A composite model of survival from out-of-hospital cardiac arrest using the Cardiac Arrest Registry to Enhance Survival (CARES)

Abstract

ObjectiveUsing CARES data, to develop a composite multivariate logistic regression model of survival for projecting survival rates for out-of-hospital arrests of presumed cardiac etiology (OHCA). MethodsThis is an analysis of 25,975 OHCA cases (from October 1, 2005 to December 31, 2011) occurring before EMS/first responder arrival and involving attempted resuscitation by responders from 125 EMS agencies. ResultsThe survival-at-hospital discharge rate was 9% for all cases, 16% for bystander-witnessed cases, 4% for unwitnessed cases, and 32% for bystander-witnessed pVT/VF cases. The model was estimated separately for each set of cases above. Generally, our first equation showed that joint presence of a presenting rhythm of pVT/VF and return of spontaneous circulation in the pre-hospital setting (PREHOSPROSC) is a substantial direct predictor of patient survival (e.g., 55% of such cases survived). Bystander AED use, and, for witnessed cases, bystander CPR and response time are significant but less sizable direct predictors of survival. Our second equation shows that these variables make an additional, indirect contribution to survival by affecting the probability of joint presence of pVT/VF and PREHOSPROSC. The model yields survival rate projections for various improvement scenarios; for example, if all cases had involved bystander AED use (vs. 4% currently), the survival rate would have increased to 14%. Approximately one-half of projected increases come from indirect effects that would have been missed by the conventional single-equation approach. ConclusionThe composite model describes major connections among predictors of survival, and yields specific projections for consideration when allocating scarce resources to impact OHCA survival.