How would that possibly work?

A double-blind study requires that both the practitioner and patient are blinded to the treatment. In this case, how would the practitioner not know that he/she was inserting a needle into a person's skin? Perhaps you could trick the patient into believing they have been injected with a needle, but I don't see how you'd trick the practitioner.

Yes, the Professor Linden is right.  But I think that it is possible to use placebo points as recommended by Equator (STRICTA) viable on http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3096500/

The so called "placebo", "sham" or "false" acupuncture points also have an effect on the patient if they are stimulated with a needle. The situation of acupuncture treatment can not be simulated  double blinded. If the person putting on the needles does not know where he/she is puncturing, the treatment is not acupuncture. To treat or not to treat - that´s the question. I´ve never seen an uneffective placebo for acupunture in the literature.

I think that it is possible. There is a method of punctuation as" KIM sham Acupuncture".

In this method there is a needle without the head that has not any penetration in the skin. SO in study there is a set of needles that neither practitioner nor patients know about the nature of needles. So just the coordinator of study knows who was punctuated by true needles and who by false needle. 

A stimulating and exciting methodological question!

A lot has been written about this over the past decades, for example the discussion paper on placebo controls for acupuncture studies by Vincent and Lewith in 1995 (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1295163/). In 2001, White and colleagues published consensus-based recommendations on the design of clinical trials of acupuncture (Complementary Therapies in Medicine, doi:10.1054/ctim.2001.0489), which include a short section on considerations with respect to blinding and design of placebo - these may give you some good starting points to delve further into the topic.

Also, as Katia has already referred to above, the STRICTA initiative has developed consensus-based standards for reporting interventions in controlled trials of acupuncture, first in 2001 (http://www.ncbi.nlm.nih.gov/pubmed/12184354). These were revised and updated in 2010 (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3002761/), and more recently further additions to STRICTA were suggested (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4474462/).

I think that there may be two wider aspects to consider: From the point of view of designing the intervention and placebo, a transparent description of both intervention and placebo to the same level of detail is crucial, so that both could be replicated, and this is also emphasised in STRICTA.

From a pragmatic clinical research point of view, I think that we should be mindful of the real world context we deliver services in; and that the inclusion of a non-treatment control group and/or a control group that receives non-acupuncture related treatment can add valuable insights.

There is certainly some ambivalence concerning a proper design of sham treatment: on one hand the trial protocol should reflect as closely as possible a commonly practiced clinical approach (which is often individualized to a particular patient), and on the other hand answer the requirements for proper control and blinding, and be reproducible as well.

Reproducibility could be achieved through adherence to STRICTA, as mentioned before, and there could be also an agreement as for the proper sham procedure. Blinding of the therapist, on the other hand is impossible for acupuncture, as it is for psychotherapy and physiotherapy. Therefore a reasonable study design could not be triple blind (of the patient, the therapist and the evaluator). Double blinding (of the patient and the evaluator) is however possible, and provided that treatment procedure is well documented (via STRICTA), it should do the job

I agree completely with Dorit. Double blinding is the real possibility.