Think about the last thirty seconds of a memory clinic appointment. The diagnosis has been said out loud, the scan has been discussed, and now there is a prescription being slid across the desk with a short sentence attached to it. Something like: about one in ten people get nausea on this. That sentence is true, it is required, and it is also, according to a review published in The Lancet Psychiatry in May 2026, an active ingredient. Not a metaphor for one. Burke and nine co-authors, among them Dartmouth's Tor Wager and the placebo researcher Ted Kaptchuk, argue that the words and the ritual wrapped around a treatment do measurable things to how that treatment is experienced, and that medicine has spent decades treating this as noise to be subtracted in trials rather than as something that could be used, carefully and without lying, at the bedside.
What the review is, precisely, matters. It is a narrative review, paper 1 of 2 in a Lancet Psychiatry collection on reconceptualising placebo and nocebo effects, running from pages 413 to 425 of volume 13. It reports no new patients, no new randomisation, no new outcome data. What it does is sort the existing literature into a menu of strategies and then ask, of each one, whether a patient could be told the whole truth and still benefit. The menu includes traditional pure and impure placebos, authorised deception, honest open-label placebo (an inert pill given with full disclosure that it is inert), reducing the dose of a real drug by pairing it with a conditioned placebo, nocebo education, nocebo reframing, shifting mindsets, and simply enhancing the therapeutic context around a treatment that already works. The authors also set out the mechanistic scaffolding beneath all this, including open-hidden paradigms, where the same drug is given with or without the patient knowing, and animal work, such as a report of placebo analgesia in mice traced to a specific brain pathway. That last item is mouse work. The review presents it as mechanism, and it should be read as nothing more.
The half of the paper that will reach a clinic soonest is the nocebo half, because it costs nothing and changes only grammar. The review's own worked example is the one in the table: rather than telling someone that 10% of people get a side effect, tell them that "90% of people do not experience side-effects". The risk is identical. The expectation planted is not. Alongside it sits an adapted blinded placebo run-in, starting treatment with a placebo phase and then escalating the active dose gradually, on the reasoning that anticipatory, expectation-driven reactions can be damped before the real drug arrives. And the authors are explicit about where the line falls. The same tables carry a column of approaches that work only without informed patient consent, including withholding information about the seriousness of a disease. Those are named, not recommended. The point of the exercise is to find what survives full disclosure.
For a family living with dementia, the relevance is not the sugar pill. It is the conversation. People with Alzheimer's disease, vascular dementia, Lewy body dementia or a mixed picture frequently arrive at the memory clinic already carrying several prescriptions, and the decision to keep or abandon a drug often turns on how the first fortnight felt and how it was described in advance. If the framing of a warning measurably shifts how many symptoms get reported, then the way a cholinesterase inhibitor or an antidepressant is introduced is not a soft skill, it is part of the intervention. That said, the honest caveat has to be loud: this review is addressed to psychiatry and medicine in general, not to dementia care in particular, and it does not present outcome data in people living with dementia. Nor does it resolve the question that would matter most here, which is what an expectation-based strategy means for someone whose capacity to hold a rationale in mind is changing.
The journal's own editorial, published in the same May issue, put the stakes plainly: some of the brain circuitry targeted by new treatments is also activated by placebos, and the ethics of placebos and consent remain, in its words, "vigorously debated". That is the uncomfortable shape of this literature. The effects are real enough to contaminate drug development and large enough that researchers now want to recruit them deliberately, yet most of the honest techniques have been tested in conditions like back pain, irritable bowel, fatigue and allergic rhinitis rather than in progressive neurodegenerative illness. Burke and colleagues are not claiming otherwise. They are claiming that the mechanistic work has matured to the point where clinicians should be given something to do with it, and that doing nothing has its own cost, paid in medications abandoned in week two for reasons nobody examined.