Placebo effect was influenced by publication year in three-armed acupuncture trials

We investigated the relationship between the placebo effect and the trial or patient characteristics.

We identified randomized clinical trials with acupuncture, sham and no-treatment groups in which no-treatment or conventional therapy was applied to the no-treatment group. Thirty-one trials in which no treatment was applied in the no-treatment group were categorised as 'strict' trials. Thirty-nine trials in which no-treatment or conventional therapy was applied to the no-treatment group were categorised as 'less strict' trials. We calculated the treatment effect, defined as the difference in the effect size between the acupuncture and no-treatment groups, and the placebo effect, defined as the difference in the effect size between the sham and no-treatment groups. Then, a random effect meta-regression analysis was performed on the two effects with respect to trial or patient characteristics.

The treatment effect was not found to be associated with any factors in both the strict and less strict trials. However, the placebo effect was found to be associated with the publication year in both the strict and less strict trials (P=0.009 and 0.005, respectively). The placebo effect increased by 0.05 in effect size per year in trials that were published more recently.

While the treatment effect was not influenced by any trial or patient characteristics, the placebo effect was associated with the publication year.