Seven RCTs (393 patients) were included.
Study quality.
Four of the seven RCTs were rated as low quality. The review found no association between trial quality and results.
Short-term results.
Acupuncture versus waiting list (2 low-quality RCTs): there was limited evidence that acupuncture was more effective than control for pain and function.
Acupuncture versus sham acupuncture (3 high-quality RCTs): there was strong evidence that acupuncture improved knee pain compared with sham acupuncture. There was inconclusive evidence for function. The RCTs did not assess the other outcomes.
Acupuncture versus physical therapy (2 low-quality RCTs): there was inconclusive evidence for acupuncture compared with physical therapy in terms of pain and function. The RCTs did not assess the other outcomes.
Adverse effects (2 RCTs): the adverse effects of acupuncture were considered to be relatively mild.
A short duration of disease was associated with a positive outcome for acupuncture.
Acupuncture treatment rating: few RCTs adhered to the guideline criteria for acupuncture treatment described in textbooks. Four of the five RCTs considered to use adequate acupuncture treatment found that acupuncture significantly improved pain compared with control.
Follow-up.
Intermediate follow-up (4 RCTs): two RCTs found that the benefits of acupuncture lessened after one month. One RCT found that the benefit of acupuncture was maintained at 3 months. The results for the fourth RCT were not reported.
Long-term follow-up: none of the RCTs followed all the enrolled patients beyond 3 months.