What can we learn from discrete-choice experiments in osteoporosis?