Acupuncture placebo better than acupuncture

The title of the article should be:

"How to turn a negative result into a positive: Pretend you always planned to perform a two-tailed test after you find the intervention doesn't work as predicted."

:)
 
I think the issue here is sample size and lack of any other data regarding the patients. with only 370 patients (equals 185 per group if seperated equally), the 11% difference is approximately 20 patients. This swing could be easily explained by a number of factors that were not mentioned in the article, such as age of mother, previous IVF attempts, etc. with such a small sample size, randon sampling that does not recognize other possible factors can easily get skewed by randomly putting mothers with common traits in these other factors into the same group.
 
Acupressure over acupoints has been shown to be an effective
intervention for a number of clinical conditions such as
alleviation of dysmenorrhoea, reduction of anxiety and
nausea and vomiting during pregnancy (Beal, 1999, Markose
et al., 2004, Hsieh et al., 2006, Jun et al., 2007, Mora et al.,
2007). Therefore, the non-invasive placebo needle used in
the present study may elicit physiological effects. Subjective
sensations like DeQi have been regarded as an important component
to evaluate whether the effect of acupuncture is properly
elicited. Unfortunately, patients were not asked to
document this point. Our results indicated that the placebo acupuncture
may indeed lead to a higher pregnancy rate. However,
we cannot draw a firm conclusion because a control arm
without any acupuncture was not included for comparison.
Click to expand...

:(
 
Search science-based medicine http://www.sciencebasedmedicine.org/ for "prior probability," research on irrational procedures is rife with false positives. Also, see R. Barker Bausell's book- publications coming from Asia are misleadingly positive. Also, cwalner has a good point. I can't do the calculation; but I wonder how many of the 20 have to be moved out to make the difference insignificant?

Also, Ivor makes a good point if I understand him correctly. When one looks at so many endpoints, the probability of one being positive is increased so the level of statistical significance must be more stringent.

D. Cheney, thanks for linking the article, I have not read it yet.
 
JJM said:
<snip>

Also, Ivor makes a good point if I understand him correctly. When one looks at so many endpoints, the probability of one being positive is increased so the level of statistical significance must be more stringent.

<snip>
Click to expand...

Adjusting for multiple comparisons is important, but I'm not sure it applies in this study, though I haven't read the full paper yet.

What I was getting at was the researchers did not set out to use real acupuncture to change in either direction the women’s chances of a successful IVF procedure. To do so would be unethical. No, their alternate hypothesis was that real acupuncture would improve the chances of a successful IVF procedure compared to the women in the control group who received sham acupuncture.

So initially their hypotheses were:

Null: Intervention success rate = Control success rate.
Alternate: Intervention success rate > Control success rate.

But when they analysed the results and found the intervention group did slightly worse than the control group (i.e. p=0.038; not exactly an amazing difference as SG and others have pointed out), they decided to change the alternate hypothesis to:

Post-Hoc Alternate: Intervention success rate <> Control success rate.

Which is called having your cake and eating it.
 
Ivor the Engineer said:
Adjusting for multiple comparisons is important, but I'm not sure it applies in this study, though I haven't read the full paper yet.

What I was getting at was the researchers did not set out to use real acupuncture to change in either direction the women’s chances of a successful IVF procedure. To do so would be unethical. No, their alternate hypothesis was that real acupuncture would improve the chances of a successful IVF procedure compared to the women in the control group who received sham acupuncture.

So initially their hypotheses were:

Null: Intervention success rate = Control success rate.
Alternate: Intervention success rate > Control success rate.

But when they analysed the results and found the intervention group did slightly worse than the control group (i.e. p=0.038; not exactly an amazing difference as SG and others have pointed out), they decided to change the alternate hypothesis to:

Post-Hoc Alternate: Intervention success rate <> Control success rate.

Which is called having your cake and eating it.
Click to expand...


(my bolding)

amazing to see that on a sceptics forum.
 
Last edited:
cwalner said:
... with only 370 patients ... with such a small sample size, randon sampling that does not recognize other possible factors can easily get skewed by randomly putting mothers with common traits in these other factors into the same group.
Click to expand...

I don't think that is a valid criticism. From reading the article, it seems clear that random assignment procedures were adequate to rule out subject selection as a confound.
 
Ivor the Engineer said:
Adjusting for multiple comparisons is important, but I'm not sure it applies in this study, though I haven't read the full paper yet.

