Akupunkturtherapie für klinische Studien, Teil 1

von Gabriel Stux  

Wie soll man Akupunktur in klinischen Studien anwenden?

Diese Frage stellt sich zur Zeit für die drei Arbeitsgruppen, die für die gesetzlichen Krankenkassen kontrollierte Studien vorbereiten. Für den nicht-kontrollierten Teil dieser Studien werden keine Leitlinien formuliert.
Der kontrollierte Teil der Studien erfordert aber eine intensive Beschäftigung mit dieser Frage. Dazu liegt bereits seit einem Jahr eine Arbeit im "Clinical Acupuncture Scientific Basis", Springer-Verlag 2001, vor, die hier in zwei Teilen ins Netz gestellt wird.
Der zweite Teil folgt in der nächsten Ausgabe am 1. November 2001.
Ein weiterer Beitrag beschäftigt sich in dieser Ausgabe mit der Frage von grundsätzlichen Leitlinien für die Akupunkturtherapie von chronischen Erkrankungen.

G. Stux, S. Birch

The quality of clinical trials of acupuncture conducted since the early 1970s, is often poor. General reviews and formal evaluations of the clinical trial literature are uniform in these findings [4, 23, 26, 27, 30, 32, 34, 37-39, 43, 51, 53, 55, 59, 66, 67, 70-74]. While the literature evaluating these clinical trials continues to grow, it has focussed largely on the methodological quality of the study designs and the soundness of the results. There has been little discussion of the adequacy of the acupuncture treatment itself. A few authors have suggested that the quality or adequacy of the acupuncture in these clinical trials may be problematic [18, 32, 57, 63], and a few have discussed guidelines to suggest what might constitute an adequate treatment [5, 64]. There is generally a paucity of systematic analysis of the adequacy of the acupuncture tested in these clinical trials. One review team merely assessed whether the treatment was reported or not [34, 66, 67]. One review team attempted to develop criteria for such an evaluation, but experienced difficulties with the method that was selected [39]. This essay will outline criteria for evaluating the adequacy of treatment in acupuncture studies, present evaluations of the adequacy of a number of clinical trials, and describe a systematic method for ensuring adequacy in future studies. Preliminary versions of some of these methods and data have been presented in previous research conferences and publications [4-7, 64].

When acupuncture treatment is administered, there are several decisions that are made that form the basis of selecting the optimal treatment for the patient. For this patient, what are the most appropriate sites of treatment? What is the optimal number of treated sites? What is the optimal number of treatment sessions? What is the optimal frequency of treatment? What techniques of treatment should be applied at each treated site, and what tools are best used to do this? Is acupuncture to be used alone or as a complement to standard therapy for this patient? Prior to asking these questions, two more questions are usually raised: who shall administer the acupuncture treatment, and what are the training requirements for that person?
The acupuncture treatment is usually selected based on answers to all these questions. However, many of these questions are not explicitly described or answered in the acupuncture literature, and hence the practice of acupuncture. Most texts describe how once the diagnosis is selected, one should apply the prescribed treatment, with certain pluses and minuses according to the patient. But this process is not always an explicit one, and, as is happening in many areas or practice, the decisions and issues are constantly evolving. Take for example the treatment of drug addiction using acupuncture. Addiction is a complex health problem for which acupuncture has been applied over the last twenty five years. The treatment was initially applied by needling the auricles with additional electrical stimulation of the needles [75]. Then, as it became more widely used, needles alone started being used, at set points that practitioners had noticed seemed to be most helpful [13]. These treatments were added into existing treatment programs, where group and individual counselling is the standard treatment [14]. Many using this approach stated that the counselling was really an essential part of therapy and that the acupuncture should be used principally as an adjunctive treatment. It was also recommended that the acupuncture be applied daily if possible, at least for the detoxification stage of treatment, and then as needed after that. Clinical trials testing the needling of the points in the auricles then started being conducted. However, a close examination of these trials reveals a certain heterogeneity in the approach. Some used three, some four, some five auricle points, and some used additional body acupoints [e.g. 15, 48]. Most used the acupuncture as an adjunctive therapy to the standard counselling therapy, but some used the acupuncture alone without the concurrent counselling [e.g. 40]. Frequency of treatment also varied. Reviewers examining the clinical trial literature of acupuncture in the treatment of addictions have expressed their frustration with the variability of approaches and the need for a more standardized approach [51].
The variation seen in this one area of acupuncture practice and trial literature is not unique. In general we can say that acupuncture is not a homogenous method, unvarying in its applications in the different countries and cultures where it is used [10]. There are many different models of acupuncture practice and many different methods and techniques that may be used [8]. Thus the answers to this series of questions are usually tailored to the training and background of the acupuncturist asking them. For example the TCM acupuncturist is trained to identify the "zheng" or pattern that gave rise to the complaint of the patient, and to treat this whilst also applying treatment to relieve the symptoms. In the TCM model there are hundreds of possible "zheng" [76]. The Japanese Keiraku Chiryo style acupuncturist is trained to identify one of only 4 patterns "sho", and to apply treatment to correct this and relieve the symptoms [61]. Some styles of acupuncture do not apply any particular assessment methods to determine a "sho" or "zheng", rather they apply treatment simply as a means to relieve the symptoms [e.g. 20]. As a consequence of this diversity, we must recognize the following if we are to proceed with an examination of the standards for acupuncture treatment in clinical studies of acupuncture:

