Skip to main content

Transforming the understanding
and treatment of mental illnesses.

Meeting Summary: Placebo Workshop: Translational Research Domains and Key Questions

July 11–12, 2024

On this Page

Day 1: NIMH Welcome Remarks
Historic Perspectives on Placebo in Drug Trials
Current State of Placebo in Regulatory Trials
Current State of Placebo in Device Trials
Current State of Placebo in Psychosocial Trials
Day 1 Panel Discussion
Day 2: Day 1 Recap and Day 2 Overview
Measuring & Mitigating the Placebo Effect
Mapping Expectancy-Mood Interactions in Antidepressant Placebo Effects
Day 2 Panel Discussion
Closing Remarks

Day 1: NIMH Welcome Remarks

Opening Remarks

Shelli Avenevoli, Ph.D., Acting Director, National Institute of Mental Health (NIMH)

Dr. Shelli Avenevoli welcomed participants and talked about the workshop's goal of bringing together neurobiology, clinical trials, and regulatory science experts to explore placebo effects in drug, device, and psychosocial interventions. She emphasized NIMH's long-standing recognition of the placebo phenomenon's impact on mental illness studies and the importance of designing and interpreting clinical trials to address significant research gaps.

Introduction and Workshop Overview

Tor D. Wager, Ph.D., Director, Cognitive and Affective Neuroscience Laboratory, Dartmouth College

Dr. Tor D. Wager welcomed attendees to the workshop, which aimed to explore how placebo and nocebo effects influence clinical trial design and outcomes; uncover the psychological, neurobiological, and social mechanisms behind placebo effects; and apply this knowledge to enhance therapeutic effects and improve clinical trial design.

Dr. Wager provided a historical perspective, tracing the placebo effect to ancient medications documented in the Ebers Papyrus. These medications often lacked active ingredients but endured because of their influence on patient beliefs. He demonstrated the challenge of the placebo effect with an example of a Parkinson’s disease clinical trial, in which both the treatment and placebo effects showed significant improvements, masking treatment efficacy. He highlighted similar findings across surgeries and medications, such as arthroscopic knee surgery and epidural steroid injections, which later proved to be no more effective than placebos.

Dr. Wager emphasized the complexity and power of placebo effects, which were pervasive across multiple outcomes such as pain, anxiety, and depression. These effects could result from many types of cues such as verbal suggestion, social cues, expectations about treatment outcomes, interpretation of symptoms, and emotions and memories that can trigger learned brain-body responses. These effects may be driven by both conscious and unconscious processes that are embedded in the nervous system and may engage the brain in forming a predictive model and causal inference for what is expected to happen.

Dr. Wager underscored the challenges in distinguishing specific treatment effects from placebo effects in clinical research. For example, there were placebo effects in chronic pain treatments and a shrinking drug-placebo gap that complicated the approval of new treatments. He advocated for improved clinical trial designs that would more effectively separate these effects and integrate placebo research with current treatments to enhance patient outcomes. He asked workshop participants to consider several open topics: 1) outcomes that should be considered when evaluating whether an intervention is meaningful, 2) how changes in decision-making and suffering could be differentiated from response biases, 3) the identification of health-relevant outcomes affected by verum treatments but not placebos, 4) the types of experimental designs that could help separate specific treatment effects from context effects, and 5) predictors of who would be a placebo “responder” in order to maximize verum treatment effects in clinical trials.

Historic Perspectives on Placebo in Drug Trials

Part 1: The Scale of the Placebo Problem

Ni Aye Khin, M.D., Executive Medical Director, Neurocrine Biosciences Inc.

Dr. Ni Aye Khin's provided a comprehensive historical perspective of the placebo response in drug trials, with a particular focus on psychiatric research. She outlined the FDA's requirements for drug approval, which mandated substantial evidence from adequate and well-controlled trials—typically necessitating two positive randomized controlled clinical trials. The FDA had a holistic approach in reviewing clinical efficacy and safety data that considered both successful and unsuccessful studies. This approach was crucial for addressing the high rates of placebo response and declining treatment effects in psychiatric trials over time, particularly as more trials were being conducted outside of the U.S.

Dr. Khin presented an exploratory analysis of pooled efficacy data from trials for major depressive disorder and schizophrenia. The analysis included 81 short-term randomized controlled trials (RCTs) for depression, involving more than 20,000 subjects who were primarily Caucasian females with a mean age of 40 to 43 years. The findings of this analysis indicated a higher and more variable placebo response in non-U.S. sites compared to U.S. sites, although drug effects were slightly larger in non-U.S. sites. Overall, however, the difference in drug-placebo effects was similar across both regions. Additionally, the analysis compared fixed and flexible dosing regimens, showing a slightly higher trial success rate for fixed-dose trials. She also reviewed subject-level data to determine responder definitions and optimal trial duration, concluding that a 50 percent change from baseline was sufficient to define responder status and that eight-week trials were generally effective.

Dr. Khin then discussed maintenance efficacy trials (i.e., relapse prevention trials), which, while not typically a regulatory requirement, could be conducted to assess long-term efficacy. These trials involved an open-label treatment phase followed by a double-blind, randomized withdrawal phase, demonstrating a significant reduction in relapse rates for the drug treatment group compared to placebo group. In schizophrenia trials, pooled data analysis from 32 RCTs demonstrated an increasing placebo response, stable drug response, and a declining treatment effect over time—particularly in trials conducted in North America. This trend persisted over 24 years, highlighting the ongoing challenge of managing placebo responses in clinical trials. She noted that treatment effects decreased as patient body weight increased, particularly in North American trials.

She concluded by emphasizing the importance of collaboration among researchers, clinicians, and regulatory bodies to address the challenges posed by placebo responses. She highlighted the need for improved data standards and technological advancement to facilitate more effective pooling and analyses of clinical trial data.

Part 2: Strategies and Results

Michael Detke, M.D., Ph.D., President, Detke Biopharma Consulting LLC

Dr. Michael Detke discussed various strategies to reduce placebo response and improve drug-placebo separation in clinical trials for central nervous system (CNS) disorders. He highlighted both historical and contemporary approaches, emphasizing the importance of trial designs such as crossover and sequential parallel design. The crossover design increased study power by using patients as their own controls could increase study power, but may be confounded by washout effects and changing baselines. In contrast, sequential parallel study design addressed placebo responses by re-randomizing placebo non-responders, thereby enhancing the reliability of trial results. He highlighted the transition from a “last observation carried forward” (LOCF) statistical method to mixed model repeated measures to reduce false positive and negatives, thereby improving effect size.

