Extracted from Law360
On Jan. 9, the U.S. Food and Drug Administration issued a draft guidance document, "Use of Bayesian Methodology in Clinical Trials of Drug and Biological Products," which clarifies how Bayesian approaches may be used in clinical trials to support regulatory decision-making. It reflects the agency's continued interest in innovative trial designs that may accelerate new drug approvals, especially for treatments addressing unmet medical needs.
The draft guidance also signals the FDA's broader openness to Bayesian methods that could support more efficient evidence generation in generic drug development. This may be especially relevant for programs involving complex generic products, where conventional approaches can be challenging due to low prevalence rates.
The FDA's Gold Standard
Clinical trials form the backbone of the evidence used to establish a drug's safety and effectiveness for regulatory approval. The FDA's current gold standard is the randomized controlled trial, which usually takes four to five years to complete.
These trials rely on frequentist statistics, which assess whether the observed results would be likely to occur by chance if the drug had no effect. Frequentist analysis starts with that assumption — the null hypothesis — and calculates the probability of getting results as extreme as those observed. If that probability is very low, the null hypothesis is rejected and the drug is considered effective.
How Bayesian Methods Work
Bayesian approaches work differently from randomized controlled trials. Instead of relying on a single hypothesis test at the end of a trial, Bayesian methods combine prior knowledge with new data as it becomes available.
Data from earlier studies or real-world evidence can be incorporated as prior information and updated as patient outcomes accrue. This process estimates the probability that a treatment works and allows trial designs to adapt as evidence grows.
The result may be greater efficiency and less patient exposure to ineffective treatments. Historically, however, Bayesian methods have been used mainly in early-phase or exploratory studies, with limited use in large Phase III trials.
FDA Experience With Bayesian Approaches
Although the FDA's traditional standard has been the frequentist randomized controlled trial, the agency has relied on Bayesian methods in certain settings.
For example, the FDA has accepted Bayesian hierarchical models that incorporate data from prior placebo-controlled trials to support the primary effectiveness inference. The FDA has also accepted Bayesian borrowing as supportive evidence in pediatric submissions, where sample sizes are limited and extrapolation is scientifically justified.
Barriers to Wider Adoption
The move toward Bayesian approaches is not without challenges, especially in terms of broad adoption in the scientific community. In 2022, the Drug Information Association Bayesian Scientific Working Group conducted a survey to identify barriers to implementation in clinical trials.
Over half of the respondents expressed interest in exposure to Bayesian methods, but reported little comfort in interpreting Bayesian analyses. Insufficient knowledge of Bayesian approaches was ranked as the most important perceived barrier.
FDA Focus on Innovation
The draft guidance reflects the FDA's growing experience and confidence in applying Bayesian methods to support innovative clinical trial designs.
As part of its commitment under the Prescription Drug User Fee Act VII for fiscal years 2023-2027, the FDA has emphasized enhancing its capacity to review complex and innovative trial designs. These efforts include continued investment in staff expertise and expansion of the paired meeting program to facilitate early and interactive engagement with sponsors.
Consistent with this commitment, the FDA held a public workshop in March 2024 to discuss complex adaptive, Bayesian and other novel clinical trial designs. The workshop emphasized the need for sponsors to address practical design and execution considerations, including prospectively defining decision criteria, evaluating trial operating characteristics through simulation and justifying any borrowing of external information.
FDA Initiatives Supporting Bayesian Methods
In recent years, the FDA has been actively promoting innovation in trial design. Since 2024, its Center for Clinical Trial Innovation has partnered with sponsors to explore Bayesian approaches.
Through the C3TI Demonstration Program, sponsors may submit proposals for nonadaptive Bayesian trials. Under the Complex Innovative Trial Design Meeting Program, sponsors may submit proposals for complex adaptive designs.
These initiatives underscore the FDA's commitment to modernizing clinical research.
FDA leadership has also publicly supported Bayesian adaptive methods. FDA Commissioner Marty Makary, has said that Bayesian methods "help address two of the biggest problems of drug development: high costs and long timelines."
In June 2025, leadership within Center for Biologics Evaluation and Research's cell and gene therapy program confirmed the acceptability of Bayesian trial designs that use real-world or historical clinical trial data to augment control populations or incorporate earlier-phase studies to inform predictions in an experimental arm.
FDA's Drive Toward International Harmonization
Bayesian principles are not new to global regulatory frameworks. ICH E9, Statistical Principles for Clinical Trials, has been in effect since 1998 and permits Bayesian methods when the rationale and robustness are clear.
ICH E11A, Pediatric Extrapolation, finalized in 2024, supports pediatric drug development using Bayesian extrapolation. Most recently, the ICH E20, Adaptive Designs for Clinical Trials, draft guideline highlights the benefits of Bayesian approaches in adaptive confirmatory trials.
This draft guidance also aligns with the European Medicines Agency's 2025-2026 workplan to accelerate clinical trials. The EMA oversees medicines for all European Union member states and countries in the European Economic Area.
FDA and EMA alignment should help reduce uncertainty, streamline trial planning and enable sponsors to design a single study that supports regulatory submissions in multiple jurisdictions.
What's Next for Sponsors
The FDA's draft guidance provides clarity for sponsors interested in using Bayesian methods in clinical trial development. It specifically addresses how Bayesian approaches may be used in Phase III settings to supplement control population data or incorporate prior data from similar trials to help evaluate current data.
The draft guidance also outlines how Bayesian analytics may guide interim adaptations and dose selection. It applies broadly to investigational new drug applications, new drug applications, biologics license applications, and related supplements, indicating FDA support for Bayesian approaches at all stages of drug development.
The draft guidance is especially relevant for sponsors in oncology and rare disease development, where patient recruitment is difficult, data sources are diverse, and trial designs must adapt dynamically.
While the draft guidance is directed at trials supporting new drug approvals, the same Bayesian toolkit may be relevant for generic drug development, particularly for complex generic drugs.
For certain locally acting topical products, the FDA's draft guidance on in vitro permeation test studies recognizes the need to control variability and ensure robust comparisons between the test product and the reference listed drug. In these settings, Bayesian methods can be helpful for incorporating prior information, e.g., pilot IVPT data, to inform decision-making when optimizing study design.
In light of these developments, companies may want to begin assessing whether their programs could benefit from Bayesian approaches and whether internal expertise and training to design and implement Bayesian methods will be needed.
Sponsors should also seek early regulatory input to assess data limitations, determine the appropriate extent of borrowing from prior information and ensure alignment with FDA expectations.