Do n-of-1 trials really tailor treatment?