Navigating stroke care: Geospatial assessment of regional stroke center accessibility

Abstract

Introduction

Reducing time between stroke onset and hospital intervention is crucial for positive outcomes in stroke patients. While EMS utilization decreases time to intervention, many US regions are not within timely proximity to an advanced-care-capable stroke center (ASC), defined as a comprehensive or thrombectomy-capable center. This study aims to utilize geographic methodology to identify regions in Wisconsin with both high stroke mortality and low physical accessibility to certified stroke centers (SCs), particularly ASCs.

Methods

Geocoded mortality records for stroke death between 2015 and 2020 were accessed from the Wisconsin Department of Health Services. Indirectly age-standardized mortality ratios (SMRs) were estimated continuously across Wisconsin using adaptive spatial filtering and mortality records at the census block group level; the surface was then averaged by census tract for tract level SMRs. Addresses for SC locations within Wisconsin and bordering states were collected, and drive times from Wisconsin census tract centroids to the nearest SC subtypes were estimated. Drive times and mortality ratios were evaluated at the tract level alongside Rural-Urban Commuting Area (RUCA) codes. Spatial error regression modeling was used to determine RUCA classifications with the highest stroke risk independent of accessibility to stroke centers.

Results

Approximately 50%, 68%, and 78% of Wisconsin residents resided within 30, 45, and 60 minutes of an ASC, respectively. Median drive time from census tract centroids to the nearest ASC were highest for rural tracts (M=90 minutes, IQR=68-115) compared to small-town (M=82 minutes, IQR=49-113), micropolitan (M=53 minutes, IQR=43-77), and metropolitan tracts (M=19 minutes, IQR=11-35; p<0.001). Clusters of high stroke SMRs were found in urban centers as well as rural areas irrespective of county declinations. Spatial regression modeling suggested small-town census tracts had the highest SMR irrespective of physical accessibility to care and spatial correlation. In small-town census tracts >45 minutes from the nearest ASC, the median stroke SMR was 1.12 (IQR=0.94-1.40) with 226,000 residents and 150 stroke deaths per year.

Conclusion

Small-town areas are associated with both long drive distance to ASC locations and high stroke mortality. Geographical analyses reveal apparent stroke care deserts and may inform strategic allocation of emergency medicine resources and coverage.