Mind Over Body

Mr. Wright (not his real name, but that's what he's referred to in the literature), had cancer and was on the edge of death. His doctor figured he only had days to live. Mr. Wright was given a serum that he believed might cure his cancer. Sure enough, in just a matter of days his tumors disappeared. He was released from the hospital, happy and cured. A couple months later he read a report that the treatment he had received was no good. His tumors promptly grew back. Then his doctor convinced him that a new formulation of the serum was shown to be effective. He took it (this new formulation was just water) and once again he was cured. Another couple of months went by and he read more reports about that serum being a scam and completely useless. He died two days later. Thus is the power of our beliefs over our physical nature. We call this super power the Placebo Effect.

We all have this super power, which means medical research faces a particular challenge. Most scientific research subjects such as geology, astronomy, or paleontology do not have a vested interest in the research. A rock or a fossil does not care if you think it's a thousand years old or a billion years old. But a human has ideas about the research being done on him. And those ideas can change the very research being done.

If I give you a pill and tell you it will make the pain go away, there's a chance that your pain will go away even if the pill I gave you is nothing but a sugar pill. Presumably, sugar itself has no effect on your pain. So what's going on?

It's kind of like Neo in the Matrix: 

your mind makes it real. 

Here's a nice video explaining some
of the strange effects of placebos:

Because of the placebo effect, research on human subjects needs to control for it. What this means, in practical terms, is that people are divided up into groups. One group, for instance, will get the placebo and the other group will get the real medicine that is being tested. It's like a contest. If the medicine helps more people then the medicine wins, but if the percentage of people who took the placebo is about the same as the percentage of people who took the medicine, then that means that, as far as we can tell, the medicine actually didn't do any good.

No Better Than a Placebo

To be "no better than a placebo" is the ultimate failure for any kind of medicine, procedure, or other medical treatment.

Homoeopathy vs Placebo - as you might expect, the placebo wins.

Tylenol vs Placebo for Lower Back Pain - Placebos also win in contests where you might not expect them to win. For instance, acetaminophen can't beat a placebo for lower back pain.

You can do your own web search to find other placebo wins. It's not my goal to list them all. Whatever your favorite medical treatment for whatever ails you, it might not be a bad idea to search for placebo controlled studies to find out how effective it is.

Double-Blind

No, double-blind does not mean that the scientist conducting the study is blind in both eyes. It really means that neither the person taking the medicine/placebo nor the person dishing out the pills knows who is getting the placebo and who is getting the test drug. This is because we've found that if the doctor knows which pill they are giving to the patient they can give subtle hints that can adversely effect the study.

Even a horse is smart enough to read the subtle non-verbal cues. There's a famous case of the horse who could do math. His name was Clever Hans. People would give the horse simple math problems like 6-3 or 4+1 and the horse would tap it's foot up to the right number and then stop. What they found was that the owner of the horse would unknowingly give the horse a signal showing the horse that it was at the right number. When the owner was hidden behind a curtain the horse's math skills disappeared. Here's a picture of Clever Hans putting on one of his shows:

XKCD is right. Not every experiment can be double-blinded.

One of my favorite blogs, Science Based Medicine, has an article on Acupuncture that perfectly illustrates why blinding is important whenever it can be done. In this experiment someone who was not involved and had no knowledge about which children were getting the acupuncture should have done the evaluations. When the experimenter evaluates her own work she's not going to be the most objective observer. Blinding is necessary in order to counter confirmation bias, which we've discussed here.

So one of the important factors you need to consider when reading or hearing about some so-called scientific study is whether it was conducted properly. Doing good empirical studies on human beings is hard. So if it's possible, the study should include a placebo control group and, if possible, blinding both the patient and the doctor so that no one really knows who is in the study group and who is in the placebo group.

Actual medical research is more complicated that this. I would just like to take a second to point out a couple mitigating factors. For example, if there is a known effective (and by effective, I mean more effective than a placebo) treatment for some condition, then it would not be a good practice to withhold that treatment from people. So new medical treatments are sometimes put up against the known effective treatment in order to see if the new treatment is at least as effective, or more effective, than the known effective treatment. This means that not every piece of research will directly control for placebos. But if the known effective treatment was previously studied using a placebo control group then the new study can safely bypass a placebo control.

The other problem is that some things we want to know about are not testable in a laboratory very easily. Sometimes we want to know about lifelong habits and although this kind of study can be done on mice that only live two years, it's hard to do it on humans who are more likely to live seventy years. We'll deal with this problem in another post.

End Notes:
You can read more about Mr. Wright in The New York Times, Scientific American, or the original paper on Mr. Wright is here: Klopfer, Bruno, "Psychological Variables in Human Cancer", Journal of Projective Techniques, Vol.21, No.4, (December 1957), pp. 331–340.