Methodological challenges in design and conduct of randomised controlled trials in acupuncture

Complex intervention

Methods to locate acupuncture points vary. In 2008, the World Health Organization regional office for the Western Pacific published a manual of standard acupuncture point locations,6 which has been well adopted. However, there are other acupoint location systems which vary somewhat from the WHO standards but are esteemed by specific schools of acupuncturists, such as Yang Jiasan’s acupoint system.7

Acupuncture theory varies, contributing to differences in treatment regimens. Practitioners using traditional Chinese medicine theory to guide acupoint selection might prefer to use meridian based or symptom based acupoint selection methods. Western medical acupuncture bases its treatment on sensory stimulation. As an acute pain stimulus, acupuncture can elicit a euphoric feeling by increasing release of opioid peptides (such as endorphins and dynorphins), thereby alleviating ongoing pain.8 Acupuncture stimulation serves the function of a stimulus and can achieve muscle reflex relaxation.9 To treat musculoskeletal pain,10 most western practitioners in primary care, rheumatology, orthopaedic, and pain clinics will likely choose Ashi or trigger points (box 1).

Current evidence for acupoint specificity varies and complicates selection of treatment regimens. Acupoint specificity refers to different treatment effects that the selection of acupoints brings. A randomised controlled trial investigated acupuncture as an adjunct to antianginal drug therapy for chronic stable angina. The study showed that stimulating acupoints on the disease affected meridians resulted in four fewer angina attacks than stimulating acupoints on non-affected meridians (95% confidence interval −2.43 to −5.71) during week 13 to 16.15 However, individual patient data meta-analysis on acupuncture for chronic pain showed no significant association between pain relief and the acupoint prescription (fixed, flexible, and individualised prescription).16

Acupuncture treatment dosage can differ significantly and impact the treatment effect.17 Duration, intensity, repetition intervals, and number of treatments and needles are important components of acupuncture dosage. Existing randomised trials of acupuncture have, however, either under-reported this critical information or used vastly different treatment dosages. For example, the latest Cochrane review18 addressing acupuncture for the treatment for depression found that only two thirds of the included randomised controlled trials reported needling duration, while in 63 included trials, treatment frequency varied from twice daily to once a week.

Existing evidence has explored the impact of acupuncture dose on treatment effect but rarely addressed the possible interaction between dose components. An individual patient data meta-analysis of acupuncture for chronic pain found a significant association between pain relief and the number of acupuncture treatment sessions but not de qi (box 1), frequency of sessions, duration of sessions, and number of needles used.16 Other studies found a larger number of needles,19 higher intensity,20 and greater frequency of treatment 21 may increase treatment effects. For example, in a randomised controlled trial on knee arthritis, electroacupuncture (>2mA) with a strong current relieved more pain.22 When designing randomised controlled trials in acupuncture, large variability and limited research evidence make it challenging for trialists to determine the appropriate treatment dosage.

Needle manipulation technique may affect the magnitude of the treatment effect.23 The complex manipulation techniques include lifting, thrusting, twirling range, de qi sensation, and frequency of electrostimulation. Devices might help to detect and standardise the needle manipulation features but are not yet widely known and rarely used in trials.24

Importance of expertise

As a non-pharmacological intervention, the expertise of the practitioner who delivers acupuncture might influence its effectiveness. As recommended in the Revised Standards for Reporting Interventions in Clinical Trials of Acupuncture (STRICTA),25 trialists should report acupuncturists’ expertise.

The evidence regarding the impact of expertise is, however, conflicting. An individual patient data meta-analysis based on acupuncture trials for pain conditions included 39 trials enrolling 20 827 patients. Almost all participants (97.3%) lived in western countries. Of the 39 trials, 14 (36%) had no requirement for acupuncturists to have clinical experience, 20 trials (51%) required between six months and four years’ experience, and five trials (12.8%) at least five years’ clinical experience. Results revealed no apparent association between experience and the magnitude of the treatment effect.16 Between trials comparisons are, however, less robust than within trial comparisons, and a single trial not included in the meta-analysis reported that senior acupuncturists achieved a larger treatment effect than their juniors.26

Used as combination therapy and highly personalised

Acupuncture regimens often include conventional treatments and multiple acupoint stimulation therapies, such as manual or electro-acupuncture or body acupuncture plus auricular therapy. With regard to personalisation in determining treatment, the acupuncturist in traditional Chinese medicine considers the patient’s syndrome pattern differentiation or meridian or body constitution (box 1). Patients with the same disease diagnosed by western medicine may show different syndrome patterns; acupuncture treatments vary accordingly.27

