Poisson Cluster Analysis of Cardiac Arrest Incidence in Columbus, Ohio

Abstract

Background. Scarce resources in disease prevention and emergency medical services (EMS) need to be focused on high-risk areas of out-of-hospital cardiac arrest (OHCA). Objective. Cluster analysis using geographic information systems (GISs) was used to find these high-risk areas and test potential predictive variables. Methods. This was a retrospective cohort analysis of EMS-treated adults with OHCAs occurring in Columbus, Ohio, from April 1, 2004, through March 31, 2009. The OHCAs were aggregated to census tracts and incidence rates were calculated based on their adult populations. Poisson cluster analysis determined significant clusters of high-risk census tracts. Both census tract-level and case-level characteristics were tested for association with high-risk areas by multivariate logistic regression. Results. A total of 2,037 eligible OHCAs occurred within the city limits during the study period. The mean incidence rate was 0.85 OHCAs/1,000 population/year. There were five significant geographic clusters with 76 high-risk census tracts out of the total of 245 census tracts. In the case-level analysis, being in a high-risk cluster was associated with a slightly younger age (–3 years, adjusted odds ratio [OR] 0.99, 95% confidence interval [CI] 0.99–1.00), not being white, non-Hispanic (OR 0.54, 95% CI 0.45–0.64), cardiac arrest occurring at home (OR 1.53, 95% CI 1.23–1.71), and not receiving bystander cardiopulmonary resuscitation (CPR) (OR 0.77, 95% CI 0.62–0.96), but with higher survival to hospital discharge (OR 1.78, 95% CI 1.30–2.46). In the census tract-level analysis, high-risk census tracts were also associated with a slightly lower average age (–0.1 years, OR 1.14, 95% CI 1.06–1.22) and a lower proportion of white, non-Hispanic patients (–0.298, OR 0.04, 95% CI 0.01–0.19), but also a lower proportion of high-school graduates (–0.184, OR 0.00, 95% CI 0.00–0.00). Conclusions. This analysis identified high-risk census tracts and associated census tract-level and case-level characteristics that can be used to target public education efforts to prevent OHCA and to mitigate its occurrence with CPR and automated external defibrillator training. In addition, EMS resources can be redeployed to minimize response times to these census tracts.

Key words: :

• cardiac arrest
• cluster analysis
• geographic information system