Development and validation of a diagnostic prediction model for heart failure-related pleural effusions: the BANCA score

Abstract

Aim. Heart failure (HF) is a common cause of pleural effusion (PE); however, its diagnosis can be challenging owing to overlapping clinical features with other causes. Measuring N-terminal pro-brain natriuretic peptide (NT-proBNP) levels is diagnostically useful, but the optimal cut-off point must be determined. This study aimed to optimise NT-proBNP cut-offs for diagnosing HF-related PEs and develop a scoring model to improve diagnostic precision. Methods. A retrospective analysis was conducted on 2005 patients with PEs, who were divided into derivation and validation cohorts. Receiver operating characteristic curve analysis identified the optimal cut-off point for NT-proBNP in pleural fluid and serum. A scoring system was developed based on logistic regression (TRIPOD statement) that included both clinical and laboratory variables. Its diagnostic accuracy was compared with that of existing methods using the area under the curve (AUC). Results. NT-proBNP ≥2500 pg·mL−1 was found to be the optimal cut-off for identifying HF-related effusions. Pleural fluid NT-proBNP levels tended to have superior diagnostic accuracy compared with serum NT-proBNP levels. A scoring system, denominated as BANCA (bilateral effusions on chest radiographs, age, NT-proBNP levels in pleural fluid, cholesterol in pleural fluid, and albumin gradient), demonstrated an AUC of 0.957, outperforming existing diagnostic criteria for identifying cardiac effusions. In cases misclassified as exudates by Light's criteria, the BANCA score accurately identified 80% of the HF-related PEs. Conclusions. The BANCA score is a reliable tool for diagnosing HF-related PEs, offering superior accuracy compared with co