Acupuncture in Medicine

Summary

To evaluate the biological effects of needling, a randomised, controlled trial group of 49 patients with angina pectoris had acupuncture while cardiological, neurophysiological and psychological observations were made in a mutually independent manner. Needling was found to improve the working capacity of the heart. In addition, acupuncture was found to activate cardiovascular autoregulatory mechanisms in 24 healthy persons.

To evaluate the effect of acupuncture in daily life, a controlled trial group of 69 patients with severe angina pectoris were followed for 2 years after treatment. The incidence of cardiac death or myocardial infarction was 7%, compared to 15-21% for the control group of published results concerning invasive treatments. Due to clinical improvement, surgery was postponed in 61 % of the patients. The annual number of in-hospital days was reduced by 90%, leading to a US$ 12,000 saving for each patient.

Keywords

Introduction

Four hundred years ago the philosopher Descartes postulated the interaction of a separated mind and body. This approach has led to technological achievements in nearly all aspects of human life. The bypass operation is such an achievement. It is based on the scientific pathophysiological interpretation of angina pectoris being the consequence of an insufficient ratio between oxygen demand and supply to the myocardium. The improved myocardial blood supply from the new vessels should cure the disease and accordingly heal the patient. When the operation fails to meet this aim, one may ask whether this scientific interpretation represents the full picture of angina pectoris. Furthermore, is the heart more than just a pump? With this in mind, two major questions may be important in evaluating the effects of acupuncture on the cardiovascular system:

1. A scientific question:
   Does needling have a biological effect of its own?
2. A clinical question:
   Does acupuncture help patients in their daily life?

In normal practice, the above scientific definition of acupuncture covers only one part of the picture. Apart from needling, acupuncture includes a social interaction between patient and physician. In this connection, acupuncture is used according to traditional Chinese theory in which acupuncture activates homoeostatic mechanisms of the body and the doctor helps the patient to develop a lifestyle that supports this effort. From the patient's point of view, the outcome of this more complex treatment may be more important than the specific biological effect of needling. Furthermore, it offers the possibility of obtaining a fair estimate for the degree and duration of the effects of treatment as applied to daily life.

I would like to address the challenges that may occur when studying the effect of acupuncture on the cardiovascular system both from a scientific point of view and from the patient's point of view.

Study design in acupuncture research

1. The groups compared are identical with regard to all factors that may influence the result.
2. The treatments compared appear identical both to the patients and to the observers.

The complexity of the challenge

Placebo has been defined as the non-specific, positive therapeutic effect of the entire patient-physician relationship (Wulff 1981, Grünbaum 1985). Nocebo is the opposite. The specific effect is the therapeutic influence attributable solely to the processes of the therapy being rendered (Grünbaum 1985). The term placebo has been used for centuries, but nocebo was introduced recently by Kennedy (1961) to describe those stimuli such as fear, anxiety and mistrust that might have a negative effect on the healing process.

Although many studies have been carried out, no special personality characteristics have been found to be associated with either effect. On the contrary, the same person may react differently under different conditions, and the disease or symptoms may change over time or from patient to patient. The placebo effect is enhanced as the need for help increases.

It is suggested that the placebo/nocebo effect is related to the expectation from the patient, the observer, and the attending physician in combination with a classic Pavlovian conditioning response activated by positive or negative anticipation of healing (Wichramacekera 1985, Rosenthal 1969, Kaada 1986, Wall 1992). They exert a reciprocal inhibition at the brainstem level: the placebo effect through activation of the endogenous opioid-serotonergic, pain-inhibitory descending system, and the nocebo effect through activation of noradrenergic neurons in the locus coeruleus (reviews: Kaada 1986, 1989).

The placebo effect may vary according to the disease or condition to be treated (Richardson 1992). It is generally agreed that the influence of the placebo effect may be negligible on the death rates of, for example, patients with cancer or cirrhosis. However, in other conditions such as angina pectoris the impact of the placebo effect is known to be pronounced (Beecher 1955, Amsterdam et al. 1969, Benson and McCallie 1979).