Dr. Detke explored several operational strategies, including centralized ratings, which offered advantages over site-based ratings by reducing variability and improving drug-placebo separation. Centralized ratings helped maintain blinding and minimize bias, often resulting in larger effect sizes and reduced sample size requirements. He also discussed the potential of technology, such as computerized patient self-report and combining both human and computer assessment, to enhance the accuracy and reliability of trial results. He addressed challenges related to “professional patients” who participate in multiple studies, suggesting strategies such as pharmacokinetic collection to confirm drug intake and artificial intelligence (AI) models (e.g., AiCure) to monitor adherence.

Additionally, Dr. Detke discussed the role of placebo lead-ins, in which patients initially receive a placebo to identify high placebo responders who might be excluded from subsequent study phases, and flexible dosing strategies that tailor treatment to individual patients needs and potentially leading to better efficacy and tolerability. He also explored the role of biomarkers and AI models in predicting placebo response and drug efficacy, presenting data showing that baseline EEGs could identify patient subgroups with distinct response patterns, therefore enhancing precision in clinical trials.

He emphasized the need for ongoing validation of biomarkers against current gold standards, as well as continued collaboration and optimization of these methods to address inherent bias and variability in CNS clinical trials. Dr. Detke advocated for a tailored application of these study strategies across specific phases of therapy development and underscored the importance of collaboration and continued optimization of these methods to address the inherent variability and bias in CNS clinical trials.

Current State of Placebo in Regulatory Trials

Placebo Response in Psychiatric Clinical Trials: A Regulatory Perspective

Tiffany R. Farchione, M.D., FAPA, Director, Division of Psychiatry, US Food and Drug Administration

Dr. Tiffany R. Farchione provided a regulatory perspective on challenges and strategies related to placebo response in psychiatric clinical trials. She presented data showing an increasing trend in placebo response rates in schizophrenia and bipolar disorder trials, underscoring the complexity placebo responses present in the FDA drug evaluation and approval process. Dr. Farchione emphasized the critical role of placebo-controlled studies in providing substantial evidence of drug effectiveness, which was crucial for determining whether a drug’s effects were genuine. She explained that various factors, including patient expectations, trial design, and subjective outcome assessments, could significantly influence placebo responses.

Dr. Farchione discussed several strategies to manage placebo response, including the potential use of non-inferiority studies and sequential parallel comparison designs (SPCD). However, she noted that non-inferiority studies were currently not feasible for psychiatric drugs due to difficulties in defining a stable non-inferiority margin, given the variability in placebo response over time. While SPCD had been used in some trials, it presented methodological challenges and had not yet been accepted for new drug approvals in psychiatry. She provided an example of a successful strategy using blinded central raters in trials for tardive dyskinesia, which helped mitigate the placebo response by ensuring more objective and consistent assessments. Dr. Farchione also drew parallels between psychiatric and other diseases, demonstrating that placebo response was a widespread issue across various medical fields.

Dr. Farchione emphasized the need for improved clinical trial designs and more precise outcome assessments to reduce variability and enhance the accuracy of drug evaluations. She noted that a significant advancement would be the development of drugs targeting the underlying pathophysiology of psychiatric disorders, which would allow for more objective measures and better-defined study entry criteria, ultimately leading to more effective treatments. She urged workshop participants to view the current challenges as a call to action, aspiring for a future in which psychiatric trials were more precise and grounded in a deeper understanding of disease mechanisms.

Current State of Placebo in Device Trials

Sham in Device Trials: Historical Perspectives and Lessons Learned

Sarah Hollingsworth “Holly” Lisanby, M.D., Director, Translational Research Division, NIMH

Dr. Sarah Hollingsworth “Holly” Lisanby discussed the complexities involved in using sham procedures in device trials, emphasizing that—unlike drug trials—sham procedures must replicate the mode of application and ancillary effects of active devices without providing actual stimulation. She highlighted different neuromodulation devices, which differ in the level of invasiveness and focality, and the unique challenges involved in creating effective sham procedures for each device type. For example, transcranial magnetic stimulation (TMS) involved different devices, each that required distinct sham configurations to match the sensory and operational aspects without eliciting a biological effect.

Dr. Lisanby provided historical context on the evolution of sham methodologies used in TMS trials. Early sham methods, such as coil tilt configurations, were found to be biologically active, as demonstrated by intracerebral recordings showing that some shams induced nearly 75 percent of the effect of active TMS. This led to the development of more sophisticated methods, such as magnetic shielding and auditory masking, though these often failed to fully replicate the ancillary effects of active devices. The optimized TMS (OPT-TMS) Trial, which used scalp stimulation and auditory masking, demonstrated improved blinding and more precise differentiation between active and sham groups. However, it underscored that even advanced sham methods had limitations, and that many pivotal trials for devices such as TMS and transcranial direct current stimulation (tDCS) often failed to meet primary endpoints, complicating regulatory approval and insurance coverage.

Dr. Lisanby also addressed challenges in sham procedures for surgically implanted devices, such as vagus nerve stimulation (VNS) and deep brain stimulation (DBS), including the potential for unblinding through procedural cue such as battery recharging and severe clinical conditions among participants. She emphasized the importance of measuring the biological activity of sham interventions and conducting rigorous blinding assessments for patients, operators, and raters. Dr. Lisanby offered future directions for device trials, including developing next-generation devices with fewer ancillary effects, validating sham interventions, and considering active stimulation of control brain areas as comparison conditions.

Dr. Lisanby underscored the need for meticulous sham methodology to ensure valid and reliable trial outcomes. She called for innovative approaches, such as quiet TMS and controllable pulse TMS, to minimize placebo responses and improve trial efficacy. Although achieving regulatory approval did not guarantee insurance coverage or adoption in clinical settings, it was nonetheless important to continue rigorous validation and innovation to improve device trial methods.

