Are control groups ethical? It depends on what type of disease process is being studied, what the study design is, and what kind of control group is being put together.
Ex 1: A new therapy for AIDS is being tried in Africa, where 85% of AIDS-progressed HIV patients dont get care. In the US, because almost all such subjects get care, it wouldnt be ethical to run a pure therapy-versus-non- pharmacologically-active-placebo trial. In Africa, it might be ethical to run such a trial, provided that it doesnt drag on forever, and patients arent being harmed. In the US, a non-inferiority trial design might be preferred, with the controls recieving the normal therapy they would or the current state-of-the-art therapy, and subjects trying out the experimental one.
Ex 2: A study is being conducted on venoclusive disease, which ends up in a fatality 95 - 99% of the time. The experimental therapy therapy has been tried before in another country, and seems to prevent fatalitites for every other subject.
Those outcome probabilities are HUGELY disparate. Best way to run the study is to run a historical/retrospective case-control study, with matched case-controls. In short, look for highly comprehensive databases with info about people who didnt get the therapy before it became available, and make them your controls. Then stratify analysis by socio-economic and biological factors in both the historic and the present subject groups.
Ex 3: You dont really want to use historical controls, because the populations differ very widely...ie it may not be very clear that a group of 30 - 50 year old Okinawans will compare in a valid way with 55 - 65 year old Zambians. You could run a cross-over design, where you recruit a certain number of subjects, observe them on a certain therapy for an amount of time, and then cross them over to the experimental therapy and vice versa (start some of the subjects on the study therapy and then watch their deterioration when they are crossed over to another type of therapy for a certain amount of time before intervening)
BOTTOM LINE:
The whole ethical debate about controls versus subjects is a whole lot more complicated than it seems on the surface of things, and can easily be side stepped by ingenious study design. This is why two weeks of epi and biostats in medical school doesnt cut the mustard, folks...it is also why in certain cases, and MD alone cannot suffice for clinical research design on the PI level.
