Composite end points in clinical trials of heart failure therapy: how do we measure the effect size?