Abstract 2918: Serial BNP Measurements Have Limited Utility in the Out-Patient Setting to Predict Death/Rehospitalization in Systolic Heart Failure
Objectives: To evaluate whether serial BNP measurements in the out-pt setting offer additional prognostic information in HF.
Methods: The STARBRITE trial randomized 130 pts hospitalized with HF to out-pt management based on clinical assessment alone vs. clinical assessment + serial BNP measurements at discharge, 10, 30, 60, and 90 days. The primary endpoint was death/rehospitalization at 90 days. The discharge BNP level, when volume status was optimized, was considered the baseline BNP. We used logistic regression to evaluate whether the ratio of the last measured BNP prior to death/readmission or at last f/u, relative to the baseline BNP, predicts outcomes. We used ROC analysis to evaluate the predictive value of different BNP ratios. BNP ratio was log transformed for the analysis.
Results: Complete BNP data was available in 94 pts. The median BNP ratio was significantly greater in the 32 pts who died or were rehospitalized compared to the 62 pts who survived without hospitalization (1.1 vs. 0.8, p=0.02). Increasing BNP ratios predicted death/rehospitalization, even after adjusting for level of congestion on clinical evaluation (p=0.02, Figure⇓). However, the sensitivity and specificity of particular BNP ratios for predicting death/rehospitalization remained poor (Figure⇓).
Conclusions: Marked elevations in BNP relative to baseline portend a poor prognosis. However, there does not appear to be a threshold increase above baseline that should trigger therapeutic changes to prevent death/rehospitalization.