Published online by Cambridge University Press: 21 May 2015
Measurement of the serum B-type natriuretic peptide (BNP) level and more recently its precursor, N-terminal proBNP (NT-proBNP), has been advocated to facilitate the diagnosis of heart failure in the emergency department (ED). We sought to determine the potential impact of adding NT-proBNP testing to the routine evaluation of emergency patients with acute dyspnea.
This prospective cohort study enrolled a convenience sample of acutely dyspneic patients at a tertiary care ED. We excluded trauma patients and those under 30 years of age. Patients underwent standard evaluation, including radiography when indicated. At the point of final diagnosis and blinded to the NT-proBNP result, physicians documented the likelihood that heart failure accounted for the patient's acute dyspnea on a 7-point Likert scale, the data from which was subsequently collapsed to 3 categories for analysis purposes. The primary outcome was the agreement between clinical impression and the NT-proBNP assay classified using manufacturer-recommended, age-specific cut-offs. Newly proposed cut-offs from a recent study were also evaluated.
One hundred and twenty-nine patients making 139 ED visits were enrolled (median age 76 years; 59% admitted). The serum NT-proBNP assay was positive in 119 (86%, 95% confidence interval [CI] 80%–91%) cases, including 75% (43/57, 95% CI 62%–86%) of the cases that the treating physician felt were not caused by heart failure, and 86% (25/29, 95% CI 68%–96%) where the treating physician was unsure. The median NT-proBNP concentration was higher in patients clinically believed to have heart failure rather than pneumonia or chronic obstructive pulmonary disease; however, the ranges of these values overlapped extensively (median 4361 pg/mL; interquartile range [IQR] 2386–10877 v. 1651 pg/mL; IQR 370–4745, respectively).
There is high discordance between the clinical impression of treating physicians and NT-proBNP concentrations, notably in patients who are believed not to have heart failure. Although the reference standard of ED diagnosis is imperfect, the broad overlap in NT-proBNP concentrations suggests poor specificity in this target patient population. The introduction of routine ED NT-proBNP testing using the current cut-offs would be expected to result in substantial indirect costs from further diagnostic testing. It remains unclear whether the introduction of this diagnostic test would have a positive impact on clinically relevant patient outcomes.