OBJECTIVETo evaluate the predictive value of admission plasma amino-terminal pro-B-type natriuretic peptide (NT-proBNP) on in-hospital mortality in patients with decompensated heart failure.

METHODSPlasma NT-proBNP levels were measured in patients with decompensated heart failure within 24 hours after admission with ELISA method. The NT-proBNP levels were compared between survivals and dying patients in hospital. ROC analyses were performed to evaluate the predictive value of admission plasma NT-proBNP on in-hospital mortality and to identify the optimal NT-proBNP cut-point for predicting in-hospital mortality. A binary logistic regress analyses was used to evaluate if NT-proBNP was an independent predictor for in-hospital mortality.

RESULTSA total of 804 patients with decompensated heart failure were enrolled in his study (293 valvular heart diseases, 219 ischemic cardiomyopathy, 141 dilated cardiomyopathy, 14 hypertrophic cardiomyopathy, 21 restrictive cardiomyopathy, 39 hypertensive heart disease, 41 chronic pulmonary heart disease and 36 adult congenital heart disease) and 96 patients were in class II, 450 in class III and 258 in cases IV according to NYHA Classification. During hospitalization, 64 deaths were recorded and the on admission plasma NT-proBNP levels of patients died during hospitalization were significantly higher than those of survivals [4321.1 (3063.8, 6606.5) pmol/L vs. 1921.6 (873.9, 3739.2) pmol/L, P<0.01]. Area under receiver operating characteristic curve (AUC) of NT-proBNP to predict in-hospital death was 0.772 (95% CI: 0.718 - 0.825, P<0.01), the optimal plasma NT-proBNP cut-point for predicting in-hospital mortality was 3500 pmol/L, with a sensitivity of 70.3%, a specificity of 72.0%, an accuracy of 71.9%, a positive predictive value of 17. 8% and a negative predictive value of 96.6%. Patients whose NT-proBNP levels were equal or more than 3500 pmol/L had a much higher in-hospital mortality (17.8%) compared with those with NT-proBNP levels of less than 3500 pmol/L (3.4%), P<0.01. Binary logistic regress analyses demonstrated that admission plasma NT-proBNP, pneumonia, heart rate and NYHA class were independent predictors for in-hospital mortality in patients with decompensated heart failure (P<0.05 or 0.01) and admission plasma NT-proBNP was the strongest predictor for in-hospital mortality.

CONCLUSIONSAdmission plasma NT-proBNP level was an independent predictor for in-hospital mortality in patients with decompensated heart failure. The optimal NT-proBNP cut-point for predicting in-hospital mortality was 3500 pmol/L in this patient cohort.