Placebo, nocebo and specific effects will influence the result of any treatment of a disease. In daily clinical practice the physician can use the placebo effect to the benefit of the patient, but in a scientific study aimed at elucidating the specific effects of acupuncture, the presence of placebo and nocebo effects imposes a bias. To eliminate the influence of these effects, the factors in Table 1 must be addressed.

Table 1 Factors influencing placebo and nocebo effect 1. Choice of control group and control of treatment 2. Sample size 3. Bias from the observer 4. Bias from the acupuncturist 5. Bias from the patient/subject 6. Influence from psychosocial factors

1. Choice of control group and control treatment

In the double-blind, randomised clinical drug trial, the effect of the active treatment is compared with that of an inactive preparation, a placebo, which appears identical to the active treatment to both patient and attending physician. In acupuncture trials no similar control treatment seems possible. Acupuncture entails penetration of the skin at specific sites related to the condition of the patient, repeated physical contact involving the hands of the acupuncturist and the skin of the patient, and a unique physical sensation when the needle stimulates the acupuncture point. This sensation is described as soreness, aching, or burning and is not to be confused with the sensation related to the penetration of the skin. Furthermore, since the whole body can be regarded as one large acupuncture point, any insertion of an acupuncture needle might have both a specific local and a general effect on the body.

Consequently, a true non-active control treatment identical to genuine acupuncture is not possible. Many attempts have been made to solve this problem (Lewith and Machin 1983, Lewith 1984, Vincent and Richardson 1986, Vincent 1989). In various ways previous trials have dealt with the selection of a proper control treatment, focusing on the penetration of the skin as well as the unique physical sensation.

When the control group received no treatment (Coan et al. 1980) or a placebo pill (Richter et al. 1991) there was no skin penetration and no physical sensation. Some physical sensation (but differing from that caused by a needle) was elicited during mock Trancutaneous Electrical Nerve Stimulation (Melzack 1975), where electrodes attached to the surface of the skin were connected to a dummy electric stimulator. Rubbing the skin is yet another way of creating some physical sensation (Jensen et al. 1979).

In sham (non-standard) acupuncture the skin is penetrated, but the needle is inserted at a site different from that of the traditional (standard) treatment points. This has been tried in various ways: far from the traditional treatment point, for example lower instead of upper limb (Jobst et al. 1986), in the adjacent dermatome (Fung et al. 1986), or within the same dermatome but outside the traditional Chinese meridian system (Gaw et al. 1975). In superficial acupuncture the needle is inserted at the same site as in the standard treatment group, but superficially, to a depth of only 2-4mm compared with up to 10mm (Hansen and Hansen 1984). The needle might be inserted without stimulation (Gaw et al. 1975), or with stimulation similar to the standard acupuncture treatment (Fung et al. 1986). Injection of medicine at an acupuncture site is yet another control treatment (Garvey et al. 1989).

Some trials used more than one control group: acupuncture has been compared with injection therapy, physiotherapy and placebo medication in shoulder pain (Fernandes et al. 1980), and traditional acupuncture was compared with dummy acupuncture and no-treatment in testing antiemetic effect (Dundee et al. 1986, Dundee 1988, Dundee and McMillan 1992).

Instead of comparing the effect of acupuncture between groups receiving different treatments, the crossover design has been used to compare, for example, traditional and dummy acupuncture (Dundee et al. 1987), or acupuncture and a placebo pill (Richter et al. 1991). This design may be useful in measuring effects with duration far shorter than the observation period.

2. Sample size
Although this issue has no specific bearing on acupuncture trials, it may be worth considering. If we postulate that traditional (standard) acupuncture helps 70% of patients, sham or dummy (non-standard) acupuncture 50%, and placebo 30%, a study that compares standard acupuncture and placebo, and has 20 patients in each group, will have a 25% probability of identifying a difference in the effect between the two treatments at a 5% significance level. This means that there will be a 75% probability of a false negative result, i.e. a treatment being rejected as ineffective although it would have proved effective in a larger scale study. A study comparing standard and sham acupuncture with 20 patients in each group has only a 13% probability of identifying a significant between-group difference; with 85 patients in each group, the probability is 75%.

