Use of B-type natriuretic peptide testing in a community teaching hospital 4 years after implementation and agreement of results with discharge diagnoses.

To the Editor: We previously reported on the use of B-type natriuretic peptide (BNP) testing in the 400-bed Hennepin County Medical Center at the inception of testing in August 2001 (1 ). We now describe its use and report the agreement of the results with diagnoses in the patient records in this urban teaching hospital 4 years later. We queried all BNP orders for June through August 2005 and found 975 test orders on 608 patients. To determine a diagnosis of heart failure, physician discharge dictations and ICD-9 codings were reviewed after Institutional Review Board approval. Final diagnoses were likely influenced by the BNP concentrations in some cases, potentially leading to overestimation of apparent diagnostic accuracy. The Biosite Triage and Beckman Coulter Access BNP assays were performed according to the manufacturers’ guidelines and were highly correlated (r 0.99; n 50), with Bland–Altman analysis demonstrating a mean (SD) difference of 16 (23) ng/L. Total imprecision (CV) near the 100 ng/L cutoff was 12% for both assays. In 2001, most BNP tests were requested by the emergency department (44%) and cardiology units (41%), noncardiac intensive care unit (ICU; 3.6%), general medicine (3.5%), and miscellaneous clinics (4.1%) (1 ). In 2005, the emergency department ordered 41% and the cardiology units only 10%, diluted by the increase in test utilization by noncardiac ICUs (16%), general medicine units (15%), miscellaneous clinics (13%), cardiology clinics (3.5%), and miscellaneous inpatient services (1.2%; Table 1). Thirty percent of patients had a diagnosis of heart failure, and 70% did not. One half (50%) of the group were female. The mean (SD) ages of the heart failure and non-heart failure groups were 63 (15) and 58 (16) years, respectively (P 0.01). The median (interquartile range) BNP concentration was significantly greater for heart failure patients than for non-heart failure patients [482 (13–1024) ng/L vs 61 (5– 161) ng/L, respectively; P 0.01]. BNP was increased ( 100 ng/L) in 92.8% of patients with a diagnosis of heart failure and was within reference values in 53.6% of those without heart failure. In patients without a diagnosis of heart failure, 272 BNP tests had values 100 ng/L, representing 28% of ordered tests in these patients (without heart failure). Many of these patients had renal, other cardiovascular, pulmonary, or liver disease and/or multisystem pathology, which can also increase BNP. In patients with a diagnosis of heart failure, 28 BNP results were 100 ng/L. Possible explanations include confounding variables such as Table 1. Mean (SD) measured BNP immunoreactivities, as percentages of baseline value, for samples subjected to different sampling and storage conditions and tested with various automated assays.