Monday, December 27, 2010

Placebo: Five Points

Most discussions of placebo exasperate me. It's one of those things, like childrearing, that everyone thinks they're an expert about, especially people who know very little about it. People make lordly pronouncements and expect to be taken seriously, and if you probe a little bit you'll find they have not read any of the basic texts on the subject, and they're familiar with neither the methodological problems nor with the research. They are, in fact, pig-ignorant, and they're just saying whatever occurs to them on the spur of moment. Remarkably enough, people who dislike the idea of placebo effects tend to downplay them and pour contempt on research about it, and people who like the idea tend to play it up and find every study full of promise. Double standards abound.

So let's start by lining up some of the things that everyone ought to know before they even start talking:

1. No. You are never going to get a double-blind study of placebo effects. If that's not immediately obvious to you, go home and stop bothering me.

2. Of course regression to the mean accounts for some of it.

3. We're not talking about cures for cancer, here. We're talking about mild but extremely interesting and puzzling benefits, across a wide spectrum of conditions.

4. There is no conspiracy against placebo research. Drug companies are anxious to understand placebo effects because they'd like to be able to get what they regard as noise out of their studies, and they're embarrassed by how often their concoctions don't outplay sugar pills. But they're not worried about placebo research putting them out of business.

5. Repeated attempts to establish the credulity of the patient as the variable that correlates most closely with the effectiveness of placebo have failed. The most closely correlated variable is the confidence of the dispenser in the efficacy of the treatment.

No comments: