Acupuncture works (as a strongly-proven placebo)
Several people have mentioned acupuncture as an example of a solid and reputable alternative medical technique which works. Sometimes I even hear that clinical trials have proven that acupuncture works. This is technically true but misleading. I'd like to clarify this once so that I can refer to my written record later.
It's important to know how well acupuncture works, what is required for it to work (if it does), and whether it can be made to work even better (once we know what is required). Since the two major components of acupuncture are the points, meridians, and channels of the body and the actual needles that get inserted into them, the best double blind studies test those variables. Some experimental groups use real needles inside opaque sheaths and others use sheaths with needles that do not pierce the skin. Some groups stick patients in the correct locations and some groups stick patients in random locations. Then the results are analyzed to determine efficacy.
The results are unambiguous: acupuncture works when you use needles, and it works just as well when you don't. It works when you insert needles into the right places, and it works just as well when you pretend to insert needles into the wrong places. In short, acupuncture works whether you do it right, wrong, or not at all. It is indistinguishable from placebo. "Both groups improved statistically from baseline, and acupuncture and placebo had similar credibility." There is "no statistically significant difference between groups". These trials have been replicated by different labs and the concensus is that "the great bulk of the randomised controlled trials to date do not provide convincing evidence of pain relief over placebo."
Experiments have also revealed some interesting anomalies. For example "placebo acupuncture works better than placebo western medication". Some work indicates that different parts of the brain light up between placebo and real acupuncture, which demonstrates that your brain knows the difference between pretending to get poked and actually getting poked even if you don't. Believing that you are getting pricked is important even though actually getting pricked isn't, which validates the claim that "It is more likely to be effective if you believe it". And ikkyu2 pointed me to a study showing that you can make acupuncture stop working with naloxone which has also been shown to mediate other placebo reactions as well as legitimate drug reactions.
It's undeniable that acupuncture and other placeboes make patients feel better, but the real question in my mind is whether it's ethical to charge someone $80-$120 per session for the equivalent of Dumbo's feather. Whether it's ethical to charge a toll on the Wile E. Coyote invisible bridge as long as you tell them not to look down.
As an aside, it frustrates me when some people say that "western science has a hard time modeling alternative medicine". It smacks of cherrypicking when negative test results are dismissed as irrelevant while tests indicating even the slightest hint of a positive result are widely touted. If you're skeptical a technique can't be falsified by "western science" then you need to be skeptical about ostensible verifications as well. And that's why I give special kudos to two alternative medical institutions, the Japan School of Acupuncture, Moxibustion & Shiatsu and the Physiotherapy Institute for Oriental Medicine Research Foundation, who helped develop a double-blind placebo needle for acupuncture research. Scientists don't know as much about these techniques as their practitioners do, and practitioners are always welcome to create their own tools for testing or suggest tests by which their claims can be demonstrated and accepted. It speaks very highly of a technique when its practitioners are willing to meet scientists halfway.
Update: Hey, check out the new data. The studies with the best results are the ones where the patients know when to report the best results. Stop letting them know what to expect and the effect evaporates.
It's important to know how well acupuncture works, what is required for it to work (if it does), and whether it can be made to work even better (once we know what is required). Since the two major components of acupuncture are the points, meridians, and channels of the body and the actual needles that get inserted into them, the best double blind studies test those variables. Some experimental groups use real needles inside opaque sheaths and others use sheaths with needles that do not pierce the skin. Some groups stick patients in the correct locations and some groups stick patients in random locations. Then the results are analyzed to determine efficacy.
The results are unambiguous: acupuncture works when you use needles, and it works just as well when you don't. It works when you insert needles into the right places, and it works just as well when you pretend to insert needles into the wrong places. In short, acupuncture works whether you do it right, wrong, or not at all. It is indistinguishable from placebo. "Both groups improved statistically from baseline, and acupuncture and placebo had similar credibility." There is "no statistically significant difference between groups". These trials have been replicated by different labs and the concensus is that "the great bulk of the randomised controlled trials to date do not provide convincing evidence of pain relief over placebo."
Experiments have also revealed some interesting anomalies. For example "placebo acupuncture works better than placebo western medication". Some work indicates that different parts of the brain light up between placebo and real acupuncture, which demonstrates that your brain knows the difference between pretending to get poked and actually getting poked even if you don't. Believing that you are getting pricked is important even though actually getting pricked isn't, which validates the claim that "It is more likely to be effective if you believe it". And ikkyu2 pointed me to a study showing that you can make acupuncture stop working with naloxone which has also been shown to mediate other placebo reactions as well as legitimate drug reactions.
It's undeniable that acupuncture and other placeboes make patients feel better, but the real question in my mind is whether it's ethical to charge someone $80-$120 per session for the equivalent of Dumbo's feather. Whether it's ethical to charge a toll on the Wile E. Coyote invisible bridge as long as you tell them not to look down.
As an aside, it frustrates me when some people say that "western science has a hard time modeling alternative medicine". It smacks of cherrypicking when negative test results are dismissed as irrelevant while tests indicating even the slightest hint of a positive result are widely touted. If you're skeptical a technique can't be falsified by "western science" then you need to be skeptical about ostensible verifications as well. And that's why I give special kudos to two alternative medical institutions, the Japan School of Acupuncture, Moxibustion & Shiatsu and the Physiotherapy Institute for Oriental Medicine Research Foundation, who helped develop a double-blind placebo needle for acupuncture research. Scientists don't know as much about these techniques as their practitioners do, and practitioners are always welcome to create their own tools for testing or suggest tests by which their claims can be demonstrated and accepted. It speaks very highly of a technique when its practitioners are willing to meet scientists halfway.
Update: Hey, check out the new data. The studies with the best results are the ones where the patients know when to report the best results. Stop letting them know what to expect and the effect evaporates.