Stroke represents a major cause of mortality and long-term disability in China, placing a substantial burden on the healthcare system. Within the tiered healthcare framework, an effective two-way referral system is crucial for ensuring timely acute care and coordinated rehabilitation. However, current research primarily focuses on aggregate metrics like referral rates, leaving the structural characteristics of the referral network underexplored. This study uses social network analysis to assess upward and downward referrals for stroke inpatients. Based on provincial inpatient data, we constructed upward and downward referral networks and analyzed patterns using social network analysis. Indicators at the network, subgroup, and node levels were assessed, including density, K -core, degree centrality, and betweenness centrality. Network visualization was performed using NetDraw 2.140 and Gephi 0.9.2. The study included 14,334 inpatient stroke cases, of which 4,948 (34.52%) involved upward transfers across 468 institutions, and 3,954 (27.58%) involved downward transfers across 490 institutions. Both referral networks were primarily characterized by transfers between secondary and tertiary hospitals, accounting for 75.68% of upward and 68.30% of downward referrals. The overall referral network density was low: 0.02 for the upward referral network and 0.01 for the downward referral network. Core subgroups (4-core in the upward referral network and 5-core in the downward referral network) were mostly composed of secondary and tertiary hospitals, with minimal participation from primary care institutions (11.76% and 16.66%, respectively). Node analysis revealed that secondary hospitals exhibited higher degree centrality and betweenness centrality compared to primary care institutions. The upward referral rate for stroke in Hunan Province slightly exceeds the downward referral rate. Both the upward and downward referral networks exhibit a clear hierarchical stratification characterized by core, intermediate, and peripheral layers identified through K -core decomposition, with tertiary hospitals occupying a central position and secondary hospitals as key receiving and coordinating hubs. However, primary care institutions show limited participation, and overall network connectivity remains sparse with insufficient regional collaboration. These findings suggest that facilitating the smooth transfer of patients during the rehabilitation phase from tertiary hospitals to lower-level institutions, strengthening the transitional role of secondary hospitals, and enhancing the rehabilitation capacity of primary care institutions may help optimize the stroke tiered care system.
Zhang et al. (Thu,) studied this question.