Aims: Red cell distribution width (RDW) is prognostic in patients with heart failure (HF), but it has not been compared with N-terminal brain natriuretic peptide (NT-proBNP). We sought to make this comparison.
Methods and results: Patients referred to a specialist HF clinic between 2001 and 2008 were assessed comprehensively including medical history, echocardiogram, and blood tests. Cox-regression was used to assess the multivariable relationship between RDW, NT-proBNP, and all-cause mortality. A total of 1087 patients were recruited; median (IQR) follow-up was 52 months (29-66); age 72 years (64-78); 74% male; 70% ischaemic heart disease; 20% diabetic; 85% NYHA >or= 2, and 63% with at least moderate LV impairment (EF < 35% equivalent). In a multivariable model, both RDW and NT-proBNP were independently prognostic (RDW: chi(2) = 21.8 vs. 49.1 both P < 0.001). In a model using quartiles of each variable, the relative risk for each was similar for the second and third quartiles compared with the first. A larger increase in risk for NT-proBNP is seen in the fourth quartile.
Conclusion: Red cell distribution width is a readily available test in the HF-population with similar independent prognostic power to NT-proBNP across the first to third quartiles. Prognostic models in HF should include RDW and further investigation is necessary to determine the pathological mechanism of the relationship.