Why Sugar Pills Keep Beating Cancer Symptom Drugs in Clinical Trials

Some of the most frustrating failures in cancer research have nothing to do with whether a drug works. In trials of treatments for cancer-related fatigue, pain, nausea, and other symptoms, patients given inert placebos often improve dramatically — sometimes nearly as much as those given the experimental therapy. When the placebo arm shrinks the gap between treatment and control, a genuinely effective drug can look useless, and years of development can collapse for reasons that have little to do with pharmacology. A new methodological review argues that this phenomenon, known as placebo-response inflation, is not random noise but a design-sensitive, measurable problem that trialists in supportive oncology and palliative care can no longer afford to ignore.

The review, published in Supportive Care in Cancer by Mellar P. Davis and Boris Kiselev of Atrium Health in Charlotte, North Carolina, synthesizes evidence from meta-analyses, meta-regressions, and methodological studies across multiple therapeutic domains. Its central claim is provocative: placebo-response inflation should be treated as a determinant of what trial designers call assay sensitivity — a trial’s ability to distinguish a true treatment effect from nonspecific improvement — rather than as an unavoidable background hum of symptom variability. In symptom-focused oncology trials, where the primary outcomes are patient-reported and the symptoms themselves fluctuate, the stakes of getting this right are unusually high.

The mechanisms behind placebo-response inflation fall into several distinct categories. The first are operational: the sheer structure of modern trials. Evidence from meta-regressions in psychiatry shows that the number of study sites and the density of recruitment at each site influence how much patients improve on placebo. Large, rapidly recruiting multicenter trials tend to dilute the apparent treatment effect, a pattern documented in antidepressant trials over three decades and in studies of antipsychotics, where meta-regression analyses have questioned whether drug and placebo responses remain connected as sample sizes grow. A 2026 meta-analysis of adjunctive trials for negative symptoms in schizophrenia found that effect sizes deflate as sample size increases, a signature of placebo-arm inflation rather than weakening drugs.

Psychological mechanisms form the second category, and they are the most intuitively powerful. Expectancy — what a patient believes a treatment will do — directly shapes symptom reporting, particularly for subjective outcomes like pain and fatigue. Blinding integrity compounds the problem. When patients or clinicians can guess who is receiving the active drug, often because of recognizable side effects, expectancy effects contaminate both arms of the trial. This has been a particular embarrassment in cannabis-based medicine research: a systematic review of trials of cannabis-based therapies for pain found that placebo responses tracked media attention, and commentators have described inadequate blinding as the Achilles heel of medical cannabis studies. In oncology, where hope is a clinical reality, expectancy effects may be especially potent.

Statistical mechanisms provide the third pillar. Regression to the mean — the tendency of extreme values to move toward the average on repeated measurement — guarantees that patients enrolled at their symptomatic worst will often improve regardless of treatment. This is why entry criteria that demand high baseline symptom severity, combined with a single baseline assessment, virtually manufacture placebo improvement. The review notes that symptom variability and chronicity act as disease-related moderators of this process: fluctuating symptoms measured at an unstable moment produce apparent treatment responses that are pure statistical artifacts. Repeated baseline assessments before randomization are one straightforward way to separate real change from regression to the mean.

The oncology-specific evidence is sobering. Studies of fatigue in patients with advanced cancer have documented clinically meaningful improvement in placebo arms of double-blind trials, and recent work comparing blinded and open-label cancer fatigue trials shows that placebo effects are substantial even under rigorous conditions. An exploratory analysis of the ABCD randomized clinical trial has even begun hunting for predictive biomarkers of placebo response in patients with cancer, suggesting that who improves on placebo may eventually be identifiable in advance. Earlier foundational work on placebo effects in oncology, dating back two decades, had already flagged the issue; the new review argues that the field has not adequately acted on it.

What makes supportive oncology and palliative care uniquely vulnerable is the convergence of all these mechanisms in a single research setting. The endpoints are subjective and patient-reported. The symptoms — pain, fatigue, nausea, dyspnea — are inherently variable and often worsening with disease trajectory. Recruitment is typically multicenter and spread across sites of uneven quality. The interventions themselves, whether drugs, behavioral programs, or procedural techniques, are expectancy-sensitive by nature. And the patient population faces progressive illness, which introduces disease-related moderators that other fields rarely confront. A trial that stacks all of these risk factors is structurally primed to produce a large placebo response and a correspondingly weak signal of therapeutic benefit.

The methodological implications are concrete. The review points to several design levers with empirical support. Controlling recruitment density and site quality can reduce operational inflation. Psychoeducational procedures that recalibrate participant expectations have been shown in all-placebo psychiatric studies to mitigate placebo response, offering a template for oncology trials. Placebo run-in periods, in which potential responders to placebo are identified before randomization, have been associated with reduced placebo response in antidepressant trials, though they raise their own ethical and interpretive questions. Active placebos — control agents that mimic the side effects of the real drug — can protect blinding integrity but are rarely used. Balanced placebo designs, which cross pharmacological status with expectancy instructions, have demonstrated that conventional double-blind trials may actually underestimate true drug effects by entangling them with expectation.

None of these fixes is free of trade-offs. Placebo run-ins select a population that may not represent clinical practice. Active placebos can introduce their own symptoms that confound reporting. Expectancy manipulation raises questions about informed consent. And in palliative care, where trial windows are short and populations are fragile, adding repeated baseline assessments or complex designs imposes real burdens on patients and sites. The review’s contribution is to frame these not as isolated tactical choices but as a coherent, mechanism-based strategy: trialists should diagnose which mechanisms are most likely to inflate placebo response in their specific setting and design against them deliberately, rather than discovering the problem in a failed readout.

The broader lesson extends beyond oncology. The same inflation dynamics have eroded effect sizes in psychiatry, where decades of meta-regression work have mapped the relationship between trial characteristics and placebo response, and in pain research, where blinding failures have repeatedly muddied the evidence base for cannabinoids. Supportive oncology, the authors argue, should learn from these fields before its own evidence base accumulates the same distortions. If placebo-response inflation is a design-sensitive and measurable quantity, then every inflated placebo arm is not just a failed trial but a data point about how trials should be built. Treating it that way could mean the difference between abandoning an effective symptom therapy and proving that it works.

Subject of Research: Mechanisms of placebo-response inflation in symptom-focused clinical trials in supportive oncology and palliative care

Article Title: Placebo-response inflation in supportive oncology symptom trials: mechanisms, methodological challenges, and design implications

Article References: Davis, M. P., & Kiselev, B. (2026). Placebo-response inflation in supportive oncology symptom trials: mechanisms, methodological challenges, and design implications. Supportive Care in Cancer, 34(10), Article 962. https://doi.org/10.1007/s00520-026-11218-4

Image Credits: AI Generated

DOI: 10.1007/s00520-026-11218-4

Keywords: placebo response, supportive oncology, palliative care, clinical trials, symptom management, regression to the mean, expectancy, assay sensitivity, blinding integrity, patient-reported outcomes, cancer fatigue, trial design