Scientic Production in Cerebrospinal Fluid Leak (Rhinorrhea and Otorrhea) from 1945 to 2018

BACKGROUND: To analyze research activity on cerebrospinal uid (CSF) leaks in general and CSF rhinorrhea and otorrhea in particular, and to identify the main topic clusters in these areas. METHODS: We identied all documents using the medical subject heading “Cerebrospinal Fluid Leak” indexed in the MEDLINE database between 1945 and 2018. We performed a descriptive bibliometric analysis and analyses of networks and research clusters in order to identify the main topic areas of research. RESULTS: From 1945 to 2018 a total of 4130 records were published with the term CSF, including 2821 documents (68.1%) with the term CSF rhinorrhea and 1040 documents (25.8%) with CSF otorrhea. The number of documents published increased from 10 in 1945-49 to 642 in 2010-14. Articles were the dominant document type (86.8% of the documents analyzed), while case reports were the main type of study (37.4%). In terms of geographic distribution, researchers from the United States of America (USA) led in the number of signatures (39.06%), followed by those from the United Kingdom (UK) (7.49%). The most active areas of research in the eld were “Postoperative Complications,” “Tomography, X-Ray Computed,” and “Magnetic Resonance Imaging.” The terms “Adults,” “Young Adult,” and “Middle Aged” were most common in CSF rhinorrhea research; and the terms “Infant,” “Child, Preschool,” “Child,” and “Adolescent” were more common in CSF otorrhea. CONCLUSIONS: Articles and case reports related with “Surgery” and “Postoperative Complications” associated diagnosis are the main of study, highlighting the importance of this document type in advancing


Scienti c Production in Cerebrospinal Fluid Leak (Rhinorrhea and Otorrhea) from 1945 to 2018
Background Cerebrospinal uid (CSF) leak is a discharge of cerebrospinal uid through a hole in the skull bone, most commonly draining from the nose (CSF rhinorrhea) or from the ear (CSF otorrhea), occurring either through the external auditory meatus or through the eustachian tube into the nasopharynx) [21]. Common etiologies of CSF rhinorrhea include trauma, neoplasms, and prior surgery, while CSF otorrhea is usually associated with craniocerebral trauma, neurosurgical procedures or other conditions; however, both conditions may occur spontaneously [9] The past few decades have seen an increase in research publications in the eld of otorhinolaryngology [8] Bibliometrics, too, has consolidated as a discipline enabling the analysis of scienti c activity in a particular eld or area of knowledge through the quanti cation of the bibliographic characteristics of scienti c publications [5] Several studies have analyzed the scienti c production of otorhinolaryngology and head-neck surgery [3,17,18] and neurosurgery [1,2,7,11,15,16,23]. Moreover, there are several publications focusing speci cally on the activity in different countries, including African countries [1].
Other manuscript analyze speci c research topics neurosurgery, like central nervous system arteriovenous malformations [15], idiopathic intracranial hypertension [16], endoscopic third ventriculostomy [23] carotid artery stenting [2] pituitary adenoma [7] o IDEAL framework in neurosurgery [11] One of the areas that has seen signi cant development within bibliometrics is the generation of visual representations, networks and maps of scienti c activity [5]. These have great analytic potential, helping to characterize the structures, groupings and interconnections between the elements under analysis, including the descriptors assigned to the documents, such as in the present study. Although bibliometric indicators have been used to approach the study of numerous diseases [5,6], this methodology has not been applied to research on the well-known condition of CSF, which encompasses surgical, diagnostic, and therapeutic-radiological aspects, among others.
The aim of this study is to describe the scienti c production on CSF leak through bibliometric indicators, using the documents indexed in the MEDLINE database between 1945 and 2018. We analyze the evolution in the number of publications, the publication types, the document categories of clinical interest, the journals of publication, the geographic distribution of the research and the topics addressed according to the different types of CSF leak (CSF rhinorrhea and CSF otorrhea) by means of co-word maps.

Identi cation of documents on "Cerebrospinal Fluid Leak"
The rst step of this study was to identify the documents about CSF leak in the MEDLINE database, the main database of reference for researchers and professionals in the health sciences. It is an open access resource that permits, through the use of the medical subject headings (MeSH) thesaurus, the precise identi cation of documents that address the concepts under analysis as well as the contents of those documents. The CSF leak descriptor was included in 2005. This term encompasses the speci c terms "Cerebrospinal Fluid Rhinorrhea" and "Cerebrospinal Fluid Otorrhea," both included in 1966. The searches of indexed documents with any of these three descriptors took place on the Clarivate Analytics Web of Science (WOS) platform, which includes the MEDLINE database, on January 21, 2019.
The Clarivate Analytics WOS platform provides information on the institutional a liation of all authors.
With this strategy, 4155 documents were retrieved, but we excluded 25 editorials, leaving 4130 documents to be analyzed. The analysis of the scienti c production by country was limited to the 3019 documents (73.1%) included in WOS database.

