Using spatial regression methods to evaluate rural emergency medical services (EMS)

Abstract

Emergency Medical Services (EMS) are acute services provided outside of the hospital. EMS are crucial in rural environments where hospitals are often far away and difficult to access. Establishing EMS performance measures is critical in improving a rural community's access to these services and eliminating systemic inequalities. However, an absence of data leads to challenges in developing objective and quantifiable service metrics. EMS data are regularly collected through the National EMS Information System (NEMSIS), yet the manner of data collection and quality of data vary across agencies. Moreover, the amount and complexity of information makes data analyses difficult, subsequently effecting EMS leaderships' ability to identify improvement needs.This study used NEMSIS data to exemplify approaches for establishing two data-driven performance measures. The measures used in this study – timely service and service coverage – are both dependent on the mobility and accessibility of the EMS transportation network. Two types of spatial models: the spatial econometric model and geographically weighted regression (GWR) model, were developed and then compared to the linear regression model to help identify response time factors. GWR performed best in terms of goodness-of-fit statistics and was chosen to help understand how factors (e.g., weather, transportation) impact the timely provision of EMS in rural areas. The GWR results provided additional insights through the particular spatial patterns of the coefficient estimates and their statistical significance to EMS practitioner for their references to reduce local response times.