Abstract TMP33: Network Analysis Of Acute Stroke Care Systems In Korea

Abstract

Purpose: Acute stroke care systems have been shifted to clusters of hospitals for cooperatively applying advanced treatment. It developed various forms of network care systems, of which measurement and evaluation are requested. Methods: Using cross-sectional national data for acute stroke care in Korea, hospital usage information after a medical contact was surveyed and transformed into node (hospital) and edge (transfer) format. Of them, dense-linked hospital groups were partitioned as a stroke community using the Louvain algorithm. It investigated the concentration, strength of connection, and functions of derived communities. Results: For six months, a total of 19113 subjects with ischemic stroke headed primarily to 1009 hospitals. They merged into 246 ASHs (acute stroke care hospitals) via 3114 (16.3%) secondary interhospital transfers. Those connections were partitioned for the 93 stroke communities (Figure A) comprising median of 9 hospitals (IQR, 5 - 14) and managed a median of 31.3 subjects per month (IQR, 18.2 - 48). They presented highly centralized structure (median group centralization degree [GDC], 1.00), which used to classify the communities (Figure B). One- or two-hospital centered communities showed similar performances of endovascular treatment (EVT) compared to lower volume or poor formatted communities. Conclusions: The network analysis method effectively contoured the high centralizing stroke communities and helped the functions on the EVT accessibility.