Standardization of bibliographic information
Using the bibliographic information from the retrieved documents, we created a relational database to standardize the bibliographic data and calculate the study variables, with special attention to the following elds: Publication type ( eld: Publication Type). Based on the information collected from this eld, we classi ed the records according to the document types (articles, reviews and letters) as well as the clinical approach (case reports, clinical trials, controlled clinical trials, evaluation studies, metaanalysis, observational studies, practice guidelines, randomized clinical trials, and validation studies).
Institutional a liation of papers included in WOS ( eld: Author Address). This eld includes information relating to the institutional a liation of the authors signing the documents. Although all institutional a liations have been available for papers published in MEDLINE since 2014, we aimed to study and present the information using a standard approach, so we limited the analysis to the papers included in the WOS.
Descriptors ( elds: MeSH Terms [MH] and MeSH Subheadings (also called quali ers) [SH]). We differentiated the MeSH from the subheadings or topic quali ers (82 auxiliary descriptors that specify the aspect being addressed in relation to the MeSH terms) and identi ed the descriptors assigned with respect to CSF leak, CSF rhinorrhea, and CSF otorrhea.

Calculation of indicators A) Bibliometric indicators on scienti c production
To analyze the scienti c activity of the area, we determined the number of documents published each year, the journal of publication, the document types, the clinical approach to the study, and the countries of the authors signing the documents. This information generates a picture of the scienti c activity, the media used (scienti c journals), and the weight of countries leading the research-in general and according to the different types of CSF (rhinorrhea or otorrhea).

B) Topic analysis
To characterize the research in the area at a topic level, we determined the frequency with which the descriptors assigned to the documents appeared for each group: the generic CSF leak group and the speci c document groups on CSF rhinorrhea and CSF otorrhea. We constructed topic maps to analyze the relationship between the descriptors and the research groups around them, that is, the existing research clusters, approaches, and specializations. This process had the following steps: Determination of the co-occurrence of the descriptors assigned to the documents and generation of a matrix of absolute values. The joint assignment of two descriptors in a single document implies a thematic a nity, as both aspects are addressed simultaneously in the same paper.
Elimination of generic descriptors. We eliminated some excessively generic descriptors that did not describe relevant information on content or that presented a high density of relationships with the rest of the descriptors.
Visual representation of the network. To establish the existing topic clusters in the area and represent them visually, we used a clustering algorithm in the VOSViewer program, which helps to detect the communities (clusters) within a network, made up of groups of homogeneous items that are strongly related to each other [21].

Statistical analysis
We compared the MeSH terms in CSF rhinorrhea and CSF otorrhea documents using the Chi-squared test. P values of less than 0.05 were considered statistically signi cant. We used statistical software SPSS version 22.0 for analysis.

Ethical considerations
Due to the nature of the study and dataset, it was not necessary to obtain informed consent or approval from an institutional ethics committee

Bibliometric indicators
Evolution of production From 1945 to 2018, 4130 articles were published with the CSF leak terms. The term CSF rhinorrhea was in 2821 documents (68.1%) and CSF otorrhea, in 1040 (25.8%). Both terms were present in 316 documents, while 269 documents used only the generic CSF leak descriptor. Figure 1 shows the evolution of the number of documents published on CSF leak.

Publication type distribution
The main publication type was article, which encompassed 86.8% (n = 3584) of the documents; followed at a considerable distance by review, which made up 9.4% (n = 390); and letter, at 3.8% (n = 156).

Geographical distribution
Of the 3019 included in WOS, 2578 showed the country of the institution writing the manuscript. Table 2 shows the number of documents published on CSF by the 15 most productive countries, both as a total and by type of CSF. Only 160 of the 2578 documents with country data (6.21%) show evidence of international collaboration. The leading country in total scienti c production on CSF-as well as speci c production on CSF rhinorrhea and CSF otorrhea-is the United States of America (USA), followed by the United Kingdom (UK). The other countries with the largest research contributions were Japan, Germany, Italy, and France, with little difference between CSF rhinorrhea and CSF otorrhea. Topic analysi

Quali ers
The most common quali ers were "Surgery" and "Etiology." The quali ers "Surgery" and "Methods" were signi cantly more common in CSF rhinorrhea, whereas "Abnormalities" was signi cantly more common in CSF otorrhea (Table 3).
Generic MeSH Terms Table 4 shows the frequency of MeSH terms referring to human/animal research, female/male, age groups and type of study, in all CSF documents and by type of CSF. Most research was in humans. The proportion of documents relating to males and females is similar in CSF rhinorrhea and CSF otorrhea. By age, the terms "Infant," "Newborn," "Adults," "Young Adult," and "Middle Aged" are more common in CSF rhinorrhea; and the terms "Infant," "Child, Preschool," "Child," and "Adolescent" were signi cantly more common in CSF otorrhea. The most common MeSH terms describing the type of study was "Retrospective Studies" and "Follow-Up Studies," with similar results in both types of CSF.  and their ties. Among the most important links were "Treatment Outcome" and "Postoperative Complications" with "Reconstruction Surgical Procedure," and with regard to diagnosis, "Tomography, X-Ray Computed" and "Magnetic Resonance Imaging." In CSF rhinorrhea, the most important MeSH terms were "Tomography, X-Ray Computed," "Endoscopy," and "Postoperative Complications" (Fig. 2b). In CSF otorrhea the most important were "Tomography, X-Ray Computed," "Retrospective Studies," "Treatment Outcome," and "Tempore Bone" (Fig. 2c).

Discussion
The number of articles published on CSF increased throughout the study period, which is normal, as most bibliometric analyses on other diseases also reveal an upward trend in the number of publications [3,13,10]. The most signi cant aspect to point out is the predominance of case reports (approx. 35%), which show similar values as for other surgical areas, such as maxillofacial surgery (31%) [10]. Individually, case reports are insu cient grounds on which to base treatment decisions, but when they are considered together and adequately codi ed and integrated into structured information systems, they can provide early insight toward characterizing rare diseases, as they allow physicians to compare cases and check diagnoses [5].
At a journal level, this study shows the important multidisciplinary approach in the area, as the three most prominent publications are journals of otorhinolaryngology, surgery, and clinical neurology. The multidisciplinary approach of the science is a scienti c reality that has been associated with greater advances in knowledge, translation of results, and impact of research [4,6].
The USA was the predominant country in research production, followed by various European countries (UK, Germany, France and Italy) and Japan. This situation is also apparent in the otolaryngology eld [3,17,18], neurosurgery [2,15,16], as in other health sciences [3,10]. The top countries did not include any from Africa, while the only Latin American country was Brazil. The top-contributing Asian countries (in addition to Japan) were India, China, and South Korea, while Turkey ranked eighth. A combination of factors can explain these observations. First of all, the USA emerges as the main point of reference in CSF, re ecting its global leadership in all scienti c research spheres. Furthermore, the need to have a surgical structure to produce research in this eld would favor the most developed countries to a greater extent [3,7,18].
The level of international collaboration was very low (6.21%), which highlights the need to develop surgical structures that favor the promotion of research in countries with less scienti c development and implement strategies that favor multidisciplinary collaboration [4].
With regard to the MeSH terms, CSF rhinorrhea was the main branch, while surgery and postoperative complications were the most important sub-topics. Relevant topics related to managing patients with CSF were otorhinolaryngology and clinical neurology [21]. Research also focused on identifying risk factors and assessing different treatments and their outcomes, as shown by the terms related to etiology ("Skull Fracture" of "Skull Base"), diagnosis (especially via "Tomography, X-Ray Computed" or "Magnetic Resonance Imaging"), treatment ("Treatment Outcome" or "Endoscopy"), and complications (meningitis).
There is difference in the patient pro le of research in CSF rhinorrhea and CSF otorrhea, with the former area concentrating on older patients ("Adult" or "Middle Age"). Spontaneous CSF rhinorrhea is associated with increased intracranial pressure and considered a manifestation of idiopathic intracranial hypertension in middle-aged people, whereas secondary CSF rhinorrhea is associated with trauma in the same age group [12]. On the other hand, CSF otorrhea research focused on the "Infant," "Child, Preschool," "Child" and "Adolescent" age groups, re ecting the fact that CSF otorrhea can be primary and in most of cases is secondary to pediatric skull base fractures [19]. Thus, even though CSF rhinorrhea and CSF otorrhea are included under the same umbrella MeSH term, the pro le of research is quite different in these sub-elds.
The main limitation of this study was that we did not analyze citations, with a focus on the journals with the highest impact and dissemination at an international level; this perspective would be necessary to reach a truly comprehensive view of research in the eld. [10] Likewise, other aspects could also be considered, such as co-authorship networks or gender disparities in scholarly productivity [6,18]

Conclusion
The present study has revealed some features that differ notably from bibliometric analyses on other clinical pathologies: the predominant interest in surgery and the weight of clinical case reports, which emerge as the primary channel for generating and disseminating knowledge in the eld. Postoperative complications, diagnostic aspects related to computerized tomography and magnetic resonance imaging, treatment outcomes, and surgery procedures such as endoscopy were also topics of interest.
The USA led research production, while European countries and emerging countries such as China contributed less than in other research areas. For this type of study formal consent is not required.

Consent for publication
All the authors give their consent for publication Availability of data and material J.M. Ramos-Rincon has full access to and is the guarantor for the data. The datasets generated are available from the corresponding author on reasonable request.

Competing interests
None declared