Can Expecting Side Effects Cause Them? Apparently Yes
By the BrainSnail editorial team. How these articles are written and checked, and how to tell us when one is wrong.
People given an inert substance and told to expect unpleasant effects report them at substantial rates, which is the mirror image of the better known placebo response. It creates a genuine problem for how medicine communicates risk.
What the effect is
The placebo response describes improvement following an inert treatment, attributed to expectation, context and the ritual of being treated. The same mechanisms operate in the opposite direction, so a person expecting harm reports harm, and the effect has been demonstrated repeatedly under controlled conditions. Participants given inert substances described as causing headache report headaches. Participants warned about a side effect report that specific side effect rather than a general malaise. The responses are not confined to what somebody says, since measurable changes in pain sensitivity, heart rate, hormone levels and brain activity accompany them, which distinguishes the effect from simple complaining.
How it is demonstrated
Several experimental designs produce it reliably:
- •Telling participants an inert substance commonly causes a particular symptom
- •Applying a harmless stimulus described as painful, which is then rated as painful
- •Telling people a device emits something harmful when it emits nothing
- •Comparing groups given identical treatment with different warnings
- •Observing that side effects in trials match those listed on the information sheet
- •Stopping an effective drug and warning of withdrawal, which produces withdrawal
Where it shows up in practice
The effect appears in ordinary medicine with measurable consequences. Placebo arms of drug trials report side effects at substantial rates, frequently matching the specific effects listed for the active drug, which indicates that a large share of reported side effects in the active arm are not caused by the drug either. Analyses of statin trials have found that reported muscle pain occurred at nearly the same rate in placebo and active groups, and a trial in which participants alternated between statin, placebo and nothing found symptoms in the placebo periods almost equal to those in the drug periods. Switching patients to a generic version of a drug produces a rise in reported side effects even when the formulation is identical.
The mass episodes
Expectation can spread through a group and produce symptoms in numbers, which is documented repeatedly and which is genuinely difficult to manage. Episodes typically begin with one person becoming unwell, frequently for an identifiable reason, after which others nearby develop comparable symptoms in the absence of any physical cause that investigation can find. Schools, workplaces and communities are the usual settings. Reported symptoms are real and are typically headache, nausea, dizziness and breathlessness rather than anything requiring an unusual mechanism. Investigation must take the possibility of a genuine exposure seriously first, since dismissing a real chemical release as psychological has happened and is indefensible, and only then consider the alternative.
The dilemma it creates
Informed consent requires telling people what might happen, and telling them makes it more likely to happen, which is a genuine ethical conflict without a clean resolution. Withholding information is not acceptable, since people have a right to know and since undisclosed effects destroy trust when they occur. Proposals that are being tested include framing information positively, so that a figure is given as the proportion who do not experience an effect rather than the proportion who do, and offering patients a choice about how much detail they receive. Evidence that positive framing reduces reported effects without reducing understanding is accumulating, and the approach remains contested precisely because it involves choosing how to present a true fact.
The takeaway
Expectation of harm produces reported harm, along with measurable physiological changes, which is the mirror image of the placebo response. Placebo arms of trials report the specific side effects listed for the active drug, and statin trials have found muscle pain at nearly equal rates in both arms. Informed consent requires disclosure and disclosure increases the effect, which is a conflict addressed by framing rather than resolved.