Development and validation of a nomogram model for predicting in-hospital cardiac arrest risk: A prospective multi-center observational study

To construct and validate a prediction model for the risk of cardiac arrest (CA) in patients in the emergency resuscitation room. This prospective multi-center observational study enrolled 936 patients in the emergency resuscitation rooms of 6 Grade III Class A hospitals in Xinjiang from 12/01/2023 to 31/08/2023. After rigorous screening, 463 patients were included in the analysis. 324 patients were included in the modeling group for predicting the cardiac arrest, 139 patients were assigned to the validation group to evaluate the model. The modeling group was further divided into a Cardiac arrest group (n = 164) and a non-CA group (n = 160). Data analysis and modeling were performed, and the model was validated through various statistical and visualization methods, comparing its predictive performance with the Modified Early Warning Score. Five risk factors were ultimately identified, and a prediction model for the risk of cardiac arrest was constructed. The evaluation and validation results of the model showed that the area under the curve (AUC) of the modeling group was 0.957 (95% confidence Intervals (CI), 0.935–0.979), and the AUC of the validation group was 0.952 (95% CI, 0.921–0.993); the positive predictive value was 0.928, and the negative predictive value was 0.951. The HL test indicated χ² =12.800, P = 1.000, while the AUC of the modified early warning score scoring model was 0.621 (95% CI, 0.662, 0.580), P < .001. A cardiac arrest risk prediction model based on the neutrophil count, tracheal intubation, potassium, Lactate, and albumin have been successfully constructed. This model exhibits good accuracy, discrimination, and clinical practical value.