Placebo

A placebo is one way trials keep people unaware of who got what. The CONSORT reporting guideline notes that random allocation can help hide which treatment each person received, possibly by using a placebo, which reduces bias after assignment [1]. For that to work, the placebo has to be hard to tell apart from the real thing. One trial report quoted in CONSORT describes a placebo matched to the study product for taste, color and size, made of fillers such as microcrystalline cellulose [1]. Another describes capsules identical in appearance [1].

The placebo effect is a separate question: how much people change just from receiving a placebo. A Cochrane review measured it using randomized trials that had both a placebo group and a no-treatment group [2]. With outcome data from 202 trials across 60 clinical conditions, the authors did not find that placebo interventions have important clinical effects in general [2]. In certain settings placebo could influence patient-reported outcomes, especially pain and nausea, although the authors note this is hard to separate from biased reporting [2]. These were clinical populations, and results may not apply to healthy adults.

This may matter for outcomes that people rate themselves, such as mood or alertness. In the same review, the pooled placebo effect was larger for patient-reported outcomes (standardized mean difference -0.26) than for observer-reported outcomes (-0.13), where a negative value favored placebo over no treatment [2]. Larger placebo effects were also linked to physical placebos such as sham acupuncture, small trials, and trials designed to study placebo [2].

That is why it helps to check what each result was compared with.

• Randomized controlled trial: the design in which a placebo group is most often used.
• Double-blind: keeping people unaware of who got the placebo.
• Open-label: a trial in which participants know their assignment.