Acupuncture: Specific and Non-Specific Effects

Abstract

As an alternative and complement to conventional or allopathic medicine, acupuncture has become a firmly established therapy in many countries. Despite this wide acceptance, the question of how acupuncture works, or whether it works at all, remains the subject of lively debate. One problem in the discussion of acupuncture and its possible mechanisms is that acupuncture is not only used for the alleviation of acute or chronic pain but also for the treatment of nausea and vomiting following chemotherapy, for addictions, asthma, depression, etc. An article on the 1998 NIH Consensus Conference on acupuncture even suggests that ‘further research is likely to uncover additional areas where acupuncture interventions will be useful’ [1]. Given such a variety of applications, efficacy of acupuncture certainly cannot be explained by one single mechanism, unless we assume it acts exclusively through non-specific (placebo) effects. Furthermore, practitioners (in Germany, mainly physicians in private practice) follow a variety of protocols with regard to point selection and needling techniques. The best evidence available to date shows that body needle acupuncture is effective as a symptomatic treatment of chronic pain. Several recent, large randomized controlled trials (RCTs) have demonstrated a strong therapeutic effect of acupuncture [2–9]. In all of these studies, patients who received sham acupuncture served as one control group, while the second control group consisted either of patients who received conventional, guideline-based treatment or of a waiting list. The first direct comparison of acupuncture and conventional treatment was made in the GERAC studies [2, 4, 5]. This approach was taken from pharmaceutical research. The procedure is similar to that involved in testing a new drug against established standard medication. Direct testing is particularly useful to compare the value of ‘new’ therapies to ‘old’ ones. Trials may be designed to test whether the new treatment is more effective than the old one (superiority design), or to prove that it is not inferior (non-inferiority design). In the latter case the new treatment must have other advantages, e.g. significantly fewer side effects. The use of sham acupuncture as a control was designed to assess non-specific effects of acupuncture, which would also be expected in verum acupuncture (following the rules of traditional Chinese medicine [TCM]). In sham acupuncture, needles are inserted at non-acupuncture points, superficially, and without needle stimulation (i.e. no elicitation of the deqi sensation that indicates sufficient intensity of stimulus in the afferent fibers). Sham acupuncture is considered the most suitable control for verum acupuncture as far as patient blinding is concerned. Yet, it is no physiologically inert ‘placebo’ treatment [10], because inserting needles into the skin (‘minimal acupuncture’) produces specific physiological effects that may well be significant, especially in studies on pain treatment. This leads to an underestimation of the specific effects due to point selection, deeper needle insertion and stimulation of nerve fibers until deqi in verum acupuncture. Consequently, the specific effects of verum acupuncture may well be underestimated in sham-controlled RCTs [1]. The GERAC studies, the largest studies on acupuncture to date (about 300 patients per treatment group) and consequently those with the highest power to reveal differences between sham and verum acupuncture, failed to show any such difference. Still, perhaps the most surprising result of these studies, even to the authors, was that acupuncture was found to be as effective as drug interval therapy for the treatment of migraine and in fact significantly more effective than the conventional treatment for knee and back pain. Yet, this was equally true for verum acupuncture and for sham acupuncture. How can this be explained? It is not sufficient to simply assume that there are additional, previously unknown active