A nomogram to predict in‑hospital mortality in post-cardiac arrest patients: a retrospective cohort study.

Abstract

Nomograms of prognosis in patients with a history of cardiac arrest (CA) have been established. However, there are some shortcomings and interferences in their clinical application. Our study aimed at developing a utility nomogram to predict the risk of in‑hospital death in post‑CA patients. We retrospectively extracted data from the MIMIC‑IV database. The least absolute shrinkage and selection operator logistic regression and multivariable logistic regression were used to investigate independent risk factors. A nomogram defined as a prediction model was established for these independent risk factors. The model performance was measured by examining discrimination (area under the receiver operating characteristic curve [AUC]), calibration (calibration curve analysis), and utility (decision curve analysis [DCA]). A total of 1724 post‑CA patients were enrolled in the study. Of those, 788 survived and 936 died. The incidence of in‑hospital death was 54.3%. In this nomogram, the predictors included age, malignant cancer, bicarbonate, blood urea nitrogen, sodium, heart rate, respiratory rate, temperature, SPO2, norepinephrine prescription, and lactate level. The internally validated nomogram showed good discrimination (AUC 0.801; 95% CI, 0.775-0.835). The calibration curve analysis and DCA confirmed that this prediction model can be clinically useful. We established a risk prediction model based on the admission characteristics to accurately predict the clinical outcome in post‑CA patients. The nomogram might help with the risk identification and individual clinical interventions.