Identification of Out-of-hospital Cardiac Arrest Clusters Using a Geographic Information System
Article first published online: 8 JAN 2008
Academic Emergency Medicine
Volume 12, Issue 1, pages 81–84, January 2005
How to Cite
Lerner, E. B., Fairbanks, R. J. and Shah, M. N. (2005), Identification of Out-of-hospital Cardiac Arrest Clusters Using a Geographic Information System. Academic Emergency Medicine, 12: 81–84. doi: 10.1111/j.1553-2712.2005.tb01484.x
- Issue published online: 8 JAN 2008
- Article first published online: 8 JAN 2008
- Received February 13, 2004; revision received May 31, 2004; accepted August 27, 2004.
- emergency medical services;
- heart arrest;
- cardiopulmonary resuscitation;
- geographic information system;
- out-of-hospital cardiac arrest
Objectives: To locate all out-of-hospital cardiac arrests (OHCAs) in Rochester, New York, and identify clusters of OHCAs, as well as clusters of patients who did not receive bystander cardiopulmonary resuscitation (CPR), in order to identify locations that may benefit from prevention efforts. Methods: The locations of all adult OHCAs of cardiac etiology occurring in the study city over a four-year period were plotted on a map using ArcGIS. Location information was obtained from the emergency medical services (EMS) medical record and included street address and zip code. Descriptive data related to patient treatment and transport were also abstracted. Kernel analysis was used to identify areas with the highest density of OHCAs. Census-defined block groups were used to calculate OHCA incidence in order to determine the effect of population density. Patients with OHCAs who did not receive bystander CPR were selected and kernel analysis was repeated to identify areas with the highest density of no bystander CPR. Results: A total of 537 OHCAs that met the inclusion criteria occurred during the study period. Ninety-four percent had sufficient location information to be plotted. Two clusters of OHCAs were identified. One cluster covered two block groups that were found to have the highest incidence of OHCA in the city (incidence: 142 [95% CI = 42 to 241] and 105 [95% CI = 35 to 175] per 10,000 people). EMS providers or first responders started CPR (i.e., no bystander CPR) for 80% of patients. Kernel analysis revealed three areas with a high density of no bystander CPR; these areas coincided with the OHCA cluster sites. Cluster communities were found to have a lower median household income and a larger percentage of people living below the poverty level, to have more residents of African American race, and to have more residents without a high school diploma compared with the city's population in general. Conclusions: Out-of-hospital cardiac arrest can be plotted by geographic location. Clusters of OHCAs can be identified, which could be used to guide resource allocation. Clusters of OHCAs in which the patients did not receive bystander CPR can also be identified and could be used to direct educational programs. Census data can be superimposed on this information to identify characteristics of cluster locations and were used to demonstrate that the identified clusters were not simply the result of population density.