A global bibliometric analysis of Plesiomonas-related research (1990 – 2017)

Plesiomonas shigelloides is an emerging pathogen with damaging effects on human health such as gastroenteritis and extraintestinal infections. Here, we carried out a bibliometric survey that aimed to examine publication trends in Plesiomonas-related research by time and place, international collaborative works, identify gaps and suggest directions for future research. The search term “Plesiomonas shigelloides” was used to retrieve articles published between 1990 and 2017 from the Web of Science database. Only primary research articles were included in the analysis. A total of 155 articles were published within the survey period, with an average of 5.54±2.66 articles per year and an annual growth rate of −0.8%. Research output peaked in 2000 and 2006 (each accounting for 7.7% of the total). The United States ranked first in terms of numbers of articles (n = 29, 18.1%) and total citations (n = 451). Cameroon, Canada, Cuba, Switzerland and Turkey co-shared the 10th position each with 2 articles (1.3%). Research collaboration was low (collaboration index = 3. 32). In addition to Plesiomonas shigelloides (n = 82, 52.9%), the top Authors Keywords and research focus included lipopolysaccharide and nuclear magnetic resonance (n = 13, 8.4%). Diarrhea (n = 43, 27.7%), Aeromonas species (n = 41, 26.5%) and infections (n = 31, 20.0%) were also highly represented in Keywords-Plus. Authors’ collaborations and coupling networks formed two mega-clusters which nodes were shared solely by authors from high-income countries. The common conceptual framework in retrieved articles determined by K-means clustering revealed three clusters with sizes of 7, 16, and 29, representing research responses focused on extraintestinal and gastroenteritis, P. shigelloides lipopolysaccharide structure, and co-infections, respectively. Our bibliometric analysis revealed a global diminishing research in Plesiomonas; greater research outcomes from high-income countries compared to others and low collaboration with developing countries.

Bibliometric analysis is a statistical method for assessing both the quantitative and qualitative scope and adequacy of research efforts attained in an area of interest [41]. It can be used to determine national and international research focus and evaluate research performance in order to identify future research priorities, funding sources, and interdisciplinary collaborations [42][43][44]. It also provides a resource to policy-makers for implementing necessary prophylactic measures in case the analysis reveals a sharp increase in case reports or articles regarding a health issue in a particular geographic area [45,46]. Bibliometric reviews can additionally help international health agencies to identify priorities (e.g. by nations) for disbursing aid [47] and awarding research grants.
Here we carried out a bibliometric analysis of studies on P. shigelloides published between 1990 to 2017. The articles were evaluated in terms of annual and country-specific output, theme, domain clusters, international collaboration networks, citations, topical evolution related to keywords and co-occurrence networks, co-authorship, and funding. The aim of the survey was to evaluate international participation in P. shigelloides research-with a special interest in regions where Plesiomonas infections have higher prevalence rates (i.e., Africa and Southeast Asia)-in order to address knowledge gaps and provide a resource that can help identify present and future research priorities. publication dataset (e.g.; authors, words, countries, references, keywords) for coupling, cocitation, collaboration, conceptual framework and multiple correspondence analyses. Bibliographic coupling occurs between two articles ⅈ and ⅉ when their reference lists cited at least one common source [62]. But, in a collaboration network, the nodes comprise authors and the links co-authorships [63]. The number of bibliographic coupling that occurs between articles ⅈ and ⅉ or co-authorship in scientific collaboration network denotes the strength of the network [61]. A network depicts relationships in a system as a set of nodes (components) and links (relationships) [64]. Co-Word or conceptual framework analysis explore K-means clustering and other dimensionality reduction techniques to identify clusters of common concepts known in a bibliographic collection. It relies on word co-occurrences in a publication dataset [61,64]. Scientific productivity or an author's contributions in a field is evaluated in term of Lotka's law [65]. The Lotka's law is an inverse square law that describes how often authors published in a field [65].

