Can a sugar pill really make you feel better?

Sometimes, yes, but far less often than the legend says. A placebo is a treatment made to look real with no pharmaceutical effect: a sugar pill, a saline injection, even a sham surgery. For pain and nausea, such fakes can genuinely help, and not just in your imagination. Since 1978 we have known that placebo pain relief depends on the brain releasing its own opioids, its homemade painkillers. For most other outcomes, though, careful studies find no important clinical effect.

The hard part is separating a real placebo effect from things that would have happened anyway. If you take something on a day you feel unusually awful, you will probably feel better later simply because an extreme measurement tends to be followed by a less extreme one. Statisticians call this regression to the mean. So a good trial compares a placebo group with a group that gets nothing at all. The change in the placebo group is the placebo response; only the difference from no treatment is the placebo effect.

Measured that way, the magic shrinks. A 1955 study called The Powerful Placebo convinced doctors that placebos were clinically important, but a 1997 reassessment found no evidence of any effect in its data, because it had ignored regression to the mean. A 2001 meta-analysis covering 40 medical conditions found a significant effect for only one of them: pain.

Where perception is involved, though, expectations are strong. More pills work better than fewer, capsules beat pills, injections beat capsules, and a kind, confident doctor makes a placebo work better. Fittingly, the word is Latin for "I shall be pleasing".