The epidemiologic transition: one many or none?