The research I've heard about basically says "placebos only have subjective effects". The classic study was on asthma. Participants receiving the placebo reported feeling better, but had no measurable improvement in lung function.
So relying on a placebo for a subjective feeling of improvement could be dangerous whenever what's needed is an actual, objective improvement -- ie, "whenever something is actually wrong".
Would you go to a doctor who you knew dispensed ineffective sugar pills to the sick?
Would it be ethical for a doctor to dispense ineffective sugar pills to the sick and conceal this fact?
Here's the official opinion of the American Medical Association: http://www.ama-assn.org/ama/pub/physician-resources/medical-...?
It depends a little on what else they could dispense. I would rather they give out homeopathic medicines then give out antibiotics, in the face of unreasonable demands from their patients. I'd more rather they successfully educate their patients. I'm not sure the degree to which that's typically possible.
[1] http://www.theguardian.com/science/2010/dec/22/placebo-effec... [1 alternate] http://www.scientificamerican.com/podcast/episode/placebos-w...
While I think that "real" placebos would be better for this than de facto placebos with bullshit pseudoscience attached to them, it's still not completely stupid to reimburse them if they do provide the same level of relief as active medicine through placebo effect.
Of course they're not effective at killing bacteria, but there are some domains in which placebos have their place.
Edit - to be clear, the reason given for reimbursing them is explicitely that they are good as placebos, they are not claimed by the French social security to be useful as active medicine.