Predischarge Lung Ultrasound B-Lines as a Robust Predictor of 90-Day Rehospitalization in Heart Failure with Reduced Ejection Fraction (HFrEF): Evidence from Clinical and Classification and Regression Tree (CART)-Based Risk Stratification

Abstract

 Background: Heart failure is a major global health burden, with high morbidity, mortality, and substantial rehospitalization rates. Residual pulmonary congestion at discharge is a key determinant of early readmission but is often underestimated using clinical assessment alone. Lung Ultrasound (LUS) offers an objective, bedside tool for detecting pulmonary congestion through B-line quantification. This study aimed to evaluate whether predischarge B-line measurements predict 90-day rehospitalization due to all cardiovascular causes in patients with Heart Failure with reduced Ejection Fraction (HFrEF).

Methods: This retrospective cohort study included 107 adults hospitalized with HFrEF at Wahidin Sudirohusodo Hospital from March to July 2023. LUS was performed prior to discharge using an eight-zone protocol, and total B-line counts were recorded. Clinical, laboratory, and echocardiographic variables were obtained from the Heart Failure Registry and electronic medical records. Patients were followed for 90 days to identify heart failure–related rehospitalization. Statistical analyses included independent t-test, Mann–Whitney U tests, chi-square test, ROC curves, multivariate logistic regression, and a Classification and Regression Tree (CART) model to identify the strongest predictors of rehospitalization.

Results: Among 107 patients (mean age 58.52 ± 12.16 years; 74.77% male), 48 (44.86%) experienced 90-day rehospitalization due to all cardiovascular causes. Rehospitalized patients had significantly higher predischarge B-line counts (25.98 ± 8.23 vs. 17.63 ± 8.03, p <0.0001). A B-line cutoff of ≥20 predicted rehospitalization with 75.7% accuracy, 77.08% sensitivity, 74.58% specificity, and an AUC of 0.754. Renal dysfunction (eGFR 58.72 ± 29.32 vs. 75.43 ± 27.49 mL/min/1.73 m², p = 0.003; creatinine 1.86 ± 2.01 vs. 1.21 ± 0.76 mg/dL, p = 0.021), lower EF (31.6 ± 8.15% vs. 36.64 ± 9.05%, p = 0.004), and higher filling pressures (PCWP, E/e′) were also significantly associated with rehospitalization. The CART model identified B-line ≥20 as the strongest primary classifier, with additional risk contributed by renal impairment and reduced EF, yielding an accuracy of 90.65% and an AUC of 0.966.

Conclusion:  Predischarge B-line quantification is a strong predictor of 90-day rehospitalization in HFrEF. Integration of B-line assessment with renal function and cardiac parameters substantially improves risk stratification. LUS provides a practical, noninvasive tool for optimizing predischarge evaluation and may guide interventions to reduce early rehospitalization.