Data integration in China's county-level medical communities: institutional dilemmas, generative mechanisms, and optimisation pathways from a holistic governance perspective

Background The compact county-level medical community (CMC) is China's principal vehicle for building an integrated county health delivery system, and data integration is the core engine of its transition from "physical formation" to "chemical integration". Yet data sharing among CMC member institutions remains limited, and the barriers are widely treated as merely technical. This study explains why data fragmentation persists despite mature technology and increasingly mandatory policy. Methods We conducted a theory-guided policy analysis combined with a secondary analysis of national health statistics. Policy documents issued between 2017 and 2026 and indicators from the China Health Statistical Yearbook (2017–2024 editions) and annual national statistical bulletins were examined through an analytical framework that couples holistic governance theory with resource dependence theory across three dimensions: governance structure, governance mechanism, and governance capacity. Results National statistics reveal three structural constraints: a supply-demand mismatch (primary care institutions hold 17.8% of inpatient beds but deliver 51.8% of outpatient visits), polarised bed occupancy (91.1–97.5% in tertiary hospitals versus 50.2–57.5% in township health centres during 2019–2023), and a near five-fold gap in average inpatient expenditure per admission. These constraints map onto three mutually reinforcing dilemmas: managerial silos, interest silos, and technology silos. We conceptualise their common driver as "data powerization", a process whereby provider organisations convert patient data into bargaining chips, defensive risk barriers, and symbols of organisational independence, analogous to the "information blocking" described in other health systems. The three dilemmas constitute a circular lock-in: structural fragmentation distorts incentives, distorted incentives motivate data powerization, data powerization starves primary care of data and patients, and the resulting siphoning effect further entrenches fragmentation. Conclusions Data fragmentation in CMCs is fundamentally an institutional rather than a technological problem. Effective integration requires a sequenced pathway comprising payment reform as mechanism leverage, high-level coordination as structural guarantee, and cloud-based capacity building as the foundation, so as to reverse the cost-benefit logic of data hoarding and return health data to its status as public infrastructure.