Recent research patterns and key topics in autoimmune thyroid inflammation from 2000 to 2022

The publications were obtained from the Core Collection database of Web of Science (WoSCC) because it is generally considered one of the most systematic, authoritative and comprehensive databases and has been widely used for scientometric analysis and visualization of scientific documents in a considerable number of studies (22–26). All searches were completed and downloaded on 1 day, March 20, 2022, to avoid the bias caused by daily database updates and verified by two authors independently. The following search terms were used to extract document data: TS= (“ autoimmune thyroiditi*”) OR TI= (“chronic lymphocytic thyroiditi*”) OR TI=(hashimoto*) OR TI= (“postpartum thyroiditis”). The selection criteria were as follows: (1) language: English; (2) document type: article or review; and (3) timespan: 2000–2022. Initially, there were 4845 documents, and 3599 satisfied these criteria. After screening each of these titles and abstracts, 2324 articles that exclusively targeted the topic of AIT were ultimately included. For the articles that satisfied these inclusion criteria, all records, including the titles, authors, abstracts, keywords and references, were exported, saved as plain text files and stored as download_txt files and then imported to CiteSpace5.8.R3. After using CiteSpace to analyze these documents, 2290 were subjected to visualization analysis.

CiteSpace is a freely available Java application for the visualization and analysis of scientific literature trends and patterns (27). It was created by Professor Chaomei Chen in 2004 and has become popular globally to present the structure, laws, and distribution of scientific knowledge and provide a comprehensive overview of the research situation (28). Therefore, CiteSpace can be used to predict the research trend of a certain discipline or field during a certain period (29). The CiteSpace parameters were set as follows: link retaining factor (LRF = 3), e for top N (e = 2), time span (2000–2022), years per slice, look back years (LBY = 23), links (strength: cosine, scope: within slices), selection criteria (g-index: k = 25), and minimum duration (MD = 1) (30). VOSviewer is a program for building and viewing bibliometric maps (31). We used VOSviewer (version 1.6.18) to visualize the collaborations between countries, institutions, journals as well as high-frequency keywords. In VOSviewer, nodes were used to represent countries, institutions, journals or keywords, and the size of the node was determined by its frequency of co-occurrence in titles and abstracts (32). VOSviewer can be used to build journal maps based on collaborative data or to build keyword maps based on co-occurrence data (33). In the figure production process, we also used another software program, Pajek, an extremely professional web analytics tool, which was greatly beneficial for us to make the relationships clearer.

We used Citespace software to generate a map of reference co-citations with corresponding clusters so that we could extract landmark references and clusters of research (Figure 1A). Eighty-seven different clusters were identified in this network of co-citation references and we filtered the clusters which contained more than 10 documents, with significant modularity and silhouette scores (Q= 0.7964; S= 0.92). Q-values range from 0-1, with values greater than 0.3 indicating a significant delineation structure; S-values greater than 0.5 indicate reasonable clustering results, and 0.7 or more are more convincing.

When two publications are cited jointly by a third publication, this is referred to as a co-citation relationship (34). Therefore, co-cited references are usually regarded as a knowledge base in a particular field (35). The more frequently a piece of literature is cited, the more crucial it is considered in a certain field. We analyzed all of these documents, and a total of 1137 co-cited references were identified from 2000 to 2022 based on CiteSpace with the time slice set as a year. Table 1 demonstrates the top 10 cited references. Each reference was co-cited at least 30 times. Fifty percent of the authors were from the United States. Among the top 10 cited references, the review entitled “Hashimoto thyroiditis: clinical and diagnostic criteria” published by P Caturegli et al. in the journal of Autoimmunity review was the most co-cited paper (n = 97), followed by Autoimmune thyroid disorders, which was also published in Autoimmune thyroid disorders (n=70). The remaining eight references had 30-44 co-citations. The literature entitled Increased circulating pro-inflammatory cytokines and Th17 lymphocytes in Hashimoto’s thyroiditis published in The Journal of Clinical Endocrinology and Metabolism had the largest centrality (0.16), indicating that it was the most acknowledged literature and had a significant influence on other researchers’ work.

The top 11 co-citation clusters were shown in Figure 1B. In the timeline view, different colors of nodes on the same line indicated different years (Figure 1C) (30). Accordingly, the nodes that were closer to the right reflected the more recent references. CiteSpace was also a convenient tool for exploring the evolution track and stage characteristics of a specific research field. The clusters were listed from 1998 to 2016: “postpartum” (Cluster 9, n=54), “interferon-alpha” (Cluster 10 n=46), “apoptosis” (Cluster 8, n=59), “regulatory t cells” (Cluster 2, n=101), “hashimoto’s encephalopathy” (Cluster 5, n=95), “CCL2” (Cluster 11, n=26), “CTLA-4” (Cluster 1, n=107), “thyroid cancer” (Cluster 7, n=67), “IgG4” (Cluster 6, n=82), “oxidative stress” (Cluster 0, n=128), “papillary thyroid cancer” (Cluster 4 n=96), and “Vitamin D” (Cluster 3, n=98).

