Don't know much about drugs and trials so I am speaking out of my ass but I'd assume if the numbers were large enough, you don't actually infect the patients.

Don't know much about drugs and trials so I am speaking out of my ass but I'd assume if the numbers were large enough, you don't actually infect the patients. I would expect you could compare the natural rate of infection in the target group vs the control group after many years taking into account the historical/expected rate of infection in a random sampling of the population involved.