In clinical trials, many researchers prefer fixed acupuncture treatment protocols that neglect syndrome patterns or semi-fixed treatment protocols based on simplified pattern differentiation, rather than complex real world treatment packages of multiple acupuncture interventions, to reduce the difficulty of treatment quality control. This simplification of the intervention in clinical trials may provide only indirect evidence for treatments actually used in practice, possibly resulting in an underestimation of acupuncture’s treatment effect as administered in (at least eastern) clinical practice.

Importance of patient-practitioner interaction

The process of establishing a confident and trusting relationship can be critical to an intervention’s impact.429 Patients have identified a contrast in their relationship between western biomedicine practitioners and acupuncturists.4 Traditional Chinese medicine diagnostic and management procedures include listening to patients speaking about their health condition and attention to patients’ feeling, mood, diet, rest, physical exercise, touching of the skin, and peaceful time spent together; each communicates that the clinician’s knowledge of their particular experience is essential. The process tends to build a close relationship between acupuncturists and patients.30

Participants’ expectations can modify treatment effect

Eight hundred and sixty four patients participating in four different trials investigating migraine, tension-type headache, chronic low back pain, and knee osteoarthritis were randomised to eight weeks of acupuncture treatment or sham acupuncture (superficial needling of non-acupoints).31 After adjustment for confounders including intervention group, age, sex, and duration of complaints, in comparison with baseline patients in both standard and sham acupuncture groups with a high expectation of benefit proved more likely to achieve a reduction in pain of more than 50% than those with a low expectation (odds ratio of high to low expectation 1.67, 95% confidence interval 1.20-2.32).31 Results are consistent with neuroscience research suggesting that anti-analgesic expectancy could block analgesia procedures and reduce pain relief.32 Trialists are increasingly making use of participants’ expectations measures, such as the acupuncture expectancy scale, a validated instrument that measures the benefits patients perceive they would gain from acupuncture.33

Prior expectations towards acupuncture may also influence patient enrolment and retention in clinical trials. Patients who hold a strong belief in or against acupuncture treatment may decline participation, either afraid of being allocated to the sham group or unwilling to accept acupuncture treatment. If they do enter a trial, their prior beliefs may influence compliance and trial retention.34

Sham acupuncture is problematic

Aside from specific effects, an intervention may also have non-specific effects. Observation, assessment, and therapeutic rituals may elicit non-specific effects.35 Trialists often employ a placebo control with blinding of patients to differentiate the non-specific effect of the target intervention. There are, however, challenges in using placebo control in acupuncture trials.

To achieve blinding, trialists have used many types of sham acupuncture. Sham acupuncture without skin penetration, such as the retractable blunt needle, is one of the earliest sham acupuncture inventions.36 Tube needles can hide the entire treatment process from view, but the procedure neither allows the angled insertion or manipulation technique after insertion nor applies to all acupoints (for instance, points on fingers and scalp).37 More importantly, it might not adequately blind the participants, particularly those who have experienced previous acupuncture treatment.38 Sham acupuncture may therefore aim to simulate the experience of optimal acupuncture more closely by more shallow insertion of needles in suitable acupoints; not undertaking subsequent manipulation; selecting acupoints in the relevant meridians but not specific to the health condition; selecting acupoints in irrelevant meridians; selecting skin areas close to but differing from the optimal acupoints; or some combination.3940 Though the acupuncturist will remain unblinded, this can blind patients more successfully.

However, a serious problem exists with sham acupuncture. Evidence suggests needle insertion or even skin touching creates specific effects.41 Thus, the effects of sham acupuncture may be similar to—though presumably lesser—optimally performed acupuncture. Additional evidence comes from the work of Harris et al, who found that acupuncture increases central μ-opioid receptor binding potential, while sham acupuncture reduces or does not change it,42 and from the work of Zucker et al, who reported a larger pain reduction when using sham over real acupuncture in fibromyalgia patients with lower pain thresholds.43 Furthermore, specific effects of different types of sham stimulation may differ: penetrating needle sham, in comparison with other shams, showed the largest chronic pain relief.16 The crucial message from this work is that trials using sham acupuncture may underestimate acupuncture’s treatment effect.