Development and validation of a nomogram to predict in-hospital cardiac arrest among patients admitted with acute coronary syndrome.

Abstract

This study aimed to develop and validate a nomogram to recognize in-hospital cardiac arrest (CA) in patients with acute coronary syndrome (ACS). This multicenter case-control study reviewed 164 ACS patients who had in-hospital CA and randomly selected 521 ACS patients with no CA experience. We randomly assigned 80% of the participants to a development cohort, 20% of those to an independent validation cohort. The least absolute shrinkage and selection operator (LASSO) regression model was used for data dimension reduction, and multivariable logistic regression analysis was used to develop the CA prediction nomogram. Nomogram performance was assessed with respect to discrimination, calibration, and clinical usefulness. Seven parameters, including chest pain, Killip class, potassium, BNP, arrhythmia, platelet count, and NEWS, were used to create individualized CA prediction nomograms. The CA prediction nomogram showed good discrimination (C-index of 0.896, 95%CI, 0.865-0.927) and calibration. Application of the CA prediction nomogram in assessments of the validation cohort improved discrimination (C-index of 0.914, 95%CI, 0.873-0.967) and calibration. The results of decision curve analysis demonstrated that the CA prediction nomogram was clinically useful. Our study generated a friendly risk score to recognize in-hospital CA with good discrimination and calibration. Further studies need to establish a pathway to guide the application of the risk score in clinical practice.