Placebos Have Side Effects Too

The placebo is the most talked-about treatment in medicine.

Everyone's heard of the "placebo effect", by which pills containing no drugs at all, just chalk and sugar, often seem to make people feel better. But if the mere expectation of improvement can produce improvement, then the expectation of unpleasant consequences, such as side effects, should make people feel worse. This is sometimes called the "nocebo" effect.

Two recently published papers tried to measure it. They looked at people who took part in randomized controlled trials of various drugs, and who were given placebos. Because different drugs have different known side effects, if the nocebo effect is real, the side effects reported by the placebo group should depend on the drug they think they might be taking. As the authors of one of the papers put it:

In a typical clinical trial, the subjects know they can receive either the active medication or the placebo and, accordingly, they are informed about the possible adverse events they may experience during the trial. ... Therefore, informing subjects about the possible adverse events they may experience, may have a significant impact on their expectations and experiences of negative effects.

Accordingly, Rief et al compared the side effects reported in the placebo groups of a large number of antidepressant drug trials. At the same time a separate group of researchers, Amanzio et al, did the same thing for trials of migraine drugs, which is a nice coincidence.

Both papers found that reported side effects do indeed depend on the drug being studied. In the antidepressant paper, people who believed they might be on tricyclic antidepressants (TCAs) reported many more "side effects" than those in trials of SSRIs. These included dry mouth, drowsiness, constipation, and sexual problems. This makes sense, because TCAs do have worse side effects than SSRIs.

Likewise, for the migraine trials, the placebo groups in trials of anticonvulsants reported more symptoms associated with those drugs, such as dizziness and sleepiness. Placebo groups in trials of NSAIDs (like aspirin) were more likely to report upset stomachs and so forth. Finally, in trials of triptans, which have very mild side effects, the placebo group reported few problems.

It's also interesting to compare the two papers. None of the migraine trial placebo patients reported experiencing sexual problems, while many of the antidepressant placebo patients did. Some antidepressants can cause sexual problems, while migraine drugs generally don't.

So, was the "nocebo effect" really making people feel worse? It could well have been, although there are other interpretations. People might just be more willing to report symptoms that they believe are drug side effects. Researchers might be more likely to write them down. And different kinds of people end up in trials of different drugs: some people might be more likely to report certain symptoms. Just as with placebos, we shouldn't rush to ascribe incredible mind-over-matter powers to the "force of suggestion" when there are more prosaic explanations.

Nevertheless, there's an important lesson here. Anecdotal evidence about drug's side effects shouldn't be accepted at face value, any more than anecdotes about their benefits. Drugs do, of course, cause adverse effects. But some drugs have worse reputations than they deserve in this regard. In such cases, nocebo effects might account for some of the reported problems...


Rief W, Nestoriuc Y, von Lilienfeld-Toal A, Dogan I, Schreiber F, Hofmann SG, Barsky AJ, & Avorn J (2009). Differences in Adverse Effect Reporting in Placebo Groups in SSRI and Tricyclic Antidepressant Trials: A Systematic Review and Meta-Analysis. Drug safety : an international journal of medical toxicology and drug experience, 32 (11), 1041-56 PMID: 19810776


Amanzio M, Corazzini LL, Vase L, & Benedetti F (2009). A systematic review of adverse events in placebo groups of anti-migraine clinical trials. Pain PMID: 19781854