Objective: Plasma concentrations of B-type natriuretic peptide (BNP) and troponin are often measured for diagnostic purposes when patients are admitted with heart failure, but their prognostic value when measured soon after admission is uncertain. We aimed to investigate the added prognostic value of admission measurements of BNP and troponins in patients with acute heart failure. Methods and results: Multivariable prognostic models for death or any worsening heart failure (WHF) or rehospitalization for WHF by 30 days, 30-day death or rehospitalization for WHF, and 90-day mortality were constructed using baseline data from the Value of Endothelin Receptor Inhibition with Tezosentan in Acute heart failure Studies (VERITAS) including BNP and troponin I. Of 1347 patients, the median (interquartile range) value of BNP was 422 (156–945) pg/mL and 855 (63%) had measurable troponin I. By 30 days, 432 patients had died or experienced WHF. Clinical variables had only moderate predictive performance that was not substantially improved by BNP or troponin I (c-indices 0.6528 and 0.6595, respectively). By 30 days, 150 patients died or were rehospitalized for WHF. The c-index using clinical variables (0.6855) was not improved by adding biomarkers. By 90 days, 135 patients had died. The c-index for mortality was somewhat better than for composite outcomes (0.7394) but improved little with biomarkers (0.7461). Conclusion: Routine clinical data recorded at the time of admission in patients with acute heart failure are poor at predicting recurrent admissions but somewhat better at predicting mortality. Neither BNP nor troponin measured at admission improved predictions; measurement closer to discharge, or of other novel biomarkers, might perform differently.
- Acute heart failure
- Statistical models
ASJC Scopus subject areas
- Cardiology and Cardiovascular Medicine