3. Bias from the observer
The blinding of the observer is easily managed both in traditional pharmaceutical clinical trials and in acupuncture trials: the patients are randomised, one group to the standard acupuncture and one to the control treatment, with the observer unaware of the treatment given. To exclude bias, the observer must believe that the trial is testing either an active against a non-active treatment, or two active treatments. Furthermore, issues must not be suggested, nor statements made that might affect the patient emotionally. Both intentional and non-intentional observer expectancy has been found to influence the performance of a test (Rosenthal 1969), the effect of placebo pills (Blackwell et al. 1972), and pain tolerance in experimentally induced pain (Zelman et al. 1991). The impact of this effect, named The Rosenthal Effect, is illustrated by the following diary entry from a patient:

"Right after the initiation of acupuncture a pronounced improvement appeared, but when I was told at the final exercise test that I ought to have a coronary bypass operation, I was knocked-out and felt terrible. For a long period afterwards I had chest pain several times daily and needed to take nitroglycerine, which caused an unpleasant pressure in my head. Months later, I consulted another cardiologist who said that an operation was not needed right now. My mood improved right away, so did my general well-being, and the chest pain declined." (Ballegaard 1989).

4. Bias from the acupuncturist
When compared with surgery and pharmacological treatment, the risk of inducing bias is far greater in acupuncture treatment due to the extensive and prolonged contact between patient and doctor: for instance, ten or twelve 45 minute sessions within a three-week period in the case of angina pectoris. Accordingly, it is of major importance to address the possible bias from the Rosenthal effect induced by the acupuncturist. Owing to the very nature of acupuncture, and irrespective of the control treatment chosen, an experienced acupuncturist can never be blinded, and an experienced acupuncturist may be needed to give a traditional treatment, which otherwise may not be correctly performed. During training as an acupuncturist, trainees develop a positive bias towards acupuncture, which may then influence attitude during a clinical trial.

A special challenge is how to deal with the possible physiological effects of human touch, which is an inherent part of acupuncture. A few examples may illustrate the problem. Rats receiving a high-cholesterol diet and a one-to-one relationship to the investigator, including touch 3 times daily, showed a 60% reduction in diet-induced arteriosclerosis when compared with an untouched group (Nerem et al. 1980). Premature infants in incubators increased their weight faster when touched daily than did non-touched infants in the same condition (Helders et al. 1988). In unconscious arrhythmic heart patients, manual pulse taking had a normalising effect on the heart rhythm (Lynch et al. 1974a,b).

5. Bias from the patient
Effective patient blinding in an acupuncture trial is difficult to achieve. In the Western world the amount of information about acupuncture has increased during the last decade. The general population, therefore,has some knowledge of the way acupuncture may be experienced and any control treatment differing from this might change the patient's expectations.

Patients attending acupuncture trials will often have a positive or at least a neutral attitude towards acupuncture. If not, they might feel participating in the study a waste of time, especially because, compared to a pharmaceutical treatment, acupuncture is time-consuming. This patient attitude may create a positive bias towards acupuncture.

6. Influence from psychosocial factors
The possible influence of the patient's psychosocial state on a clinical trial is generally believed to be eliminated by the use of the double-blind randomised trial (Wulff 1981). Unless the factors are identified and a stratification of the patients is performed, it is, however, unknown whether the distribution of characteristics is even between the compared groups. The use of a crossover design, where possible, may solve this problem.

The complexity of the issue was already apparent in 1958, when Wolf wrote:

"It is probable that most adaptive functions of the cardiovascular system are responsive to stimuli that owe their force to their special significance to the individual".

In the pathogenesis of coronary heart disease the influence of psychosocial factors (personality, social support system and life situation) is widely recognised (Kringlen 1986). The results of treatment of these patients were similarly affected (Blumenthal et al. 1982, Williams et al. 1986, Levenson et al. 1989, Diederiks et al. 1991).

Conclusion
It can be concluded that the conditions for traditional double-blind randomised trials of acupuncture in heart disease cannot be met as:

1. The patient cannot be truly blinded.
2. The acupuncturist cannot be truly blinded.
3. Psychological and social factors have an influence on any treatment in heart patients.

