Background Models to stratify risk for patients hospitalised for acute decompensated heart failure (ADHF) do not include the change in N-terminal pro-B-type natriuretic peptide (NT-proBNP) levels during hospitalisation.
Objective The aim of our study was to develop a simple yet robust discharge prognostication score including NT-proBNP for this notorious high-risk population.
Design Individual patient data meta-analyses of prospective cohort studies.
Setting Seven prospective cohorts with in total 1301 patients.
Patients Our study population was assembled from the seven studies by selecting those patients admitted because of clinically validated ADHF, discharged alive, and NT-proBNP measurements available at admission and at discharge.
Main outcome measures The endpoints studied were all-cause mortality and a composite of all-cause mortality and/or first readmission for cardiovascular reason within 180 days after discharge.
Results The model that incorporated NT-proBNP levels at discharge as well as the changes in NT-proBNP during hospitalisation in addition to age ≥75 years, peripheral oedema, systolic blood pressure ≤115 mm Hg, hyponatremia at admission, serum urea of ≥15 mmol/L and New York Heart Association (NYHA) class at discharge, yielded the best C-statistic (area under the curve, 0.78, 95% CI 0.74 to 0.82). The addition of NT-proBNP to a reference model significantly improved prediction of mortality as shown by the net reclassification improvement (62%, p<0.001). A simplified model was obtained from the final Cox regression model by assigning weights to individual risk markers proportional to their relative risks. The risk score we designed identified four clinically significant subgroups. The pattern of increasing event rates with increasing score was confirmed in the validation group (BOT-AcuteHF, n=325, p<0.001).
Conclusions In patients hospitalised for ADHF, the addition of the discharge NT-proBNP values as well as the change in NT-proBNP to known risk markers, generates a relatively simple yet robust discharge risk score that importantly improves the prediction of adverse events.