> There was that famous meta-analysis from 2001 that showed no significant effects, but that averaged over a _lot_ of trials where we wouldn't expect much of an effect (cancer, particularly).
This is the exact meta-analysis that I was referring to.
I'd need to go back to my PhD to remember all the flaws I found with that approach, and the vast number of studies (including meta-analyses) that found statistically significant placebo effects following this.
Particularly for pain, there's biochemical evidence of placebo effects being created due to sugar pills, and these effects being reversed by opioid antagonists (naloxone, particularly), so the evidence is particularly strong for this.
If I get a chance this week, I'll review my thesis for the citations that I knew back then (but it's been over 10 years since my viva and I no longer work in the area so can't recall off the top of my head).
(This is standard for Cochrane reviews, btw and definitely a good thing). I fundamentally disagree with the approach of this review and analysis (placebos are not a generic effect (apart from pain)) but love Cochrane in general).
They find no strong placebo effect, and when there is a placebo effect, it is difficult to distinguish statistically from no-treatment.