The designation “References with citation bursts” indicates that the relevant literature was cited at a high frequency during a certain period. The top 25 references with the strongest citation bursts are presented in Figure 2A. As illustrated, the blue line segment indicates the time interval, while the red line segment reflects the frequently cited time. This figure also indirectly reflected that the research in Hashimoto thyroiditis has a stable progression because each year has references with citation bursts, and many of them continue to 2022. The analysis of burstness revealed that the top 3 references with the strongest citation burst were Hashimoto thyroiditis: Clinical and diagnostic criteria, Autoimmune thyroid disorders, and Hashimoto Encephalopathy: Syndrome or Myth. When focusing on the last 5 years, as illustrated in Figure 2B, the top 3 references with the strongest citation burst were Hashimoto thyroiditis: Clinical and diagnostic criteria, The incidence and prevalence of thyroid autoimmunity, The association between papillary thyroid carcinoma and histologically proven Hashimoto’s thyroiditis: a meta-analysis.

There were 2153 articles (92.64%) and 171 reviews (7.36%) among the 2324 documents, including 62 clinical trials, 28 RCT as well as 43 meta-analysis or systematic reviews. The distribution of publication number by year, presented in Figure 3, shows a relatively upward tendency, indicating a steady development of this field. The quantity of published documents demonstrates how quickly knowledge is updated in this subject and is an essential indicator for grasping the trends in this field. Only 47 publications about AIT were published in 2000, but 23 years later, the number of documents has approximately quadrupled.

According to the VOSviewer analysis, a total of 2290 articles were from 120 countries. As shown in Figure 4A, the top 3 countries/regions were China (n = 364, 15.90%), Italy (n = 351, 15.33%) and the United States (n = 329, 14.37%). Strikingly, the top 3 countries contributed 40.85% of the total number of publications. Moreover, as illustrated in Figure 4B, each node represents a country or region, and the size of the node is proportional to the number of documents published. The lines between nodes represent cooperation between countries. We find that extensive cooperation between many countries/regions was observed, and the most obvious cooperation was between China and the United States. What is noticeable is that the total link strength of the United States was significantly higher than that of other countries, indicating that it cooperated more closely with other countries. This is also reflected in the fact that the number of documents in China was more than that in the United States, but Chinese citation and total link strength were far less than those of the United States, as shown in Figure 4C.

When restricting the timeframe to the last 5 years (2016–2021), the top five most productive countries were China (n=241), Italy (n=113), Turkey (n=82), Poland (n=64) and the United States (n=60).

As shown in the top ten countries/regions with the strongest citation bursts (a significant change in documents in a short period) (Figure 4D), China showed the highest burst strength (54.88) since 2017, indicating that there were many scholars studying AIT in China during this period. This has continued up to 2022, suggesting that this field of research will continue to grow.

These documents were contributed by 561 institutions, with the top 20 institutions contributing a total of 577 articles, accounting for 25.20% of the total (Figure 5A). As shown in Figure 5C, the top 3 institutions/affiliations with the most documents were the University of Messina (n=67), University of Pisa (n = 65) and China Medical University (n = 51). According to the number of citations, the order from highest to lowest was the University of Pisa (citations = 2060), the University of Messina (citations = 1414) and China Medical University (citations = 1389) (Figure 5D). Remarkably, six of the top 20 institutions are from China. They are China Medical University, Jiangsu University, Fudan University, Shanghai Jiao Tong University, Zhejiang University and Capital Medical University. The University of Pisa had the highest total link strength of 716, followed by the China Medical University, which had 525. Many Chinese institutions collaborated closely with each other. For burst monitoring of institutions (Figure 5B), the top three ranked institutions were University of Missouri, bursting from 2000 to 2008, followed by Wayne State University, bursting from 2001 to 2009, and Mayo Clinic, bursting from 2001 to 2010.

The WoSCC search showed that the 2290 documents included in the current analysis were published in 723 different journals over the last 23 years since 2000. The top 10 most-cited journals were listed in Table 2. As shown in Figure 6A, among these journals, Thyroid had the highest number of output and impact factors (122, 5.33%, IF=6.568), followed by the Journal of clinical endocrinology and metabolism (62, 2.71%, IF=5.958). The influence of journals depends on the number of times they are co-cited, which reflects whether the journal has essential influence in a specific topic (33). Six journals have been cited more than 1000 times. Specifically, the journal with the highest number of citations was the Journal of Clinical Endocrinology and Metabolism (5053), followed by Thyroid (3721) (Figure 6B). We also found that there was a strong link between Thyroid and the Journal of clinical endocrinology and metabolism (Figure 6C). According to the 2022 Journal Citation Reports (JCR), 50% of the top 10 journals were located in the Q1 region. Simultaneously, we noticed that 9 of 10 journals are based in the United States or the United Kingdom (Table 3).

