Evaluating treatment efficacy by multiple end points in phase II acute heart failure clinical trials: analyzing data using a global method.

BACKGROUND: To assess concomitant simultaneous effects on multiple end points using global statistical methods in phase II acute heart failure studies. METHODS AND RESULTS: Using simulations we have assessed different statistical methods to evaluate concomitant effects of a new intervention on dyspnea relief (using 2 measures), length of hospital stay, worsening heart failure to 5 days, mortality, and heart failure readmission to 30 days. Treatment effect scenarios included large (20% to 28% relative improvements) and very large (30% to 43% relative improvements) effects among others. Placebo responses and correlations among end points typical in recent acute heart failure clinical trials were used. Powers for the average Z score exceeded 70% with ≥75 patients per group for 35% relative improvement across all 6 end points. Assessing dyspnea alone generally provides lower power than the average Z score approach, with power deducted ≈50% under most of scenarios. Other approaches generally provide lower power than the average Z score method. CONCLUSIONS: Assessing the effects of new therapies on multiple clinical end points using the average Z score enables detection of therapeutic efficacy using sample sizes of 100 to 150 patients per group, approximately double the power achievable assessing the effects on dyspnea alone.

Duke Scholars

Author Michael J Pencina Biostatistics & Bioinformatics, Division of Biostatistics