Placebo and Placebo Control

A placebo is an intervention with no active ingredient, matched to the test treatment in appearance, route, schedule and device so that participants cannot tell them apart. Its purpose in a trial is subtractive. The placebo arm absorbs everything that changes over the study period regardless of pharmacology, which includes the natural course of the condition, regression to the mean, the co-interventions everyone receives, the effect of being observed and measured, and genuine expectation.

For injectable peptides the control is a sham injection with matched diluent, needle and visit schedule, which is why blinding survives longer here than for an oral tablet. The magnitudes are instructive. In the 68-week STEP 1 obesity trial the placebo arm lost roughly 2 percent of body weight on the same lifestyle programme, while in analgesia and depression trials the placebo response can approach half the response seen on active drug. Placebo arms also generate adverse events at substantial rates, the nocebo counterpart.

Only the difference between arms is attributable to the drug, so the placebo arm effectively defines the effect size that gets quoted. Whether a placebo control is even permissible is an ethical question rather than a statistical one: where an effective treatment exists, withholding it is usually unacceptable and an active comparator or add-on design is used instead, which changes what the trial can conclude.

The phrase placebo effect is routinely stretched to mean the whole improvement observed in a placebo arm, most of which is not psychology at all but natural history and measurement artefact. The other common move is describing an uncontrolled before-and-after study as placebo-controlled, or dismissing an inconvenient trial result as just placebo when the trial contained a placebo arm and reported the difference.