Challenges and Strategies in Implementing Effective Sham Stimulation for Noninvasive Brain Stimulation Trials

Zhi-De Deng, Ph.D., Director, Computational Neurostimulation Research Program, NIMH

Dr. Zhi-De Deng discussed the critical role of valid sham techniques in ensuring the reliability of brain stimulation studies. To maintain blinding, effective sham procedures must look and sound like active TMS, produce similar somatic sensations (i.e., coil vibration, scalp and muscle activation), and prevent actual brain stimulation. He reviewed different sham techniques, including coil tilt techniques, two-coil configurations, and dedicated sham systems. Each method had trade-offs, with none achieving an ideal balance of sensory mimicry without residual brain stimulation. For instance, coil tilt techniques often remained biologically active; two-coil techniques and dedicated shams often lacked any sensation to the scalp.

Dr. Deng also examined the potential of concurrent cutaneous electrical stimulation (e-stim) to enhance the realism of shams by placing electrodes on the scalp to mimic or match the sensation of active TMS. However, this technique had challenges with decaying waveforms associated with the intensity of electrical pulses, as well as current amplitudes that were not linear with intensity settings. Further, different pulse chapes preferentially activate different nerve fibers than the active TMS, leading to discomfort and an inaccurate sensory experience.

Additionally, Dr. Deng reviewed clinical literature to identify factors influencing sham responses in clinical trials for depression. There were correlations between the number of sessions and sham responses (i.e., the longer the treatment course, the higher the expectation of improvement) and between active response and sham response (i.e., the higher the active response, the higher the sham response). These sham responses have increased in brain stimulation trials over time, potentially due to changing attitudes about certain treatments and a greater awareness of medical research and healthcare information. Additionally, clinical trial participants may have become more health conscious, leading to stronger expectations from treatment. Dr. Deng noted that responses to active treatment had also increased over time, although not as much as the increase in sham response.

Dr. Deng talked about similarities in placebo responses across drug and device treatments. For example, sham TMS showed a stronger early placebo response than placebo pills in late-life depression. However, by 12 weeks into the trial, the placebo response was similar for both. Dr. Deng highlighted the need for more refined sham techniques and rigorous trial designs to better understand and mitigate placebo effects in brain stimulation trials.

Current State of Placebo in Psychosocial Trials

What is the Psychosocial “Placebo Pill”? Transferring the Placebo Construct to Psychosocial Trials

Dr. Winfried Rief addressed the complexities of applying the placebo construct to psychological therapies and trials. Psychological treatments often used mechanisms such as conditioning and expectation, which were also placebo mechanisms in medical treatment. However, unlike drug trials with clear active ingredients, psychological interventions lacked the same specificity that can complicate the development and implementation of effective placebo conditions.

To demonstrate these challenges, Dr. Rief reanalyzed a meta-analysis of 50 placebo-controlled trials for depression and anxiety, highlighting the difficulty in creating effective placebo conditions in psychological treatments. Many trials used psychoeducation, supportive therapies, or interventions shown to be effective for other conditions—such as cognitive restructuring—as controls, blurring the lines between active and placebo conditions. Similar to patient expectations in medical interventions, patient expectations in psychological interventions were also powerful predictors of treatment outcomes—patients with positive expectations tended to sustain higher success rates than those who did not expect improvement.

Pre-treatment experiences also played a crucial role in shaping patient expectations. Dr. Rief provided an example of a study on heart surgery patients, in which they optimized treatment outcome expectations prior to surgery, moving patient expectation from being just noise from a placebo effect to a target mechanism. In the study, those who experienced expectation optimization had better outcomes than those who did no. Similarly, therapists could work with patients to develop positive outcome expectations and improve outcomes. Dr. Rief noted that nocebo effects—that is, unwanted side effects such as increased conflicts, new or worsening symptoms, or even suicidal ideation—were commonly reported but rarely assessed in psychological treatment

Dr. Rief emphasized the importance of the clinician-patient relationship in shaping expectations. For example, how much improvement a patient experienced during the last treatment could be associated with how much improvement a patient expects to see in the future. Therapist behaviors such as warmth and competence significantly also impacted treatment effectiveness.

Dr. Rief highlighted the need for different types of placebo-controlled trials and control conditions in psychosocial interventions to identify mechanisms of change. It was also important to evaluate both specific and non-specific mechanisms to assess overall treatment efficacy. A comprehensive assessment of nocebo effects and a better understanding of how different treatment mechanisms interact were crucial directions to advance psychosocial treatment development.

A Social Neuroscience Approach to Placebo Analgesia

Lauren Y. Atlas, Ph.D., Faculty, Cellular and Neurocomputational Systems Branch, NCCIH/NIMH

Dr. Lauren Y. Atlas discussed the need to incorporate social neuroscience into the study of both placebo analgesia and placebo effects in general. While traditional research had focused on intrapersonal factors such as patient expectations and learning history, it was important to examine interpersonal processes (i.e., expectations, learning, history) and the social context in which treatment occurred. For example, Dr. Atlas proposed that teaching clinicians how to enhance patient outcomes could be more effective than relying on patient-specific factors. Affective neuroscience could help improve the understanding of the shared and unique mechanisms of different types of placebo effects and examine the role of social context in treatment.

A meta-analysis of clinical trials showed that placebos had no effect on binary or objective outcomes, but significantly impacted continuous subjective outcomes, particularly on the perception of pain. The authors of the meta-analysis suggested that the placebo effect on subjective outcomes was a reporting bias This would suggest that, instead of a placebo model in which social context provided cues that led to changes in sensory processing and a subjective response, social context might bypass sensory processing altogether and instead lead directly to a subjective response. Dr. Atlas reviewed neuroimaging studies showing that placebo effects were not associated with domain-specific brain circuits associated with pain (i.e., the neurologic pain signature), but were mediated by domain-general brain circuits associated with salience, affect, cognitive control, and value-based learning. In other words, the mechanisms of the placebo effect were not unique to specific clinical conditions, but rather more general affect and cognitive processes.

Dr. Atlas also discussed how social context impacted patient expectations, provider assessment of pain, and pain outcomes. For example, patient-provider interactions, such as the provider’s level of warmth, competence, and perceived similarity to the patient, could significantly influence pain expectations and treatment outcomes. She highlighted research indicating that first impressions of provider competence, including those mediated by racial and gender stereotypes, could impact patient expectations. These stereotypes could not only impact pain perception, but also provider beliefs. For example, a study showed that providers were less likely to ascribe pain to Black individuals than white individuals. These race-based differences were correlated with scores on a modern racism scale.

