The observable is that many antibiotics increase "feed efficiency". It's an aggregate measure, so perhaps it just reflects lower incidence of subclinical infection. And it permits high-density management, which increases the risk of infectious diseases.