The dual-map overlay of journals represents the topic distribution of academic journals (31) (Figure 7). The citing journals are located on the left, while the cited journals are on the right, and the colored paths illustrate the citation relationships (39). As shown, the yellow and green paths indicate that studies published in health/nursing/medicine or molecular/biology/genetics journals are usually cited in the studies published in medicine/medical/clinical or molecular/biology/immunology journals.

We used VOSviewer to construct a network map of keywords (Figure 8A). Excluding hashimoto’s thyroiditis (574) and some unmeaningful words, the keywords appearing at high frequency among these documents were hypothyroidism (299), expression (242), prevalence (203), antibodies (195), cancer (191), diagnosis (162), pathogenesis (114), dysfunction (104), iodine (89), susceptibility (89), T-cells (85) and so on. These keywords have been used in research in previous studies or are the current focus of research. Figure 8A shows the blue, purple, red, orange and yellow clusters, indicating five research categories. The blue clusters included the keywords hypothyroidism, dysfunction, prevalence, age, adolescence, natural history and so on. The purple cluster included the keywords levothyroxine, selenium, vitamin D, women, postpartum thyroiditis and type 1 diabetes. The red clusters included the keywords hashimoto’s thyroiditis, management, classification, nodules, cancer and so on. The orange clusters included the keywords mice, Th 17, IFN-gamma, growth-factor beta, gene, susceptibility, polymorphism, tolerance T-cells, regulatory T-cells, differentiation, genes, receptor, expression, susceptibility and so on. The last set of clusters, the yellow ones, included the keywords guidelines, quality of life and so on. We also used CiteSpace to cluster keywords chronologically, and we showed the largest cluster in Figure 8B. They were “autoimmune thyroiditis”, “IFN-γ”, “papillary thyroid cancer”, “Graves’ disease” and “Hashimoto disease”.

CiteSpace was also used to illustrate the keywords with strong citation bursts. As Figure 8C demonstrates, we screened out top four keywords whose citation bursts continued to 2022, including “Vitamin D”, “oxidative stress”, “polymorphism” and “association”.

In this era of information explosion, it is comparatively more difficult to grasp the emphasis in a specific field and accurately obtain cutting-edge information as well as identify the research trend and hotspots. Knowledge management and bibliometric analysis are often used as methods to address the concerns of scientists (40–42). In this article, we aimed to employ an innovative method to analyze and visualize the knowledge structures of the AIT research field by employing software such as CiteSpace and VOSviewer, which are bibliometric tools. In the present study, we performed a systematic literature search of the Web of Science databases for articles published in the last two decades about AIT (2000–2022). After excluding studies that did not meet the screening criteria, this scientometric study included 2290 English papers published in 723 journals with 39661 co-cited references from 561 institutions in 120 countries/regions. Among 28 RCTs, 8 have focused on the effect of selenium supplementation on autoimmune thyroiditis (43–46), and 7 meta-analyses were conducted on this contradictory topic (38, 47, 48). The effect of levothyroxine and other nutrients such as vitamin D on thyroid function has also been studied in many documents (49–53). Also, meta-analyses have also summarized the relationship between autoimmune thyroiditis and other diseases such as chronic urticaria (54), systemic lupus erythematosus (55), thyroid cancer (36, 56–58) as well as depression and anxiety (59).

Based on the results of reference analysis and keyword extraction, those dominating were “apoptosis”, “insulin resistance”, “encephalopathy”, “IFN-γ” and others mentioned in the top-ranked items may have been popular research topics over the past years but may have slowed down with the development of other novel directions such as “papillary thyroid cancer”, “Vitamin D”, “oxidative stress”, “polymorphism” and “association”. Through the analysis of references and keywords, we can understand the current research status of the field and scientifically predict the future research trends to provide scientific researchers with the direction to move forward. In the following sections we will specify the current research status of the first four keywords.

An analysis based on bibliometric tools such as CiteSpace and VOSviewer provides a better illustration of evolving research priorities and trends, as well as relatively comprehensive and objective data analysis, compared to that of a traditional overview. However, there are some limitations in the research design. First, the publications we included in the analysis had a cut-off date of March 20, 2022, but the WOS Core Collection data continues to be updated and some of the 2022 literature is already online, which was not included in our work so that our paper does not fully reflect the reality of 2022. Second, only English documents were included; thus, contributions in other languages may be ignored. Third, because of the format requirement of CiteSpace, only publications in the WoSCC database were finally included in our statistics, which may ignore articles that are only in other databases, such as PubMed, Medline and Scopus. However, because of the significant cross-replication of the literature in the various databases and the authority of the WoSCC database, we consider that this work still can be applied to present the overall situation and general trend for this field.