Dr. Atlas advocated for more research into how social factors influenced clinical outcomes and placebo effects and placebo analgesia. She emphasized that understanding the interplay between social dynamics and placebo responses could improve patient-provider interactions, reduce health disparities, and improve patient outcomes. Further exploration of domain-specific and domain-general mechanisms of placebo effects would help distinguish the broader clinical context from disease-specific outcomes, thus enhancing the effectiveness of treatment across diverse populations.

Panel Discussion

Moderators:
Carolyn Rodriguez, M.D., Ph.D., Professor, Stanford University
Alexander Talkovsky, Ph.D., Health Scientist Administrator, NIMH

Question: Are studies in animals free from a placebo effect?

Answer: Dr. Wagner explained that while it was challenging to determine animal expectations, there was evidence of conditioned or learned placebo effects in animals. There had been several studies indicating that animals might develop internal mental models, place cues, or other expectation and response patterns, suggesting that animal studies can produce placebo effects.

Dr. Atlas added that there was a study showing that dopamine was necessary for learning the association between context and pain relief, while opioids were necessary for experiencing pain relief.

Question: How does naloxone influence the neurologic pain signature (NPS)?

Answer: Dr. Atlas noted that there was limited research on naloxone's effect on NPS responses. Although there were studies showing an effect of opioid analgesics on the NPS, studies blocking the opioid system to observe NPS responses were limited.

Question: Does the process of determining the motor threshold in TMS unblind participants?

Answers: Dr. Lisanby said that TMS intensity was determined by determining the motor threshold using single magnetic pulses on the primary motor cortex, regardless of whether the participant would receive active or sham TMS, which could therefore unblind the participant. However single pulses caused less discomfort than repetitive pulses and therefore participants might think they were receiving active repetitive TMS.

Dr. Deng added that different TMS protocols felt very different, and being non-naive to one protocol does not necessarily break the blind.

Question: How did COVID-19 impact clinical trial assessments and should sites with unusual response patterns be excluded?

Answer: Dr. Detke presented data showing that activities of daily living (ADLs) were significantly impacted by the peak of COVID-19 cases. While something like a pandemic can create unusual outcomes, he noted that excluding sites with unusual response patterns such as large placebo responses would be challenging and may increase false positives if done post hoc. However, sites changed over time and varied in terms of the number trial participants. Although evaluating performance by clinical sites was challenge, it was still important.

Question: How significant is a two-point difference on a 52-point scale in clinical trials?

Answers: Dr. Khin answered that a two-point change was the difference between a drug and a placebo and would be enough for most drugs to be approved. Whether there was real-world significance related to that change remained a question.

Dr. Farchione added that the FDA would ask investigators using new indications or endpoints for an a priori definition of a clinically meaningful change at the within-patient level, not the between drug and placebo group level.

Dr. Rief said that the smaller the benefit between drug and placebo, the more important the question about the benefit-harm ratio.

Dr. Lisanby talked about outcome measures and the importance of determining how sensitive and proximal they were to how the intervention engaged mechanisms. This was part of the rationale for the Research Domain Criteria (RDoC) to assess domains of function. It would be important to discuss regulatory pathways for different types of outcome measures and how one would select a measure that was more effective at differentiating real treatment versus placebo effects.

Dr. Wager said that it would be important to understand FDA perspectives on what counted as mechanisms of action, specifically whether fMRI, EEG, or other indirect measures counted. Dr. Farchione answered that they did not yet count. These modalities showed group differences, but could not predict individual patient responses. The FDA’s Biomarker Qualification Program, for instance, sought to understand in what way a drug action or interaction could help inform the trial or assessment of effect. However, putting these into a drug label was a long way off.

Dr. Lisanby talked about device trial design, in which instructions for using the device for individual physiology was already on the label. While EEG or fMRI measures may be useful for identifying treatments and optimizing dosing, there was still a need to demonstrate to the FDA that the intervention improved quality of life and clinical outcomes.

Discussion

Question: Was there a perceived trade-off between warmth and competence and could they be manipulated as variables in a clinical trial?

Answer: Dr. Rief said that it was necessary to control these variables or there was a risk of creating differences across the different arms of a clinical trials. Although minimizing warmth and competence could minimize potential placebo effect, it could also reduce motivation to participate. If warmth and competence were maximized, there was a risk of increased placebo effects. It was a trade-off and maintaining an average level of both across different treatment arms was important.

Dr. Atlas said that it depended on the goal of the trial. If the goal was to reduce placebo effect and identify the drug benefit, then trials in which participants never saw the same rater would reduce the likelihood of building a relationship. On the other hand, if the goal was to have the best patient outcomes, then the opposite would apply in order to improve patient wellbeing.

Dr. Rief added that most investigators assume that they have to reduce placebo effects to maximize the difference between placebo and drug effects. But it was an assumption, not a known. Some interventions needed some minimum placebo effect to show its full action. The best practice might be to maintain an average amount of placebo mechanism.

Dr. Wager noted that there was a tendency to screen out placebo responders, but that it often did not work ill in clinical trials. If, for example, there was a synergetic interaction, one would not only be screening out placebo responders, but also drug responders. There was an opportunity to test the effects of active treatment with factors such as expectation or perceived warmth and competence. He asked whether there were any such studies in neurostimulation.

Dr. Lisanby answered that even if a device trial was not explicitly measuring the effect of the device operator, the effect was nonetheless occurring. It would be helpful for investigators of device trials to partner with other investigators with expertise in the context of care and psychosocial aspects. Cross-pollination of fields would also help select appropriate measures to test the integrity of the blind, as well as expectancy and other contextual factors.

Question: How should trials involving psychedelics handle the challenge of blinding?

Answers: Dr. Farchione said that blinding trials of psychedelics or empathogens was practically impossible. However, it was still important to have a placebo-controlled design in order to assess the safety of the drug. It was a challenge to determine the best placebo design because even when an active comparator was used, people still know. Complementary study designs, such as a dose response study, could be one approach because it could measure dose effects. An unblinding questionnaire might also be helpful, and the FDA was starting to ask for an assessment from the raters. A pre-dose expectancy assessment could also be helpful. Although they did not have a good solution now, they were rapidly learning and adjusting.

Dr. Lisanby agreed that it was a very complicated situation. She expressed concern about therapist unblinding and suggested that recording the therapy could provide helpful data to interpret whether therapist unblinding was impacting the psychosocial aspects of the intervention.

