PLACEBO EFFECT OR THERAPEUTIC EFFECT?
A META-ANALYSIS OF ACUPUNCTURE FOR CHRONIC PAIN

Shu ZHANG1, Nan M. LAIRD1, Roger B. DAVIS2
1Department of Biostatistics, Harvard School of Public Health.
2Division of General Medicine and Primary Care, Beth Israel Hospital.

Objective: To evaluate the effect of acupuncture in treating chronic pain, by using randomized, controlled, mostly single-blind(patient blind)studies.
Data Sources: Articles published in English after 1970, searched through MEDLINE and related references.
Study Selection: (a) Treatment: needle insertion into classic acupoints. (b) Control: sham acupuncture, waiting list, or mock TENS. (c) Study Design: parallel studies (two-group comparison) or first periods of cross-over studies. (d) Outcome Measure in Each Study: improvement rate after treatment(patient self-assessment).
Thirteen of 69 studies evaluated met selection criteria. Eight of the 13 used sham acupuncture as the control group, in which patients were blinded from the treatment.
Data Extraction: improvement was used directly if reported, otherwise was derived from the raw data.
Results: The overall treatment difference (improvement rate in acupuncture group improvement rate in control group) was 0.429 (0.324, 0.534) for all 13 studies combined (p<0.001), and 0.413 (0.287, 0.538) for the 8 patient-blind studies (p<0.001).
Conclusion: The key question for the effect of acupuncture was whether it was therapeutic or placebo effect. This issue was raised largely because many trials were unblind. Since the majority of the trials in this meta-analysis used sham acupuncture as the control group, credible patient blindness was provided, and thus the conclusions from this analysis provide convincing evidence to distinguish therapeutic effect from placebo effects. Our results strongly support the theory that acupuncture had a therapeutic effect in treating chronic pain.