Skip to content


Open Access

N-terminal pro-B-type natriuretic peptide and the diagnosis of left ventricular systolic dysfunction: what is the optimal cut-off?

  • A Turley1,
  • A Roberts1,
  • B Kunadian1,
  • A Davies1,
  • M de Belder1,
  • J Drury1 and
  • M Stewart1
Critical Care20059(Suppl 1):P331

Published: 7 March 2005


Public HealthPeptideConfidence IntervalLeft VentricleClinical Symptom


The neurohormone N-terminal pro-B-type natriuretic peptide (NT-proBNP) is predominantly released from the left ventricle in response to increasing wall tension, and is suggested as a useful screening test for suspected left ventricular systolic dysfunction (LVSD).


To evaluate the predictive value of different NT-proBNP reference cut-off values in patients with clinical symptoms and signs suggestive of LVSD.


NT-proBNP was measured in 1054 patients with clinical symptoms and signs suggestive of LVSD over a 1-year period (September 2003–October 2004). We retrospectively reviewed our digital archive echocardiographic database.


An echocardiogram was performed in 549/1054 (52%) patients. NT-proBNP was >150 ng/l in 744/1054 patients (71%), 42% male, median age 76 years (33–100 years) and > 500 ng/l in 411/1054 patients (40%), 50% male, median age 76 years (33–100 years). ROC curves are shown for NT-proBNP cut-off values of 150 ng/l, positive predictive value (PPV) 0.22 (confidence interval [CI] 0.19–0.25), negative predictive value (NPV) 0.97 (CI 0.94–0.98), and 500 ng/l, PPV 0.3 (CI 0.26–0.35), NPV 0.93 (CI 0.9–0.94), for the detection of LVSD.

At a cut-off value of 500 ng/l, NT-proBNP has a sensitivity 0.85 (CI 0.76–0.91), specificity 0.67 (CI 0.64–0.69), PPV 0.18 (CI 0.15–0.22), and NPV 0.98 (CI 0.97–0.99) at detecting moderate to severe LV impairment.


NT-proBNP is a useful tool in the 'rule-out' of LVSD. An NT-proBNP level of 500 ng/l has a similar negative predictive value to an NT-proBNP level of 150 ng/l with improved positive predictive value. This is particularly important in this population where other confounding comorbidities are common.
Figure 1
Figure 1

(abstract P331)

Figure 2
Figure 2

(abstract P331)

Authors’ Affiliations

James Cook University Hospital, Middlesbrough, UK


© BioMed Central Ltd 2005