Am Fam Physician. 2001;64(11):1886
Placebo treatments have been reported to help patients with a wide range of conditions, including pain, asthma, and high blood pressure. The quality of evidence supporting these findings has not been thoroughly evaluated. The majority of reports on placebo have estimated the effect of the placebo as the difference from the baseline in the condition of patients in the placebo group of a randomized trial after treatment, and the effects cannot be distinguished from the natural cause of the disease and other factors. Hróbjartsson and Gøtzsche systematically reviewed clinical trials comparing placebo with no treatment.
The authors searched the American and European medical literature for trials that contained placebo and no-treatment groups, often as part of a study with a third group comparing an active treatment. Of the 727 potentially eligible trials that had data for placebo and no treatment, 114 investigations met the meta-analysis inclusion criteria of effective randomization, blinded assessment of outcomes, no excessive dropout rates, and no paid or asymptomatic volunteers. Only the methods section of the article was used for obtaining outcome data, to avoid subjective interpretations within the article. Most of the studies used for the meta-analysis (88 of 114) were not specifically designed to compare placebo with no treatment.
Trials were divided into binary and continuous outcomes. The median number of patients in each trial was small (27 patients in continuous outcome studies and 51 patients in binary studies), and a wide range of placebo effectiveness was noted.
When the data from each of the trials were pooled, there was no evidence of a beneficial effect of placebo in studies with binary outcomes, or in continuous outcome trials that used objective scoring criteria. There was a positive effect for placebo in continuous outcome trials with subjective assessment criteria (especially treatment of chronic pain), but this effect diminished with increasing sample size, leading the authors to suspect biased assessment in smaller, subjective studies.
The authors conclude that there was no obvious beneficial effect to placebo versus no treatment, and discouraged the use of placebo treatment outside the setting of a controlled clinical trial.