320 A Model for Examining Local Geographic Variations in Out-of-Hospital Cardiac Arrest Outcomes

Abstract

Out-of-hospital cardiac arrest (OHCA) is a leading cause of death in the United States. Although global and national trends have been examined, geographic disparities of OHCA outcomes at local and community levels is less understood. In this study, we developed and tested a replicable, community assessment strategy aimed to identify spatial variations in OHCA outcomes using socioeconomic, out-of-hospital, and inhospital factors. Emergency medical services records of OHCA within Alachua County between 01/2012 and 01/2017 were retrospectively reviewed. Adult non-traumatic OHCA that had recorded dispatch locations and dispatch time stamps were included. These records were individually matched to medical records at the University of Florida (UF). Out-of-hospital-inhospital data and other patient-level variables were extracted from electronic medical records. Socioeconomic demographics for census tracts were extracted from 2010 Census data. Location of cardiac arrest was geocoded using ArcGIS. Primary outcomes were survival to emergency department (ED), survival to admission, survival to discharge, and discharge to home. Multiple imputation was implemented to account for incomplete records. Individual cases were nested into census tracts to develop multilevel logistic regression models in order to assess the geographic variance and probabilities of survival. Two sets of adjusted models were created for each outcome by adding patient-level and then census-level variables to test associations with survival. Chi square deviance test was applied to assess model fit of after the addition of census-level variables. Of the 1562 OHCA cases extracted from out-of-hospital records, 1335 (85.5%) cases were included and geocoded. Predicted probability of survival to ED was 0.57 (95% CI: 0.51 - 0.63). Three-hundred seventy-two patients transferred to UF and 318 (85.5%) were successfully matched to inhospital records. Predicted probabilities of survival to admission, survival to discharge, and survival to home discharge of these patients were 0.42 (95% CI: 0.36 - 0.48), 0.16 (95% CI: 0.11 - 0.23), and 0.07 (95% CI: 0.04 - 0.13), respectively. Census tracts accounted for significant variability in survival to ED, discharge, and discharge home outcomes. There was no significant geographic variation in survival to admission outcomes. Addition of census-level variables in adjusted models significantly improved model fit for survival to ED, discharge, and discharge home outcomes. Multiple modifiable patient and census-level variables of interest were identified in unadjusted and spatially adjusted models, including rural-urban differences. We identified important geographic differences that exist in outcomes of OHCA at the community level, especially in rates of survival to ED, discharge, and discharge with good neurological outcome. The methods used in this research not only identifies specific locations within a county that vary in survival rates but also explains this variation through potentially modifiable socioeconomic and out-of-hospital factors. This research represents a replicable schematic for assessing geographic variations in survival outcomes and factors that influence these outcomes in local communities.