Possible scientific acupuncture study designs

Angina pectoris patients

First, the patients underwent a psychosocial test that included inquiring about their expectations as to the outcome of the treatment for angina pectoris (Ballegaard et al. 1995). The hypotheses tested were established through a retrospective testing of the patients participating in an initial study (Ballegaard et al. 1986).

Second, the patients were randomised to traditional or non-standard acupuncture (Ballegaard et al. 1990) for their angina pectoris. This made blinding of the observers possible. By correlating the results of this trial with the psychosocial testing, it was possible to detect whether such factors influenced the outcome of acupuncture used to treat an illness. The traditional acupuncture was given according to traditional Chinese medicine, each patient receiving ten treatments in a supine position within three weeks. The needles used were Chinese stainless steel, 30 gauge and 1.5in long. After obtaining needling sensation (de qi), the needles were left in place for 20 minutes. No electrical or mechanical stimulation was given. In the control group (non-standard acupuncture), needles were inserted superficially through the skin, with no attempt to obtain needling sensation, within the same spinal segments as the acupuncture points but outside the Chinese meridian system and not at trigger points. The needles were then left untouched.

Third (Ballegaard et al. 1991, 1995), the patients received traditional acupuncture from a different acupuncturist, while the changes in skin temperature, pain threshold (PT) and pain tolerance threshold (PTT) were recorded on the index finger, close to the acupuncture site, and on the hallux, distant from the site. These two measurement sites were selected because the local effect of acupuncture has been found to exceed the general one, both for skin temperature (Cao et al. 1983) and pain threshold (Andersson 1979). This is in contrast to the uniform influence on the body that would be expected from a placebo effect, when neither patient expectation nor conditioning is involved (Rosenthal 1969, Wichramacekera 1985, Kaada 1986). The patients were told that this experimental set-up was designed exclusively for the purpose of increasing our understanding of the mechanism by which acupuncture works and thus had no therapeutic aim. In this respect the trial examined only the effects of acupuncture on skin temperature and pain thresholds. A computerised test programme and automatic monitors were used to minimise the communication between acupuncturist and patient, and both were told not to discuss acupuncture during the procedure. Furthermore, although the patient and acupuncturist were together for approximately one hour, it was assumed that unexpressed acupuncturist expectancy was not likely to produce different effects on the index finger as compared to the hallux. In this study the needles were inserted in point Hegu (LI.4) bilaterally. The needles were stimulated electrically at 2Hz at an intensity sufficient to produce visible muscle contractions of the first dorsal interosseous muscle, but at well below pain threshold. The anode was connected to the left Hegu point (on the measurement side). However, Peter Nathan (personal communication), recently of the National Hospital for Neurology and Neuro Surgery in London, pointed out that the two stimulation sites preferably should be on the same arm. Accordingly, this was changed in the study of healthy people (Ballegaard et al. 1993).

Skin temperature was used to reflect the activity in the sympathetic nervous system in order to examine the relationship between the anti-anginal effect and change in sympathetic tone. PT and PTT were recorded in order to examine the relationship between the pain-inhibitory and the anti-anginal effects of acupuncture.

By correlating the results from the two separate acupuncture trials, we believe that acupuncturist bias was eliminated. The risk that acupuncturist bias from two individually blinded acupuncturists could interfere with a result in which anti-anginal effect was related to neurophysiological change when using the index finger as a site for measurement, but not when using the hallux, was considered insignificant when compared to the known sources of error such as inaccurate reading or transfer of data. Furthermore, we believed that acupuncturist bias would have a general effect, and thus produce no difference between hallux and index finger. The design helped to blind the patients as far as possible and patient expectation was already accounted for in the psychosocial questionnaire. Excluding patients who previously had received acupuncture treatment for their angina pectoris eliminated the development of any conditioning response. With this background, the study had the potential for differentiating between a placebo effect, the specific effect due to non-standard needling and the specific effect of traditional acupuncture.

