Acute Heart Failure
The clinical diagnosis of HF can be challenging, particularly in patients presenting with acute shortness of breath in the urgent care setting. Information obtained from clinical history and physical examination, as well as from the electrocardiogram and chest radiograph, may provide valuable clues as to whether HF is the cause of symptoms, but additional diagnostic tests, including echocardiography, may be required to obtain a more definite diagnosis. Currently, the best documented and most widely used clinical application of BNP testing is for the emergency diagnosis of HF in patients presenting with acute dyspnoea. Following the publication of the results of The Breathing Not Properly Multinational study in 2002, BNP measurements have rapidly entered the clinical arena. This multicentre diagnostic test evaluation trial, which included 1,586 patients who visited the emergency department (ED) with a main complaint of acute dyspnoea, used a rapid point-of-care fluorescence immunoassay for BNP determination.² Diagnosis of HF was adjudicated by cardiologists blinded to the BNP results. BNP levels were found to provide strong and incremental diagnostic information to conventional historical, clinical or other laboratory tests and to have greater diagnostic accuracy for the diagnosis of heart failure than the ED physician using all other available information. Importantly, BNP performed well in patients with an intermediate (20–80%) pre-test probability of HF as evaluated by the ED physician.³ In the subsequent N-terminal Pro-BNP Investigation of Dyspnea in the Emergency Department (PRIDE) study, similar results were published for NT-proBNP. This study included 599 patients presenting to the ED of Massachusetts General Hospital in Boston with acute dyspnoea.⁴ The primary aim was to compare the diagnostic accuracy of NT-proBNP with that of clinical assessment for diagnosing acute HF. In this trial, NT-proBNP again performed slightly better than clinical assessment alone as a diagnostic tool. However, the combined use of NT-proBNP and clinical assessment performed better than either alone. A decision threshold for NT-proBNP of 900pg/ml was found to provide optimal discrimination for diagnosing heart failure, with an overall diagnostic accuracy of 87% (sensitivity 90%, specificity 85%). Lowering the decision threshold to 450pg/ml resulted in increased sensitivity (98%), but at the cost of decreased specificity (76%) and a slight reduction in overall diagnostic accuracy (83%). Based on a pooled analysis of studies evaluating the diagnostic value of NT-proBNP in acutely dyspnoeic patients, optimal confirmatory and exclusionary cut-off values for acute HF have been presented.⁵

Measurement of BNPs on presentation to the ED may result not only in an enhanced initial diagnostic approach to the patient, but also in a shorter hospital stay and reduced cost. The effect of rapid point-of-care BNP testing on time to discharge and total cost of treatment was first evaluated in the B-type Natriuretic Peptide for Acute Shortness of Breath Evaluation (BASEL) study, a prospective, non-blinded, randomised trial of 452 acutely dyspnoeic patients.⁶ BNP measurement in the ED resulted in reduced need for intensive care and reduced length of hospitalisation, which together translated into reduced total cost of treatment. The results of the BASEL trial have recently been corroborated and extended by the Canadian IMPROVE-CHF study, a randomised trial of 501 patients presenting with shortness of breath to seven EDs across Canada.⁷ Knowledge of NT-proBNP levels was associated with a shorter stay in the ED (on average 42 minutes), but the likelihood of being admitted to hospital, duration of hospitalisation or in-hospital mortality were not significantly different between the NT-proBNP and the conventional care groups. Importantly, re-hospitalisation rates at two months following discharge and direct hospital costs were significantly reduced when NT-proBNP levels were known to providers.

Although this study suggests that NT-proBNP testing provides clinically useful diagnostic information, it is not clear whether the better outcomes and savings were related to better and more rapid triage and initiation of adequate therapy or to less use of diagnostic testing. Although serial sampling of NT-proBNP was performed in hospitalised patients (at 72 hours), these results have not yet been published. The value of repeated testing therefore remains uncertain. Despite the shortcomings of IMPROVE-CHF, including unblinded design and relatively modest statistical power, its results strongly support the routine use of BNPs in the initial evaluation of patients presenting with acute dyspnoea.