Introduction: Acute heart failure is a serious condition. Atrial fibrillation is the most frequent arrhythmia in patients with acute heart failure. The occurrence of atrial fibrillation in heart failure patients worsens their prognosis and leads to a substantial increase in treatment costs. There is no tool that can effectively predict the onset of atrial fibrillation in patients with acute heart failure in the ICU currently.
Materials and methods: We retrospectively analyzed the MIMIC-IV database of patients admitted to the intensive care unit (ICU) for acute heart failure and who were initially sinus rhythm. Data on demographics, comorbidities, laboratory findings, vital signs, and treatment were extracted. The cohort was divided into a training set and a validation set. Variables selected by LASSO regression and multivariate logistic regression in the training set were used to develop a model for predicting the occurrence of atrial fibrillation in acute heart failure in the ICU. A nomogram was drawn and an online calculator was developed. The discrimination and calibration of the model was evaluated. The performance of the model was tested using the validation set.
Results: This study included 2342 patients with acute heart failure, 646 of whom developed atrial fibrillation during their ICU stay. Using LASSO and multiple logistic regression, we selected six significant variables: age, prothrombin time, heart rate, use of vasoactive drugs within 24 h, Sequential Organ Failure Assessment (SOFA) score, and Acute Physiology Score (APS) III. The C-index of the model was 0.700 (95% CI 0.672-0.727) and 0.682 (95% CI 0.639-0.725) in the training and validation sets, respectively. The calibration curves also performed well in both sets.
Conclusion: We developed a simple and effective model for predicting atrial fibrillation in patients with acute heart failure in the ICU.
Keywords: Acute heart failure; Atrial fibrillation; Disease severity; Nomogram; Prediction model.
© 2022. The Author(s).