The bibliometric analysis relies on literature databases. At present, widely used databases include Scopus, Web of Science, Pubmed, Embased, Cochrane Library et al. Among them, the Web of Science database contains large-scale multidisciplinary, high-impact, international, and comprehensive academic journals. Meanwhile, evidence has shown that the Web of Science database provides a better knowledge map effect when CiteSpace is used for visual analysis (Falagas et al., 2008; Martín-Martín et al., 2018). Therefore, it is rational and effective to select the Web of Science database as our study data source. Specifically, the data were collected from the Web of Science Core Collection (WoSCC), including the SCI-EXPANDED, CCR-EXPANDED, and Index Chemicus. Due to the daily database updates, to avoid bias, we searched the literature retrieval from WoSCC on a single day, June 19, 2021.

The search strategy included the topic “pain catastrophizing” with the literature type limited to “ARTICLE” or “REVIEW.” Articles published from 2010 to 2020 in English were retrieved and we checked the relevance of the results. Finally, a total of 1,576 references were obtained. We saved the document data in the form of full records and cited references as plain text format. Some preparatory work is needed before data is imported into the CiteSpace. We created a new folder and created four folders in the folder named input, output, data, and project. Saved the file previously exported in WoSCC to input and name it “download _ ^**.txt” format recognized by CiteSpace. The ^** represents the number. Then opened the CiteSpace to perform data format conversion and deduplication, visualization, and other operations. The results remained unchanged after data duplicate checking in the CiteSpace software.

CiteSpace is a bibliometric analysis visualization software developed by Professor Chaomei Chen that uses the Java platform (Chen, 2004). It is an interactive analytic tool enabling visualization tasks in science mapping through a combination of bibliometrics, visual analytic methods, and data mining algorithms (Zhu et al., 2021). CiteSpace provides kinds of function selection for bibliometric studies, including collaboration network analysis, co-citation analysis, and co-occurrence analysis and can generate visual maps (Chen, 2017). By generating a series of visual knowledge maps, CiteSpace explores the research status, research hotspots, research frontiers, and evolution process of a scientific field, to reveal the research direction, research stage, and frontier characteristics of institutions and authors, and finally judge the development trend of this field.

The version of this software is constantly being updated. The version used in this research is 5.7 R1 (64-bit). The parameters of CiteSpace were as follows: time-slicing was performed from January 2010 to December 2020 (1 year per slice), all options in the term source were selected, one node type was selected at a time, selection criteria (top 50 objects), and pruning (pathfinder). Nodes and links were used to generate visualization knowledge maps. Each node in the map represented an element to be analyzed, such as a cited journal, country, or author. The size of the node represented the frequency of the citation, and nodes of different colors indicated different years. The connection lines between nodes were regarded as the co-occurrence or co-citation relationship; the lines’ thickness meant the strength of the relationship, and the color corresponded to the first co-occurrence or co-citation time of nodes. Colors from cool to warm represent early to recent. The centrality was also named betweenness centrality. Nodes with high centrality (> 0.1) were usually considered turning points or pivotal points in a field. When the CiteSpace default number of network nodes was greater than 350, the centrality calculation function would be closed. We need to manually click the “compute node centrality” function in the node menu.

A total of 1,576 publications were retrieved. Among these, 1,510 (95.81%) were research articles, 66 (4.19%) were review articles. The number of annual publications is shown in Figure 1. The number of publications increased from 58 in 2010 to 294 in 2020, but with some fluctuation. As shown in Figure 1, the trend of increase in publications can be divided into three stages: the initial stage (2010–2011), with a slow rate of publications; the smooth growth stage (2012–2016), with a slight continued upward trend; and the rapid growth stage (2017–2020), with accelerating output, and reached a peak in 2020.

The number of journals that published 1,576 papers on pain catastrophizing research was 342. Many journals were pain-related specialized journals, other remaining journals were physical, psychological, and specialized diseases journals. The top five journals are listed by the number of publications in Table 1 and publishers of these journals were mostly located in the United States. The five journals all belong to the pain journal category. Pain ranked first in the frequency of the journals and the IF ranking. Only one journal’s impact factor exceeded 5, whereas the average impact factor of the remaining journals was approximately 3.38.

Co-citation analysis, one of the most important indicators, has been widely applied in bibliometrics.

Co-cited journals were those cited together by other researchers. Through co-citation of journal analysis, we can obtain a distribution of key knowledge sources in a field. Table 2 presents the top five cited journals with the highest frequency and centrality of pain catastrophizing research. The most frequently cited journal was Pain, followed by Clinical journal of pain, Psychological assessment. In terms of centrality, the top five journals were Physiotherapy research international, Biological psychiatry, Gastroenterology, American journal of industrial medicine, European spine journal.

A distribution map of countries was generated; the merged network comprised 62 nodes and 118 links (Figure 2). The nodes and links between them reveal the countries and cooperative relationships, respectively. The large the node, the more publications. In addition, the wider the line, the stronger the relationships. In total, 1,576 references were published by research groups in 62 countries; the top five countries were the United States, Canada, Netherlands, Australia, and Belgium (Table 3). In total, 642 articles were published in the United States, therefore, it was in the top position in terms of the research on pain catastrophizing. Pain is a significant public health concern in the United States. The United States, Canada, Netherlands, Belgium were the most prolific countries in Northern America and Europe, while Australia was the most productive country in Oceania. The top five countries in terms of centrality were England (1.05), Sri Lanka (0.98), Scotland (0.91), Australia (0.89), and China (0.86). The centrality of these five countries is greater than 0.1, indicating that these five countries have a certain influence in the study of pain catastrophizing.

