The Diagnostic Efficacy of TFC in Combination With E/e′ Ratio and LVEF for Pulmonary Congestion in Heart Failure Patients With Permanent Atrial Fibrillation

Background: Pulmonary congestion is a key manifestation of decompensated heart failure (HF) and contributes to adverse outcomes, especially in patients with permanent atrial fibrillation (AF). However, practical tools enabling bedside, repeatable, and real-time assessment of pulmonary congestion remain limited. To create and internally validate a non-invasive predictive model for pulmonary congestion in patients with HF and permanent AF utilizing hemodynamic, echocardiographic, and clinical parameters. Methods: This retrospective study included 66 patients with HF and permanent AF, classified into pulmonary congestion and non-congestion groups based on standardized chest radiography criteria. Cardiography (ICG) parameters, echocardiographic indices, and laboratory markers were evaluated. A multivariable logistic regression model was developed using a backward elimination approach guided by the Akaike Information Criterion (AIC). Model performance was evaluated using the area under the curve (AUC), calibration, and decision curve analysis, with internal validation conducted through bootstrap resampling. Model interpretability was further assessed by comparing AUCs of individual predictors and examining risk stratification based on model-derived scores. Results: The final model identified independent associations of pulmonary congestion with left ventricular ejection fraction (LVEF) (OR = 0.934, 95% CI = 0.879–0.992), E/e′ ratio (OR = 1.229, 95% CI = 1.029–1.467), and thoracic fluid conductivity (TFC) (OR = 1.237, 95% CI = 1.070–1.431). The model showed strong discriminative ability (AUC = 0.865, 95% CI = 0.773–0.956), satisfactory calibration (Hosmer-Lemeshow test, p > 0.05), and clinical utility. Internal validation using 500 bootstrap resamples confirmed these results, with robust discrimination (corrected AUC = 0.853, 95% CI = 0.763–0.942), consistent calibration, and maintained net clinical benefit. Model interpretability analysis confirmed its added discriminative value over individual predictors and supported its utility in stratifying pulmonary congestion risk. Conclusions: The proposed model provides a practical, non-invasive approach for identifying pulmonary congestion in patients with HF and AF. It may facilitate early bedside detection and support dynamic clinical decision-making.