Ahhh, acupuncture for pain. Great big messy topic. :-)
The short answer is that the best studies in terms of design, controls for bias, etc. suggest it is a placebo. This means it changes people’s perception of their discomfort so that when asked they rate their pain as lower (usually ~15% or so, though it varies a lot) on pain scales, though nowhere near the degree of effect most pain control drugs exhibit. It does not appear to have any measurable effect on the actual disease process for osteoarthritis (as this woman probably has). The analagesic effect is the same for “fake” acupuncture given at places not considered acupuncture points by some acupuncturists (though there are some acupuncturists who don’t believe in points anyway), and for essentially random pricking of the skin with toothpicks or any other mild local irritant.
Since pain is inherently subjective, this benefit is “real,” but again it doesn’t seem to be significant in terms of measurable indices of disease or as strong as the benefits of medication. The process is pretty close to harmless (a few reports of infections, hepatitis, and the sorts of things that you get from breaking the skin with foreign objects), so I personally have no objection to people doing it so long as it doesn’t interfere with pursuing therapies that really affect the disease (such as the total hip replacement). The problems come up when people begin to believe that the effect on their perception of discomfort is somehow evidence supporting all the metaphysical claims and other purported benefits for acupuncture. Apart from pain, nausea, anacupuncture pretty consistently fails to show any real benefits in good studies.
Of course, the real answer is much longer and more nuanced. There is a lot of bandying about of terms like “endorphins” with regard to acupucture, and there’s no question that poking needles into people causes all kinds of chemicals to appear in the blood or tissues. the problem is translating this into a clinically significant therapeutic effect, and after about 40 years of pretty voluminous research, we haven’t been able to do so, which is part of why I think the placebo answer is the most likely. Really good, detailed discussions of this can be found in:
Trick or Treatment: The Undeniable Truth about Alternative Medicine by Edzard Ernst and Simon Singh
Snake Oil Science by R. Barker Bausell
You can also check out the topic reference section at Science-Based Medicine