Page 53 - Biomarkers for risk stratification and guidance in heart failure

Chapter 3
All events were adjudicated by a blinded event committee, consisting of medical specialists in cardiology, nephrology, vascular medicine, pulmonology, and neurology. Serious adverse events included admissions to the emergency room, hospital admissions, and death. Treatment in the NT-proBNP group was considered to be protocol adherent when 1 of the following actions was undertaken upon an elevated outpatient NT-proBNP level: starting or intensifying HF medication according to the ESC guidelines, all therapeutic and diagnostic actions searching for underlying causes of HF such as hypertension, ischemic heart disease, valvular heart disease, anemia, and cardiac arrhythmias; hospital admission (for decompensated HF); or registering for heart transplantation.
Statistical analysis
Based on previous studies and observations, it was estimated that, with an event rate of 20%, 480 patients would be needed to reach a relative risk reduction in number of events of 50% in the NT-proBNP-guided group compared with the clinically-guided group at an α level of 0.05 and a power level of 0.80. Although the primary end point changed during the start-up phase of the study, power analysis remained the same. One year after the first patient being included, a pre- specified interim analysis was performed, with a difference in events (p<0.01) as criterion to preliminary stop according to lan-DeMets alpha spending rule. The interim analysis demonstrated a pooled event rate of 65%. Thereupon the power analysis was re-evaluated. It was calculated that 364 patients were needed to demonstrate a minimum reduction in pooled events of 30%.
Results are presented as frequencies, mean (SD), or median (interquartile range \[IQR\]), where appropriate. Between group comparisons were performed using the t test, Mann-Whitney U test, or chi-square test where appropriate. Event rates for all-cause mortality were estimated by the Kaplan-Meier method. Hazard ratios were calculated using Cox regression analysis. Time-dependent Cox regression analysis was performed to analyze the prognostic impact of elevated NT-proBNP levels above target value at the outpatient clinic. All calculations were performed with the use of the SPSS statistical package version 15.0 (SPSS Inc., Chicago, Illinois).
52