Probably the control should be some other psychoactive substance that also has some noticeable hallucinogenic effect, but is not believed to have curative properties.

The other way (which is close to what they have done) is to compare and contrast the therapeutic effects of different dosages - if the patient feels the expected sideeffects in both cases, but the normal dose achieves measurably better results than a minimally low dose, then that's some evidence about the effectiveness of the active ingredient.