Background: In patients suspected of new slow-onset heart failure, data on the comparative diagnostic performance of automated B-type natriuretic peptide (BNP) assays are scarce.
Methods and results: Two hundred patients referred to a heart failure outpatient diagnostic facility underwent standard diagnostic work-up including echocardiography. The reference standard for the diagnosis of heart failure was an expert panel conclusion. N-terminal pro-BNP on Elecsys and BNP on Axsym and Centaur machines were measured in a single batch. Data were available for 172 patients; 51 had heart failure (29.7%). All 3 tests had high c-statistic values. An intermediate-risk subset of 111 patients (34% with heart failure) was created by excluding patients with very high or very low probability based on history and physical examination, the subgroup most in need of an additional test. Applying different thresholds for ruling heart failure in or out, the positive predicted values in this "gray zone" group were 75%, 76%, and 72%, respectively, and the negative predictive values 83%, 71%, and 85%, with the remaining 50% of patients having ∼18% probability of heart failure.
Conclusion: In practice, a valid diagnosis in patients suspected of slow-onset heart failure remains elusive for many in the absence of echocardiographic imaging.
Copyright © 2011 Elsevier Inc. All rights reserved.