This was a study done by UCL/University of Southampton jointly, published in the journal NeuroScience, using PET (positron emission tomography) scans to examine brain activity of arthritis sufferers undergoing acupuncture. 14 volunteers each had 3 kinds of treatment in random order: blunt needles, which didn't pierce the skin and which the volunteers realised wouldn't help them, collapsible trick needles which appeared to pierce the skin but didn't, and real acupuncture.
The 3 approaches produced 3 separate patterns of brain activity: the blunt needles just activated the sense of touch; the trick needles stimulated opiate (natural pain-killing activity), and so did the real acupuncture, which also activated a region of the brain known as the insular, thought to be involved in the modulation of pain.
So on the face of it, real acupuncture has a measurable effect on brain activity differing from needles which don't pierce the skin (even if the patient thinks they do). I'm not going into that here, as what interests me is the comment at the bottom of the article from Professor Henry McQuay, professor of pain relief at the University of Oxford: '...it is extremely difficult, technically, to study acupuncture and tease out the placebo effect." That makes perfect sense, as randomised controlled trials are an attempt (among other things) to remove or reduce individual variation in measuring the effectiveness of an outcome.
Now in Western medicine, that works, so far as it goes, because symptoms lead to a diagnosis of a particular illness or disease, and a drug, for instance, can be evaluated in terms of its effectiveness in curing that disease. You are trying to establish that a drug is safe and effective at standard dosages which can be systematised and referred to and applied across the country or the world. However, with acupuncture, homeopathy and many other forms of natural healing, it is primarily the current state of the individual that is evaluated.
With acupuncture the quality of the pulse and the appearance of the tongue are of particular importance; at the end of a session the acupuncturist may well check the pulse again to see whether it has balanced. In other words there is an attempt to bring the patient's body/organism back towards a state of balance or homeostasis. The aim, therefore, is to stimulate the body into rejecting the illness and possibly strengthen the immune system; merely a reduction of stress alone would in many cases go some way to accomplishing that.
So to say whether or not acupuncture 'works' is to answer the wrong question. Presumably no one would argue that it works any worse than placebo? In which case if you add in an element of relaxation, which acupuncture will undoubtedly produce in most people after the needles are in, there is already a therapeutic effect beyond the placebo effect, but one which isn't generally recognised as such.
A better question might be, for example, "How can we maximise the effectiveness of acupuncture (for example) treatment in terms of producing that relaxation response?" That is irrespective of any other benefit from the treatment.
Naturally, all these therapies are limited to stimulating only what the body can accomplish on its own, but in some cases, this can be near miraculous. I exclude herbs which, like drugs, affect the body chemistry from the outside. Herbs, however, can be prescribed on their energetic signature. In Western herbalism this is the tradition of vitalistic prescribing versus scientific phytotherapy, which is strongly grounded in biochemistry.
If we add in the vital factors of the skill, rapport and approach of the individual practitioner, that will make a huge difference on the outcome of treatment. Hence, applying the rules of drug evaluation to healing arts where much depends on the individual practioner and patient, is as ludicrous as evaluating the quality of a singer by pulling people off the street and seeing whether the singer performing live had a greater than average emotional impact on the audience compared to the singer miming along to a recording of themselves.