Results

There was no significant influence from patient expectation and psychosocial factors on the anti-anginal effect of acupuncture (Ballegaard et al. 1995).

There was no significant difference between the effect of the traditional and non-standard acupuncture on angina pectoris, both effecting a reduction in nitroglycerin consumption and in anginal attacks. Thus, it was concluded that the clinical improvement was due either to a specific effect of both methods or a placebo effect (Ballegaard et al. 1990).

In the neurophysiological trial it was found that traditional acupuncture increased pain thresholds locally, but not distantly (Ballegaard et al. 1991), which supports the findings by Andersson (1979). Accordingly, it was concluded that the effects observed during the neurophysiological trial were due to acupuncture and not to bias from the acupuncturist.

Concerning skin temperature, there was a significant correlation between the change in skin temperature locally and the anti-anginal effect, but none with the change in skin temperature distantly. These findings supported the suggestion that the effects observed during the neurophysiological trial were due to acupuncture rather than to bias from the acupuncturists. Furthermore, the findings suggested that a mutual mechanism was underlying the anti-anginal effect of acupuncture and the local increase in skin temperature. One such mechanism could be a decrease in sympathetic tone.

Although acupuncture was found to increase pain thresholds locally, this effect was not significantly correlated with the anti-anginal effect, and accordingly an acupuncture-induced increase in pain threshold and pain tolerance threshold was not found to play any significant role in the anti-anginal effect. To test the validity of these findings, the change in exercise tolerance during exercise tests was related to the change in length of time for which myocardial ischaemia was noted, and to the change in myocardial oxygen consumption. The former was measured as duration of ST-depression on ECG thus indicating change in pain threshold, and the latter was measured as change in Delta PRP (systolic blood-pressure heart-rate product). The correlation was significant only for the increase in Delta PRP and it was found to be significantly greater than that for time with ST-depression. Accordingly, these findings suggest that the anti-anginal effect may be due to positive haemodynamic alteration rather than to an increase in pain threshold.

The correlation between anti-anginal effect and change in skin temperature on the index finger was found to be significant, both for the group receiving non-standard acupuncture and for that receiving traditional acupuncture. Accordingly, both treatments were taken to have a specific effect.

Conclusions

Healthy subjects

From a methodological point of view, the situation is more simple than when testing the effect in diseased persons. Using healthy individuals with no past acupuncture experience, the influence from subject expectation was eliminated. In this trial (Ballegaard et al. 1993) the effect of acupuncture was compared with the effect of a placebo pill in a randomised crossover design.

Participants were told that the effect of the pill was expected to be the same as that of acupuncture, but the exact effects were not suggested, as this could induce certain expectations and thus influence the result. The randomised crossover design helped to eliminate a possible influence of bias from the observer. Furthermore, the observer was separated from both the subject and the acupuncturist by a curtain, all measurements were performed by automatic machinery, and there was no verbal contact between subject and observer or between subject and acupuncturist. As the electric stimulator produced an unavoidable noise, it was turned on throughout all sessions. The acupuncturist stayed with the subject during the entire test procedure and always performed the same physical movements, including finger touch, whether the subject was having a placebo pill or acupuncture. The influence of the acupuncturist's expectation was eliminated by the hypothesis of the study: the effect of acupuncture being determined by the pre-treatment physiological state of the subject. As this information was not available to the acupuncturist, there was no chance of anticipating any direction of response of that particular person on that particular day. Similarly, there was no possibility of the subject developing a conditioning response.

There is a risk that an observed modulating effect of acupuncture was reflecting the well-known artefact: regression towards the mean. This was eliminated by comparing the effect of acupuncture to that of a placebo pill, which in this trial represented the natural regression towards the mean. For acupuncture treatment the points Hegu (LI.4) and Shousanli (LI.10) were used bilaterally. The needles were inserted perpendicularly to the skin into the underlying muscle to a depth of approximately 5mm. After obtaining needling sensation, the needles were stimulated electrically at 1.6Hz for 20 minutes at a level of intensity sufficient to produce visible muscle movements, but well below the pain threshold. The anode was connected to point Hegu.

