Not all side effects are equally negative. Some negative side effects are acceptable and ethical.
Let's take DMSO as an example. Treatments with DMSO leave a particular garlic-like taste in the mouth. This is a negative side-effect. But DMSO is not a poison during normal treatments any more than ethyl alcohol is a poison.
A good placebo for DMSO would have the same side-effects as otherwise a patient and perhaps even the doctor could tell the difference based on taste/smell compared to the anticipated taste/smell. I believe this still isn't possible, which makes testing DMSO difficult. One solution, for example, is to look for a dose-response curve, which is a more complicated signal to extract.
Doesn't that mean that it's poison by nature if said side effects are negative?
Screw wanting to take part in that.
"Here's a pill that will cause oily discharge and nothing else" Yum!