Open Placebo in Clinic: Little Benefit, Big Expectations

Can you honestly tell a patient, “This is a placebo,” give them a capsule… and still get a benefit? A recent meta-analysis in Scientific Reports collected 60 randomized open-label placebo (OLP) trials and provided the most comprehensive answer to date: On average, OLPs produce a small but statistically significant effect across a wide range of outcomes. The effect is stronger in clinical patients and almost exclusively on self-reports, while the effect on objective metrics (physiological/behavioral measures) is tiny and inconclusive.

Background

The classic placebo effect in the clinic has always run up against ethics: you can’t deceive a patient to relieve symptoms, and without “masking,” placebos don’t seem to work. Against this background, the idea of an open-label placebo (OLP) emerged: giving capsules or a treatment ritual, honestly informing them that there is no active substance in them, but explaining how expectations, conditioned reflexes, and the ritual itself can trigger natural mechanisms of relief. Over the past 10-15 years, dozens of small RCTs of OLP have appeared for lower back pain, irritable bowel syndrome, allergic rhinitis, insomnia, hot flashes, anxiety, and fatigue. The pattern of results is repeated: self-assessed symptoms improve, sometimes noticeably, but objective markers (hormones, steps, lung function, etc.) change little or inconsistently. Due to small samples, variable quality of instructions and heterogeneous controls, the field remained “loose”: it was not clear what the actual effect size was, who had a higher effect (clinical patients or healthy volunteers), what role the suggestiveness of explanations played and for which outcomes (subjective vs. objective) one should expect benefit. This created a demand for an updated, large meta-analysis: to collect all OLP RCTs, separate them by types of populations and outcomes, assess the risk of systematic errors and understand where an “honest placebo” is a meaningful, ethical tool and where nothing can be expected from it.

The main thing is in the numbers

• The review included 60 RCTs / 63 comparisons (≈4.6 thousand participants), the search was conducted in 8 databases until November 9, 2023, the protocol was registered in PROSPERO and designed according to PRISMA-2020.
• Overall effect of OLP: SMD 0.35 (95% CI 0.26-0.44; p<0.0001; I²≈53%) - small but stable.
• Clinical vs. non-clinical samples: SMD 0.47 vs. 0.29 - the difference is significant (OLPs “work” more in patients).
• Self-reports vs. objective outcomes: SMD 0.39 vs. 0.09 - that is, the effect lives almost entirely in self-assessments of symptoms, and on “hard” indicators it is close to zero.
• Suggestiveness of the instruction (how vividly the power of the placebo was explained to the participants) moderates the effect: without the "inspiring" rationale, there were no results, with it - there were, although formally the differences between the levels of suggestibility did not reach significance. The predictive intervals for "high suggestibility" almost did not include zero.
• The type of control (waiting, usual therapy, hidden placebo, no treatment) did not fundamentally affect the magnitude of the effect - significant small-medium effects were observed everywhere.

What's new? The authors have directly compared the effectiveness of OLP between clinical and non-clinical groups and between outcome forms for the first time. Previous meta-analyses either considered these sections separately or did not combine them into a single model. Here, thanks to the increased test base, it was possible to test both hypotheses at once - and confirm that "honest placebo" is especially sensitive to who and how we measure.

How It Was Done (and Why the Method Is Important)

• We collected RCTs of OLP from 2001-2023: from pain, anxiety and allergic rhinitis to fatigue and academic stress; 37 non-clinical and 23 clinical trials, duration - from 1 to 90 days (median 7). Self-reports and objective outcomes were analyzed separately; heterogeneity is moderate.
• We checked for publication bias (funnel plot, Egger test - no evidence of systematic publication bias; Fail-Safe-N ≈ 3111). We performed sensitive analyses: we excluded outliers and studies with a high risk of systematic error, and also calculated a three-level model (effects are nested in studies) - the conclusions held.

What does this mean for practice?

• Where it is appropriate to try OLP:
  • conditions with leading symptoms according to self-assessment (pain, anxiety, fatigue, functional complaints),
  • when deception is unacceptable, but one wants to use expectations/ritual of treatment without an ethical conflict,
  • as an addition to standard care (TAU), and not instead of it.
• How to present an “honest placebo”:
  • thoughtful instructions (that the placebo triggers natural mechanisms, a positive attitude is not necessary, commitment is important),
  • ritual and format (tablet/capsule/spray) - as anchors of expectations,
  • transparency and shared decision-making with the patient.

And yet there should be no illusions. Where the outcomes are objective (hormones, steps, physiology), in total across the fields of meta-analysis, OLPs change almost nothing. This is not “magic without an active substance,” but the management of expectations and attention, which is more pronounced on the subjective side of the disease experience.

Limitations that the authors themselves write about honestly

• Small sample sizes in many RCTs ⇒ risk of “small study effect”. Large and long trials are needed, especially in clinical groups.
• The lack of blinding for OLP and the prevalence of self-reporting increase the risk of bias - even with a good design.
• Repeatability and independence: a significant proportion of the work is from the same research teams; the field needs more independent groups.

Where should researchers look next?

• More objective outcomes in clinical RCTs of OLP (sleep, activity, biomarkers).
• Tests for the sustainability of the effect (follow-up after months), and not just “today-tomorrow”.
• Comparison of "honest placebo" with ritualized activities (breathing, journaling, digital rituals) to separate the contribution of instruction and ritual.

Conclusion

"Placebo without deception" is not a trick, but a technological work with expectations. It really alleviates subjective symptoms, especially in patients, if presented with a clear and respectful explanation. But do not expect miracles in objective indicators: here "honest placebo" is still weak.

Source: Fendel JC et al. Effects of open-label placebos across populations and outcomes: an updated systematic review and meta-analysis of randomized controlled trials. Scientific Reports, August 15, 2025. Open access. https://doi.org/10.1038/s41598-025-14895-z