Global Trends and Current Status in Osteonecrosis of the Femoral Head: A Bibliometric Analysis of Publications in Recent 30 Years

Background: Osteonecrosis of the femoral head (ONFH) is a progressive and disabling disease with heavy socioeconomic burdens. The purpose of our study was to summarize the global trends and current status in ONFH. Methods: Publications related to ONFH from 1991 to 2020 were searched from the Web of Science (WOS) core collection database. The data were analyzed with bibliometric methods. Microsoft Excel was used to statistical analysis and draw bar charts. SPSS was applied to perform linear regression analysis. VOSviewer was used to conduct bibliographic coupling analysis, co-authorship analysis, co-citation analysis, and co-occurrence analysis. Results: A total of 5,523 publications were covered. The United States consistently ranked rst in total publications, sum of times cited, average citations per item, and H-index. Kyushu University was the main contributor to ONFH. Clinical Orthopaedics and Related Research was the major publishing channels for ONFH-related articles. Takuaki Yamamoto published the most ONFH-related articles. Studies regarding ONFH could be divided into 5 clusters: mechanism study, treatment study, complication study, radiological study, and etiological study. Mechanism study might become the hot spot in the future. Conclusions: This study identied the last 30 years’ articles in ONFH and summarized their global trends and current status, which classied them by country, institution and author, publication, funding agency, and direction. This study will help researchers understand the research perspectives, hot spots, and trends of ONFH.


Introduction
Osteonecrosis of the femoral head (ONFH) is a progressive and disabling disease with heavy socioeconomic burdens [1]. Severe ONFH leads to the collapse of subchondral bone and eventually the damage of the hip joint, resulting in a loss of labor capacity [2]. Patients with advanced ONFH often do not respond well to medical treatment and require total hip arthroplasty (THA). However, THA, especially for young people, may cause a series of complications, including dislocation, periprosthetic fracture, infection, and prosthesis loosening [3]. The etiology of ONFH is various, among which the overuse of glucocorticoids (GCs), trauma, and alcohol are the most critical ones, while the pathogenesis of ONFH remains to be determined. A growing number of studies have focused on the pathogenesis and therapeutic approaches of ONFH worldwide. However, global trends and current status in ONFH have not been summarized yet.
Bibliometrics is a method to cognize the development tendency of a certain eld and evaluate the contribution of a collection of research results such as all publications of the same scholar, institution or country by collecting literature metrology characteristics of the publications [4][5][6]. In addition, bibliometrics can also be used to guide policy formulation [7]. Nowadays, bibliometric analysis has been used in a wide range of elds, including anesthesia, cancer, orthopedics and, neurology, to compare the contributions of different research ndings [5,[8][9][10].
Therefore, the purpose of this study was to evaluate and summarize the global trends and current status in ONFH during the last 30 years, helping researchers understand the research perspectives, hot spots, and trends of ONFH.

Data sources
The search was conducted using the Web of Science (WOS) Core Collection database, including SCI-

Bibliometric analysis
Microsoft Excel 2017 was used for statistical analysis and the graphing of all the bar charts in the study. In addition, SPSS (v.26.0, IBM, New York, USA) was applied to perform linear regression analysis on the trends of total publications in the past 30 years. P < 0.05 is considered statistically signi cant. R 2 represents the degree to which the linear regression model explains the overall variance [11].
VOSviewer is a software for plotting maps based on network data. In the network visualization, items are represented by circles. The size of the circle is determined by the number of publications of the item. The distance between two circles approximately indicates the relatedness of the items. The color of an item is determined by the cluster to which the item belongs. In this study, VOSviewer was used for conducting bibliographic coupling analysis, co-authorship analysis, co-citation analysis, and co-occurrence analysis.

