Benefits of combining prevalent and incident cohorts: An application to myotonic dystrophy