It was found that compared with placebo, during treatment acupuncture had a significantly greater homoeostatic power in affecting local skin blood-flow and systolic blood-pressure heart-rate product (PRP), which is a measure of myocardial oxygen consumption, while heart rate differences were insignificant. During the 30 minute post-treatment observation period, the difference was significant as to local skin blood-flow only.

From this background, it was concluded that needling had an enhancing effect on existing homoeostatic mechanisms of myocardial oxygen consumption and local skin blood flow. The findings support the result of the angina pectoris study (Ballegaard et al. 1995), suggesting a significant effect of acupuncture on myocardial oxygen consumption/supply ratio, measured as PRP, and sympathetic tone, measured as local skin temperature (skin blood flow).

The evaluation of acupuncture in daily life

Angina pectoris patients

In order to obtain a fair estimate for the size and duration of the effect of the acupuncture treatment complex in daily life, 69 consecutive patients with advanced angina pectoris were followed for up to 2 years after treatment. The patients received 12 acupuncture treatments over a four-week period, given according to traditional Chinese theory. The needles were inserted with the patient in a supine position and, after obtaining needling sensation, were then left in place for 20min. No electrical or mechanical stimulation was given. The patients were instructed to perform acupressure twice daily on the middle of the sternum at the level of the fourth intercostal space: Shanzhong (CV.17) and, if possible, aided by the spouse, on the back between the shoulder blades 1.5in lateral to the spinal processes of the fourth and fifth thoracic vertebrae: Jueyinshu (BL.14) and Xinshu (BL.15). Furthermore, the patients were informed about adjustments of lifestyle and attitudes, stress-coping techniques, daily relaxation exercise, daily physical exercise, diets rich in potatoes, vegetables, fruit, bread, nuts, fish, garlic, olive oil and a moderate intake of red wine (Ballegaard et al. 1996).

Among the 69 patients, 49 were candidates for coronary artery bypass grafting (CABG), while bypass had been rejected in the remaining 20 patients. We compared our endpoint findings with those of a large, prospective, randomised trial comparing CABG with percutaneous transluminal coronary angioplasty (PTCA) (King et al. 1994).

During the 24 months observation period the incidence of death or myocardial infarction was 21% among the patients undergoing CABG, 15% of those having PTCA and 7% for our patients. No significant difference was found in pain relief between the three groups. An invasive treatment was postponed in 61% of our patients owing to clinical improvement, and the annual number of in-hospital days was reduced by 90%, bringing about an estimated saving of US$12,000 for each of our patients. The result suggests that the combined effect of acupuncture, acupressure and lifestyle adjustments may be highly cost-effective for patients with advanced angina pectoris.

Conclusions

In order to get an understanding of the underlying physiological mechanism of needling in the treatment of a disease, it may be worthwhile to study the effect in healthy persons of physiological variables that are related to that disease. In doing so, the presented studies suggest that needling enhances the existing homoeostatic mechanisms influencing the myocardial oxygen consumption/demand ratio and the local tone of the sympathetic nervous system, and that this effect helps the patient with angina pectoris by increasing the working capacity of the heart. Consequently, the patients experience fewer anginal attacks during their daily lives.

To provide an evaluation method for potential clinical applications of acupuncture, it is suggested that quality controls may be useful, as this design enables measurement of the effect in a situation equivalent to normal clinical practice. In doing so, the presented study indicates that acupuncture, acupressure and lifestyle adjustments applied according to classical Chinese philosophy may be cost-beneficial for patients with advanced angina pectoris.

Does this work help to answer the puzzle presented in the introduction: Why are one third of the patients not healed, when impaired myocardial blood supply is restored by a coronary artery by-pass operation? It may be that, apart from epicardial blood supply, the pathogenesis of angina pectoris is influenced by stimulations of sensory, emotional, social and psychological origin. Furthermore, a treatment strategy addressing these pathogenic aspects seems to be equally as efficient as highly technological, invasive procedures to restore epicardial blood supply. These findings indicate that further research in this direction may be helpful.

Søren Ballegaard MD
Acupuncture Centre, Lemchesvej 1
DK-2900 Hellerup, Denmark

References