Acupuncture: The better the study, the worse the result

My column on SEEDMAGAZINE.COM today discusses recent research on acupuncture:

The science behind acupuncture is dubious. It's difficult to properly control an acupuncture study because its practitioners--and those receiving treatment--are heavily invested in the results. In a Norwegian study of acupuncture as treatment for hot flashes during menopause, 80 out of 535 volunteers dropped out because they were randomly assigned to the "no treatment" group. As Euan Lawson, a general practitioner in Cumbria, UK, explained in his analysis of the research, acupuncture is quite popular in Norway, with nearly a third of the population having received the treatment at some point in their lifetimes. With this level of popular acceptance of acupuncture, it's no wonder that a small apparent benefit was found: The women who received acupuncture reported experiencing slightly fewer hot flashes than those who remained in the randomly assigned "self-treatment" group. This result is easily explained as a placebo effect: The women and their practitioners both want the treatment to work and believe it will, so therefore it does, albeit only very slightly.

A better way to do acupuncture research is to use a sham acupuncture control group that leads both patients and practitioners to believe they are getting real acupuncture. Accordingly, researchers have designed "needles" that appear even to practitioners to be real but don't puncture the skin. This is the same way careful drug trials are done: Patients and physicians don't know whether they receive the experimental drug, an alternative drug, or a placebo.

You can read the rest here.

Also, here are my picks for neuroscience and psychology for If your post is featured, make sure you click through: there's a new icon you can display to show off the fact that your post was selected!

  • Steve Genco offers a compelling critique of fMRI and the many ways it can be abused.
  • On the other hand, Kevin Mitchell discusses some amazing new technology that allows us not only to see brain activity, but also how different parts of the brain interact.
  • On the other other hand, dlPFC discusses some research that effectively uses regular old fMRI to show why we have so much trouble multitasking, and even suggests how we might be able to learn to multitask better.
  • Finally, a completely unrelated -- but fantastic -- post about the amazing nose of the star-nosed mole.

More like this

This is an interesting article, and states a problem in acupuncture research, but comes to the wrong conclusions. Acupuncture is not a drug and, like surgery and psychology, cannot be effectively tested as a drug. "The better the study, the worse the result" is true, as far as it goes, but the so-called better studies all rely on things like standardized protocols for western diagnoses. Until patient groups are controlled for general TCM diagnostic groups, these studies will continue to give poor results. A 5'4" 200lbs. depressed woman with migraine headaches, for example, will not receive the same acupuncture points as a 6'2" 180lbs. work-stressed man. Whether sham acupuncture is used or not isn't the point. The basic design of these studies is flawed from the beginning. Acupuncture can be tested, and often is - especially outside the U.S. - but adapting models of study created for pharmaceuticals will not give useful results.

(And yes, I am an acupuncturist)

@ #1: This is why randomization is important. In a randomized trial, on average, those in different conditions will not be systematically different, even if any single individual in one condition is different from a single individual in another.

a trial design suggestion, just for poster #1: each patient is evaluated, and a real accupunturist determines the best possible treatment points (but doesn't tell the patient). The patient is then randomized to one of two groups. One group gets the treatment decided by the "expert" and the other group gets accupuncture, at randomly selected treatment points (that don't match those selected by the expert). The treatment points assigned to the patient are communicated to a treating accupuncturist, without telling him/her which group the patient belongs to.

Whether sham accupuncture is used is exactly the point (contrary to what #1 writes). The witchdoctor gets results by putting on the mask and doing the dance. If the patient is aware that he/she is assigned to get "real" accupuncture versus the control group, the placebo effect will kick in.

David: I can't believe that hasn't been tried yet.

#1: You are suggesting that acupuncture works for some people/treatments and not for others. I've never seen any quantified data that purports this, and don't think it exists. It is as valid to blanket treat a symptom with acupuncture (as the experiment does), varying the treatment protocol based on the subject, to the satisfaction of the 'expert' as it would be with any other experiment. You vary dosage of a medicine based on the weight/condition of a subject. I fear, you are grasping at straws.

By Walter Sear (not verified) on 05 Sep 2009 #permalink