You feel better. Is the disease better?
“I feel better” is an important observation. It is also an unfinished sentence. Better in what way, compared with what, and for how long?
Those questions can sound unfriendly when someone is describing relief. Nobody wants a good afternoon cross-examined. But the distinction matters the moment a personal experience becomes a claim about a treatment. A calmer evening, a lower symptom score and a changed disease process are different things to establish. One deserves attention without automatically proving the others.
A trial with two very different results
In a 2011 asthma crossover pilot, 46 people were randomized; 39 completed repeated visits involving albuterol, a placebo inhaler, sham acupuncture or no intervention. Researchers measured airflow and asked about improvement. The two measures told different stories.
| Intervention | Mean maximum increase in FEV1 | Mean self-rated improvement |
|---|---|---|
| Albuterol | 20.1% | 50% |
| Placebo inhaler | 7.5% | 45% |
| Sham acupuncture | 7.3% | 46% |
| No intervention | 7.1% | 21% |
FEV1 measures air forcibly exhaled in one second. These are two-hour visit outcomes and different scales, not percentages of participants helped. Albuterol improved airflow more than the other conditions; subjective ratings did not significantly distinguish the three interventions. Source: Wechsler et al., Figures 3–4 and Results.
The useful question is therefore not which column to believe. It is which question each column can answer. A report about the experience of breathing cannot stand in for a lung-function measurement. Nor does a lung-function number contain the whole experience of living with symptoms.
A symptom is not an imaginary thing
“Subjective” is often used as a put-down. In research, it can simply mean that the information comes from the person experiencing it. A pain scale does not become worthless because a machine cannot fill it in. The FDA explicitly includes patient reports among clinical outcome assessments. What people feel and can do belongs in the evidence.
The catch is precision. “Participants reported less interference with daily life” and “the intervention changed the underlying disease” are different claims. The first does not need to impersonate the second to matter. By the same token, a biological measurement is not automatically the final answer: the FDA distinguishes clinical outcomes from surrogate endpoints, which need evidence to support what they predict.
This gives a reader a more useful habit than sorting studies into “real” and “psychological.” Name the outcome. Keep the claim the same size as the measurement.
What if everyone knows it is a placebo?
Open-label placebo studies make the question stranger. Participants are told that the pills lack an active drug ingredient. Can the act of taking them still change an outcome?
A trial published in October 2025 randomized 120 adults with migraine to usual care with or without open-label placebo for three months. It did not find a statistically significant advantage on the prespecified primary outcome: change in monthly headache days. Some secondary measures, including physical quality of life and pain-related disability, favored the placebo group.
That is a more specific result than “placebos treat migraine.” Participants knew their assignments, recruitment fell short of the planned 150, and the authors caution about interpreting secondary results after the primary result was nonsignificant. Headache days were recorded in diaries, too: this was not a clean division between subjective feelings and an objective laboratory test. Different patient-reported outcomes can diverge.
The practical reading lesson is about headlines. If an article announces “improvement,” look for the noun that follows it. An improvement in disability scores is not the same result as fewer headache days. A nonsignificant difference also does not prove the two approaches identical.
And sometimes the additional benefit does not appear
In a trial published online in July 2026, 122 cardiac-rehabilitation inpatients were randomized to usual care, usual care plus self-monitoring, or both plus an openly described placebo spray. Over two weeks, the groups improved on cardiac anxiety, sleep quality and well-being questionnaires. The study found no significant between-group differences on those primary outcomes.
This is a particular inpatient setting, not a verdict on every placebo intervention. It is nevertheless a useful counterweight to the idea that an openly labeled placebo reliably adds something everywhere. Here, improvement over time and evidence of an additional placebo benefit were not interchangeable.
Improvement is an observation. Cause needs a comparison.
There is another trap hiding in the term “placebo response.” It can mean everything that changes in a placebo group. That bundle may include fluctuating symptoms, concurrent care, measurement effects and regression toward the mean, as well as effects of expectations or treatment context. The NIMH’s 2024 placebo workshop makes this distinction explicit.
Consider a hypothetical sleep diary begun after an unusually bad week. The next week might have been less awful even without a new purchase. That possibility does not disprove the person’s improvement; it limits what “I started this, then improved” can establish. A fair comparison asks what happens over the same period to people assigned to a different condition.
Even a placebo-versus-no-intervention comparison needs care: knowing that one has received nothing can affect expectations and reporting. The comparison estimates a result under a particular design; it does not turn a personal experience into a laboratory assay of one mental mechanism. Our short guide to placebo groups unpacks that distinction.
The next question after “it helped”
No percentage in these studies is a universal “placebo rate.” None establishes the effectiveness of an unrelated supplement, a commercial blend or a family remedy. Different conditions, measures and comparisons require their own evidence.
The better conversation can make room for two statements: the relief matters, and the explanation remains open. Instead of asking someone to choose between believing their experience and reading the research, ask what improved, how it was measured and what the comparison showed. A claim that survives those questions is clearer and more useful than a dramatic answer about whether placebos “work.”
Sources and limits
- Wechsler et al., NEJM, 2011: randomized crossover pilot; short-term asthma outcomes, not long-term disease control.
- Kleine-Borgmann et al., JAMA Network Open, 2025: randomized migraine trial; primary outcome nonsignificant, some secondary outcomes favorable; participants unblinded.
- Tanzmeister et al., Journal of Psychosomatic Research, 2026: randomized cardiac-rehabilitation trial; details here checked against the authors’ published abstract, not a full-text appraisal.
- FDA: patient-reported outcomes and surrogate endpoints: official explanations of measurement, not evidence for a product.
- NIMH placebo workshop, 2024: expert methodological discussion, not a new clinical trial.
Sources checked September 30, 2026. Independent clinical review remains pending.
Published by Bencao Notes Sources checked: 2026-09-30
Review status
Editorial demonstration · Independent clinical review pending
Named author and reviewer credentials have not yet been provided. Source verification is not clinical review. Editorial standards and disclosures
Image credits
Stock photograph; not medication used in the studies.
cottonbro studio / Pexels · Image license · Resized and converted; thumbnails cropped by layout.