Double blind acupuncture study?

[hgc]

One of my sisters is an MD, with a master's degree in public health, well acquainted with scientific method. She is also a major woo-woo in some regards. We were talking about another family member with a health problem, and she mentioned that she had recommended a reiki healer friend of hers. This set me off into a long tirade of mockery-tinged pointed questioning about the value of such things. I have given her Demon Haunted World to read, but she hasn't started it yet.

So anyway, in our conversation yesterday, I countered her assertion that it doesn't matter that the mechanism for accupuncture, since it works and that's good enough the the argument that it has NEVER been shown to work. By that I meant that it hasn't been shown through methods that seek to remove subjective bias, such as double-blinded studies with control groups. She said you can't double-blind accupuncture, because what are you going to do with the control group? How do control subjects think they're getting accupuncture when they aren't? I suggested sticking the needles in the wrong place in control subjects, but she objected that it may be dangerous. Now, I don't think sticking needles in the "wrong" place is any more dangerous than sticking them in the "right" place, but that's just my assertion.

Does anyone have any idea how you would design a double-blinded study for the effectiveness of accupuncture? This also assumes the effect you are looking for is inherently subjective (pain relief).

 

[aerocontrols]

Does anyone have any idea how you would design a double-blinded study for the effectiveness of accupuncture?

must... resist... urge... to make joke about 'only needing two needles'

Whew... glad I was able to keep that under control

 

[Paul C. Anagnostopoulos]

How about a needle that gives a tiny shock when you touch the skin? That might feel like a needle prick, without actually penetrating the skin. If you couple this with subjects who have never had acupuncture, you might pull it off.

~~ Paul

 

[Javalar]

Easy.

Just ask the "real" acupunturists for spots where sticking a needle does nothing. Use those spots for the control group.

Now everybody is happy!

 

[BobM]

Tricky...

You could design relative effectness studies easily enough though, for specific conditions.

Group A: Nothing
Group B: Acupuncture
Group C: some other treatment

And measure the results of each group. That would at least place acupuncture somewhere.

....

Idea... complicated perhaps but you could design a table with clamps for each needle. Each needle placed into the victim.. er patient would be held in place (just lightly) by a clamp.

You could then design needles that instead of actually being driven into the skin, are just resting against the surface.

I imagine a system could be designed so it would be difficult for a patient to know whether the needles were actually penetrating, or just pressing against them.

 

[Guest]

Do the acupuncture on animals.

 

[hgc]

Whodini:

Do the acupuncture on animals.

And how would you collect the results? I can imagine it now: OK, duck, quack once if you feel better and quack twice if you feel the same.

 

[hgc]

BobM:

You could design relative effectness studies easily enough though, for specific conditions.

Group A: Nothing
Group B: Acupuncture
Group C: some other treatment

And measure the results of each group. That would at least place acupuncture somewhere.

It doesn't do much to account for placebo effect, which I suspect, in acupuncture, is everything.

 

[Guest]

And how would you collect the results? I can imagine it now: OK, duck, quack once if you feel better and quack twice if you feel the same.

I don't know, maybe measure something in the blood, pulse, or something else.

 

[Gebeker]

So anyway, in our conversation yesterday, I countered her assertion that it doesn't matter that the mechanism for accupuncture, since it works and that's good enough the the argument that it has NEVER been shown to work. By that I meant that it hasn't been shown through methods that seek to remove subjective bias, such as double-blinded studies with control groups. She said you can't double-blind accupuncture, because what are you going to do with the control group? How do control subjects think they're getting accupuncture when they aren't?

At least two forms of control have been used in acupuncture studies. One strategy involves trying to "fool" the subject into thinking a needle has been inserted when it hasn't. The most promising scheme for doing this has involved a recent development of a "fake" acupuncture needle. Read about it here Supposedly, the fake needle fools subjects pretty well.

I suggested sticking the needles in the wrong place in control subjects, but she objected that it may be dangerous.

