The Scoring Model to Predict ICU Stay and Mortality After Emergency Admissions in Atrial Fibrillation: A Retrospective Study of 30,206 Patients

Abstract

Abstract Background Atrial fibrillation (AF) imposes a significant burden on the emergency department (ED); the rapid assessment on the conditions and subsequent interventions are crucial for the prognosis of AF patients admitted to the ED. We aim to derive and validate a more accurate and simplified scoring model to optimize the triage of AF patients in the ED. Methods We conducted a retrospective study using data from the MIMIC-IV database and developed scoring models employing the Random Forest algorithm. The area under the receiver operating characteristic (ROC) curve (AUC) was used to measure the performance of the prediction for ICU stay, and the death likelihood within 3, 7, and 30 days following the ED admission. Results The study included 30,206 AF patients, in which 53.6% (16,184) are females. The median age is 75 years old (95% CI 65–83). All patients were randomly divided into training, validation, and testing cohorts at a ratio of 7:1:2. The training set consisted of 21,145 patients, the validation set included 3,020 patients, and the remaining 6,041 patients were classified as the validation set. Across the three cohorts, 9,444 patients (31%) necessitated ICU transfers, and mortality rates were 1% at 3 days, 2% at 7 days, and 6% at 30 days. In the testing set, the scoring models exhibited AUCs of 0.737 (95% CI 0.710–0.765) for ICU stay, 0.730 (95% CI 0.666–0.759) for death at 3 days, 0.748 (95% CI 0.710–0.786) for death at 7 days, and 0.740 (95% CI 0.713–0.768) for death at 30 days. Conclusion We derived and validated novel simplified scoring models with good discriminative performance to predict the likelihood of ICU stay, 3-day, 7-day, and 30-day death in AF patients after ED admission. This novel tool has a promising prospect to optimize the triage of the ED.