ArticlesAlmost three years ago, I pondered whether you could harness the placebo effect in a pill, and how you would test that.
Then, in the comments, I wondered if you could get people to knowingly take a placebo in a trial. In a later comment, I found an exploratory study that suggested it might be possible – at least among neurotics.
Now I have discovered a report of a new “proof-of-principle” study suggesting that openly and honestly providing placebos may still have a placebo effect.
Disclaimer: I haven’t read the source article, but have perused the abstract.
There are small surprises here: yes, you can get people to take sugar pills even if they know they are sugar pills; placebos might still work even if you know they are placebos; and the National Center for Complementary and Alternative Medicine sponsored a trial that wasn’t widely attacked for poor design!
But there was an mild poke from the Not Exactly Rocket Science blog, pointing out that the trial wasn’t double-blinded because the patients knew they were receiving the placebo (as opposed to no treatment) – the scientists, quite rightly, did not.
It is not often you will see me defend a medical study against the need for double-blindness, but this comes back to a very similar issue to my original point in my original post. When you are testing the effectiveness of an openly-revealed placebo, then double-blindness is surely meaningless. I think having a single-blind test (for the researcher) is the most appropriate design possible in this case.
The other mild criticisms, acknowledged it seems, by the author, seemed more fair, and I think we all agree it needs more substantial study before a conclusion can be drawn.
Comment by Aristotle Pagaltzis on December 24, 2010
Maybe we need a trial where some patients receive the real drug and some a placebo, and some patients are told they are receiving the real drug and some a placebo, but what patients get and what they are told are uncorrelated. That would be double-blinded, of course.
Comment by Julian on December 26, 2010
Aristotle,
I need to think that suggestion through.
Let’s call those tests:
That would give us six different comparisons.
In some of them, two variables changed at the same time, so it is difficult to get a clear picture of the cause:
Now, two of the remaining three don’t test the placebo effect but the real drug.
So that leaves GPTR versus GPTP, which may be an interesting measure. It suggests how much belief in the placebo affects the placebo effect. It wouldn’t surprise me if some experiments in this area have been done.
If humans were robots, it would be reasonable to believe that there would be no difference in this test, because those robots would not get better just because they believed that they would.
If belief was the entire part of the placebo effect, you would expect the difference to be large – i.e. account for the whole of the placebo effect. But I have several provisos.
The first is that we have no way of telling from this experiment if it does account for the full placebo effect. To tell that, we would have to compare with no treatment. That’s what the experiment described in the post does, but at a cost of double-blindness.
The second is that the experiments (as described) don’t distinguish between what the patients were truthfully told, and what they irrationally believed. If they, deep down, thought the doctors might be lying and/or that the pill really might have an effect, then experiment has failed to control for belief.
The third is that this “proof-of-principle” study (which may well be wrong or flawed and needs further exploration) has suggested that our expectations – that the placebo effect would be destroyed by knowing the truth – could be wrong.
Comment by Julian on December 26, 2010
I used to play a variant of roller-hockey with a group of friends. Occasionally, one would fall over and bruise themselves. In these cases of no serious injury, to distract them from the pain and indignity, they would be offered a “medicinal jube“. After eating a sweet, they would return to the game, happier.
It wasn’t long, however, before the sweet-tooths in the team started demanding “preventative jubes” before the game and at half-time, to reduce the risk of injuries during the game.
I like to think this study on placebos somehow justifies
mytheir demands.Comment by Julian on December 26, 2010
Sweet-tooths? Sweet-teeth? Sweets-tooth?
Comment by Aristotle Pagaltzis on December 26, 2010
I don’t think it’s of primary interest to split the comparisons out in this way. If it were, you’d just run a regular trial for that one constellation. The more interesting conclusions would seem to follow from considering triplets of these constellations and from comparisons of some such triplets to other (where they differ by only one member).
Your objection about controlling for belief is a good point, though.
Comment by Julian on December 26, 2010
Aristotle, I can’t say I understand how that would work…
Comment by Julian on December 30, 2010
The Science-Based Medicine blog wasn’t overly impressed with the study I cited, and the hype that surrounded it.
They challenged whether the effect measured was big enough to be clinically relevant, especially given the subjectiveness of the reporting – a topic I didn’t address.
There is an aspect here, though, that I think they are missing. Homeopaths get a lot of credit for the power of their placebos. In situations where there genuinely isn’t any treatment, it would be nice if doctors could fall back on placebos, for the non-trivial-but-still-limited benefit they can give. However, there are severe ethical issues involved with doctors lying to patients. This study has embedded within it the idea that there may actually be an ethical way to prescribe placebos.