A distribution map of institutions was generated; the merged network comprised 361 nodes and 431 links (Figure 3). The 1,576 references were published in 361 institutions; the top five institutions were the University of Washington, McGill University, Ghent University, University of Florida, and Maastricht University (Table 4). The top five institutions in terms of centrality were The University of Queensland (0.41), Queen’s University (0.38), Auckland University of Technology (0.37), Brigham and Womens Hospital (0.33), and Aarhus University Hospital (0.29). Generally speaking, institutions were mainly concentrated in universities and a few in hospitals.

Concerning the number of publications, Jensen MP from the University of Washington was the most prolific author. One of his articles identified a biological factor that may be associated with greater vulnerability to pain-related catastrophizing, which may contribute a new possibility for treating catastrophizing (Jensen et al., 2015). George SZ, Edwards RR, Meeus M, and Ring D were also active in the field of pain catastrophizing research (Table 5). As shown in Figure 4, the map of the authors comprised 439 nodes and 1,251 links. The size of the nodes in the cooperation graph represents the number of articles published by the authors, and the thickness of the lines between them reflects the strength of their cooperation relationship.

CiteSpace was used to generate a co-author map comprised of 653 nodes and 829 links. Sullivan MJL ranked the highest about citation counts (1,196), followed by Edwards RR (335), Osman A (334), Vlaeyen JWS (330), and Crombez G (307) (Table 5). From the cluster summary, the authors devoted their mind to scale development, mechanism exploration, related outcomes report, intervention research of pain catastrophizing. Sullivan MJL, a specialist in psychology from McGill University, focused on the development of the PCS and the outcome of pain catastrophizing, such as poor response to disease and poor quality of life. The top five cited authors in centrality were Staud R, Katz J, Walker LS, Arendt-NielsenlI L, and Zale EL.

The top five cited references about frequency and centrality are shown in Tables 6, 7. According to the ranking of frequency and centrality in cited references, most were review papers, a few were original research papers. The first ranked citation in terms of frequency was the article published in Expert Review of Neurotherapeutics titled, “Pain catastrophizing: a critical review” (Quartana et al., 2009). In this review, Quartana PJ provided a detailed explanation on the conceptualization of pain catastrophizing, focused the discussion on a number of theoretical mechanisms of action and offered evidence to show that pain catastrophizing represents an important process factor in pain treatment.

About the centrality of the cited references, the study that was ranked first was conducted by Asmundson, a Ph.D. in psychology from Canada. In this review, the author described the current state-of-the-art regarding the co-occurrence of anxiety disorders and chronic pain and pointed out that pain-related catastrophizing may maintain or exacerbate clinically significant symptoms of anxiety (Asmundson and Katz, 2009). The network map of the cited references comprised 697 nodes and 961 links (Figure 5). Over the past decade, a comprehensive analysis referring to multiple studies mainly focused on the following: the concept of pain catastrophizing analysis; its related mechanism exploration and outcomes report; the development, reliability, and validity test and localization of the PCS to adapt it to the cultures of various countries in the world; and the tentative exploration of intervention measures to reduce pain catastrophizing.

It was believed that research frontiers could be identified by analyzing the frequency and centrality of keywords. The map of keyword co-occurrence was generated; it consisted of 502 nodes and 777 links (Figure 6). We found that “low back pain” was the most popular keyword after removing the “pain catastrophizing” word. Low back pain is one of the most frequent problems worldwide, with an estimated global prevalence of 40–85%, and is the leading cause of functional decline and disability (Tagliaferri et al., 2020). A survey by Coyne also demonstrated that among 809 participants, the most common pain condition was lower back pain, with an incidence rate of 76.6% (Coyne et al., 2021). It is commonly believed that most people with low back pain can recover in a few weeks. However, reoccurrences are common and considerable fractions may go on to develop chronic low back pain. The occurrence of low back pain episodes mainly accounts for biomechanical factors, but psychosocial factors seem to take an important place in chronicity (Ibrahim et al., 2021). One important psychological factor linked with chronic low back pain is catastrophizing. Pain catastrophizing can predict the development of chronic back pain within one year after painless baseline and chronification of acute back pain (Taub et al., 2017). According to the frequency and centrality, the other popular keywords were “validation,” “disability,” “chronic pain,” “depression,” “quality of life,” “anxiety,” “attention” (Table 8).

“Burst words” mean that words are cited frequently over some time. We can predict the research frontier according to the distribution of keywords with the strongest citation burst. The top 22 keywords with the strongest citation burst from 2010 to 2020 are shown in Figure 7. The red bars demonstrated that the keyword was cited frequently, the green bars showed that the keyword was cited infrequently. Functional disability, population, and total hip would be potentially cited frequently over the coming years, which represent the emerging trends.