E Ernst* MD PhD FRCP (Edin.)
*Department of Complementary Medicine
School of Postgraduate Medicine and Health Sciences
University of Exeter
25 Victoria Park Road
Exeter EX2 4NT UK
¡§TCM has stood the test of time and therefore does not require testing through scientific methods¡¨. This belief is as widespread as it is wrong. Today it is generally considered essential that all therapeutic interventions are tested rigorously to determine their efficacy, regardless of whether they are mainstream or alternative, modern or traditional. Many insufficiently tested mainstream treatments are now being evaluated. Only if this process comes to a positive conclusion, will a given therapy continue to be a part of routine healthcare. Those treatments which are as yet outside routine care also require proper evaluation. Only if this process comes to a positive conclusion, will such a therapy be considered for integration into routine care.
Put into a nutshell, these are the principles of evidence-based medicine (EBM). For evaluating treatment modalities, EBM does no longer rely on criteria such as plausibility of mechanisms of action nor does it listen to the opinion or experiences of those individuals who dominate a given field of therapeutics. EBM rather relies on the results of rigorous clinical investigation and analysis. This is perhaps best illustrated through the ¡§hierarchy of evidence¡¨ which can be viewed as the backbone of EBM (Fig. 1).
According to this hierarchy, the basis of all clinical knowledge and research is represented by clinical observations (e.g. case studies, observational data etc). Experience and traditions are therefore not devalued by EBM. On the contrary, they are the basis for everything that follows: they lead us to formulating a hypothesis (e.g. acupuncture is effective in the treatment of pain). However, they are never sufficient to test, prove or disprove a hypothesis. For this we need controlled experiments, preferably randomised clinical trials. Only with controlled trials can we hope to determine whether an observed effect was actually due to our intervention rather than, for instance, spontaneous recovery (i.e. natural history of the disease). And only with randomised trials can we be reasonably sure that we are comparing groups of patients who were in all respects comparable.
Thus the randomised trial is the best available method to assess the efficacy of therapeutic interventions. This is true for conventional therapies as much as it is true for TCM. It does, however, not mean that the randomised trial is without flaws and problems. The fact is that it has many pitfalls but, at present, we have no better method for exclusion of bias in efficacy studies.
Even if the randomised trial were flawless, we would be ill advised to rely on one single such study. We always require independent replications. More often than not, such replications yield a slightly different result. This can have many reasons, for instance, slightly different types of patients treated with slightly different treatments. Eventually we can end up with several trials, some of them may show that acupuncture is effective others that it is ineffective. In such a situation it is tempting to select the evidence which fits ones beliefs. A proponent of acupuncture could therefore publish a review which only contains the positive evidence. Likewise, an opponent could write a paper entirely based on the negative evidence. Both reviews would seem to be evidence-based but both might be overtly misleading.
This simple (and somewhat simplistic) example makes it clear that only systematic reviews can hope to create a fair picture. Such reviews have to demonstrate that the totality of the evidence on a given subject (and fitting certain predefined inclusion/exclusion criteria) has been included. In this way, systematic reviews minimise selection bias and random error.
Again few people would say that this is a flawless approach to determine the truth. In fact, it has numerous problems and has been rightly criticised by several authors. Most experts nevertheless agree that it is the best available method of arriving at a fair judgements regarding the efficacy of therapies.
In TCM a sizeable number of systematic reviews have been published; all relate to acupuncture (Table 1). According to this type of evidence, there are now four well documented indications for acupuncture, two indications where acupuncture does not seem to be more effective than control interventions and a relatively large list of indications where randomised trials are available but the systematic review of these data is inconclusive. The reasons for inconclusively can vary. In the case of stroke, for instance, all studies are positive yet the methodological quality of these trials is so poor that no firm conclusions are permissible. In the case of osteoarthritis, on the other hand, some rigorous trials imply efficacy of acupuncture while other, equally rigorous studies suggest the opposite. Thus no firm conclusions are again permissible.
In summary, the methods of EBM can and should be applied to TCM. If this is done, we will identify areas for which the evidence is compelling and where (applying the logic and standards of EBM) TCM should be considered for integration into routine healthcare. We will also identify areas where more and better studies are needed to determine the efficacy of our treatments. EBM is no threat to TCM, on the contrary it provides a unique opportunity and a formidable challenge.
*published by our Department
(an up-dated list of our own systematic reviews can be obtained on request)