Your friend is mistaken. This is, in fact, probably the most common control used in acupuncture studies. For the record, most studies have failed to find any difference between the "real" acupuncture points and sham acupuncture. This suggests that the analgesic effects of acupuncture are probably due to nonspecific factors, such as intense sensory stimulation.

The approach to blinding the subject can involve things like selecting subjects that know nothing about acupuncture. It is much more difficult to adequately blind the researcher. One approach some studies have used is to employ single blind methods (subject is supposedly blind to condition but the researcher is not), and use questionaires to attempt to determine whether or not subjects' expectations differed between conditions.

If you have access to medical journals, here is an article you might check out that discusses some of the methodological issues related to acupuncture research: Vincent and Richardson (1986). Pain The evaluation of therapeutic acupuncture: concepts and methods. Vol 24: pp 1-13.

There is a large body of acupuncture research. As is usually the case with alternative medicine research, the overall quality of the studies is poor. Most studies use either don't use controls at all, or use controls that are inadequate to assess efficacy. A great deal of nonsense has been written about acupuncture, even by researchers that really should know better. For example, the NIH in their 1997 Consensus Development report on acupuncture concluded that there was convincing evidence that acupuncture is effective against nausea. This conclusion was based almost entirely on studies conducted by a single laboratory. (And some of those studies had designs that were inadequate to assess efficacy.) The positive results seem to come almost exclusively from that one laboratory, while some other groups found no effect. So, while is enough evidence there to warrant further research, it is (in my opinion) absurd to say that "there is clear evidence that needle acupuncture is efficacious" for nausea.

David Mayer did a detailed analysis of the literature and found that the NIH report was repeatedly misleading and overly optimistic about acupuncture. See Mayer, 2000. Acupuncture: An evidence based review of the clinical literature. Annu. Rev. Med. Vol 51: pp 49-63.

Finally, I disagree with your friend when she says that it doesn't matter why acupuncture works. It matters a great deal. If, as may be the case, acupucture produces pain relief through nonspecific mechanisms (like intense sensory stimulation), then it is completely pointless to train practitioners in "proper" acupuncture methodology, and it is even more pointless to send anyone to an acupuncturist for treatment.

Furthermore, even if acupuncture has specific effects, there is still the issue that acupuncturists are not qualified to make proper medical diagnoses, so no one should go to an acupuncturist instead of a medical doctor.

I wouldn't be all that surprised if it were shown that acupuncture is effective against some forms of pain. However, for now, my opinion is that the available data are insufficient to recommend acupuncture for treatment of any disorder.

 

[rwald]

While I completely agree with you about using fake points for a control study, there's still one issue that needs to be mentioned. In theory, it's possible that the "official" acupuncture points actually are better than a random series of points. After all, after doing this stuff for however-many thousand years, it's possible that the acupuncturists have points that work. That isn't to say that they work for the reasons the acupuncture people think they do; the whole "meridian" theory is clearly bogus. I'm just saying that it's possible that it might work for some other reason.

For the record, I think the above scenario is very unlikely, and based on the studies referenced above, almost certainly untrue. But that's why some real, double-blind studies need to be done. And the way to apply the second blind is as follows: The researcher who asks the questions about effectiveness isn't the one who actually applied the acupuncture; after the subject is treated, he is taken into a separate room, where a researcher, who doesn't know which group the subject is in, asks about the procedure.

 

[RichardR]

And the way to apply the second blind is as follows: The researcher who asks the questions about effectiveness isn't the one who actually applied the acupuncture; after the subject is treated, he is taken into a separate room, where a researcher, who doesn't know which group the subject is in, asks about the procedure.

This still would not be a true double-blind study, since the person inserting the needles would know if he is putting them in the right place or not. It would therefore be possible that sensory clues are being passed to the patient.

 

[shemp]

Re: Re: Double blind acupuncture study?

must... resist... urge... to make joke about 'only needing two needles'

Whew... glad I was able to keep that under control

You also missed the opportunity to say "Jebus Krist! I'm not letting some blind guy stick needles in me!"

