Is maximum positive predictive value a good indicator of an optimal screening mammography practice?