1. To test a particular style or method of practice, care must be taken to ensure that relevant methods be used. Thus in answer to the series of questions above, it will be important that answers to each are congruent with that style or method of practice. If testing the TCM style of acupuncture, the characteristic "de qi" sensation should be obtained at each site of insertion as is prescribed by the literature [17], similarly the "zheng" or pattern should be determined to be able to administer optimal treatment [12].
2. Additionally, the conclusions of the study should discuss that style of acupuncture rather than "acupuncture", as such generalizing may be misleading. For example it would be misleading to discuss how acupuncture can treat addictions since the treatment is principally of the auricles, it would be more precise to discuss the effects of auricular acupuncture.
3. If one wants to test "acupuncture" and be able to start generalizing across the field of acupuncture, it will be necessary to take the diversity of practice approaches into account. This will require coming up with a systematic method of examining the literature to find appropriate answers to the above series of questions.

This series of questions and issues related to the diversity of practice approaches has not been addressed in most clinical trials, creating many problems in the interpretation of these studies.

In an effort to understand the scope of the problem of adequacy of treatment in acupuncture studies, Birch conducted a review of the acupuncture literature describing the treatment of chronic headache, low back or neck pain. Twenty six Japanese, Chinese and English language texts were examined. Based on this review thirty four controlled clinical trials of acupuncture for chronic headache, low back and neck pain were examined to see if treatments approximated what was found in the literature review. This literature review found a conservative minimum treatment for each condition involves the treatment of 10 treatment loci for a course of 10 treatment sessions [5, 7], see below for details and examples. Table 1 shows the results from applying these criteria to the thirty four controlled headache, back pain and neck pain studies see [7] for references.
This review found that none of the thirty four studies clearly administered adequate treatment according to these basic criteria for adequate treatment of the three chronic pain conditions. Some may have done, but because of inadequate reporting of treatment methods, it is not possible to determine this one way or the other [6]. When these minimum adequate criteria were relaxed it was found that half the studies administered this approximately adequate treatment, see table 1.

Table 1 - Adequacy of treatment in 34 controlled studies of acupuncture for head, neck and back pain