 

[Plutarck]

This still would not be a true double-blind study, since the person inserting the needles would know if he is putting them in the right place or not. It would therefore be possible that sensory clues are being passed to the patient.

I have some ideas then.

First, you make a paddle with fake needles in it, and you place it appropriately, for instance, against someone's foot.

The needles are placed in/on the paddle in various places, some of which are the "right" ones, and some of which aren't. A computer decides which needles are used and which are not, and records the data accordingly.


In another similar idea, doctors who know little/nothing about acupuncture are instructed in the proper uses of needles and so trained in using them - however, they don't know where they are supposed to stick them.

They are given various needle placements randomly, and they insert the needles as their non-human instructor says.

That way neither the patient nor the doctor knows what is going on.

The results are rather certain to be that it doesn't matter where you stick the needle, but there is some apparent psychological/placebo effect. Pain is notorious for being caused or cured by thought, rather than by some literal physical occurance outside someone's own brain.


That would sufficiently debunk the idea that acupuncturists know something special, and thus show that their theories about how it work are clearly wrong, or they just somehow managed to develop the theories and supposedly know so much about them, yet they do absolutely nothing special. They might as well just stick needles in people nearly randomly.

 

[heath]

Some good suggestions here, but isn't the simplest way to double blind the study to have 2 groups of accupucture naive doctors and
- give one a list of "correct" locations to insert needles
- give the other group a list "incorrect" locations to stick needles.

Full accupuncture training not required. Just for some specific ailement that accupuncture claims to cure.

They will not know which group they are in, even if they compare notes (they could read up but you need to isolate them or get some agreement they would not - messy but better than nothing). You could even split the doctors into sub groups of (pre test) believers and non-believers within each group. Same for the patients. Giving all 8 possible combinations of believing/not doctor, believing/not patient and correct/incorrect location.

Simple, effective double blinding.

You'd need a failry large set size to get any significance our of it with 3 variables but the data would be pretty comprehensive (assuming honesty of the participants)

 

[Akots]

Why not pick ten people with a disease A, and ten more with a completely unrelated disease B, and treat them all with the actupuncture cure for disease A, all the while making sure the acupuncturists and patience are unaware of who's getting what?

TELL them it' srandom, but make it a 50/50; so that exactly 5 people in each group get the correct treatment.

 

[hgc]

Akots:

Why not pick ten people with a disease A, and ten more with a completely unrelated disease B, and treat them all with the actupuncture cure for disease A, all the while making sure the acupuncturists and patience are unaware of who's getting what?

TELL them it' srandom, but make it a 50/50; so that exactly 5 people in each group get the correct treatment.

That's good. Assuming there are defined treatment variations, this is a very doable experiment.

I also believe the experiment procedures around putting needles in the "wrong" place for the control group is fine too, but easily objected to by those who think putting needles in the wrong spot is a bad thing.

 

[rwald]

As far as putting needles in the wrong spot being a "bad thing," we obviously don't have any proof of that. So if they want us to believe them, then we'll need proof...like a double-blind study...

 

[hgc]

rwald:

As far as putting needles in the wrong spot being a "bad thing," we obviously don't have any proof of that. So if they want us to believe them, then we'll need proof...like a double-blind study...

I don't think it's that simple. If you were willing to design an experiment to test the validity of acupuntures claims, then you are implying that the claim is at least possible. You then automatically grant a possibility to the companion claim that sticking needles in the wrong place is harmful (if the claimant makes that claim) and you can't design a test that will knowingly harm your control subjects.

 

[heath]

As far as putting needles in the wrong spot being a "bad thing," we obviously don't have any proof of that. So if they want us to believe them, then we'll need proof...like a double-blind study...

This would be a very easy and irrefutable proof of accupuncture, if putting needles in the wrong place made people unwell.

I'm sure there are enough believers willing to risk it. Same for non believers of course