Right, but how to do this while preserving the placebo effect? If the patient knows it’s “just” water, the effect is lost. Some sort of deliberate deception seems to be necessary, which raises ethical problems.

I've found placebos work really well on me even when I know they're placebos.