A randomized clinical study on acupuncture should be designed by the investigators with the involvement of a biostatistician to ensure the quality of the study.

Selection of the patient

The patients included in the study should represent the future group of patients to whom the findings of the study will be applied. The illness should be defined exactly. The source of patients recruited, and the criteria for patients’ inclusion and exclusion, should be considered carefully and stated in the protocol.

If, in the proposed study, acupuncture will be used based on a knowledge of traditional diagnosis, the patients should be selected according to diagnosis and differentiation of syndromes within traditional medicine. This should also be stated in the protocol.

Size of the study

The size of the study should be decided according to the requirements of accepted statistical analysis. Sufficient sample size will be needed in order to provide adequate statistical power to detect a clinically significant difference between two treatment groups.

Site(s) of investigation

Clinical research must be carried out under conditions which ensure adequate safety for the subjects. The site selected for the clinical research must have adequate facilities, including laboratories and equipment, where necessary, and sufficient clerical, medical and allied health workers to support the study as required.

Facilities should be available to meet possible emergencies.

A multicentre study may be necessary, and this may require a special administrative system to ensure that the study is conducted simultaneously and adequately at different sites by several investigators following the same protocol. The training of investigators from different sites to follow the same protocol and standardization of methods for selection of patients, termination of participation, administration, data collection and evaluation will be necessary.

Blind techniques

Blind techniques can be used in randomized controlled clinical trials. Blinding techniques may apply to patients, investigators and outcome assessor. Whenever possible, the patient should not be aware to which treatment group which he or she has been allocated. It is difficult for the investigator who gives acupuncture to the patient to be unaware of the treatment. It is essential that assessment should be made blind to treatment. The assessor should be responsible for practitioner and recording the details of reactions and results of the treatment received by patients. It is acknowledged that an unblinded therapist may affect the patient’s response.

Randomization

In clinical trials, there are two meanings of randomization. One is random sampling of the study population from the parent population. The other is random allocation which assigns the patient to one of the treatment groups by using a chance mechanism.

A randomized controlled (clinical) trial (RCT) is a study method which uses a random allocation method. Using this method comparability of groups will be maintained. Although an RCT is the most powerful method of reducing bias in the comparative evaluation of treatment options, it may become impractical when recruiting patients to some studies in the field of acupuncture, particularly when the patient has a strong preference for acupuncture treatment. In other words, the randomization mechanism may affect the outcome both positively and negatively.

Control groups

RCTs require one or more control groups for purposes of comparison. The control group may be (not in order of priority):

• mock TENS;
• sham acupuncture;
• non-treatment;
• standard therapy;
• real acupuncture; and
• minimal acupuncture.

The selection of the control group depends on the hypothesis being tested.

Crossover

Crossover studies are usually inappropriate in acupuncture. In acute self-limiting conditions, the natural resolution of the illness confounds tile concept of crossover technology. In chronic conditions acupuncture may work for a variable time after completion of treatment (days or years). A very prolonged washout period is required if a crossover model is to be adopted and this in itself poses ethical problems.

Strategic approaches to randomized controlled clinical trials

There are no established rules for systematically choosing the most appropriate controls in RCTs. The currently available scientific evidence suggests that the closer one gets to a purely endorphin-mediated effect, the less relevant it is to think in terms of point location and the more misleading a real versus sham comparison may be within an RCT. Conversely, the more that acupuncture treatment is autonomically mediated, such as in the management of non-painful conditions, the more relevant it may be to use a sham versus real acupuncture model when evaluating its clinical effectiveness (see Annex 1).