Dr. Wager asked whether a sensitivity analysis design that could independently manipulate expectations or context could induce a certain experience. Then one could measure whether outcomes were sensitive to the patient experience. Expectation could be induced by suggestion about how the drug might impact them.

Dr. Farchione said that giving different people different instructions would introduce another source of variability, but that the FDA was open to creative solutions as long as participants were being monitored appropriately.

Question: Are we confident that placebo effects and specific drug effects additive and not interactive?

Answers: Dr. Atlas explained that the relationship between placebo and specific drug effects can vary depending on the endpoint, drug dose, and study context. While some studies show additive effects, others reveal interactions, indicating the complexity of this relationship.

Day 2: Day 1 Recap and Day 2 Overview

Cristina Cusin, M.D., Director, MHG Ketamine Clinic

Dr. Cusin summarized the first day of the workshop, in which panelists explored the history, innovations, and complexities of placebo effects in clinical trials and psychological therapies. Key themes included:

  • Challenges in developing appropriate shams for device-based trials
  • Ensuring effective blinding in trials
  • Addressing the complexities involved in different sham methods
  • Examining the neurocircuitry underlying placebo effects, including salience, affect, and cognitive control
  • Disentangling psychosocial factors from biological treatment outcomes

There were also several calls to action for future research, including the need for:

  • Innovative trials designs such as sequential parallel comparison design to minimize placebo responses
  • Statistical approaches such as mixed model repeated measures to improve trial outcomes
  • Objective measures for psychiatric trials that worked similarly to measures in other medical fields
  • Extended biomarker discovery in psychiatry
  • Improved sham technologies and methods for device trials to achieve true blinding
  • Studies that consider the impact of psychosocial factors and environment on patient outcomes
  • Understanding the specific and non-specific effects of treatment
  • Recognition of the importance of patient perceptions and needs in clinical research.
  • Interdisciplinary collaboration to better understand the complexities of placebo effects

Dr. Cusin previewed the agenda for the second day of the workshop and welcomed panelists.

Measuring & Mitigating the Placebo Effect

Placebo and Nocebo Effects: Predictive Factors in Laboratory Settings

Luana Colloca, M.D., Ph.D., M.S., Professor, University of Maryland, Baltimore

Dr. Luana Colloca provided an overview of the complex neurobiological mechanisms underlying placebo and nocebo effects, particularly in pain treatment. She explained that placebo effects were influenced by multiple factors such as verbal suggestion, prior therapeutic experience, social learning, contextual and treatment cues, and interpersonal interaction. Studies showed that these placebo factors were associated with brain activity in regions such as the frontal cortex, nucleus accumbens, and ventral striatum. Conversely, the mechanisms of nocebo effects were driven by external factors such as negative expectations, prior negative experiences, and witnessing negative outcomes in others, as well as internal factors such as mood, maladaptive cognitive appraisal, and personality traits. Neuroimaging studies showed associations between the nocebo effect and brain regions such as the brainstem, spinal cord, and hippocampus.

Dr. Colloca talked about pharmacological conditioning studies that demonstrated how drugs could mimic or block placebo effects. For example, one study found that administering apomorphine followed by saline reduced clinical symptoms in Parkinson’s disease, but only in some patients. In another study, participants receiving morphine followed by a placebo had increased pain tolerance, suggesting that there was the potential for dose-extending placebo strategies for opioid tapering. Similarly, animal studies indicated that conditioning with ketamine followed by saline could replicate antidepressant-like effects.

Dr. Colloca noted that there could be gender differences in the placebo analgesic effect. For example, studies with vasopressin and oxytocin showed that vasopressin enhanced this effect in women but not in men. Dr. Colloca’s team investigated whether manipulating the intensity of pain stimulation and expectation of pain could serve as an alternative to medication for studying placebo and nocebo effects. By using matched and mismatched anticipatory cues before administering each level of pain intensity, they found strong placebo and nocebo effects when expectation matched the anticipatory cue.

She noted that social observation was another important model for observing the placebo effect. For instance, studies showed that merely observing someone else receiving painful stimulation or not could influence perception of pain, which was associated with activation in the temporoparietal junction, amygdala, and periaqueductal gray. Interestingly, another study showed that when chronic pain patients were studied by experimenters of the same race, there were larger placebo effects. However, sex concordance had a different effect. Women studied by a male experimenter showed larger placebo effects—a finding that was not replicated among men. Dr. Colloca noted that personality traits such as neuroticism could also predict placebo effects. She also talked about specific genes associated with placebo responses, providing potential targets for personalized treatments.

Discussion

Question: Is it possible to extend or sustain placebo boosting effect? What is the dose response relationship with placebo or nocebo?

Answer: Dr. Colloca answered that the goal was to enhance placebo effects, which was demonstrated by using intranasal vasopressin to extend the relationship with placebo. It was shown that a minimum of three to four administrations was necessary to boost pharmacological memory. The longer the active drug was administered before being replaced with a placebo, the greater the placebo effects. Similar relationships were observed for nocebo effects, indicating that longer conditioning resulted in stronger placebo or nocebo effects.

Question: Do you have any theories on the potential for transmitting placebo responses socially between observers?

Answer: Dr. Colloca said that the concept of transferring placebo responses was not entirely new, as studies had shown that pain could be transferred in both animal models and humans. Transfer analgesia was a natural continuation of this research. Mimicking behaviors observed in others was a fundamental learning mechanism from childhood and an evolutionary trait for protection from predators. Observation was a strong mechanisms for boosting behaviors, including placebo effects.

Genetics and the Placebo Response in Clinical Trials and Medicine

Kathryn T. Hall, Ph.D., M.P.H., Senior Vice President, The New York Academy of Medicine

Dr. Kathryn T. Hall explored the intricate relationship between genetics and the placebo response in clinical trials and medicine, challenging the traditional assumption that placebo responses were simply additive to drug effects. She highlighted that physiological drivers of placebo effects, such as opioid and dopamine signaling, could be influenced by genetic variation. Understanding these genetic interactions was crucial for interpreting clinical trial outcomes, rather than relying on conventional methods of subtracting placebo effects from drug effects.