Global trends of publications in ONFH
A total of 5,523 ONFH-related articles published between 1991 and 2020 was identi ed in our study. The total publications of ONFH generally increased over time (R 2 = 0.872, P < 0.001). The annual publications regarding ONFH have grown nearly sevenfold over the three decades from 61 in 1991 to 481 in 2020 ( Figure 1A). A total of 98 countries or regions have published studies on ONFH ( Figure 1B). Among them, the United States published the most articles (1,369, 24.787%), followed by China (1,246,22.560%), Japan (544, 9.850%), Germany (303, 5.486%), England (283, 5.124%) and South Korea (281, 5.088%). The United States and China published more than twice as many articles as Japan ranking third. Moreover, the total number of ONFH publications in each country or region also increased over time: the United States (R 2 = 0.732, P < 0.001), China (R 2 = 0.661, P < 0.001), Japan (R 2 = 0.525, P < 0.001), Germany (R 2 = 0.755, P < 0.001), England (R 2 = 0.679, P < 0.001) and South Korea (R 2 = 0.805, P < 0.001) ( Figure 1C). Figure 1D shows the number of publications from individual countries or regions visually in the heat map.
Quality analysis of global publications Country Figure 2A illustrates the sum of times cited, average citations per item, and H-index of the top ten countries with the most publications associated with ONFH. Among the ten countries, sum of times cited (45,314), average citations per item (33.1), and H-index (94) of the United States are all higher than the other nine countries. In terms of the total publications, China ranked second. However, China ranked ninth in average citations per item (11.33), just above India. Other developed countries, including Japan, Germany, England, South Korea, France, Canada, and Italy, all had high average citations per item and Hindex despite the small number of publications.

Institution
Over the past three decades, approximately 3,721 institutions worldwide have published ONFH-related literature. Figure 2B  Author Analyzing quality of publications by the author, the top ten contributors to ONFH are presented in Figure  2C. Of the ten authors, ve were from Japan (Takuaki Yamamoto, Yukihide Iwamoto, Goro Motomura, Nobuhiko Sugano and Yasuharu Nakashima), two were from China (Dewei Zhao and Changqing Zhang), two were from the United States (Michael A Mont and Harry K W Kim), and the remaining one was from South Korea (Kyung-Hoi Koo). Moreover, eight of these authors were from the top ten institutions, four were from Kyushu University (Takuaki Yamamoto, Yukihide Iwamoto, Goro Motomura and Yasuharu Nakashima), Nobuhiko Sugano was from Osaka University, Dewei Zhao was from Dalian University, Changqing Zhang was from Shanghai Jiao Tong University and Kyung-Hoi Koo was from Seoul National University. The greatest contributor was Takuaki  Co-authorship analysis Country Co-authorship analysis is a measure to determine the connectivity of items based on the number of coauthored publications. Figure 5A shows the relationship of 59 identi ed countries (the minimum number of documents of a country is over ve) in total link strength using VOSviewer.

Co-citation analysis Journal
Co-citation analysis refers to a method presenting the relatedness of items based upon the number of times they are cited together. Figure 6A

Co-occurrence analysis
The relatedness of items is determined by the number of documents in which they occur together. The aim is to determine hot research directions and topics critical for tracking the development of science [18]. As illustrated in Figure 7A, 515 identi ed keywords (the minimum number of occurrences of a keyword in titles and abstracts is over ten) are classi ed into 5 clusters: "Mechanism study", "Treatment study", "Complication study", "Radiological study", and "Etiological study". In the "Mechanism study" cluster, the most used keywords were nontraumatic osteonecrosis, steroid-induced osteonecrosis of the femoral head, mesenchymal stem cell, and apoptosis. For the "Treatment study" cluster, the frequently used keywords were core decompressing, follow-up, total hip arthroplasty, and replacement. In the "Complication study" cluster, the most used keywords were children, femoral neck fracture, complication, and management. For the "Radiological study" cluster, the major keywords were MRI, diagnosis, bone marrow edema, and transient osteoporosis. In the "Etiological study" cluster, the main keywords were riskfactor, natural-history, systemic lupus erythematosus, and bone mineral density. These results demonstrated the distribution of research areas on ONFH in the last 30 years.
The overlay visualization is identical to the network visualization except for the colors of items. As presented in Figure 7B, keywords are colored differently depending on the average time they appear in the publications. The blue color keyword appeared earlier while the yellow color keyword appeared later. Before 2010, most studies focused on the "Complication study" and "Radiological study". The results of co-occurrence analysis indicated that "Mechanism study" may become the hot spot of future ONFH research.

Global trends in ONFH research
Bibliometric analysis could be used to evaluate the current status and forecast the future directions [18]. Therefore, our study was conducted to evaluate and summarize the global trends and current status about ONFH in the last 30 years. As demonstrated in this study, the number of ONFH-related publications has increased signi cantly over the past three decades from1991 to 2020. And we can predict that ONFHrelated publications will continue to increase over time in the coming years. Researchers worldwide participated in ONFH research, especially in North America, East Asia, and Europe from the heat map. Not only the relatively high morbidity rate, the degrees of economic growth might also explain this result. China's contribution was lower than that of the United States before 2013, and gradually surpassed that of the United States after 2013, even though the total number of publications of the United States was still more than that of China. The reasonable explanation may attribute to the rapid national development and the corresponding annually increased research funding for ONFH.

