New acupuncture trial

[darkwinter]

http://www.sciencedaily.com/releases/2008/05/080531185839.htm


New data from a randomized, controlled trial found that acupuncture provided significant reductions in pain, dysfunction, and dry mouth in head and neck cancer patients after neck dissection.

Found this in my RSS feed from ScienceDaily this morning. Here's a salient passage:



Seventy patients participated in the study and were randomized to receive either acupuncture or usual care, which includes recommendations of physical therapy exercises and the use of anti-inflammatory drugs. For all of the patients, at least three months had elapsed since their surgery and radiation treatments. The treatment group received four sessions of acupuncture over the course of approximately four weeks. Both groups were evaluated using the Constant-Murley scale, a composite measure of pain, function, and activities of daily living.


Pain and mobility improved in 39 percent of the patients receiving acupuncture, compared to a 7 percent improvement in the group that received usual care. An added benefit of acupuncture was significant reduction of reported xerostomia, or extreme dry mouth. This distressing problem, common among cancer patients following radiotherapy in the head and neck, is addressed with only limited success by mainstream means.Now, I'm not terribly well-read in the acupuncture literature, so I'm quite willing to be corrected here; but I saw no evidence in the article that this study had used anything that would account for the "placebo" effect of the treatment. It's my understanding that any trial of acupuncture - and indeed any other "SCAM" (Supplementary, Complementary and Alternative Medicine - thanks to Mark Crislip for the wonderful acronym) should include placebo control if it's going to comment on the efficacy of the approach.

What do you think?

[Mongrel]

Without seeing the paper, it's bunk.

Not a large enough trial and there should have been a sham acupuncture group as well (I think there was a section in Trick or Treatment that had a check list of what makes a rigorous trial).

Oh, probably worth keeping an eye on Respectful Insolence (http://scienceblogs.com/insolence/) this week - I'm sure Orac will do a thorough dissection of it ;)

[darkwinter]

That was my first reaction too - though ScienceDaily doesn't normally spit out bunk in my experience, which is why I was surprised to see it in there. Let's hope credulous science reporting isn't actually spreading to the dedicated scientific news outlets.

Thanks for the tip, too - I've been meaning to add Respectful Insolence to my RSS roll for a while now and kept putting it off.

[Pebble]

You would really need to see the full published trial here to draw any conclusions. As already pointed out sham acupunture is a critical requirement in any trial as the impact of researcher and patient bias is huge. Many non blinded trials have suggested benefit in cancer patients, but when blinded the effect largely disappears.

Seehttp://www.ncbi.nlm.nih.gov/pubmed/18065731?ordinalpos=16&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsP anel.Pubmed_RVDocSum

Also; the Constant-Murley scale is not reproducible, with substantial interobserver variability: so how this was performed, whether nomralised for age and sex and whether a normal shoulder was available for comparison all influence the credibility of these observations:

See: http://www.jbjs.org.uk/cgi/reprint/78-B/2/229.pdf

[Graham Lappin]

Knowing a bit about clinical trials I would make two observations, that would equally apply to any trial, be it acupuncture or a conventional drug.

1) You have to see the data, usually as a peer reviewed published paper. The devil is often in the detail.

2) You have to view any clinical trial in context of others asking the same questions.

I am saying nothing new here but I thought the point needed to be made.

There is one significant difference between clinical trials with acupuncture and similar CAMS and conventional medicines that I think does get forgotten. Conventional medicine is mechanistic. For example, the drug concentration is measured in the body over time (pharmacokinetics) in relation to pharmacologic effects (pharmacodynamics). Clinical trials are often accompanied with surrogate measurements - the examination of effects mechanistically associated with the clinical effect, such as an enzyme activity. (For the record, surrogates can be problematic). Acupuncture trials only measure a targeted symptomatic output without having an array of of related evidence.

CAMs tend to be isolated whilst real science has to be joined up.

[Mongrel]

Well some indicators can only be measured subjectively, even in EBM trials, pain and nausea etc.

The main sticking point from the article for me is that it compared the normal post chemo therapy, which they'd all had experience of, and a new more 'interactive' therapy. There are reasonably effective sham acupuncture techniques so why didn't they use them in a three armed trial (normal, sham, acupuncture).

In normal clinical trials the only reason to not use a placebo treatment is when it would be unethical to withhold a known, effective treatment with potentially terminal conditions.

[JJM]

{snip} The main sticking point from the article for me is that it compared the normal post chemo therapy, which they'd all had experience of, and a new more 'interactive' therapy. There are reasonably effective sham acupuncture techniques so why didn't they use them in a three armed trial (normal, sham, acupuncture). {snip}There is a thread on the JREF forum that discusses the idea of a three-armed trial (The forum is down and will not be back till tomorow, at the earliest.). A third arm makes no sense; especially for such a small study. A standard treatment arm lessens the power of the trial to detect a difference due to acupuncture.

The proper question to study is how acupuncture compares to a sham procedure (many are available). Since acupuncture (real or sham) typically shows a strong placebo effect, this study is fatally flawed. The authors probably chose the design because they knew a proper study would show acupuncture to be merely placebo. If this study is submitted for publication, it will not be accepted by a good journal.

[Mongrel]

There is a thread on the JREF forum that discusses the idea of a three-armed trial (The forum is down and will not be back till tomorow, at the earliest.). A third arm makes no sense; especially for such a small study. A standard treatment arm lessens the power of the trial to detect a difference due to acupuncture.

