Abstract 17309: Validation of a Novel Prediction Tool for Circulatory Etiology Death After Cardiac Arrest

Introduction: Post-resuscitation cardiac arrest (CA) triage to urgent angiography, percutaneous intervention, and mechanical circulatory support is hampered by inconclusive risk stratification, especially among patients without ST elevation myocardial infarction (STEMI). We analyzed registry data to develop a prediction tool to determine the risk of circulatory-etiology (CV) death in patients without STEMI, and validated it in a separate cohort. Methods: Using the International Cardiac Arrest Registry (INTCAR)-Cardiology data set and stepwise linear regression with an inclusion rule of P≤0.1, we determined demographic and clinical factors independently associated with CV death, and created a weighted prediction model for patients presenting after CA without STEMI. The model was then validated in a separate, larger cohort from INTCAR. This project was approved by the Maine Medical Center IRB. Results: Of 468 patients in the derivation cohort, 90 met criteria for the endpoint. In the multivariable model, age greater than 65 (OR=2.4, p=0.0001), preexisting coronary disease (OR=1.9, P=0.0065), diabetes (OR=1.8, P=0.01), in-hospital arrest (OR=1.5, P=0.1), time from collapse to return of circulation (TTROSC) greater than 25 minutes (OR=1.7, p=0.02), shock at presentation (OR=3.9, P<0.0001), and EF<30% on first echo (OR=1.6, P=0.05) were independently associated with CV death. Using weighted predictors (age>65 =1, prior CAD =1, diabetes =1, in-hospital arrest =1, TTROSC>25 =1, admission LVEF<30% =1, shock =2,), an additive score of 0-2 predicted CV death in 8.5% and ≥3 in 34% in the derivation cohort. In the validation cohort, which comprised 1197 patients, of whom 263 met criteria for CV death, a score of 0-2 was associated with 13.1% and ≥3 with 35.1% CV death, respectively. Conclusions: A simple bedside prediction tool can predict high (34-35.1%) vs. low (8.5-13.1%) risk of circulatory-etiology death in cardiac arrest survivors without STEMI. This model could be used to risk-stratify cardiac arrest survivors, and aid in the triage of patients to appropriate and cost-effective post-resuscitation treatments.