Expectancy in real and sham electroacupuncture: does believing make it so?

BACKGROUND The large placebo effect observed in prior acupuncture trials presents a substantial challenge for interpretation of the efficacy of acupuncture. We sought to evaluate the relationship between response expectancy, a key component of the placebo effect over time, and treatment outcome in real and sham electroacupuncture (EA). METHODS We analyzed data from a randomized controlled trial of EA and sham acupuncture (SA) for joint pain attributable to aromatase inhibitors among women with breast cancer. Responders were identified using the Patient Global Impression of Change instrument at Week 8 (end of intervention). The Acupuncture Expectancy Scale (AES) was used to measure expectancy four times during the trial. Linear mixed-effects models were used to evaluate the association between expectancy and treatment response. RESULTS In the wait list control group, AES remained unchanged over treatment. In the SA group, Baseline AES was significantly higher in responders than nonresponders (15.5 vs 12.1, P = .005) and AES did not change over time. In the EA group, Baseline AES scores did not differ between responders and nonresponders (14.8 vs 15.3, P = .64); however, AES increased in responders compared with nonresponders over time (P = .004 for responder and time interaction term) with significant difference at the end of trial for responders versus nonresponders (16.2 vs 11.7, P = .004). CONCLUSIONS Baseline higher response expectancy predicts treatment response in SA, but not in EA. Divergent mechanisms may exist for how SA and EA influence pain outcomes, and patients with low expectancy may do better with EA than SA.