Data retrieval
Published peer-reviewed articles on P. shigelloides were retrieved from the Web of Science (WoS) database on August 19, 2018. The WoS is among the most reliable and comprehensive databases for bibliometric studies and hosts a wide range of quality and high-impact scientific studies (12 million articles in over 12,000 journals) [44]. We used the search term "Plesiomonas shigelloides" to identify primary research articles published between 1990 and 2017. All available information was retrieved. To obtain subject-specific results and for the sake of accuracy (in order to avoid false-positive results), only article titles were searched. A title-specific search has been reported to increase recovery and specificity with a minimal loss of sensitivity compared to a topic search [44,45,66]. Articles were downloaded in the BibTeX file format. In order to account for variations in country population on rate of scientific production, the world population was retrieved from the World Bank website (https://data.worldbank.org/indicator/SP.POP.TOTL) and the midperiod population corresponding to the top 19 countries was extracted for calculation of article per million populations. England population data was retrieved from Office for National Statistics website (https://www.ons.gov.uk/peoplepopulationandcommunity/populationandmigration/ populationestimates/timeseries/enpop/pop).

Data processing and analysis
We analyzed the retrieved data for bibliometric indicators using Rstudio v.3.4.1 software (2017-06-30) with bibliometrix R-package (http://www.bibliometrix.org) [61]. Data were imported into RStudio and converted to a bibliographic data frame and normalized for duplicate marching. The duplicated article was reduced to one record in the analysis. The data frame as typical columns named after the standard ISI WoS Field Tag codify.
Further, authors' names, authors' keywords (DE), and Keywords-Plus (ID) were extracted for standardization. Authors' names were extracted twice as two different sets (A and B). Each set was checked for variant names, spelling errors and matched with affiliations. We achieved normalized authors' names when |A \ B| � |A [ B|. For keywords (DE) and Keywords-Plus (ID), a primary term was assigned to words with similar meanings (e.g., "Plesiomonas shigelloides", "Plesiomonas", "shigelloides", "Aeromonas-shigelloides", and "Plesiomonas-shigelloides" were allotted to "Plesiomonas shigelloides"). Multiple occurrences of a keyword or a similar keyword in an article were regarded as one. Co-occurrence of a term in authors' keywords (DE set) and Keywords-Plus (ID set) in the dataset was assessed as a set made of the intersect of the two sets (DE \ ID||). All set-based test was performed using a Venn diagram software (http://bioinformatics.psb.ugent.be/webtools/Venn/). An annual number of articles and total citations were also graphed.
Data were analysed for descriptive output, citation analysis, authors' h-index and scientific productivity using the relevant functions of the bibliometrix R-package. Bibliometric networks (e.g., citation, author, country, author keyword, and Keywords-Plus networks) and bibliographic coupling (co-citation and keyword co-occurrences) were computed and visualized from bibliometric two-way (bipartite) network of rectangular matrices of Articles × Attributes. A typical bibliometric network is expressed as Network(N) = X × N T where X is a bipartite network matrix composed of Articles × Attribute (e.g. Authors, keywords, citations, and Countries) and N is a symmetrical matrix N = N T .
We created a graphic model of all networks using force-directed algorithms (Fruchterman) implemented in the networkPlot function of the bibliometrix R-package. All networks were standardized using the Simpson's coefficient (inclusion index), proximity index (association strength), the Jaccard's similarity index, and the Salton's cosine coefficient among nodes of a network [61]. In addition, k-means clustering were performed on keywords to evaluate concepts in Plesiomonas field of research using the function conceptualStructure of the package. The function implements Porter's stemming algorithm [67] to modulate inflected words to their root form. For detailed search Boolean for articles identification from WoS, see supplementary file (S1 Appendix). Other bibliometric indicators such as language and affiliation were determined using the content search of BibTeX file.

Results
A total of 155 articles were published within the survey period; their attributes are presented in Table 1. The studies involved 493 authors, with 0.31 article/author (3.18 authors/article), 4.34 co-authors/article, and a collaboration index of 3.32. With the exception of two authors publishing solo, all 491 authors were involved in multi-author articles. An average of 11.49 citations/article was recorded during the study period. The scientific output related to P. shigelloides research by Lotka's law showed a beta coefficient and constant of 2.30 and 0.44,  Table 2 shows the top 20 most productive authors in the field. K. Krovacek (Sweden) ranked first, co-authoring 12 (7.7%) articles; and I. Ciznar (Slovak Republic) was second with 11 (7.1%) articles. The h_index (total citations) was 7 (162) for K. Krovacek, and 6 (148) for I. Ciznar. It is worth noting that the topmost active authors were affiliated with institutions in developed nations, including Sweden (n = 6), USA (n = 5), Japan (n = 3), Spain (n = 5), Poland (n = 3), Czech Republic (n = 1), and Slovak Republic (n = 1).
