Using Bibliometric Data to Define and Understand Publishing Network Equity in Anesthesiology

Abstract

BACKGROUND: Anesthesiology departments and professional organizations increasingly recognize the need to embrace diverse membership to effectively care for patients, to educate our trainees, and to contribute to innovative research. 1 Bibliometric analysis uses citation data to determine the patterns of interrelatedness within a scientific community. Social network analysis examines these patterns to elucidate the network’s functional properties. Using these methodologies, an analysis of contemporary scholarly work was undertaken to outline network structure and function, with particular focus on the equity of node and graph-level connectivity patterns. METHODS: Using the Web of Science, this study examines bibliographic data from 6 anesthesiology-specific journals between January 1, 2017, and August 26, 2022. The final data represent 4453 articles, 19,916 independent authors, and 4436 institutions. Analysis of coauthorship was performed using R libraries software. Collaboration patterns were assessed at the node and graph level to analyze patterns of coauthorship. Influential authors and institutions were identified using centrality metrics; author influence was also cataloged by the number of publications and highly cited papers. Independent assessors reviewed influential author photographs to classify race and gender. The Gini coefficient was applied to examine dispersion of influence across nodes. Pearson correlations were used to investigate the relationship between centrality metrics, number of publications, and National Institutes of Health (NIH) funding. RESULTS: The modularity of the author network is significantly higher than would be predicted by chance (0.886 vs random network mean 0.340, P < .01), signifying strong community formation. The Gini coefficient indicates inequity across both author and institution centrality metrics, representing moderate to high disparity in node influence. Identifying the top 30 authors by centrality metrics, number of published and highly cited papers, 79.0% were categorized as male; 68.1% of authors were classified as White (non-Latino) and 24.6% Asian. CONCLUSIONS: The highly modular network structure indicates dense author communities. Extracommunity cooperation is limited, previously demonstrated to negatively impact novel scientific work. 2,3 Inequitable node influence is seen at both author and institution level, notably an imbalance of information transfer and disparity in connectivity patterns. There is an association between network influence, article publication (authors), and NIH funding (institutions). Female and minority authors are inequitably represented among the most influential authors. This baseline bibliometric analysis provides an opportunity to direct future network connections to more inclusively share information and integrate diverse perspectives, properties associated with increased academic productivity. 3,4