If the physician can effect the trial by knowing to whom they're administering the placebo it seems that the trial lacks in its design. Someone else said that the person administering the drug/control could repeat a test because they felt the result was wrong - why is such a person involved in a drug trial? If they lie about that then it seems they can pervert the trial any number of ways.
How do the patients know how the other members of the study are being affected? If they don't get a dry mouth, but the drug company thinks they should, how does the patient know?
Surely you need the physicians and patients to record exactly what happened, no more and no less.
It seems that you're doomed to fail if you don't know that this is what's happening: suppose you use an active control that dries the mouth (for 95% of patients), if you don't have the side-effect with the drug on test then you know you're part of the active group with a high enough confidence level to alter the results in a qualitatively similar way to using a non-active control (placebo).
So what you need is a placebo that matches exactly the profile of the side-effects of the drug. I'd warrant that's impossible - you don't know the side-effects of the drug that are expected in a wide scale longitudinal study because that's what you're testing isn't it?
What is entirely nefarious is any use of "no significant side-effects beyond that experienced under placebo" when in fact they don't mean placebo but "active-control".
Placebo may be synonymous with active control in Pharma circles but it's not in common usage and so this is still a fraud.
Also, I'm not a huge fan of more regulation, but scientific integrity seems to demand the placebo ingredients be disclosed.
From the discussion, it seems clear the doctor-scientists have to think about and carefully consider what to use for the control. A single, simple answer cannot work.