Quantity of relevant literature
For literature screening, we set up the keywords as “spinal stenosis” and “language = English”. And our retrieval ranged from January 1st, 2000 to December 31st, 2018. A total of 5147 publications involved in 4651 articles and 496 reviews (Figure 1). It was able to see that the number of publications increased generally from 108 in 2000 to 512 in 2018, only declined in 2011 (Figure 2). And the growth rate was also getting better.
Distribution characteristics of literature
Most active countries or regions
Counties or regions distribution of publications in bibliometric analysis contributed to focus on the hotspots of research in spinal stenosis by refining data sources. In our study, we demonstrated the top ten countries proceeding in the field of spinal stenosis according to the publication counts (Figure 3). It was seen that the United States was far ahead of other counties with 1803 literatures followed by Japan (711), China (566) and South Korea (479). There was a noteworthy phenomenon that research level in Northeast Asian countries had gradually increased and surpassed Europe over time. We attributed it to economic development. In addition, the centrality index also reflected the core influence to evaluate research value more exactly. The United States remained a leading position with 0.71 centrality. All of the other countries were in a lower influence, among which Germany was the highest for the centrality equaled to 0.23 (Table 1). It was seen that the cooperative network had been formed between many countries and the level was relatively higher in leading countries (Figure 4).
Top ten active institutions
We also demonstrated the top ten institutions for spinal stenosis research ordered by the publication number. They were Seoul Natl Univ, Univ Washington, Univ Toronto, Harvard Univ, Yonsei Univ, Kuopio Univ Hosp, Cleveland Clin, Univ Calif San Francisco, Vanderbilt Univ, Johns Hopkins Univ (Table 2). Among them, more than half were from the United States, two institution were from Korea and only one belonged to Finland and Canada in line with the distribution characteristics of countries. Seoul Natl Univ (231), Univ Washington (169), Univ Toronto (155) and Harvard Univ (127) published more than a hundred articles and their average citations were 3.14, 20.17, 6.14 and 23.63 respectively. It was clear to see that the institution in the United States had relative higher average citation and centrality, which determined the credibility (Table 2). With extensively academic communication among scholars, it was necessary to develop closer research collaboration between various institutions for the lower level (0.015 on density map) (Figure 5).
Distribution characteristics of authors and journals
There were total 15718 authors involved in the field of spinal stenosis by our retrieval. We paid more attention on ten most active authors including Lee SH, Kim HJ, Manchikanti L, Lee JH and Konno S at all with the order of published counts (Table 3). The top three among them were Lee SH, Kim HJ and Manchikanti L with the number of 73, 60, 54. They came from Pusan National University Yangsan Hospital in Korea, Seoul National University Bundang Hospital in Korea and Pain Management Center of Paducah in American. It followed that they contributed most to the development of scientific study, though there was not a large gap with other authors, which also showed the research momentum of spinal stenosis. We draw an associated network based on the integration of cited and co-cited authors by CiteSpace at the same time (Figure 6 and Figure 7). Deyo RA was the co-cited author with the most co-citation(812). Katz JN(659) and Weinstein JN(630), close to each other, ranked second and third place. And the centrality index were 0.18, 0.10 and 0.10 which reflected the credibility and authority of their research (Table 3).
In total, there were 735 journals involved in this field. SPINE and EUROPEAN SPINE JOURNAL are the top two in order of article counts and total number of citations. They were 695, 8847 for SPINE and 426, 2908 for EUROPEAN SPINE JOURNAL. Both of them were Q2 in JCR 2018 standards. The other journals in top ten had been shown in Table 4. On the other hand, average number of citations and impact factor were important evidence to evaluate the influence of journals. PAIN PHYSICIAN(15.91) was the highest in terms of average number of citations, followed by SPINE(12.73) and JOURNAL OF NEUROSURGERY-SPINE(7.09). Based on impact factor, JOURNAL OF SPINE JOURNAL was highest of 3.196, second for JOURNAL OF NEUROSURGERY-SPINE(2.998) and third for PAIN PHYSICIAN(2.942). The count of articles published in this ten journals accounted for 42.47%. It was obvious that this ten journals laid a solid foundation for subsequent research on spinal stenosis.
Research hotspots of spinal stenosis based on MeSH clusters
Through statistical and quantitative analysis of the publications from January 1 in 2000 to December 31 in 2018, 3810 major MeSH and MeSH subheading terms had been concluded with a total frequency of 23785 times. After repeatedly proofreading and balancing, the major MeSH and MeSH subheading terms with occurrence frequency of 56 times and above were regarded as high-frequency ones. Altogether, fifty-seven terms were high in appearance with total 11754 frequency, accounting for 49.42%(11754/23785) (Table 5). The major MeSH terms/subheading MeSH terms, Spinal stenosis/surgery and lumber vertebrae/surgery, appeared more than a thousand even close to two thousand times, which reflected the predilection segment of lumbar vertebrae and treating method of surgery. Biclustering helped us make a classification of these high-frequency major MeSH and MeSH subheading terms, convenient for refining main topics of spinal stenosis. As the mothed introduced, the results were visualized in the form of matrix and mountain graph. The color was the core property to reflect the raw matrix value in matrix visualization. The balanced point was white, red color represented positive value and green represented negative value, both of which expressed a positive correlation. In the matrix graph, clusters are separated by black horizontal lines and each row was an individual category. Thus there were seven clusters showing in our matrix (Figure 8). And the row and column tags were the high-frequency major MeSH terms/MeSH subheading terms and the PMIDs of articles. When the matrix were exhibited in dendrogram, the top and left layers would appear to reveal the association with different indexes. In Figure 8, the top tree indicates the relationships among articles, and the left tree indicates the relationships among high-frequency major MeSH terms/MeSH subheading terms. gCLUTO also could rearranged the attitude of rows in initial matrix to integrate the similar rows (Table 6). In our mountain visualization, there were seven peaks existing form 0 to 6, revealing total seven clusters after biclustering (Figure 9). In the 3D landform, it was visualized the properties of each cluster by comparing the location, volume, altitude and color of corresponding peaks. Location of peaks represented the relative similarity among different clusters with the higher for the closer. Both of the volume and altitude referred to the internal features in clusters. The greater of volume reflected more objects of major MeSH and MeSH subheading terms containing and the altitude demonstrated the internal similarity in direct proportion. Furthermore, the color of peak reflected the deviation degree of dataset, equaling to standard deviation. Cluster5 with a relative red peak referred to lower value and higher with the trend of bluing in our study (Figure 9). As our analysis above, the major MeSH and MeSH subheading terms had been refined into seven categories. These seven clusters are:
Effect of medical imaging in diagnosis of spinal stenosis(Cluster 0),
Etiology of spinal stenosis with different pathological symptoms(Cluster 1),
Pathogenesis of spinal stenosis(Cluster 2),
Options of surgical method in particular conditions and adverse effects(Cluster 3),
Role of instrumentations in rehabilitation of patients(Cluster 4),
Surgical indications of spinal stenosis(Cluster 5),
Non-operative therapeutic means of spinal stenosis(Cluster 6).