Dissertation
Impact of differences in placebo response for platform trials
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Summer 2024
DOI: 10.25820/etd.007682
Abstract
Traditional, two arm randomized clinical trials are considered the gold standard for evaluating the efficacy of an experimental treatment. In order to evaluate numerous investigational treatments more efficiently, there has been a great deal of interest in more flexible clinical trial designs, such as the platform trial. Platform trials are governed by an overarching study protocol, the master protocol, which utilizes common study infrastructure to answer multiple research questions. In particular, the platform trial evaluates multiple interventions, allowing for experimental interventions to enter and exit the study over time. A platform trial increases efficiency by creating a shared infrastructure, such as study sites and procedures, and allowing each experimental therapy to be compared to a shared control arm. This shared control arm is appealing to potential trial participants because it ensures higher odds of receiving an active treatment.
Although many features of a platform trial can be extremely appealing, some features, in particular the shared control arm, can introduce unique challenges. In a platform trial, trial participants are typically randomized in a two-stage fashion. First, participants are randomized to a specific regimen of the platform, with there being a regimen for each experimental therapy of interest. Secondly, within a regimen, participants are then randomized to receive either active treatment or a matching placebo. In order to maintain blinding to treatment group assignment within a regimen, the matching placebo for a given treatment regimen is designed to be as similar as possible to the active treatment. This includes consideration for the mode of administration (e.g., oral, intravenous, surgical), and a challenge arises if these different modes of administration elicit different placebo effects. If unaccounted for, these differences in placebo response could impact the operating characteristics when comparing any active treatment to shared control, since the shared control arm will be composed of control participants from across regimens.
To assess the impact of differential placebo response on operating characteristics, we considered several hypothetical platform trials, investigating type 1 error rate and power for trials with both binary and continuous primary outcome measures. For the continuous outcome measure, the investigation of differential placebo response was motivated by an upcoming trial in Parkinson’s Disease, the Path to Prevention (P2P) platform trial. We assessed scenarios in which one treatment regimen within the platform trial was experiencing a differential placebo response while other regimens experience an assumed placebo response. For both types of outcome measures, a calculation approach was developed by modifying the traditional power calculation to account for differential placebo response within the shared control arm. This modified calculation involved using a weighted average of the placebo responses, and for the binary outcome, the calculation was further modified to use an updated variance estimator accounting for the two distinct groups within the shared control arm. The estimated operating characteristics from these modified calculation approaches were compared to operating characteristics estimated empirically through simulation studies. Overall, the calculation and simulation approaches provided similar results for both types of outcome measures, indicating that a simulation study may not always be required in order to estimate the potential impacts of differential placebo response in a platform trial.
Operating characteristics for a comparison of each active treatment group to their concurrent shared control group were investigated. The composition of the concurrent shared control arm for a given comparison in a platform trial is dictated by the timing of different regimens throughout the course of the study. Across the hypothetic trials and differential placebo response scenarios considered, there were some common findings. There were settings in which the operating characteristics were only minimally impacted by differential placebo response. As the composition of the shared control arm changed across the settings explored, patterns emerged in how the operating characteristics were impacted. These patterns depended on whether the analysis regimen was the one experiencing a differential placebo response or not. Further, the degree of impact on the operating characteristics was greater when there was a more drastic differential placebo response (i.e., more of a difference between the assumed response and differential response). Overall, this work provides a framework for the evaluation of differential placebo response within a platform trial across a variety of scenarios.
Details
- Title: Subtitle
- Impact of differences in placebo response for platform trials
- Creators
- Megan Elizabeth McCabe
- Contributors
- Emine O. Bayman (Advisor)Christopher S. Coffey (Advisor)Emily Roberts (Committee Member)Kert Viele (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Biostatistics
- Date degree season
- Summer 2024
- Publisher
- University of Iowa
- DOI
- 10.25820/etd.007682
- Number of pages
- xiii, 159 pages
- Copyright
- Copyright 2024 Megan Elizabeth McCabe
- Language
- English
- Date submitted
- 07/22/2024
- Description illustrations
- Illustrations, tables, graphs, charts
- Description bibliographic
- Includes bibliographical references (pages 153-159).
- Public Abstract (ETD)
Clinical trials are critical for establishing the safety and efficacy of an experimental treatment. Utilizing innovative clinical trial designs, such as platform trials, can make this process more efficient. Platform trials evaluate multiple experimental treatments using a single study infrastructure, with each experimental treatment being compared to a shared control arm to determine if it is efficacious. The shared control arm is advantageous because it means that fewer trial participants need to receive inactive control treatments in comparison to traditional clinical trial designs. If multiple experimental treatments were evaluated in separate clinical trials, each trial would require its own control participants to compare to experimental treatments. Even though this shared control arm is advantageous, it can also introduce challenges in the evaluation of treatment efficacy. To prevent trial participants or study staff from knowing a participant’s treatment group assignment, each experimental treatment requires a matching placebo with the same mode of administration (e.g., oral, intravenous) as the active treatment. If different types of placebos elicit distinct placebo effects, this could lead to a shared control arm composed of participants experiencing different responses to placebo. This could impact our ability to correctly identify an ineffective experimental treatment or discern meaningful treatment effects for an effective experimental treatment. This work investigates approaches for estimating these critical statistical properties of a platform trial across a variety of settings and establishes patterns in these characteristics according to the percentage of the shared control participants who experience differential placebo effects.
- Academic Unit
- Biostatistics
- Record Identifier
- 9984698053202771
Metrics
1 File views/ downloads
2 Record Views