley R.E. (17,658), and Bäckhed F. (15,518); details of citation analysis are represented in Fig. 3. Among the top 20 most cited authors, Batoo, J-M, Levenez, F and Renault, P had the greatest citation numbers per publication (1,303.8), followed by Tap, J (1,297.5) and Parkhill, J (1,291.5), more details are shown in Fig. S1B.

Publications analysis based on countries/regions

The geographical distribution of publications involved 79 countries/regions over six continents (Fig. 4). There were 20 countries that just published only one article, and 46 countries that published at least five articles. The top 20 most productive countries are shown in Fig. S1C; the US ranked first with 995 publications, followed by China (243 publications), France (242 publications), United Kingdom (223 publications), and Italy (196 publications). However, the GDP-adjusted ranking according to their productivity scores was different, the most prolific countries were Finland, Denmark, Ireland, Belgium, and Swaziland (Fig. S1D). Publication productivity scores of countries/regions were calculated by dividing the number of publications by their GDP and then multiplying the result with by 1E + 12. As shown in the Fig. S1E, the most productive continents according to the publication numbers were Europe, North America, and Asia.

In order to achieve a better understanding of the diverse publication numbers in different countries/regions, the relationship between the publication numbers and multiple factors for each country was investigated. The analyzed factors were population number (PN), gross domestic product (GDP), gross domestic product per capita (purchasing power parity, PPP), and human development index (HDI). Significant correlation was found between the publication number of intestinal microbiota in obesity research and GDP, HDI, and PN (Fig. 5). GDP demonstrated the highest correlation with the publication number (r = 0.89, p < 0.0001), followed by HDI (r = 0.32, p < 0.001) and PN (r = 0.27, p < 0.05). GDP was the most important factor that could benefit the publication productivity. Therefore, this analysis can interpret the reason why the US and China were the most prolific countries in this domain.

Analysis of citation counts for countries showed that the US had the most citations followed by France, UK, Belgium and the Netherlands (Fig. S3). The international collaboration analysis based on these countries is shown in a network visualization map (Fig. S4). As is indicated in the collaboration analysis, the United States (US) had collaborations with other 38 countries and followed by Italy (35 collaboration links), Spain (33 collaboration links), United Kingdom (UK) (32 collaboration links), and Germany (31 collaboration links). For the US, collaboration was mostly with China (relative link strength = 64) and Canada (relative link strength = 45).

Publications analysis based on institutions

Prolific institutions in publishing papers on gut microbiota and obesity are presented in Table 2. The most active institution was University College Cork in Ireland (122 publications), followed by The Institut national de la santé et de la recherche médicale (Inserm) in France (109 publications), and Universite Catholique de Louvain in Belgium (95 publications). A total of 160 institutions published at least 10 articles. Among the top 20 most active institutions, 14 are in Europe, four and in North America, and two are in China. A total of 79 organizations have been cited at least 1,000 times, the citation analysis was shown in Fig. S5. The Department of Chemistry and Biochemistry, University of Colorado obtained the highest citation number (7,930 citations), and followed by Center for Genome Sciences, Washington University in St. Louis (5,032 citations) and Danone Research, in France (4,652 citations).

Publications analysis based on journals

All the retrieved documents were published in 940 different journals. The top 20 active journals in publishing articles on gut microbiota and obesity are shown in Table 3. The most prolific journal in this field was PLOS ONE (106 publications), followed by Scientific Reports (46 publications), and British Journal of Nutrition (43 publications). The total number of articles published in top 20 journals was 728, accounting for 23.8% of total retrieved documents. Citations analysis of the included 143 journals with at least five publications indicated that Nature has the highest citation numbers (N = 22, 139), followed by Proceedings of The National Academy of Sciences of The United States of America (N = 10, 195), and PloS One (N = 5, 117) (Fig. 6). Regarding the academic journals publishing articles on gut microbiota and obesity, the Bradford model was applied. The division of Bradford’s zones in this study is shown in Table 4. A total of 940 journals were involved; the average number of articles in each zone was 305.6. The first zone included six journals: Plos One, Scientific Reports, British Journal of Nutrition, Nutrients, Gut Microbes, and Cell Metabolism.

Publications analysis based on terms frequency

A density visualization map of most frequently encountered terms is shown in Fig. 7. This analysis was performed based on the terms extracted from the title and abstract fields of retrieved publications; a number of 195 terms met the threshold with a minimum number of occurrences as 100. The term with the highest frequency was obesity (N = 4, 677), followed by gut microbiota (N = 3, 437), and diet (N = 2, 267).