Quality of global publications
In terms of total publications, sum of times cited, average citations per item, and H-index, the United States ranked rst in the world, identi ed as the leading country on ONFH. China was in the second place by the number of total publications and sum of times cited, ranked third in H-index, while only surpassed India in average citations per item among the top ten countries. That means China has held a leading position in ONFH research among developing countries. However, there is still a signi cant gap between China and developed countries with regard to the academic level of ONFH research. The top ten institutions contributing the most to ONFH were from China, Japan, the United States, and South Korea, all of which belong to the top ten countries, indicating that the establishment of outstanding institutions is the prerequisite to improve the academic level of a country. The top ten authors who contributed most to ONFH were also from the top ten countries, and eight of them worked for top 10 institutions. It demonstrated the complementary role between excellent scienti c platforms and rst-class scholars. The excellent scienti c platforms provided su cient scienti c funds and advanced experimental equipment for top scholars, who made great contributions to the further development of the platforms. Among the top ten journals, Clinical Orthopaedics and Related Research published the most ONFH-related articles, nearly twice as many as the publications of the second-ranked journal. Also, Clinical Orthopaedics and Related Research had the most total citations, though the average citations per item were relatively low. Journal of Bone and Joint Surgery-American Volume and Journal of Bone and Joint Surgery-British Volume both had relatively high average citations per item and H-index, which indicated the high quality of their publications. From above, we could track the latest advances of ONFH by monitoring the latest research of these institutions, authors, and journals.
Bibliographic coupling analysis was used in this study to establish the similarity relationship among publications from three dimensions of country, institution, and journal. Bibliographic coupling was established when two items cited the same article, determining the relatedness between two items. To some extent, the total link strength of a particular item could also explain its in uence worldwide. The United States, Kyushu University, and Clinical Orthopaedics and Related Research were the leading country/institution/journal on ONFH globally. Co-authorship analysis was used to evaluate the cooperation between items by counting the number of co-authored publications. Items with higher total link strength indicated that the countries/institutions/authors were more willing to cooperate with others. Therefore, the United States, Seoul National University, and Goro Motomura were more cooperative than others. Co-citation analysis was established when two items were cited by another article together. Compared with bibliographic coupling analysis, co-citation analysis could more scienti cally highlight the in uence of items worldwide. We could conclude that Clinical Orthopaedics and Related Research and Mont et al. were the most in uential journal/publication on ONFH globally.

Keywords of ONFH study
The co-occurrence network visualization was created by analyzing the number of articles in which keywords occurred together in the titles or abstracts. Our results suggested that nontraumatic osteonecrosis drew the most attention worldwide, in line with the reality. Other keywords with the highest total link strength were of great signi cance in their respective elds. The combination of the concluded directions with keywords could provide new thoughts to the research contents of ONFH. And further attention and high-quality studies were needed in these elds. In overlay visualization, items were colored differently from network visualization to highlight their average appearing time. This kind of visualization was pivotal in identifying the hotspot directions related to ONFH in the future. In accordance with our results, mechanism study might become the next prevailing direction. Several frontier keywords related to the pathogenesis of ONFH were listed in the overlay visualization, such as mesenchymal stem cell, apoptosis, oxidative stress, adipogenesis, osteogenic differentiation, and endothelial progenitor cell [3,[19][20][21][22]. The balance between osteogenic and adipogenic differentiation of mesenchymal stem cells, angiogenic differentiation of endothelial progenitor cells, and oxidative stress and apoptosis of endothelial cells and osteoblasts were critical impact factors in the pathogenesis of ONFH. Therefore, these directions deserve more time and funds for more in-depth and comprehensive research in the future. Moreover, our research results also provided some help for several countries or funding agencies to make more scienti c and reasonable investment plans and talent introduction plans.

Strengths and limitations
Our study gave a novel insight into the global trends and current status of ONFH-related publications by using bibliometric methods. However, there are some limitations to the study that have to be pointed out. Due to the differences in the publications covered by the major databases, including WOS, Pubmed and Cochrane library, we may omit several publications from analysis, leading to database bias. In addition, we only analyzed English publications and excluded non-English publications, which may result in language bias. Moreover, the effect of publication time on the sum of times cited has not been considered. Several latest articles with high quality might not attract our attention because of the low sum of times cited.     (C) Network visualization of the 629 identi ed authors on ONFH. In the visualized network, each item is represented by a circle. The size of the circle is determined by the number of publications of the item. The distance between two circles approximately indicates the relatedness of the items. Item color is determined by the cluster to which the item belongs. Figure 6