Berman and colleagues conducted a formal review of forty seven pain studies incorporating the above criteria for adequate and approximately adequate minimum treatment [4]. Using logistic regression analyses to examine adequacy of treatment as a variable in treatment outcome, Berman and colleagues found that both the adequate and approximately adequate minimum treatment were significantly associated with a positive treatment outcome when controlling for study quality [4]. This finding suggests that the conservative minimum treatments derived from the literature reviews are valid judgements.
A further examination of the acupuncture clinical trial literature has found other related problems. Some reviewers have found that the actual treatments are not described at all in some studies and that there is no source referencing of the applied treatments thereby not allowing a determination of what treatments were actually applied [6, 32, 34, 66, 67]. Kersken found in an analysis of eighty six controlled studies that no description of the therapy was given in eighteen studies (21%), and that no description of number of treatment sessions or other important parameters were given in nineteen studies (22%), [32]. Additionally Kersken found that while adequate description was made of the treatment principles in fifteen of the eighty six studies, thirty four studies (40%) did not give adequate description of treatment principles [32]. It is hard to imagine a study of a particular analgesic drug where the actual drug used and doses of prescription are not described, nor references given so that one can check elsewhere what was done, but the equivalent problem can be found in acupuncture studies [e.g. 18, 19, 21, 31].
In his analysis of the quality of the acupuncture in eighty six studies, Kersken found that only 33% describe obtaining a "de qi" response [32]. Pomeranz also noted that in many studies of acupuncture the specific needle technique recommended, the elicitation of the "de qi" response (obtaining the qi), was not applied, and that thus many studies did not adequately apply treatment [56]. While "de qi" is a very common method of needling, and is a characteristic of some but not all forms of acupuncture [8], it is important that when testing those models of practice in which "de qi" is normally indicated, that the study utilizes this technique. Some have also argued that the optimal form of acupuncture should use the traditional diagnostic methods of assessing patients to appropriately individualize treatments for patients in clinical studies [18, 44], while this is not necessarily a valid claim given the diversity of practice approaches [6, 8], it is important that when testing models of practice that usually employ these assessments to individualize treatments, then the study should employ those methods. However, the utilization of these methods will require prior testing to establish the reliability of the methods [6, 69]. To date very few studies of traditional diagnostic assessment methods have been conducted [69]. While evidence of the reliability of those assessments is slowly growing [9] it is not yet sufficient to eliminate the need to incorporate further reliability studies into future prospective trials that wish to utilize the diagnostic assessments methods.
Additionally, the training of the person administering the treatments can be an issue. In their three reviews, Ter Riet and colleagues addressed the issue of whether the training was mentioned in the eighty six studies they reviewed [34, 66, 67], finding that the "good quality of the acupuncturist" was mentioned in only thirty studies. Hammerschlag and Morris examined twenty three studies to find that the qualifications of the acupuncturist was mentioned in only one study [27]. However, mentioning the good quality of the acupuncturist or simply stating the qualifications of the acupuncturist are clearly not enough since to mention that someone had done some training in acupuncture is not the same as saying that they were really qualified to administer the treatment in a clinical study. For example, how long was their experience in practicing acupuncture? Obviously more needs to be specified. Here we find a very contentious issue, as the training requirements for acupuncturists around the world vary considerably [3, p.177-205]. There are places where physicians can practice without any requirement of training. There are places where the practice is unregulated and anyone can practice. However many countries and states within those countries have come up with some minimum training standards, but these vary tremendously from location to location. One of the principle variables is to do with whether the acupuncturist is also a physician or not. Typically physicians study acupuncture in shorter programs than non-physicians. Training programs for physicians can run from one or more weekends to several hundred hours, a review in Australia found programs that ran from 50-390 hours [3, p.168]. The recommended length of training programs for physicians in Germany is proposed to be 350 hours, 50% theoretical and 50% practical. The World Federation of Acupuncture and Moxibustion Societies (WFAS) recommends a minimum of 200 hours of training for physicians who wish to practice acupuncture [28]. Training for non physicians is usually at least 400 or so hours, and can be substantially more. For example in Australia, training programs run from 1562-3824 hours [3, p.158], and in the US, the accredited programs contain a minimum of 2175 hours, with the average program having 450 hours more than this [22]. It is not easy to ascertain a clear minimum standard of training for acupuncturists in studies of acupuncture, but at the least a minimum needs to be established in the context of the country of the study. Ideally an approved course of study should have been completed. We would not feel comfortable to have a physician not experienced in a particular medical procedure be the person administering that procedure in a clinical trial.
While we cannot generalize from these reviews to all clinical trials of acupuncture, it does appear that there is a significant problem in how acupuncture treatments are tested in clinical trials. Putting aside questions about what some proponents believe are the correct styles of treatment [e.g. 2, 18, 44], in simple numerical terms (numbers of points treated and number of treatment sessions), it appears that many clinical trials did not apply sufficient treatment. This is akin to testing a drug in a controlled clinical trial at an inadequate dose level for an inadequate length of time. If the results of such a trial were positive for the drug, many might consider the drug to be very effective, capable of working at lower doses than expected. Were the results of the trial negative for the drug, few would argue that the drug had been properly tested. However, knowledge of this issue has not yet been adequately incorporated into literature reviews of the acupuncture clinical trial literature. In the absence of such analyses, many reviewers assumed that the treatments were adequate, and upon finding other problems with study design or not much robust data, concluded that acupuncture does not appear to be very effective [e.g. 34, 66, 67, 71]. Were an analysis of the adequacy of treatment incorporated into such reviews, it may turn out that the conclusions of many of these authors might need to change.
In the following, we outline methods for determining what might constitute a minimum adequate treatment in a clinical trial of acupuncture. These methods and their results can be used in both a reassessment of the published literature, and by researchers planning prospective studies. Based on the minimum adequate treatment, we then describe what might be the optimal treatment, that which would be expected to produce optimal results for the patient or group of patients.

