Page 132 - Biomarkers for risk stratification and guidance in heart failure

Prognostic impact of change in NT-proBNP versus change in eGFR.
(overall P<0.05), all groups were individually compared with each other in order to assess which groups caused this difference. Glomerular filtration rate (GFR) was estimated by using the Modified Diet in Renal Disease equation (MDRD).13 Kaplan Meijer survival curves were assessed and compared using the log-rank test and Cox proportional-hazard regression analysis. Univariate Cox proportional- hazard regression analyses was performed in order to assess clinical covariates associated with the combined endpoint. We checked for interaction between all covariates and found interaction between urea baseline and a previous history of myocardial infarction. In multivariate analyses we corrected for this interaction. Multivariate Cox proportional hazard regression analysis was performed using all covariates associated with outcome to assess the clinical model. Variables were added in a stepwise fashion with p < 0.05 and P <0.1 as the cut-off for entry or retention, respectively. After assessment of the clinical model, NT-proBNP, eGFR, Urea, change in renal function (decrease versus increase in eGFR) and change in NT-proBNP (increase versus decrease) were added to form the cardiorenal model. Finally, the prognostic impact of severe worsening of renal function defined as a decrease in eGFR >20% and improvement of renal function (IRF) defined as an increase in eGFR >20% compared with a change in eGFR within 20% was analyzed in a multivariate manner.
Calculations were done using SPSS 21.0 (IBM corp, Armonk, New York, USA). 6
RESULTS
In 271 out of 345 patients included in the PRIMA study, NT-proBNP and creatinine levels at discharge and at the outpatient visit one month after hospital discharge were available. Patient characteristics at hospital discharge and one month after hospital discharge are depicted in table 1. Overall, patients were elderly, predominantly male, more than half of patients had a history of coronary heart disease, and most patients had been admitted because of de novo acute HF. Patients were divided into 4 groups based on increase versus decrease in NT- proBNP and eGFR. Overall group differences were seen in incidence of coronary artery disease, myocardial infarction, PCI or CABG, ICD, previous episode of HF and renal function at hospital discharge (table 1).
131