Measuring and appraising placebo effects in clinical trials: contemporary challenges and approaches in psychiatry

Taylor and colleagues introduce placebo and nocebo effects as complex neurobehavioural phenomena shaped by expectations, beliefs, prior experience, emotion, and other features of therapeutic context. They note that although placebo-controlled trials are central to evidence-based medicine, there has been comparatively little detailed examination of how placebo effects are measured, interpreted, and appraised, particularly in psychiatry. The authors also distinguish the overall placebo response in a trial from placebo effects alone, emphasising that the observed response can include spontaneous improvement, regression to the mean, the Hawthorne effect, and other non-specific influences. The review aims to update approaches for navigating placebo effects in randomised controlled trials and to highlight contemporary issues that have made the topic especially important in psychiatric research. These include blinding and expectancy problems in psychedelic trials, unusually large placebo responses in interventional psychiatry studies involving devices or procedures, and the possibility that placebo effects and psychiatric treatments share overlapping neurobiological mechanisms. The authors frame the paper as part 2 of 2 on reconceptualising placebo and nocebo effects, with this paper focusing on clinical trials rather than clinical practice.

This is a narrative review rather than an original empirical study. Taylor and colleagues searched PubMed, Google Scholar, and their personal files for English-language articles published between 1 January 1950 and 15 July 2025, using terms related to placebo, nocebo, psychiatry, clinical trials, blinding, and expectancy. They also reviewed relevant references from retrieved articles. The paper synthesises evidence from systematic reviews, meta-analyses, randomised controlled trials, and methodological studies across psychiatric disorders and trial designs. It draws particularly on data about placebo response rates across diagnoses, the assessment of blinding integrity, trial designs intended to reduce placebo responses, and comparative effectiveness issues in psychedelic, behavioural, and device-based interventions. Because the paper is a conceptual and methodological review, it does not report a single analytic model or pooled estimate generated by the authors themselves. Instead, it organises the literature around several trial-design and interpretation problems, including how to measure placebo response, how to assess effective blinding, and how to handle expectancy and possible overlap between placebo mechanisms and active psychiatric treatments.

The review reports that placebo responses vary substantially across psychiatric disorders. In a cited systematic review and meta-analysis of 90 randomised controlled trials including 9985 participants, placebo response was highest in major depressive disorder, followed by generalised anxiety disorder, panic disorder, ADHD, post-traumatic stress disorder, social phobia, mania, obsessive-compulsive disorder, and schizophrenia, with women showing greater placebo improvement than men in meta-regression. Another synthesis of 20 meta-analyses covering 1691 randomised controlled trials and 261,730 participants found a similar ranking. The authors emphasise that placebo responsiveness is not uniform and may change over time within diagnoses. For trial methods, the paper reports that placebo run-in designs have not clearly improved antidepressant trial outcomes: a recent systematic review found similar likelihood of response and similar between-group effect sizes in studies with run-in periods versus those without them. Sequential parallel comparison designs are presented as potentially more powerful than placebo run-in designs, but the authors note that there is no systematic review of such trials. They also describe emerging approaches that use baseline placebo-responsiveness scores or novel objective outcome measures, but these are presented as developing methods rather than established solutions. The most striking findings concern blinding in psychedelic research. The authors state that in most psychedelic-assisted psychotherapy trials, correct treatment-guess rates are around 90%, and that one Phase III MDMA trial had 82 of 96 participants correctly guessing allocation. They contrast this with traditional antidepressant trials, where correct-guess rates are much lower, around 60%. A network meta-analysis found that placebo-group improvement in psychedelic studies was 3.8 points lower on the HAMD17 than in traditional antidepressant trials, and one phase II psilocybin study even found worsening in the placebo group at interim analysis. The authors suggest that this pattern may reflect unblinding, disappointment, or nocebo effects rather than a simple absence of placebo response. The review also summarises large and sometimes unexpected placebo responses in interventional psychiatry. It reports sham remission rates of 37% in one large sham TMS trial and 42% in a sham cranial electrical therapy trial, as well as a vagal nerve stimulation study in treatment-resistant depression in which the sham group showed a late surge and no significant difference from active treatment on the primary outcome. A meta-analysis of placebo responses in treatment-resistant depression trials is cited as showing a large pooled placebo effect size of Hedges' g 1.05, corresponding to roughly a 30-50% reduction in symptom scores. The authors also note high heterogeneity across studies, including some TMS trials with no sham improvement and others with about 50% improvement. Finally, the paper argues that placebo effects and active psychiatric treatments may overlap mechanistically at genetic, neurotransmitter, and brain-circuit levels. It highlights convergent findings in the dorsolateral prefrontal cortex and subgenual anterior cingulate cortex, including overlap between circuits implicated in placebo response and targets of neuromodulation such as TMS and deep brain stimulation. The authors also note that positive expectations may be associated with better outcomes in some open-label neuromodulation studies, and they extend this reasoning to psychedelic therapy, where set and setting may interact with drug effects.

Taylor and colleagues argue that placebo and nocebo effects should be treated as central methodological issues in psychiatric trials rather than as nuisance variables. Their main interpretation is that contemporary challenges in psychiatry are no longer limited to placebo-pill comparisons: they include compromised blinding in psychedelic studies, large and variable placebo responses in device- and procedure-based interventions, and the possibility that some psychiatric treatments share neurobiological substrates with placebo effects. In this framework, standard placebo-controlled trial results may either overestimate efficacy because of unblinding or underestimate efficacy when the treatment itself overlaps with placebo-linked mechanisms. The authors place these observations in the context of earlier research showing that placebo response differs across diagnoses and can be substantial in depression and anxiety disorders. They also note that although some design strategies, such as placebo run-in periods, were intended to reduce placebo effects, evidence that they materially improve trial efficiency is limited. Similarly, while active placebos, incomplete disclosure, balanced placebo designs, open-hidden paradigms, and other approaches may help in specific contexts, none is presented as a perfect solution. A major limitation emphasised throughout is the difficulty of separating placebo effects from other non-specific effects and from true drug-related effects, especially when blinding is weak or impossible to verify. The authors also point out ethical and practical constraints, including the need to balance scientific rigour against informed consent, recruitment feasibility, and the acceptability of withholding active treatment. They stress that blinding integrity is rarely assessed empirically and that this reporting gap limits interpretation of many trial results. The paper closes by arguing for more systematic collection of expectancy and blinding data, greater use of placebo-focused designs when ethically feasible, better critical appraisal of placebo groups in training and guidelines, and routine reporting of blinding integrity. The authors also call for neurophysiological biomarkers and objective digital phenotyping measures to help characterise placebo effects more precisely, particularly in psychiatric and psychedelic research.