Network News

X My Profile
View More Activity

The Ethics of Placebos


An analysis of over 4000 wines entered in 13 U.S. wine competitions shows little concordance among the venues in awarding Gold medals. ... An analysis of the number of Gold medals received in multiple competitions indicates that the probability of winning a Gold medal at one competition is stochastically independent of the probability of receiving a Gold at another competition …

For the 375 wines entered in five competitions, one would expect by chance alone (for p = 0.09), 234 wines receiving no Golds, 116 receiving a Gold in just one competition, 23 receiving Golds in two competitions, two receiving Golds in three competitions and no wine receiving Golds in more than three competitions. The observed frequencies closely mirror these numbers.

"The more I look at empirical studies such as this one," writes Felix Salmon, "the more I’m convinced that if you’re tasting blind, there’s no correlation between perceived quality and just about anything." Which isn't to say that people don't really prefer the high-quality wines when they order them: The placebo effect is real, and it's powerful. If you think a wine is tastier due to its prices, or a pill is a powerful antidote to your ailment, your experience is likely to match your expectations.

The ethics of this have always interested me. Take a gander at the graph atop this post. It's from a CBO paper that contains many more graphs like it, and it compares the relief patients felt after a surgery meant to relieve angina pain and a skin incision conducted while the patient was under anesthesia. Relief was effectively the same. The medical community responded by abandoning "internal mammary artery ligation," though you could argue that they should have responded by increasing the use of skin incisions. The pain relief, after all, was real, and significant, even if the surgery was fake.

There are no end of situations that might lend themselves to placebo treatments: patients demanding an antibiotic when they have the flu, or painkillers when they have nothing but unspecific aches. The placebo could do the patient some good and avoid them some potential bad (side-effects, antibiotic resistance, etc). This is, of course, unethical, not to mention an invitation to some serious lawsuits, not to mention ineffective once word gets out. But the evidence in favor of the placebo effect is really quite tremendous: It's hard not to wonder if there's not some way to marshal that cheap, safe power.

By Ezra Klein  |  September 1, 2009; 5:03 PM ET
Categories:  Health Reform  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: The Great Centralizer?
Next: Opposing Afghanistan With Emphasis


Have you seen this study? It's all about perceived value yo...

Posted by: ThomasEN | September 1, 2009 5:47 PM | Report abuse

People often misunderstand graphs like the one from CBO reproduced here. It doesn't actually measure the effect of a placebo. Indeed, without another control, there is no way to tell what the true causal effect is of skin incisions. Rather, the placebo is simply showing what would have happened in the absence of treatment. Perhaps because people are likely to report getting better without any intervention.

So it's incorrect to posit skin incisions would have made people better. To do that, you'd need to devise a different experiment.

Posted by: ArininSF | September 1, 2009 5:49 PM | Report abuse

Well done, ArininSF. Ezra, I think you got this completely wrong. The most likely explanation is that people tend to improve some small amount after a heart attack, "placebo" effect or no placebo effect. You would have to compare it to no skin incision, no surgery, no nothing to determine the power of the placebo effect.

And we do make use of the placebo effect. There's entire industries (alternative medicine or health/beauty products) that for the most part are built on placebo effects. Furthermore, a lot of primary care is essentially placebo; you walk in with a flu, the doctor goes "there there it'll be all right," you go home feeling a little bit better and a week later its all gone.

Posted by: CarlosXL | September 1, 2009 6:13 PM | Report abuse

Unrelated to the placebo effect, but somewhat related to the politics of health care reform.

There was once a study in which the researchers went to two factories and asked the employees what could be done to improve productivity through lighting. In one factory, the suggestions were implemented, in the other, they were not. Productivity improved in both factories. The researchers ultimately concluded that it was the asking, not the lighting, that improved productivity.

Posted by: rmgregory | September 1, 2009 6:13 PM | Report abuse

Sorry, I see that this has to do with angina pain, not pain after a heart attack. The point, however, remains the same: this is more likely the natural history of the disease or patient reporting patterns than "placebo effect," although one could do a study to test this, exactly as ArininSF said. I think studying the placebo effect is extremley worthwhile.

Posted by: CarlosXL | September 1, 2009 6:15 PM | Report abuse

Here's a novel idea: Let's perform the operation on the people who pay for their own insurance, and everybody in the public option gets the incision.

Posted by: whoisjohngaltcom | September 1, 2009 6:32 PM | Report abuse

The truth is that relatively few treatments have been shown to be effective by high quality randomized placebo controlled studies. In many of those studies the effect is small or short-lived. The placebo effect is the rule not the exception.

Posted by: bmull | September 1, 2009 6:43 PM | Report abuse

Here you go Ezra...

Posted by: wisewon | September 1, 2009 6:50 PM | Report abuse

Short version of this post: experience is subjective.

Not sure why this surprises anyone. Wines do not exist in a vacuum.

Posted by: roquelaure_79 | September 1, 2009 7:14 PM | Report abuse

"Stochastically independent" has no meaning. Translated into ordinary words, it means "randomly independent" which makes no sense. Sometimes independent, sometimes not? Unpredictably independent?

I didn't read any further than that. Anyone who uses Big Words as decoration, without regard to meaning, has nothing to say that is worth listening to.

Posted by: pj_camp | September 1, 2009 7:25 PM | Report abuse

Ezra still hasn't learned his lesson, when he's reposting graphs that were fairly shouted down at his old blog. That surgery versus incision bit is from 1959.

If you want to read the original comments beatdown, it's here:

Posted by: Klug | September 1, 2009 8:52 PM | Report abuse

Excellent point. To answer your question, faith healing, and why not?

It is perfectly ethical, and works as long as folks believe in it.

Posted by: CorkExaminer | September 1, 2009 8:55 PM | Report abuse

Ezra, you say placebos would be "ineffective once word gets out" but I'm not so sure. In part that's because I think people are widely different in their response to placebo -- just as they are different in their response to any particular drug. Some people are really susceptible to placebos and I dare say they would still have a placebo effect even if they knew it was a placebo. I can't recall if a pscyh friend told this to me or if it's just something I am theorizing myself, but there is/may be a connection between how easily someone can be hypnotized and how susceptible they are to placebos. My own personal annecdote: even after having read studies that showed anti-depressent drugs to be no better than placebos, I did give the drugs a try when I was in a bad way. As I talked to the doctor to get the prescriptions, we talked about the study that I'd read; I had "this is no better than a sugar pill" running through my head as I took it. And lo and behold, it did seem very much to help me out. And I don't just mean the way some other commentors have suggested -- the way some things get better with time and no other treatment whatsoever. I mean, an absolute change in my feeling and perception that came on pretty rapidly and was different from any feeling I've ever had before or since going off the drug.

Final thought: don't forget that the placebo effect goes both ways, both with good effects and bad, like placebo side effects. That's additional reason to be wary of the ethics of giving placebos. But worse than the ethics of advising people to vote? We know that statistically it is an absolute waste of anyone's time to actually go to the polls. Different from giving them a placebo?

Posted by: JonathanTE | September 1, 2009 10:53 PM | Report abuse

But the evidence in favor of the placebo effect is really quite tremendous: It's hard not to wonder if there's not some way to marshal that cheap, safe power.


Posted by: Senescent | September 2, 2009 4:24 PM | Report abuse

Aw, dang, does wapo not do italics? How about blockquotes.

test test test

Posted by: Senescent | September 2, 2009 4:25 PM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2010 The Washington Post Company