Scientometric overview of nursing research on pain management

ABSTRACT Objective: to analyse research articles on pain and nursing issues using bibliometric and scientometric methodologies. Method: articles in the Web of Science database containing pain and nurse and pain and nursing were analyzed using scientometric methods through data visualization techniques and advanced text analytics. Result: among the 107,559 research articles found in the field of nursing, 3,976 of them were written based on the keywords pain and nursing, and were considered in conformity with the scope of this study. Preliminary analyses indicated that the publications have increased through the years with minor fluctuations. Titles, keywords, and abstracts were analyzed through text analytics to reveal keyword clusters and topic structures. Studies on oncology and pain in the field of nursing have a relatively higher frequency. Conclusion: the results of the analyses revealed the characteristics of the current literature in a broad range of areas by considering the particular dimensions. Therefore, the findings may support present and future research in this field by shedding light on the networks, trends, and contents in the related literature.

about pain applications, and 25% were about pain evaluations. Additionally, the most cited articles about pain (13) have been analyzed using bibliometric methods on types of pain, such as acupuncture studies (14) , orofacial pain research production (15) and migraine research studies (16) .
The published research on pain in the related literature has increased, not only in the field of medicine, but also in the field of nursing. The general demographics and emergence of the hidden patterns over the years in the related literature can be extracted through scientometric and bibliometric approaches, with the help of their detailed and analytical techniques. In this context, the studies on pain and pain management can be decomposed into different components concerning the information provided by publishers, e.g., title, author(s), abstract, references, publication information, funding, publication impact factors, location, and keywords. Even though the field of nursing has often been addressed in scientometric or bibliometric studies, pain and pain management have not been investigated from this perspective to date.
Bibliometrics can be considered an essential methodology, used to evaluate the academic performance of nursing studies (17)(18) . Quantitative bibliometric measures are used to assess the impact of research outputs, and can also be used as tools by librarians to manage collections and provide relevant resources to users (19)(20) . The bibliometric tools can reveal the trends in nursing terminology and include analyses of core journals, indicators of scholarly output, and the co-author network associated with journal articles (21) . Furthermore, various insights about the intellectual and social structure of a field, as well as research performance and dissemination of ideas can be extracted from the available data with respect to different dimensions, such as authors, documents, journals, words, indicators, metrics, and techniques.
That is, counts, correlations, clustering, and network analyses can reveal information about authorship, types of documents cited, journal distributions, and how works are connected by highlighting the patterns, trends, identified interests and spreading (22)(23) .
The necessary datasets can be retrieved from numerous online databases, such as Web of Science (WoS) or Scopus (24) . Network analyses and text analytics techniques in the scope of scientometrics are especially useful as a way of mapping a research field, and although they have been widely used in many fields, they have not yet been used in the field of pain and nursing. This motivation determined the goal of this study, which is the investigation of research articles on pain and nursing issues using bibliometric and scientometric methodologies. This study presents the covered topics, trends in the cited journals and authors, the funding in countries, and the status of organizations and works, and thus illuminates the development in this field, providing a broader perspective on the current status of the literature.

Method
This descriptive and exploratory study can be classified as both bibliometric and scientometric research, since it includes tables and graphs to present descriptive statistics and uses advanced text analytics and network analyses to reveal hidden patterns in the content of abstracts, with regard to the relationships among the terms. The methodology was designed as an end-to-end process, beginning with the dataset retrieval For the analytical stages of the methodology, several software tools were utilized, i.e., the VOSviewer (25) and Microsoft Excel for descriptive statistics, network/ density visualizations, and clustering on the networks; RapidMiner (26) for text preprocessing; and a java application for Hierarchical Latent Tree Analysis (HLTA), for topic modeling in the abstracts.
Bibliometric mapping is a quantitative approach that aimed at visualizing various bibliometric aspects of scientific publications, performed in the form of different networks. In this case, the authors induced scientific landscapes, used for content analysis, and bibliometric networks to present co-authorship and co-citation. For this purpose, the VOSviewer, a software package for analyzing and visualizing large bibliographic datasets (25) , was preferred for the graphical representations of this study concerning various dimensions, such as journals, authors, countries, organizations, and individual publications. The network representations can be built by co-authorship, co-citation, or other bibliographic relationships.
For the text analytics on titles, keywords, and abstracts in the dataset, a text preprocessing data model was performed for tokenization, filtering stop words, and part-of-speech-tags, including names and verbs, stemming, and other filters required in the further analyses. The data model was built using the RapidMiner 7.6 software platform. The data model produced preprocessed text data that were stored in a spreadsheet in which the text for each article was saved in a single cell. To provide the proper inputs for HLTA implementation to obtain the topic structure in the data, the text data in each of the cells of the spreadsheet was converted into a single text file using the Visual Basic for Applications (VBA) as coding tool. The group of text files was then processed through a progressive expectation maximization algorithm for topic detection using the suggested parameters (27) . HLTA provided insights from the content of the articles and shed light on the main topics that appeared in the pain studies that were within the scope of nursing research.

Results
During this study, 189,885 publications in the nursing category were reached through the WoS databases. This number indicated that the nursing field has a particular place in health research. When this comprehensive dataset was filtered, it was found that 107,559 (56.64%) documents were research articles, and of these articles, 3,976 (3.55%) contained the words nursing* and pain* or nurse* and pain*, in the scope of their topics. This ratio, which is related to the concepts of pain and pain management, constitutes 3.69% of the articles that exist in the field of nursing. Bibliometric      (24) . The top three universities that contributed the most to this field in the world Keywords were also analyzed to map the distribution of the articles containing these words. The authors of the articles defined 5,745 keywords in total; 194 of these words were repeated 10 times or more. The articles were subjected to a cluster analysis concerning their keywords, and eight clusters were obtained, as depicted in Figure 2.

Discussion
Research articles about pain constituted 3.55% of all nursing publications, and this percentage indicated that the subject, pain in nursing, is an intensively studied subject in the related field, and had an increasing trend during the period from 1975 to 2017. This trend can be explained by the fact that pain is a problem that many patients experience and is subject to various research attempts in the evaluation and management of pain (4) .
The H-index value of the topic is 67, and high enough Rev. Latino-Am. Enfermagem 2018;26:e3051.
to support these observations. Thus, the past and recent trends can provide much useful information to researchers.
In this study, 9.28% of pain-related articles in the field of nursing were related to the field of oncology. This result supports the findings that the first three words