The 20 top cited articles on P. shigelloides are listed in S1 Table. These studies spanned the fields of infection, immunity, clinical microbiology, and biochemistry. The total no. of citations of the top-cited articles ranged from 35 to 111; most of these were the result of funded research. Research output related to P. shigelloides for the top 20 most active countries is shown in Table 3. United States ranked first in terms of total number of articles (n = 29, 18.7%) and citations (n = 451), followed by Sweden (n = 14, 9.0%) and Germany (n = 10, 6.5%). The frequency of publication varied among the top countries from 1.3 to 18.7%. Sweden had highest productivity (1.563 article/million population) when normalized for population size using mid-period population (2003). The rank order of these countries changed when productivity was mea- The top 20 journals with the most published articles on P. shigelloides are listed in S2 Table. These journals cover a range of subjects including carbohydrates, microbiology, food science, infectious disease, immunology, and biochemistry, reflecting active areas in Plesiomonas research. Carbohydrate Research ranked first (n = 9, 5.8%), followed by Journal of Clinical Microbiology, Folia Microbiologica and Food Biotechnology each with 6 articles (3.9%). Table 4 shows the most relevant keywords related to Plesiomonas studies, including both author keywords (DE) and Keywords-Plus (ID). Both Author Keywords (DE) and Keywords-Plus (ID) have 10 keywords in common (lipopolysaccharide, Aeromonas species, antigens, oligosaccharide, disease, children, diarrhea, Shigella, virulence, and water). Fourteen keywords were unique to Author Keywords (Plesiomonas shigelloides, nuclear magnetic resonance Ranking based on the number of articles; TC, total citations. (NMR), PCR, structure, MALDI-TOF, fish, pathogenicity, media, meningoencephalitis, resistance, enterotoxin, gastroenteritis, serotyping, and sepsis), and 13 keywords were unique to Keywords-Plus (infections, Escherichia, septicemia, environments, in-vitro, polysaccharide, Vibrio species, bacteria, biological repeating unit, iron, meningitis, humans, and strains). The unique Author Keywords primarily described medium of transmission (fish) and methods involved in isolation and characterization of the microorganism (NMR, MALDI-TOF, PCR, enterotoxin, resistance, pathogenicity, serotyping, and structure) and specific infections (meningoencephalitis, gastroenteritis, and sepsis). Author keyword terms associated with identification methods of P. shigelloides included polymerase chain reaction (PCR, n = 8,  septicemia (n = 24, 15.5% (ID)), meningoencephalitis (n = 4, 2.6 (DE)), Sepsis (n = 3, 1.9% (DE)), and meningitis (n = 10, 6.5% (ID)), which ranked 5th, 11th, and 11th respectively. The common conceptual frames in retrieved articles determined by K-means clustering with three clusters of 8, 14, and 39 elements showed research responses focused on neonates (children) extraintestinal infections and gastroenteritis, elucidation of cell wall structure (lipopolysaccharides, core oligosaccharide, o-specific polysaccharide etc.), and co-infections of P. shigelloides with other pathogens, respectively (Fig 2). The 33-element cluster explained the co-occurrence and co-infection of P. shigelloides with other bacteria. Other indicators of frequently represented concepts and frameworks related to P. shigelloides included co-occurrence of terms and keywords. S1 Fig shows the co-occurrence network of the top 20 terms associated with P. shigelloides studies, while S2 Fig shows the co-occurrence networks of keywords. These concept-related frameworks or terms included virulence, meningoencephalitis, Aeromonas, newborn, antigen, pathogenicity, sepsis, diarrhea, infections, Escherichia coli, septicemia, biological repeating unit, diarrheal disease/gastroenteritis, lipopolysaccharide, water, iron, polysaccharide, bacteremia, meningitis, septicemia, lipid-A, strains, and aquatic environments.