Fair do :)

[Cuddles]

The proper question to study is how acupuncture compares to a sham procedure (many are available). Since acupuncture (real or sham) typically shows a strong placebo effect, this study is fatally flawed. The authors probably chose the design because they knew a proper study would show acupuncture to be merely placebo. If this study is submitted for publication, it will not be accepted by a good journal.

The main problem with sham acupuncture is that there doesn't appear to be any decent placebo. I've looked into this before, but unfortunately my posts are lost on the JREF forum somewhere. Basically, sham acupuncture has two main problems - it doesn't fool everyone and it isn't actually inactive. I can't remember the exact figures, but I believe in the studies I looked at it was around 20% of people who realised they were having fake acupuncture. A similar number experienced a pain reaction, meaning the fake needle induced a physical response, and so was not acutally a placeno.

The conclusions of the trials almost all say that sham acupuncture is good enough to be used as a placebo in a trial, but looking at the actual data I'm not at all convinced.

However, the placebo effect is the least of the problems of this latest study. The most important reason for using a placebo control is not the placebo effect at all, it is blinding. If you don't give a placebo, then both the doctor and the patient know which group they are in, and there cannnot be any blinding at all. This makes the whole trial bascially worthless.

[Graham Lappin]

In normal clinical trials the only reason to not use a placebo treatment is when it would be unethical to withhold a known, effective treatment with potentially terminal conditions.

Actually that's not entirely accurate. It's generally true in phase III and some phase II trials but placebo is rarely used in phase I and never in phase 0.

This is partly the point I was making about getting other indicators other than clinical affect. If for example the primary question is one of pharmacokinetics (on a phase I study) then the output is drug concentration data and therefore a placebo is pointless.

I don't want to appear to be too pedantic about this and I take your point about comparing effective clinical outcome with placebo control (or comparator). Nevertheless, I think there is sometimes a bit of an over-assumption about the use of placebo and there are occasions where it is not possible or applicable on a conventional study. Mongrel's point about ethics is one such case but there are others as well.

[Pebble]

Actually that's not entirely accurate. It's generally true in phase III and some phase II trials but placebo is rarely used in phase I and never in phase 0.

This is partly the point I was making about getting other indicators other than clinical affect. If for example the primary question is one of pharmacokinetics (on a phase I study) then the output is drug concentration data and therefore a placebo is pointless.

I don't want to appear to be too pedantic about this and I take your point about comparing effective clinical outcome with placebo control (or comparator). Nevertheless, I think there is sometimes a bit of an over-assumption about the use of placebo and there are occasions where it is not possible or applicable on a conventional study. Mongrel's point about ethics is one such case but there are others as well.

I think what everyone else here is referring to are clinical trials, hence the emphasis on placebo. The mechanistic data sometimes advances in parallel in conventional medicine (as opposed to leading) - thus while PDE5 inhibitors were produced as a consequence of basic science work in the penis, most of the mechanistic work on PDE5 expression and actions in the lung has been undertaken because of the recognition that PDE5 inhibitors were successful in treating pulmonary hypertension rather than the science leading directly to the clinical trials.

If acupuncture were successful in a decent designed clinical trial, the basic science would then become possible. But at present no one knows where to begin looking.

While we pride ourselves on setting up trials where the mechanism is plausible, one only has to go back 50 years to where much of what we presently take for granted, progressed by serendipity rather than cold logic.

[Graham Lappin]

I think what everyone else here is referring to are clinical trials, hence the emphasis on placebo. The mechanistic data sometimes advances in parallel in conventional medicine (as opposed to leading) - thus while PDE5 inhibitors were produced as a consequence of basic science work in the penis, most of the mechanistic work on PDE5 expression and actions in the lung has been undertaken because of the recognition that PDE5 inhibitors were successful in treating pulmonary hypertension rather than the science leading directly to the clinical trials.

If acupuncture were successful in a decent designed clinical trial, the basic science would then become possible. But at present no one knows where to begin looking.

While we pride ourselves on setting up trials where the mechanism is plausible, one only has to go back 50 years to where much of what we presently take for granted, progressed by serendipity rather than cold logic.

I take your point and to be clear, I am no defended of acupuncture, I can assure you. Take a look at my Blog on http://oryctolagus.wordpress.com/ to see my outrage at claims that an acupuncturist made to me. However, I was getting the message, perhaps incorrectly, that unless a clinical trial was a perfect double blinded placebo controlled study, then it was invalid.

For example, Cuddles said "If you don't give a placebo, then both the doctor and the patient know which group they are in, and there cannnot be any blinding at all. This makes the whole trial bascially worthless."

This is not really correct. Such studies are often constrained by practicalities and ethics be it with acupuncture or a conventional drug therapy. I don't agree that if the placebo is imperfect (because of necessary constraints) then the whole trial is worthless. This is too much of an absolutist interpretation I feel. The data have to be viewed with the shortcomings in mind of course, but this does not preclude a degree of interpretation.

I also think you are saying that it would only be feasible to look for a mechanism for acupuncture, after a real effect was convincingly shown in a clinical trial. Usually, the mechanism of action is shown, if only as a proof of principle, for example in vitro, then efficacy in humans are investigated. The problem with acupuncture is that the indications are so wide (ie cure all) then it's impossible to know where to start.