Birch has previously pointed out that one of the principal reasons for finding inadequate treatment was that many research teams appear to have read little of the available acupuncture literature, as evidenced by the lack or inadequate citation of sources in published studies [6]. Some studies even go so far as to cite no acupuncture literature at all [18]. The solution to this problem is obviously that more acupuncture literature needs to be read before embarking on clinical trials of acupuncture. But here there are still a number of issues that will need to be dealt with. i) There are often problems with the literature itself [6]. ii) There are problems stemming from the nature of the question to be addressed and the related issue of generalizability of results [6]. iii) There is a problem with variability in the literature [6, 12]. What we propose is the use a systematic and extensive literature review in each relevant area to address these issues.
Table 2 illustrates how the literature review process is applied. This table shows the findings from a review of eighteen texts that describe the treatment of low back pain by acupuncture. The recommended treatments of low back pain in each text are compiled so that we might see what the range and averages of recommended numbers of treatment points and treatment sessions might be. The review included a variety of texts from different approaches so that when averaging the recommended numbers of treatment points and sessions, we can increase the generalizability of results. The numbers of points recommended by each text for low back pain, and the various sub categories of low back pain are recorded. By examining the specific points that were mentioned, the total number of discrete sites to be treated could be extrapolated. If the text gave an indication of the number of treatment sessions necessary, this is also recorded. 6/6/5 indicates 3 alternative treatments (including treatments according to differential diagnostic patterns) with 6, 6 and 5 points listed respectively. (10-11) indicates that from 10-11 discrete sites are treated.

Table 2: Recommended treatment outlines for low back pain

According to these sources, low back pain is treated with from 2 - 14 points, or from 3 - 28 needles, at an average range of 5.1 - 7.8 points and 9.5 - 13.8 discrete sites per treatment. According to these sources, at least 10 treatments need to be given, which matches clinical experience [64]. It thus appears that at least 10 discrete sites treated for at least 10 treatment sessions can be considered as the minimum adequate treatment for chronic low back pain. Hence in a clinical trial of chronic low back pain, treatment should involve the application of treatment at each session to at least 10 discrete sites during a course of at least 10 sessions.

It is not common to find acupuncture texts discussing the number of treatment sessions and frequency of treatment. In table 2 only three of eighteen texts discussed the recommended number of treatments. Frequency of treatment is recommended even less often. Since this is a common problem, it will be necessary to use additional methods to determine what would be an appropriate number and frequency of treatments. Since the literature itself is not forthcoming on this, it would be useful to conduct practitioner surveys to ascertain actual practice habits, and to augment this with expert panel reviews.

It is possible to use the same overall process to examine the recommended treatments within a specific style or tradition of practice, such as zhong yi acupuncture, or traditional Chinese medical (TCM) acupuncture. Table 3 summarizes the findings from a review of sixteen different TCM acupuncture related texts [12]. Quite a wide range of diagnostic patterns were found (3-19 patterns, median 3.5).

Table 3 - Summary of low back pain treatments in TCM related acupuncture texts (From Birch, Sherman, in submission).

