Thirteen RCTs (n=2,596) were included in the review, of which eight (n=2,340) were included in the meta analysis.
Six studies were described as high quality (scoring between 6 and 8 points) and two were described as low quality (scoring 3 points).
Acupuncture compared with sham acupuncture.
Acupuncture was significantly more effective than sham acupuncture for pain reduction in studies assessing short-term follow-up (5 RCTs, n=1,334; WMD 1.54, 95% CI: 0.49, 2.60, p=0.004). The analysis of combined high-quality only studies made no significant difference to the results. Statistical heterogeneity was high. However, after the removal of one outlying study the result remained statistically significant in favour of acupuncture, but the WMD changed to 0.87 (95% CI: 0.40, 1.34) and heterogeneity was removed. Acupuncture was also found to be more effective for pain reduction in studies assessing long-term follow-up (3 RCTs, n=1,178; WMD 0.54, 95% CI: 0.05, 1.04, p=0.03); there was no evidence of statistical heterogeneity.
Acupuncture was significantly more effective than sham acupuncture for improvement in function at the short-term follow-up (5 RCTs, n=1,333; WMD 4.32, 95% CI: 0.60, 8.05, p=0.02). The analysis of combined high-quality only studies made no significant difference to the results. Statistical heterogeneity was high. However, after the removal of one outlying study the result remained statistically significant in favour of acupuncture, but the WMD changed to 2.41 (95% CI: 0.60, 4.21) and heterogeneity was reduced. Acupuncture was also found to be more effective at the long-term follow-up of function (3 RCTs, n=1,178; WMD 2.01, 95% CI: 0.36, 3.66, p=0.02).
The results were consistent with the other included studies assessing pain and function which were excluded from the meta analysis.
Acupuncture compared with no additional treatment (usual care).
Acupuncture was significantly more effective than no additional treatment for studies assessing pain reduction at the short-term follow-up (4 RCTs, n=927; WMD 3.42, (95% CI: 2.58, 4.25, p<0.00001). There was evidence of moderate heterogeneity. Acupuncture was also found to be more effective in studies assessing function at the short-term follow-up (3 RCTs, n=907; WMD 11.65, 95% CI: 6.48, 16.81, p<0.00001), though evidence of statistical heterogeneity was high. The analysis of combined high-quality only studies made no significant difference to these results. The results were consistent with the one study not included in the meta-analysis.
Acupuncture compared with other interventions.
Acupuncture was significantly more effective than an education programme for reducing pain and improving function in one RCT (n=379)(data not reported).