To illustrate the genetic impact on placebo responses, Dr. Hall reviewed studies on specific genes, such as catechol-O-methyltransferase (COMT), and their influence on placebo outcomes. For example, a study on irritable bowel syndrome found that individuals with a particular COMT gene polymorphism exhibited different placebo responses, with those having a weaker form of the COMT enzyme showing a greater placebo effect. Similar patterns were observed in trials for chronic fatigue syndrome and cognitive function, in which genetic variations significantly influenced outcomes in both placebo and drug treatment arms. These examples underscored the importance of considering genetic differences to fully understand the effects of treatments and avoid misinterpretations that could lead to incorrect conclusions about drug efficacy.

Dr. Hall also highlighted the role of genetic variation in drug efficacy. Data from clinical trials of medications such as clonidine and aspirin showed that genetic factors influenced outcomes differently across placebo and drug arms. For instance, in the Women’s Health Study on aspirin versus placebo, different COMT genotypes resulted in different cardiovascular outcomes. This demonstrated the need to consider genetic diversity in clinical trials to identify subpopulations for whom treatments may be more or less effective.

Dr. Hall proposed a shift in clinical trial methodologies to better account for genetic and epigenetic factors, suggesting that understanding gene-drug-placebo interactions could reveal subpopulations that benefit from or are harmed by specific therapies. This approach could ultimately improve trial outcomes by identifying ways to enhance placebo responses or mitigate adverse effects—leading to more effective and personalized medical treatments. She advocated for a more nuanced understanding of genetic influences on treatment responses, which could help transform clinical practices through tailored treatment based on genetic profiles. She also noted the potential for drugs to either boost or block placebo responses.

Discussion

Question: Would examining the fraction of high responders in placebo arms reveal an enrichment of genetic markers for enhanced placebo response?

Answer: Dr. Hall said that the potential for genetic markers to influence placebo response was plausible, though it had not been extensively studied. The variability in naloxone's effect on placebo response highlighted the importance of better understanding trial participants. Differences in placebo response may be influenced by gender, race, ethnicity, and geographic location. Addressing this heterogeneity in clinical trials was crucial to avoid significant financial waste.

Question: How do you decide between using a candidate gene approach versus a genome-wide association study (GWAS) approach to identify genes with potential implications in neurophysiological pathways?

Answer: Dr. Hall noted that both approaches were necessary because GWAS could uncover unexpected findings. testing assumptions and replicating results was also crucial. Once a GWAS identified a gene, such as the BBS9, replicating or testing it in another cohort would be valuable. She acknowledged that conducting another clinical trial was challenging due to the high cost and lengthy duration. Despite these challenges, replication remains important and both methods should be pursued.

Disentangling the Physiological, Psychological, and Neural Mechanisms Supporting Mindfulness-Based Analgesia from Placebo

Fadel Zeidan, Ph.D., Professor, UCSD Department of Anesthesiology

Dr. Fadel Zeidan explored the distinct mechanisms supporting mindfulness-based analgesia compared to placebo. He highlighted the chronic pain epidemic and discussed his research team’s long-term focus on developing non-pharmacological approaches to pain relief, particularly through mindfulness meditation. Traditional mindfulness-based interventions (e.g., yoga, mental imagery, silent retreats), though effective, required significant time and financial commitment, limited accessibility. To address this, his team developed a brief, user-friendly mindfulness intervention focused on non-religious, non-spiritual training to develop cognitive flexibility and emotion regulation by maintaining awareness of sensory events without internal judgment.

Their brief intervention consisted of four 20-minute sessions that emphasized breath awareness. In comparing the neural mechanisms of mindfulness meditation with a sham mindfulness control and a book-listening control, the team used noxious heat stimuli to the back of the calf to elicit a pain response. They found that mindfulness meditation significantly reduced pain intensity and unpleasantness more effectively than the sham mindfulness control. Neuroimaging revealed that the meditation engaged multiple brain regions, including the anterior insula, the anterior cingulate cortex, and the thalamus, which were associated with higher-order cognitive flexibility, emotion regulation, and deactivation of the thalamus, which mediates sensory input to the brain. In contrast, sham mindfulness primarily reduced pain through slower breathing rates, indicating different underlying mechanisms and reflecting bottom-up versus top-down pain management approaches.

Dr. Zeidan also included findings from machine learning studies to analyze brain signatures associated with pain. Mindfulness meditation reduced both nociceptive-specific and negative affect-related pain markers but not placebo-related markers, suggesting that mindfulness meditation was distinct from placebo responses. This provided further evidence that mindfulness engaged unique neural processes beyond those involved in placebo effects. Additionally, research indicated sex differences in opioid engagement during pain relief, with men showing opioid-mediated analgesia and women showing non-opioid-mediated analgesia—a novel finding in human pain research. Multiple studies showed that mindfulness-based pain relief was not mediated by endogenous opioids, suggesting that different systems were engaged in pain management between sex. These insights contribute to understanding how mindfulness and placebo effects differ, providing a foundation for developing targeted pain management strategies.

Mapping Expectancy-Mood Interactions in Antidepressant Placebo Effects

Marta Peciña, M.D., Ph.D., Associate Professor of Psychiatry, University of Pittsburgh

Dr. Marta Peciña focused on the complex interplay between expectancy, mood, neural dynamics, and long-term placebo effects in antidepressant treatments, emphasizing the high prevalence and chronic nature of major depressive disorder. She highlighted the significant challenges in achieving complete remission with current treatments, noting that while response rates to antidepressants were approximately 50 percent, complete remission was only achieved in around 30 percent of individuals. Despite these challenges, placebo responses remained surprisingly high at approximately 40 percent, which had impacted the number of clinical trials for CNS disorders due to the difficulty of demonstrating drug efficacy over placebo.

Dr. Peciña discussed the role of placebo effects not only in medication, but also in surgical interventions such as deep brain stimulation (DBS) for depression. She highlighted a study in which patients with treatment-resistant depression showed significant improvement with both active and sham DBS, resulting in no significant differences between groups. This contributed to the FDA not approving DBS for depression, despite its use for other conditions—underscoring the challenges that placebo effects posed on psychiatric device FDA approvals.

To better understand these placebo effects, she developed a novel neuroimaging task to manipulate antidepressant expectancies and dissociate expectancy and mood dynamics. Participants were led to believe they were receiving either a fast-acting or conventional antidepressant, although both were saline. The study revealed that antidepressant expectancies and mood responses were higher during positive neurofeedback and infusion cues, which was moderated by depression severity. Neural data showed increased activation in attention-related brain regions during infusion cues and positive reinforcement, indicating that participants processed these cues as rewarding. Further, computational reinforcement learning models helped predict how expectancies and mood were updated, identifying specific brain networks involved in these processes.

