Probability of Resuscitation of Sudden Cardiac Arrest Score for the Evaluation of Cardiac Arrest Outcomes

Introduction—Evaluation of paramedic care, particularly regarding out-of-hospital cardiac arrest (OHCA) patients, is an integral part of any EMS system. Quality management tools are needed for benchmarking that control for individual patient, scene and EMS response factors. However, no such standards exist that would enable an EMS system to compare actual to predicted field resuscitation outcomes.

Objective—To develop a predictive model of the probability of resuscitation of sudden cardiac arrest (pROSC) score based on patient, scene and EMS response factors that could be used to evaluate EMS system performance.

Methods—All adult OHCA patients in the North Carolina Prehospital Medical Information System (PREMIS) database from January 2012 to June 2014 who received resuscitative efforts were randomly divided into model development and validation data sets of equal proportions. Using the development data set, a logistic regression was used to identify predictors of field resuscitation, which was defined as return of spontaneous circulation. pROSC was defined as 1/(1 + e-x), where x is the weighted sum of the independent predictors from the logistic regression equation. Using the validation data set, an area under the receiver operating characteristic (AUROC) curve analysis was used to assess the accuracy of the pROSC model.

Results—A total of 11,503 patients met inclusionary criteria. Factors influencing pROSC included male gender (-0.12); Caucasian race (-0.19); age (-0.002 per year); traumatic (-0.57) and respiratory (0.54) causes of arrest; asystole (0.36); PEA (0.83) and VF or VT (1.29) as presenting ECG rhythms; layperson- (0.45) or healthcare provider (0.70)-witnessed arrest; EMS response time (-0.05 per minute); and bystander/first responder CPR (0.68). The AUROC of the validation set for pROSC was 0.71 (p<0.01).

Conclusion—The pROSC score predicts the outcome of field resuscitation efforts with moderate accuracy using variables that are easily measured. Such predictions can be used to assess the quality of EMS performance by comparing actual to expected outcomes. In addition, the pROSC score could potentially be used for risk-adjusting patients in future studies of OHCA. Further studies are needed that use national data and incorporate additional variables, such as comorbidities and prearrest medications, to improve predictive accuracy.