Alternative (complementary) medicine: a cuckoo in the nest of empiricist reed warblersCommentary: A warning to complementary medicine practitioners: get empirical or else

Editor

Vicker's tardy response to my letter deserves a prompt reply.

Yes, the wealth of Chinese research (and corresponding data
publication) into Traditional Chinese Medicine (TCM) is to be applauded as
compared to Medline, Cochrane etc. The latter are to be applauded for
their special values as a fount of excellence in many fields but, to me,
not TCM at the moment; this is obviously being addressed as I note that
more and more studies borne of true Chinese TCM-based research are
appearing amongst Western Medicine (WM) studies.

Isn't "the lack of negative findings in Chinese literature" you cite
really a myth? As I already said, much research is aimed at 'proving' to
WM science facts already known in TCM.
If I had been prescribing paracetamol for 2,000 years on my patients I may
have sufficient understanding and evidence that it works; how, when and if
to prescribe it for each individual - if I was then asked to prove this
through RCT I would apply the same methodology that had worked for me
after 2,000 years refinement - and would expect to be right most of the
time, wouldn't you?

I do not believe Prof. TANG et al say that ALL Chinese literature is
biased; the study says "MOST trials claimed that the tested treatments
were effective....". "Most" could be from just over 50% to 100%.

As to Vicker's 3 comments, the answers are:-

1. No
2. Isn't "Universally positive findings" a delusion?
3. TANG's 'funnel plot' was of 49 acupuncture studies for treatment of
stroke. The plot compares sample size with effects, and therefore ideally
requires treatments to be similar (eg all TCM or all non-TCM) and for
there to be a good spread of sample size. This was not the case. TANG says
only a few studies had sample sizes of 300 or more.
TCM-based studies ought to show a consistently better effect as each
diagnosis and treatment is patient specific; WM based studies are more
medication specific so effects would be more varied. I think that the plot
spread would be affected by the percentage of each type of trial in the
49; in this case a 'chaotic' pattern may have emerged which one could read
as proof of bias as one is not comparing like-with-like. What was the mix?

I note from the plot that 'acupuncture' was generally favoured over
control - that's a pretty expected result as far as we TCM practitioners
are concerned, even with WM-based acupuncture (when carried out by Chinese
WM/TCM medics as is usually the case).

According to 'Ernst and Resch' ALL non-RCT trials are potentially
biased. How many of the 49 were RCTs (TANGs report suggests many were non-
RCT) and does this factor affect funnel plot, which serves only to assess
likelihood of bias?

Whatever the state of bias or not in this funnel plot of 49 trials,
how does Vickers extrapolate that result to all of the other almost 3000
studies which formed part of Tang et al's project for which no funnel plot
was done and for which he intimates are 'universally flawed'?

Last time I communicated with prof. TANG he confirmed he did not have
original study papers; indeed his paper states that Journals were used for
data collection, and that many studies were published as only 'short
reports' thus having little detail. Most were of herbal medicine, and of
Western Medicine-based methodology. This reduces the impact of the
findings but clearly suggests more research is necessary. I would add that
better protocols, more use of original study papers, funnel plots based on
like methodology, and more attention to TCM-based studies (if one really
wishes to assess Traditional Chinese Medicine modalities) are advisable as
these are assessing best practice; most of these points I have already
taken up with Prof. Tang.

Answer to Q1; yes
Answer to Q2; I didn't.

Regards

John H.