The top 20 authors' collaboration and coupling networks on P. shigelloides studies were divided into two mega-clusters or spheres with nodes occupied solely by researchers from high-income countries (Fig 3A and 3B). The first sphere of the authors' network comprised 13 nodes (authors) with no fewer than 10 linkages, while the second sphere included 10 nodes (authors) with the number of collaboration linkages ranging from nine to 10. Similarly, the two separate authors' coupling network spheres included 13 and 17 authors' networks, respectively; collaborative conjugation ranged from eleven nodes (authors) in the former and 18 in the latter. Fig 4 shows 50 countries' collaboration networks on P. shigelloides studies. Collaboration pathways ranged from 1 to 9. The Sweden had a high number of collaborations (n = 9), followed by United States (n = 8), Slovakia (n = 8), and Italy (n = 7). Other countries had no collaboration networks. Prominent network color code: green, USA network; purple, Spain network; light green, Sweden network; pink, Cuba-Brazil network; blue, Japan-China network.

Discussion
The present bibliometric analysis of P. shigelloides examined global research trends between 1990 and 2017 based on data retrieved from WoS. We found that the number of research articles on P. shigelloides increased non-linearly from 5 to 155 articles. However, a negative trend in rate of increase was noted (−0.8%) suggesting that research on P. shigelloides has not been of broad interest in the past 27 years, likely due to discontinuation of Plesiomonas-related research by certain authors and differences in regional distribution of the microorganism. Furthermore, the function estimates and goodness-of-fit indicated that scientific output on P. shigelloides does not follow Lotka's law, suggesting that the number of articles related to Plesiomonas research will further decline in the future. In general, emerging and re-emerging bacterial pathogens are not accorded the same degree of attention as their viral counterparts. Health emergencies (e.g., outbreaks of infection) relating to emerging viral pathogens including Zika and Chikungunya viruses have driven the generation of new scientific knowledge, resulting in a significant increase in the number of research articles on these subjects [68]. For instance, in 2005 only eight articles were published on Chikungunya virus, but by 2014, the number had reached 302 [69]. Similarly, only 43 articles on Ebola virus were published in 2013 prior to the Ebola outbreaks in West Africa, but this increased to more than 600 articles in 2014 [70].
As is the case with other research areas, most of the leading authors in the Plesiomonas research field were from developed nations such as the United States, Sweden, Austria, Japan, Spain, Poland, Czech Republic, and the Slovak Republic, with few from low-income countries, thus follow the similar trend of low productivity of the region in other research areas. It has been suggested that the economic strength (growth) of a nation influences the research output [43,[71][72][73]. The higher prevalence of Plesiomonas infections in developing countries [74]  should motivate researchers in countries most affected to carry out more studies on this pathogen.
Some of the most frequently cited studies pertained to health and identification of the pathogen-for example, efforts to produce a multivalent vaccine using O-specific polysaccharides from Shigella spp. and P. shigelloides; iron uptake in Plesiomonas shigelloides; and comparative analysis of P. shigelloides and E. coli (Shigella) sonnei O-antigen gene clusters. Others focused on P. shigelloides hemolytic expression; serological analysis of the Plesiomonas core; and multilocus sequence typing of Plesiomonas and its pathogenic potential. The United States and Sweden dominated the list of top 20 countries most actively researching Plesiomonas in terms of numbers of articles and citations. In addition to economic strength and availability of research facilities and funding [43,[71][72][73], this productivity can be ascribed to a high level of intranational and possibly multinational collaboration with other institutions, which can impact research visibility and citation frequency [43,73,75]. In particular, the dominance of the United States has been noted in other fields of research [42,54,76,77]. Also, authors' multiple affiliations influence country collaboration network. Conversely, the low contributions from developing countries including countries from Africa characterized by a high frequency of selffunded or independent studies [78], mirror the situation of research in other fields. The shift in rank among the top 20 nations most active in the Plesiomonas research field when productivity was measured based on total citation per country ought not be regarded as a precise measure of productivity. Citation rate does not reflect publication output of an author or country [76], since the smaller the number of articles used for estimation, the larger the impact of a few frequently cited articles [76]. Self-citations and inaccurate citations can also provide false quality metrics [76].
