If Placebos Work, Why Placebo-Controlled Trials?
Placebos have been a hot-button issue in human-subjects research ethics for years. Many regard placebo-controlled trials as the “gold standard” for clinical research. Others see the use of placebos in clinical research as typically-unjustified (at least in cases where a decent treatment exists). (To get a sense of this debate, check out this article by Charles Weijer: “Placebo trials and tribulations”.)
So it has been interesting to see placebos popping up as a topic of discussion lately. Placebos are supposed not to work. That’s the whole point of a placebo-controlled trial — to compare an experimental drug against something “known” not to work, in order to make sure that any positive effects observed aren’t merely the result of the “placebo effect,” i.e., roughly the psychological result of patients feeling cared-for.
But if placebos are (by whatever mechanism) more effective than previously thought, that kind of throws a monkey wrench into the equation.
As a starting point, see this Wired article: Placebos Are Getting More Effective. Drugmakers Are Desperate to Know Why.
For a more scholarly look at a complication, see this PLoS One article by Kaptchuk et al: Placebos without Deception: A Randomized Controlled Trial in Irritable Bowel Syndrome.
And for an overview of the whole issue, see this PLoS blog entry: Meet the Ethical Placebo: A Story that Heals.
Much of the latter blog entry is about clinical (not research) use of placebos. But still, it’s something those of us with an interest in Research Ethics need to think about. As knowledge grows about the size, limits, and exact mechanism behind the placebo effect, we may well need to rethink the role of placebos in randomized controlled trials.
Research Ethics Blog 
I think you have it exactly backwards. If placebos did not work, you would not need placebo controlled trials. IF placebos did nothing, you would not have to use them explicitly in a trial; you could just compare the action of a drug to NOT having that drug. It is precisely because placebos have an effect on people that you need to make sure the drug’s effect is NOT due to the placebo’s power.
The real paradox about placebos is precisely the opposite: clinicians have for years argued that placebos do NOT work against disease, which makes it paradoxical that the scientific establishment would spend billions of dollars designing studies designed to control for the placebo’s effects.
Paul:
Thanks for your comment. I should have been clearer. I meant to say that the traditional assumption (and do correct me if I’m wrong) is that placebos are not supposed to be biologically active. They have of course long been recognized to “work,” in terms of having some more-or-less short-term effect in some applications.
So, I take it that the force of these more recent findings is that the situation is rather messier than previously thought, which makes it less straightforward to include placebos as part of a ‘clean’ trial methodology.
Chris.
I do not think the clarification proffered here helps all that much, since it is false to say that placebos are not biologically active. The example of pain falsifies the statement, since the placebo pathway is a well-recognized process by which neurochemical (read: biological) changes occur that mediate pain experiences.
Although I am nothing more than an amateur on placebo issues in general, one thing that has always struck me is the silliness of using placebo-controls in pain trials, since the evidence I am aware of is overwhelming that placebo effects can be quite significant and lasting for many kinds of pain experiences. Thus, while I can certainly and obviously see the legitimacy of seeing whether a solid tumor shrinks upon application of an intervention compared to placebo administration, the same reasoning applies poorly to pain studies, where I would imagine that pain relief is quite properly deemed the most important outcome.
Given the significant power of placebos for the relief of pain, using it as the standard for efficacy in pain interventions strikes me as odd.