Abstract
BACKGROUND: A mechanism to stratify patients resuscitated from a cardiac arrest according to the likelihood of an acute coronary lesion would have significant utility. We thus sought to develop and validate a risk prediction model for the presence of an acute coronary lesion among patients resuscitated from an arrest.
METHODS AND RESULTS: All subjects undergoing coronary angiography after resuscitation from a cardiac arrest were identified in an ongoing institutional registry from 2009 to 2014. Backwards stepwise selection of candidate covariates was used to create a logistic regression model for the presence of an angiographic culprit lesion and internally validated with bootstrapping. A clinical point score was generated and its prognostic abilities compared with contemporary measures. Among 247 subjects undergoing coronary angiography after resuscitation from a cardiac arrest, 130 (52%) had an acute lesion in a coronary artery. A multivariable model-including angina, congestive heart failure symptoms, shockable arrest rhythm (ventricular fibrillation/ventricular tachycardia), and ST-elevations-had excellent discrimination (optimism corrected C-Statistic, 0.88) and calibration (Hosmer-Lemeshow P=0.540) for an acute coronary lesion. Compared with electrocardiographic findings alone, a point score based on this model more accurately predicted the presence of an acute lesion among patients resuscitated from a cardiac arrest (integrated discrimination improvement, 0.10; 95% confidence interval, 0.04-0.19; P<0.001).
CONCLUSIONS: Patients with a cardiac arrest can be risk stratified for the presence of an acute coronary lesion using 4 easily measured variables. This simple risk score may be used to improve patient selection for emergent coronary angiography among resuscitated patients.