We explored the evolution process, focuses and characteristics of academic narratives of 10-year development of hierarchical medical system in China using social network analysis, aiming to provide a reference for improving the implementation of hierarchical medical system.

On March 21, 2021, we searched the database of CNKI for studies regarding hierarchical medical system in China published in academic journals in Chinese during January 1, 2009 to March 21, 2021 using "hierarchical medical system" "seeking healthcare in an orderly way" "seeking diagnosis and treatment services in an orderly way" "division of healthcare" "division of diagnosis and treatment" "first contact in primary care" "treating acute and chronic diseases separately" "cooperation between tertiary and secondary/primary hospitals" "bi-directional referrals" as keywords. Bibliometric methods, SATI 3.0 and UCINET 6.0 were used to explore the process, focuses and characteristics of academic narratives of hierarchical medical system through time series analysis, high-frequency keywords analysis, centrality analysis and small group analysis, and to generate visualized co-occurrence network diagrams of high-frequency keywords.

A total of 4 882 studies were included for analysis. According to the results of time series analysis, the development of the hierarchical medical system could be roughly divided into three stages: embryonic stage (2009—2013) , acceleration stage (2014—2017) and standardization stage (2018 to present) . In terms of the use of frequency, the top five high-frequency keywords were "hierarchical medical system" "medical consortium" "primary healthcare institutions" "bi-directional referrals" and "public hospitals". In terms of closeness centrality, the top five high-frequency keywords were "hierarchical medical system" "medical consortium" "primary healthcare institutions" "bi-directional referrals" and "general practitioners". Small-group analysis yielded three groups of word clusters: core word clusters (including "medical consortium" "reform of public hospitals" "bi-directional referrals" "general practitioners" and "first contact in primary care") , secondary word clusters (including "the internet" "internet-based healthcare" "internet-based hospital" and "telemedicine") , and marginal word clusters (including "diabetes" "chronic disease" "cognition" and "influencing factors") . Health policy rhetoric was an important part of the core word clusters. And secondary and marginal word clusters were constantly evolving and generating.

The academic narratives of hierarchical medical system have gradually derived from the mutually-promoted development of policies and real implementation regarding the hierarchical medical system, generally demonstrated various structures and relatively homogeneous research direction. Available academic and policy narratives of the hierarchical medical system are still limited to the scope of guiding principles, and the implementation of the concept and system still relies on the top-down push from the policy-making level, while the self-motivated micro-practices of hierarchical medical system in primary care have not yet emerged.

借助社会网络分析方法，探究当前分级诊疗学术叙事的演进过程、焦点和特征，旨在为分级诊疗现实实践的优化提供参考。

于2021-03-21，以"分级诊疗""有序医疗""有序诊疗""分工医疗""分工诊疗""基层首诊""急慢分治""上下联动""双向转诊"等作为关键词，计算机检索中国知网（CNKI），获取发表于中文学术期刊上的、有关分级诊疗的研究，检索时限为2009-01-01至2021-03-21。将文献的题录以EndNote格式导出，基于文献计量学方法，使用SATI 3.0与UCINET 6.0软件，通过时间序列分析、高频关键词分析、中心性分析与小团体分析，探究分级诊疗学术叙事的进程、焦点和特征，并生成高频关键词的可视化共现网络图。

共纳入文献4 882篇。基于时间序列分析结果，分级诊疗学术叙事的演进过程可以划分为3个阶段，分别是萌芽阶段（2009—2013年）、加速阶段（2014—2017年）和规范阶段（2018年至今）。出现频次排在前5位的高频关键词依次为"分级诊疗""医联体""基层医疗机构""双向转诊""公立医院"；接近中心度值排在前5位的高频关键词依次为"分级诊疗""医联体""基层医疗机构""双向转诊""全科医生"；小团体分析可得3组词群，即由"医联体""公立医院改革""双向转诊""全科医生""基层首诊"等高频关键词组成的核心词群，由"互联网""互联网医疗""互联网医院""远程医疗"等高频关键词组成的次生词群，以及由"糖尿病""慢性病""认知""影响因素"等高频关键词组成的边缘词群。卫生政策修辞是核心词群的重要组成部分，次生词群与边缘词群正在不断演化、生成。

分级诊疗学术叙事在与政策实践的互构中逐渐生成，其整体结构不均衡，研究向度较为单一。目前，分级诊疗的学术与政策叙事仍局限于纲领的范畴，理念与制度的落实仍依靠决策层由上至下强推，基层自发、自为的分级诊疗微观实践尚未兴起。