Validation of the CREST score for predicting circulatory-aetiology death in out-of-hospital cardiac arrest without STEMI

Abstract

Aims: The CREST tool was recently developed to stratify the risk of circulatory-aetiology death (CED) in out-of-hospital cardiac arrest (OHCA) patients without ST-elevation myocardial infarction (STEMI). We aimed to validate the CREST score using an external cohort and determine whether it could be improved by the addition of serum lactate on admission. Methods: The study involved the retrospective analysis of consecutive patients admitted to a single tertiary centre with OHCA of presumed cardiac origin over a 51-month period. The CREST score was calculated by attributing points to the following variables: Coronary artery disease (CAD), non-shockable Rhythm, Ejection fraction <30%, cardiogenic Shock at presentation and ischaemic Time ≥25 minutes. The primary endpoint was CED vs neurological aetiology death (NED) or survival. Results: Of 500 patients admitted with OHCA, 211 did not meet criteria for STEMI and were included. 115 patients died in hospital (71 NED, 44 CED). When analysed individually, CED was associated with all CREST variables other than a previous diagnosis of CAD. The CREST score accurately predicted CED with excellent discrimination (C-statistic 0.880, 95% CI 0.813-0.946) and calibration (Hosmer and Lemeshow P=0.948). Although an admission lactate ≥7 mmol/L also predicted CED, its addition to the CREST score (the C-AREST score) did not significantly improve the predictive ability (CS 0.885, 0.815-0.954, HS P=0.942, X2 difference in -2 log likelihood =0.326, P=0.850). Conclusion: Our study is the first to independently validate the CREST score for predicting CED in patients presenting with OHCA without STEMI. Addition of lactate on admission did not improve its predictive ability.