The most frequently mentioned keywords and research areas (including publication outlets) associated with Plesiomonas studies reflect the research hotspot during the survey period, which included cell wall (Carbohydrate research), co-infection, and extraintestinal infections such as septicemia, bacteremia, and meningitis. These findings reveal the most pressing health issues related to Plesiomonas-induced gastroenteritis and extraintestinal infections, and coinfections with other pathogens and effort to gain an understanding of the structural architect of the pathogen's cellwall; this was supported by other conceptual framework indicators such as co-words or keyword co-occurrence networks. However, important topics such as strainbased delineation and identification, including detection of pathogenic and non-pathogenic strains, that are necessary for infection management were lacking and were not evident from the bibliometric analyses. Newer research themes such as molecular and genomic-level studies as an alternative or complementary to traditional experimentation (which has some limitations) necessary to clarify the pathogenesis of Plesiomonas infection were not apparent throughout this study. A bibliometric survey complemented with a narrative review or metaanalysis may be beneficial in Plesiomonas research. Future research is needed to answer questions related to what particular strain of the microorganisms are pathogenic and how to differentiate pathogenic variants from non-pathogenic ones.
The two mega-clusters or spheres in the top authors' collaboration and coupling networks on P. shigelloides studies showed collaboration pathways mainly among authors from highincome countries, which is similar to trends observed in collaboration network analyses of human immunodeficiency virus and human papilloma virus studies [78]. Alliances between developing and developed countries are rare in a number of scientific areas [78]. Among researchers in the United States, collaboration pathways were largely intra-national, as suggested by a large number of publications but 6 multiple country publications. In contrast, collaborations by authors in Cuba, Finland, Czech Republic, Italy, Slovakia, and Sweden tended to be multi-national, which is more valuable for the epidemiological control of pathogens. The absence of collaboration pathway in Venezuela, India, Nigeria, and other African nations is consistent with the low number of publications from these countries. Intra-and international collaborations between developed and developing nations could provide opportunities for the division of labor and resources to address important scientific questions.
There were some limitations related to the bibliometric survey adopted in this study, including the use of a single database (the WoS), the low sensitivity and strictness of the search terms and search strategy used, and the exclusions of other document types (e.g., meeting abstract, note and proceeding papers etc.) and articles published in non-English Language (e.g in Chinese). Also, the current analysis did not allow for a narrative review and judgment on contents and results of the articles [79]. As noted earlier, emerging themes and recent research focus in Plesiomonas are not easily recognized in bibliometric studies due to low frequency of appearance in keywords. Notwithstanding these limitations, this is the first bibliometric study on Plesiomonas-related research contributing to the evidence base and would help direct future research. Also, the WoS has a larger coverage compared to other database, reliable indexing technology that minimizes the ''indexer effect" and is well accepted among scientific communities [80].

Conclusion
Our bibliometric analysis revealed a global diminishing research in Plesiomonas, greater research output from high-income countries compared to low-and middle-income countries and limited collaboration with developing countries. The low productivity in developing countries in Plesiomonas research mirror the state of affairs in other research fields. A better understanding of the clinical features, epidemiology and Plesiomonas-associated diseases is needed in countries with high infection rates. Emerging themes and recent research focus in Plesiomonas research are not easily recognized in bibliometric studies due to low frequency of appearance in keywords and, hence, the need for future studies guided by narrative reviews.
Supporting information S1 Fig. Co-occurrence network of top 20 terms associated with P. shigelloides studies. Each node in the network represents a different term. The node's diameter corresponds to the frequency of co-occurrence with other terms. Lines depict co-occurrence pathways between terms. (TIF) S2 Fig. Keyword co-occurrence networks related to P. shigelloides studies. Each node in the network represents one of the top 20 keywords. The node's diameter corresponds to the keyword's frequency of co-occurrence with other keywords. Lines depict co-occurrence pathways between keywords. (TIF) S1