Abstract Body: Background
Stroke transfer networks have rapidly evolved with limited central organization in response to the emerging and evolving evidence base for endovascular thrombectomy (EVT). Simultaneously, the US healthcare system has undergone a period of rapid consolidation with large hospital systems purchasing smaller hospitals and physician practices, an evolution termed “vertical integration” (VI). The effect that vertical integration has on emergency stroke care is unclear. This analysis longitudinally characterizes VI and non-VI stroke transfer networks to assess their evolution.

Methods
Using the 2016-2018 US Centers of Medicare Fee-For-Service files, we captured patients with ischemic stroke who had an ED-hospital transfer, defined as a beneficiary with: 1) any outpatient ED claim the day of or prior to the inpatient stroke claim, 2) an ED discharge status indicating transfer to an acute care hospital, and 3) originating and receiving facilities with different NPI numbers. We defined an ED-hospital dyad to be vertically integrated if the ED and hospital shared the same health system ID in the Agency for Healthcare Research and Quality's Compendium of US Health Systems. We characterized the vertical integration of the stroke transfer networks using descriptive statistics and the structure using multiple edge directed graphs.

Results
We identified 13287, 14217, 15169 stroke transfers in 2016, 2017, and 2018, respectively. In 2016, 29% of VI stroke transfer network dyads had ≥1 transfer with EVT compared to only 17% of non-VI dyads. While the overall fraction of dyads with ≥1 transfer with EVT increased over time (2016: 20%, 2017: 25%, 2018: 28%), the relatively higher fraction in VI dyads with ≥1 transfer with EVT compared to non-VI dyads remained stable (2017: VI 32%, non-VI 22%; 2018: VI 35%, non-VI 26%).

In VI stroke transfer networks in 2016, EDs sent stroke transfers to 1.2 hospitals on average, while hospitals received stroke transfers from 2.1 EDs. In non-VI stroke transfer networks, EDs sent stroke transfers to 1.9 hospitals on average, while hospitals received stroke transfers from 3.8 EDs. There was no substantial change in 2017 or 2018.

Conclusion
From 2016 to 2018, a greater proportion of VI stroke transfer dyads had transfers for EVT than non-VI stroke transfer dyads. VI EDs on average transferred to fewer receiving hospitals than non-VI EDs, while VI hospitals on average received patients from fewer EDs than non-VI hospitals.