Ischemic Stroke Transfer Patterns in the Northeast United States

Background: Little is known about how hospitals are connected in the transfer of ischemic stroke (IS) patients. We aimed to describe differences in characteristics of transferred versus nontransferred patients and between transferring and receiving hospitals in the Northeastern United States, and to describe changes over time.

Methods: We used Medicare claims data, and a subset linked with the Get with the Guidelines-Stroke registry from 2007 to 2011. Receiving hospitals were those with annual IS volume greater than or equal to 120 and greater than or equal to 15% received as transfers, and transferring hospitals were nonaccepting hospitals that transferred greater than or equal to 15% of their total (ED plus inpatient) IS patient discharges. A transferring-to-receiving hospital connection was identified if greater than or equal to 5 patients per year were shared. ArcGIS 10.3.1 was used for network visualization.

Results: Among 177,270 admissions to 402 Northeast hospitals, 6906 (3.9%) patients were transferred. Transferred patients were younger with more severe strokes (78 versus 81 years, P < .001; National Institutes of Health Stroke Severity 7 versus 5, P < .001), and were as likely to receive tissue plasminogen activator as nontransferred (P = .29). From 2007 to 2011, there were more patients transferred (960 [3%] to 1777 [6%], P < .001), and more transferring hospitals (46 [12%] to 91 [24%], P < .001), and receiving hospitals (6 [2%] to 16 [4%], P < .001). Most transferring hospitals were exclusively connected to a single receiving hospital.

Conclusions: From 2007 to 2011, hospitals in the United States Northeast became more connected in the care of IS patients, with increasing patient transfers and hospital connections. Yet most hospitals remained unconnected. Further characterization of this transfer network will be important for understanding and improving regional stroke systems of care.