Creation of a Novel National Dataset Through Linkage of Emergency Medical Services (EMS) Transport Destination and Verified Emergency Department (ED) Capability.

Abstract

Objective: Emergency department (ED) capabilities, such as trauma center or stroke center designation, are key to understanding the effects of emergency medical services (EMS) transport destination decisions on patient outcomes. In current EMS datasets, ED capabilities are self-reported by the EMS clinician or agency. The reliability and validity of the EMS-reported ED capabilities is unknown. Our objective was to link EMS transport destinations with verified ED capability data to develop a novel national dataset to better understand prehospital routing practices.

Methods: We linked two cross-sectional databases: the 2021 ESO Data Collaborative and the 2021 National Emergency Department Inventory (NEDI)-USA. The ESO Data Collaborative contains de-identified prehospital patient care records from nearly 2,000 participating EMS agencies across the United States. The NEDI-USA is a survey of all nonfederal, non-specialty U.S. EDs open 24/7/365 (including freestanding EDs), with verified stroke, trauma, and burn capability data. From EMS records, we obtained all unique destinations designated as "hospital" as of 2021. After verifying addresses were NEDI-eligible EDs (i.e., providing emergency services 24/7/365), we performed a 3-step linkage process to NEDI-USA: (1) name/address exact matches; (2) probabilistic matching on name/address based on bigrams, accepting adequate (>85%) match scores after review; and (3) hand-matching using Google Maps. We calculated descriptive statistics to describe the linkage process.

Results: Of the 9,420 unique "hospital" destinations in the EMS dataset, 2,714 (29%) were non-hospital facilities (e.g., nursing home) or were non-NEDI-eligible (e.g., specialty hospital such as a psychiatric facility). We linked 98% (n = 6,605/6,706) of NEDI-eligible EMS hospital transport destinations to EDs in NEDI-USA. Excluding duplicate addresses for a single hospital (e.g., ED address versus main entrance address), the linked addresses represented 3,877 unique EDs in 49 states, which included 68% (n = 3,821/5,580) of the EDs included in the 2021 NEDI-USA database.

Conclusions: We successfully linked 98% of EMS ED transport destinations to verified ED capability information. This novel linked dataset now includes rich destination capability information associated with each EMS transport that can be leveraged for describing and improving routing practices for specific patient conditions, such as patients with stroke-like symptoms to stroke centers or major traumas to verified trauma centers.