A new approach for sizing trials with composite binary endpoints using anticipated marginal values and accounting for the correlation between components | Publicación