Assessing drug effect from distributional data: A population approach with application to Duchenne Muscular Dystrophy treatment