Objectives:
To assess the diagnostic accuracy of B-type natriuretic peptide (BNP) and N-terminal proBNP (NT-proBNP) for detecting heart failure (HF)
To determine whether BNP and NT-proBNP are independent predictors of mortality and morbidity in HF and whether they add to the predictive value of other markers
To ascertain whether treatment guided by BNP or NT-proBNP improves outcomes in HF compared with usual care
To assess the biological variation of BNP and NT-proBNP in HF and non-HF populations
Data sources: Medline®, Embase™, AMED, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and CINAHL from 1989 to June 2012. Reference lists of included articles, systematic reviews, and gray literature were also searched.
Review methods: Studies were evaluated for eligibility and quality, and data were extracted on study design, demographics, diagnostic test characteristics, predictor factors, interventions, outcomes, and test-performance results.
Results: In emergency settings, BNP (51 studies) and NT-proBNP (39 studies) had high sensitivity and low specificity, and were useful for ruling out but less useful for ruling in HF. Similar results were shown in primary care settings for BNP (12 studies) and NT-proBNP (20 studies). The majority of studies assessing prognosis (183 studies) showed associations between BNP and NT-proBNP and all-cause and cardiovascular mortality, morbidity, and composite outcomes across different time intervals in patients with decompensated and chronic stable HF. Most of these were early-phase predictor-finding studies rather than model-validation or impact studies. Incremental predictive value was assessed in decompensated acute HF (7 studies) and chronic HF (15 studies). Almost all studies showed that calibration and discrimination statistics confirmed the added incremental value of BNP and NT-proBNP. Fewer studies used reclassification and model validation computations to establish incremental value. In the general population (seven studies), an association exists between NT-proBNP and mortality (all-cause, cardiovascular, and sudden cardiac) and morbidity (HF and atrial fibrillation). Overall, therapy guided by BNP/NT-proBNP was shown to reduce all-cause mortality but was graded as low strength of evidence. Seven studies assessed biological variation. The difference in serial results was higher for BNP than NT-proBNP, and the index of individuality for BNP and NT-proBNP was very low.
Conclusions: BNP and NT-proBNP had good diagnostic performance for ruling out HF but were less accurate for ruling in HF. BNP and NT-proBNP had prognostic value in HF and the general population. Therapeutic value was inconclusive. Data on biological variation expressed the differences in results and individuality expected in patients, suggesting that serial measurements need to be interpreted carefully.