1. [Kwok et al, 1991]. 7 patterns, mean 5.7 points (10.3-10.5 discrete sites)
2. [Manaka, 1970]. 3 patterns, mean 3.7 points (6.3 discrete sites)
3. [Shanghai Inst. of TCM, n.d.]. 5 patterns, mean 8.2 points (14.6 discrete sites)
4. [O'Connor, Bensky, 1981]. 3 patterns, mean 6 points (11-12 discrete sites)
5. [Anon, 1980]. 3 patterns, mean 4.7 points (7.7 discrete sites)
6. [Chang, 1987]. 4 patterns, mean 5 points (8.3-8.5 discrete sites)
7. [Feit, Zmiewski, 1990]. 4 patterns, mean 6 points (10.3-10.5 discrete sites)
8. [Liu, 1988]. 4 patterns, mean 3.8 points (6.3-7.3 discrete sites)
9. [Qiu, Su, 1985]. 3 patterns, mean 4 points (6.7-7.3 discrete sites)
10. [Wiseman, Feng, 1998]. 19 patterns, mean 7.6 points (12.9-13.8 discrete sites)
11. [Ying, De, 1997]. 4 patterns, mean 5 points (9.3-9.5 discrete sites)
12. [Liu, 1996]. 4 patterns, mean 9.5 points (17-17.5 discrete sites)
13. [Geng, Su, 1991]. 3 patterns, mean 5.7 points (9.3-10.3 discrete sites)
14. [Maciocia, 1994]. 3 patterns, mean 12 points (15-19 discrete sites)
15. [Zheng, 1990]. 3 patterns, mean 6 points (11-11.3 discrete sites)
16. [Wu, Fischer, 1997]. 5 patterns, mean 6.2 points (10.6-10.8 discrete sites)

Four TCM diagnostic patterns of chronic low back pain were commonly discussed. The recommended numbers of points and discrete sites of treatment are [12]:

• cold damp: mean 6.6 points (range 3-12), mean 9.6-11.5 discrete sites (range 5-21)
• kidney yang vacuity: mean 6.4 points (range 5-9), mean 10.3-10.4 discrete sites (range 8-14)
• kidney yin vacuity: mean 6.3 points (range 5-8), mean 11.7-11.9 discrete sites (range 9-15)
• blood stasis: mean 6.4 points (range 2-11), mean 11.2-11.8 discrete sites (range 3-20)

Thus in a study that wished to test TCM acupuncture in the treatment of low back pain, considerable variation in the literature exists making treatment selection somewhat complicated. However, these four TCM diagnostic categories could be used, with the conservative minimum treatment for each category as ten sessions with treatment of 10, 10, 12 and 11 discrete sites respectively, (for more details of these treatments see discussions in [12])

The literature review methods that we have described above can be used across traditions of practice or within traditions of practice to examine what might constitute an adequate treatment in any condition for which acupuncture is used. Table 4 summarizes results of similar reviews in the acupuncture treatment of neck pain, headaches, and asthma.

Table 4 - Summary of neck pain, headache and asthma treatment recommendations

A similar review of the treatment of osteoarthritis by acupuncture showed that this literature review method can encounter difficulties. These difficulties arise in part because of considerable inconsistency in the acupuncture literature, in part from difficulties with terms, and in part because this is a condition with many manifestations, some of which are described separately. In a review of thirty four texts from the worldwide literature, only fifteen had any actual description of the treatment of arthritis, including osteoarthritis. In these texts from at least 3-16 sites of treatment were recommended depending upon location of the arthritis and text.

These difficulties with the osteoarthritis literature suggest that it may only be possible to use the number of treatment sessions as the cut off point for determining adequacy of treatment in each clinical trial that is examined, since the selection of treatment points is quite varied. However, were there an effort to incorporate an analysis of the amount of treatment in each treatment session, then it may also be advisable to use the above recommendations as tentative guidelines to be augmented by discussions among a group of carefully selected experts. These recommendations could then serve as a starting point for the discussion of this group of experts. It would be better if the group had broad representation of the major traditions of practice so that its considerations could be more generalizable.

It should be noted that the methods used above in tables 2, 4 and 5, yield an average range of number of acupoints and discrete sites to be treated for each condition. We assumed that the lower end of that range should constitute the "minimum adequate treatment" for that condition. In their review of pain studies, Berman and colleagues found that the cut off point for minimum adequate treatment was a significant predictor of treatment outcome [4]. It thus appears that this guideline is useful in the analyses of published studies. However, when designing prospective studies, it is advisable to define adequate treatment in more positive terms, such as the "optimal treatment", so that there can be no doubt that the treatment tested was a fair test of acupuncture. Defining this "optimal treatment" can be somewhat difficult. It is possible to use the higher ranges of average numbers of acupoints and discrete sites of treatment as optimal. For example, the data in tables 2 and 4 would therefore suggest using 14, 14, 15 and 17 sites for the treatment of low back pain. neck pain, headache and asthma, respectively. It may also be necessary to process this data with surveys of practitioners and expert panels to achieve a greater level of agreement about what the appropriate. "optimal" treatment in a particular condition should be.