Derivation and Validation of the CREST Model for Very Early Prediction of Circulatory Etiology Death in Patients without STEMI after Cardiac Arrest

BACKGROUND—No practical tool quantitates the risk of circulatory-etiology death (CED) immediately after successful cardiopulmonary resuscitation in patients without ST-elevation myocardial infarction (STEMI). We developed and validated a prediction model to rapidly determine that risk and facilitate triage to individualized treatment pathways. METHODS—Using the International Cardiac Arrest Registry (INTCAR), an 87-question data set representing 44 centers in America and Europe, patients were classified as having had CED or a combined endpoint of neurological-etiology death or survival. Demographics and clinical factors were modeled in a derivation cohort, and backward stepwise logistic regression used to identify factors independently associated with CED. We demonstrated model performance using area under the curve (AUC) and the Hosmer Lemeshow test in the derivation and validation cohorts, and assigned a simplified point scoring system. RESULTS—Among 638 patients in the derivation cohort, 121 (18.9%) had CED. The final model included preexisting Coronary artery disease (OR=2.86, CI 1.83-4.49, p=