Dr. Peciña further explored how expectancy-related neural activity could predict long-term placebo effects, showing that greater salience attribution to treatment cues in the salience network predicted better mood regulation through the default mode network (DMN). Enhanced connectivity between these networks was linked to improved responses during a clinical trial of placebo and antidepressant administration. Additionally, the study tested whether non-invasive brain stimulation could enhance placebo effects, finding that specific stimulation techniques increased expectancy-related neural activity and mood responses. These results suggest that understanding and manipulating expectancy-mood interactions can significantly impact the effectiveness of placebo treatments in depression.

The Appetitive Side of Placebo Effects: Findings from Functional Magnetic Resonance Imaging and Computational Modeling of Placebo Effects on Motivation

Liane Schmidt, Ph.D., Researcher, INSERM

Dr. Liane Schmidt discussed the appetitive side of placebo effects, focusing on their impact on cognitive processes such as motivation, biases, and belief updating. She emphasized the importance of these processes in clinical settings and their role in how patients respond to treatments. She highlighted the role of the brain’s reward system, particularly in the ventromedial prefrontal cortex (vmPFC), which encoded expected and experienced rewards and was involved in both placebo effects and reward sensitivity.

Dr. Schmidt reviewed how placebo effects could influence taste experiences and reward learning. She cited examples from consumer psychology, such as how wine price labels affected perceived taste—demonstrating that expectations could alter sensory experience. In her team’s research, participants were given water with different suggestions about its effects on hunger. They found that the different suggestions significantly influenced hunger ratings and food choice, which was reflected brain activity in the vmPFC. These findings highlighted how higher-order beliefs about internal states could modulate basic motivations.

She talked about the neurocognitive mechanisms underlying placebo effects, showing that expectations could modulate neural activity in the vmPFC and its connectivity to the dorsolateral prefrontal cortex (dlPFC)—a region involved in decision-making and self-control. Using computational models such as drift-diffusion models, Dr. Schmidt demonstrated how placebo mechanisms could influence decision-making processes during food choices. This attentional filtering process was linked to neural activity between the vmPFC and dlPFC, demonstrating their role in valuation and self-control.

Dr. Schmidt provide insights into the clinical implications of these findings, particularly in how the synergy of expectations and drug experiences could influence treatment response in depression and Parkinson’s disease. She introduced the “good new bias” study, which examined belief updating in individuals with depression. This study found that these individuals typically lacked the optimism bias typically found in individuals without depression, who tend to favor updating their beliefs in response to positive information. After treatment with ketamine, individuals with depression demonstrated increased optimism bias, which was correlated with clinical improvement. This finding suggested that ketamine could enhance cognitive flexibility and positive belief updating.

Dr. Schmidt note that placebos could normalize reward learning deficits associated with Parkinson’s disease by triggering endogenous dopamine release—highlighting the interplay between dopamine, reward learning, and placebo responses. She advocated for further empirical testing of these models to enhance treatment adherence in clinical practice and suggested that motivational processes and expectations could play a key role in improving patient outcomes, providing a deeper understanding of the mechanisms underlying placebo effects and the potential application in personalized medicine.

Expectation-Driven Control of Pain Neural Circuits: Preclinical Modeling of Placebo Analgesia

Gregory Corder, Ph.D., Assistant Professor, University of Pennsylvania

Dr. Gregory Corder provided an overview of recent advancements in understanding placebo analgesia through animal studies, focusing on how expectation-driven control of pain was mediated by neural circuits and cell types. He highlighted the significant role of human fMRI studies in mapping key brain circuits related to expectations and belief systems, particularly those involved in endogenous opioid systems.

Animal models offer a unique opportunity to study specific neural circuits involved in placebo effects, using advanced techniques such as RNA sequencing, electrophysiology, optogenetics, and chemogenetics to provide a precise manipulation of neural connections and observations of their effects on behaviors. Dr. Corder’s research used these models to replicate human studies in rodents, demonstrating placebo analgesic response pharmacological conditioning with morphine followed by saline administration. One study showed that when naloxone was administered, it blocked the analgesic effects of the placebo, suggesting that endogenous opioids were actively involved in mediating the placebo response. This highlighted the critical role of opioid cell types in expectation-driven pain control.

In another novel approach, Dr. Corder and his research team developed an experimenter-free, drug-free paradigm to instill placebo effects in mice by conditioning them to associate environmental cues with pain relief. This approach led to significant placebo responses, with the mice showing reduced pain-related behaviors and a preference for zones associated with analgesia. This study also highlighted the importance of the periaqueductal gray in mediating opioid-based pain relief and demonstrated how mu-opioid receptor expressing neurons were modulated by placebo conditions.

By using optogenetics and calcium imaging, Dr. Corder’s research demonstrated the ability to directly manipulate neural circuits to induce analgesic effects. His team’s study findings demonstrated the complex interactions between neural circuits and placebo responses and offered insights into potential clinical applications for pain management. The findings also pave the path for a deeper understanding of the mechanisms underlying placebo effects and the potential for developing targeted pain management approaches.

Panel Discussion

Moderators:
Ted Kaptchuk, M.D., Professor of Medicine, Harvard Medical School
Matthew Rudorfer, M.D., Associate Director, Treatment Research, NIMH

Dr. Kaptchuk commented on the need for a more integrated approach to understanding placebo effects, particularly in clinical trials. He highlighted the importance of considering natural history and patient experiences in research, advocating for more diverse and contradictory viewpoints to be acknowledged. He also stressed the need to incorporate ethical discussions and patient feedback into future conferences.

Question: The recent emergence of intravenous ketamine for resistance depression has introduced the active placebo. What are the merits of using an active placebo in clinical trials, such as benzodiazepine midazolam for ketamine trials?

Answers: Dr. Kaptchuk referenced past meta-analyses of atropine for depression that showed mixed conclusions for the use of active placebos.

Dr. Detke highlighted the challenge of creating an ideal placebo that mimics side effects without efficacy. He mentioned the use of midazolam as an active placebo for ketamine as a good attempt, noting that different drug classes had varying levels of unblinding due to their side effects.