What I was getting at was the researchers did not set out to use real acupuncture to change in either direction the women’s chances of a successful IVF procedure. To do so would be unethical. No, their alternate hypothesis was that real acupuncture would improve the chances of a successful IVF procedure compared to the women in the control group who received sham acupuncture.

So initially their hypotheses were:

Null: Intervention success rate = Control success rate.
Alternate: Intervention success rate > Control success rate.

But when they analysed the results and found the intervention group did slightly worse than the control group (i.e. p=0.038; not exactly an amazing difference as SG and others have pointed out), they decided to change the alternate hypothesis to:

Post-Hoc Alternate: Intervention success rate <> Control success rate.

Which is called having your cake and eating it.
Click to expand...

I am not sure their initial hypothesis was directional, but if it was, it was clearly falsified by the significant results in the other direction. If so, they were acting like scientists should and published results that contradicted their original hypothesis.
 
quixotecoyote said:
Ivor's not supporting it, just reporting it.

Nothing wrong with that.
Click to expand...

pls?
he is reporting?
he is commenting, and that based on a yahoonews article while the full paper is avaible for free but he did not read it yet.

i am openmindedly interested in Acupuncture, but i read the full studie, and pointed out what i find flaw about the studie.
before comming to an oppinion, one should atleast read the studie.
But sure, when the prejudice is to big before reading it, it makes not really sence to read it.
 
Last edited:
Dictator Cheney said:
pls?
he is reporting?
he is commenting, and that based on a yahoonews article while the full paper is avaible for free but he did not read it yet.

i am openmindedly interested in Acupuncture, but i read the full studie, and pointed out what i find flaw about the studie.
before comming to an oppinion, one should atleast read the studie.
But sure, when the prejudice is to big before reading it, it makes not really sence to read it.
Click to expand...

I read the abstract of the study and looked at table II before posting my comments. The Yahoo article contains useful info. too:

"So far, there is no evidence that real acupuncture would adversely affect IVF outcomes because, in a previous meta-analysis of several acupuncture studies, the pregnancy rate was higher in the acupuncture group than in the control group..."
Click to expand...
 
Jeff Corey said:
I am not sure their initial hypothesis was directional, but if it was, it was clearly falsified by the significant results in the other direction. If so, they were acting like scientists should and published results that contradicted their original hypothesis.
Click to expand...

Rather than simply say that the real acupuncture group did not perform significantly better than the sham acupuncture group (as expected) and in fact did slightly worse, they feel the need to explain a (just about) statistically significant difference in the other direction.
 
Jeff Corey said:
{snip} If so, they were acting like scientists should and published results that contradicted their original hypothesis.
Click to expand...
Yes, the data they reported did not support the, obvious, null hypothesis. But, unlike scientists, they did not say that- instead, they moved the goalposts to include acupressure; and the possibility that acupuncture "works," just not as well as acupressure.

ETA: cross-posted with Ivor.
 
Last edited:
Ivor the Engineer said:
Rather than simply say that the real acupuncture group did not perform significantly better than the sham acupuncture group (as expected) and in fact did slightly worse, they feel the need to explain a (just about) statistically significant difference in the other direction.
Click to expand...

So? Speculation is allowed in the Discussion section of a scientific article. But the data are still there - the sham acuputure group was significantly better, if not sizably better.
 
One has to be careful about any meta-analysis re traditional Chinese medicine, especially acupuncture. Unfortunately due to cultural quirks not yet overcome in China, it is not socially (including in professional research) correct to find fault with traditional Chinese medicine. It has to do with saving face. As a result, the field of acupuncture is flooded with bad research.
 
Jeff Corey said:
Skepti'
What do you think about this study?
Click to expand...
This quote from the abstract says it all for me.
Placebo acupuncture may not be inert.
Click to expand...
Rather than looking for objective evidence acupuncture works or doesn't work, these 'believers' are looking for confirmation of their pre-existing conviction in magical medicine. What is needed but is missing here is the 'no treatment' arm of the study.
Patients were then randomized according to a computer-generated randomization list in sealed, opaque envelopes into two groups: real acupuncture and placebo acupuncture groups.
Click to expand...
This is a common flaw in double blind studies, failing to compare the 'no treatment' group.
 
You must log in or register to reply here.

Back
Top Bottom