Thirteen RCTs were included (866 patients).
The trials were disparate in terms of trial design, patient groups, needling points, acupuncture intervention, and how and when pain outcomes were measured.
The majority of trials were of poor quality and trial findings were often compromised by unclear or incorrect methods. A number of trials did not standardise rescue analgesia.
Acute pain.
1. Multiple acupuncture sessions (2 double-blind RCTs, 66 patients with acute back pain).
One RCT reported that acupuncture was significantly better than sham acupuncture for some measures only and concluded that there was an overall benefit for acupuncture. The review's authors disagreed with the original authors' conclusion for the following reasons: for the most clinically-relevant outcome (pain intensity at rest using VAS scale) there was no significant difference between treatment groups; and the authors reported significant groups differences for 3 out of the 12 measures with no corrections for multiple testing. OPVS score 11.
The other RCT concluded there was no significant difference between treatment groups but the outcome was not defined clearly. OPVS 10.
Chronic pain.
1. Single acupuncture session (2 single-blind RCTs, 100 patients with chronic neck and back pain). Inconsistent results were reported with 1 RCT reporting significant benefit for laser acupuncture and the other reporting no significant difference between treatment groups, OPVS score of 10 and 9.
2. Multiple acupuncture sessions (9 RCTs, 371 patients with chronic back and neck pain, including 2 double-blind RCTs, 1 single-blind RCT and 6 non-blind RCTs).
The double-blind RCTs both reported no significant differences between treatment groups. Problems included high drop-out rates, lack of intention to treat analysis, outcomes rated by physician and not by the patient, and outcomes not defined clearly. OPVS scores 14 and 8. The single-blind RCT reported no significant difference. OPVS score 8.
Problems with the non-blinded RCTs included: poorly-designed control group, use of physician rating, lack of statistical tests, and lack of definition of 'improvement'. Results were inconsistent among studies.
OPVS scores.
Trials had validity scores ranging from 4 to 14 (out of a maximum score of 16).
Most common problems were small group sizes and lack of blinding. Other flaws included insufficient baseline pain, lack of internal sensitivity, and the use either of inappropriate statistical techniques or no statistics. The authors of the review disagreed with the original authors' conclusions in two reports.
Relationship between the original authors' conclusion (positive or negative) and the validity score: no significant relationship was found (Mann-Whitney U-test=15, p=0.46).
Relationship between the review authors' conclusion and validity score: more valid trials were found to be significantly more likely to have a negative conclusion (U=4.5, p=0.023).
Adverse effects (3 RCTs). Adverse effects were reported by 3 acupuncture patients in one trial (increased pain at acupuncture site in 2 patients and fever in 1 patient) and 1 sham TENS (transcutaneous electrical nerve stimulation) patient in one trial. The third RCT reported no adverse effects.