Dr. Schmidt expressed concern about the dissociative effects of ketamine and how midazolam might not replicate these effects, possibly impacting patient expectations and treatment outcomes.

Dr. Cusin emphasized the lack of long-term placebo studies and the potential need to investigate the role of placebos in maintaining antidepressant responses.

Dr. Rief mentioned a study comparing active to passive placebos and the questions it raised about blinding techniques used in antidepressant studies. Many patients in the active group will perceive onset effects that those in the placebo group do not experience.

Dr. Peciña said that it was important to clarify what exactly investigators were trying to blind.

Dr. Kaptchuck referenced a meta-analysis on blinding that was replicated recently, showing that blinding with placebo was much more complicated than investigators realize.

Question: Considering the presentations on the importance of provider-patient interactions, what are your thoughts on shortened clinical visits and telehealth, in which a person might never meet their provider in person?

Answers: Dr. Colloca discussed ongoing research comparing placebo effects induced by in-person interactions versus virtual reality avatars, noting that while placebo effects persisted in both settings, empathy was only elicited in face-to-face interactions.

Dr. Hall referenced large studies involving mailed medications that had minimal provider interaction, suggesting this approach might not diminish placebo effects. Dr. Atlas noted that first impressions of providers—even through online resources such as Zocdoc—impacted patient expectations and interactions, which could influence treatment outcomes.

Dr. Lisanby talked about the importance of telemedicine for equity. While there may be differences in active and sham responses in virtual settings, was still important to leverage technology to improve access to mental health care.

Dr. Kaptchuk shared a replication study he recently published, noting that he found opposite results in which the less intrusive the provider-patient relationship, the better the effect. He noted the complexity of cultural constructs in placebo effects.

Dr. Colloca agreed that cultural context significantly impacted placebo effects and emphasized the importance of larger, cross-country collaborations.

Dr. Schmidt mentioned research on human-robot interactions, suggesting that social dynamics might differ when interacting with algorithms, which could affect placebo effects in the future.

Dr. Rief highlighted the interaction between non-specific, placebo, and drug effects, stressing the need for study designs that modulate both drug intake and psychological factors to better understand these interactions.

Dr. Rudorfer noted that rodent studies provided a lot of opportunity for control that was not possible in human trials.

Dr. Wager added that the interactions between psychosocial and expectation effects and specific effects required more research. It was challenging to do these studies, but it could be a pathway for NIMH to fund rather than pharmaceutical companies.

Question: What are your thoughts on studying sex-differences or other diversities in placebo trials?

Answer: Dr. Colloca said that location and socio-economic factors also influenced placebo responses. She stressed the importance of improving access to care and opportunities for diverse populations.

Question: What are the challenges in psychiatric genetics and the use of candidate gene approaches in pharmacogenomics?

Answers: Dr. Hall acknowledged the challenges involved in psychiatric genetics but stressed the importance of identifying genetic markers of placebo response. She emphasized looking at the placebo arms of clinical trials to understand natural history and potential interactions.

Dr. Cusin pointed out that participants in clinical trials often differ significantly from those patients in everyday clinical practice, highlighting the need to consider patient diversity and psychosocial factors in research.

Dr. Rudorfer talked about the history of who participated in clinical trials and how there was a threshold of severity required. Much of this was based on older versions of the DSM, which may have set too high a threshold. He noted the implications of using RDoC criteria in clinical trials and their potential to enrich DSM-type criteria for better differentiation of depression subtypes.

Dr. Wager talked about the challenge of predicting individual differences in placebo effects when large samples sizes were needed to make those associations. At the same time, there has been a trend toward questioning the reproducibility of most studies that likely has merit. One way forward would be to pool preregistered studies to test placebo effects.

Dr. Hall agreed with pooling resources and to seek to understand why something is not replicated.

Dr. Peciña noted the lack of consistency in placebo studies and the understanding that there was not a single placebo effect but rather complex effects interacting with multiple variables.

Dr. Rief suggested that there may be other factors to study design today that impacted the placebo effect such as more control visits or publication bias.

Dr. Kaptchuk suggested that the perceived increase in placebo effects over time might be influenced by changes in patient recruitment and natural history rather than actual increases. He also noted the role of media in shaping these perceptions.

Question: What was the physiological basis for the turning point in mice behavior observed at 90 seconds during experiments?

Answers: Dr. Corder explained the variability in mice responses and noted that the breaking of antinociceptive effect was likely related to peripheral nociceptor sensitization. He acknowledged the need for larger studies to better understand the underlying mechanisms.

Dr. Rodriguez said that obsessive-compulsive disorder (OCD) had one of the lowest placebo rates and suggested exploring the reasons underlying this to better understand OCD pathology and refine placebo effect measurements.

Question: What are the research gaps and future study directions?

Answers: Dr. Detke talked about the need for third-party reviewers in clinical trials and the potential for NIMH to fund studies exploring different blinding methodologies.

Dr. Lisanby agreed that understanding why some disorders had higher placebo responses than others was a gap area. She also noted the role of preclinical models in studying placebo mechanisms.

Dr. Hall emphasized the need to explore how modern information dissemination, such as early press releases on drug findings, influenced placebo responses in clinical trials.

Dr. Rief discussed the importance of varying context conditions in trials to better understand drug effectiveness.

Dr. Peciña pointed out the dynamic nature of expectancy and the need for improved measurement techniques.

Dr. Schmidt suggested using cognitive psychology tools to better understand the processes behind placebo effects.

Dr. Colloca agreed that there was a tendency to consider expectation as a static measure, but that the reality was that it evolved over time. 

Closing Remarks

Cristina Cusin, M.D., Director, MHG Ketamine Clinic

Dr. Cusin expressed their gratitude for a successful symposium and proposed organizing a future summit with all panelists, presenters, and other interested parties to foster collaboration. She addressed ongoing questions in the field, such as the nature of placebo responders across different studies and the impact of chronicity on placebo response rates. She noted the lack of comprehensive methodologies for clinical trial design and the need for better outcome assessments. She also expressed interest in exploring mindfulness in pain management and the role of devices in influencing food choices and managing eating disorders. She emphasized the potential for cross-disciplinary collaboration, particularly between animal researchers and clinical trialists. Finally, she thanked the panelists and collaborators for their dedication and contributions to the workshop.