Improving methods for the determination of minimal important difference estimates for patient-reported outcomes and their application in clinical research and decision-making