Abstract 4144059: Predictors of Sudden Cardiac Arrest Presenting with Pulseless Electrical Activity

Background: The prevalence of sudden cardiac arrest (SCA) manifesting as pulseless electrical activity (PEA) has been rising, survival outcomes are poor and there are no known treatments. Little is known about the clinical profile of these individuals who present with PEA-SCA. Hypothesis: Individuals presenting with PEA-SCA have unique clinical profiles that can be determined from their lifetime clinical records and resting 12-lead ECG. Methods: From an ongoing prospective, population-based study of out-of-hospital SCA in the Northwestern US (catchment pop. ≈1 M, 2002-2020), we identified adults with PEA-SCA. Comparisons of lifetime medical records and resting 12-lead ECG closest and unrelated to the SCA event were made with a control group of which 70% had known coronary artery disease (CAD). The data archive was divided, a priori, into training and validation datasets. Prediction models were constructed using backwards stepwise logistic regression, with a p value of 0.05 to retain variables in the training dataset and applied to the validation dataset. Results: A total of 604 PEA-SCA cases (male 64%, age 72±14yr) and 1418 controls (male 68%, age 66±2yr) had medical history and ECG available. The full dataset was divided into training (67%) and validation (33%) subsets. A multivariable logistic regression model in the training dataset showed that cases were older, had higher odds of using anti-psychotic medications, and higher odds of stroke, heart failure, kidney disease, chronic obstructive pulmonary disease, seizure disorder, syncope, anemia and hyperthyroidism. ECG heart rate >75 bpm, Tpeak to Tend interval ≥90 ms and delayed QRS transition zone were also associated with higher risk of PEA SCA (Table). Cases had lower odds of hypothyroidism and myocardial infarction. The model displayed very good discrimination in the training dataset (C statistic 0.86 [0.836-0.88]) and was successfully validated (C statistic 0.81 [0.77-0.84]). Conclusion: A model that combined clinical profile with ECG variables showed very good discrimination for prediction of PEA-SCA vs. controls with CAD. These findings have implications for future investigations targeted at improving prevention and management of SCA presenting with PEA.