The U.S. Food and Drug Administration has released a new draft guidance outlining how sponsors can use Bayesian statistical methods to support the safety and effectiveness of drugs throughout clinical development. The guidance applies broadly to investigational new drug applications (INDs), new drug applications (NDAs), biologics license applications (BLAs), and certain supplemental applications.
The release reflects FDA’s continued effort to modernize clinical trial design and encourage the use of innovative statistical approaches that can improve efficiency without compromising scientific rigor.
Why FDA Is Encouraging Bayesian Approaches
Traditional frequentist statistical methods rely exclusively on data generated within a single clinical trial. Bayesian methods, by contrast, allow sponsors to formally incorporate information from prior studies or external data sources into trial design and analysis.
FDA notes that Bayesian approaches can be particularly useful in adaptive trial designs, dose-finding studies, pediatric extrapolation, and situations where patient populations are limited. When applied appropriately, these methods may support more informed decision-making earlier in development and reduce unnecessary trial burden.
The guidance fulfills a commitment made under the sixth reauthorization of the Prescription Drug User Fee Act (PDUFA VI), which called for greater use of complex innovative trial designs, including Bayesian and adaptive approaches. FDA’s Complex Innovative Trial Design (CID) initiative continues to serve as a key vehicle for advancing these methods.
Key Elements of the Draft Guidance
The draft guidance outlines several core considerations sponsors should address when proposing Bayesian methods:
- Appropriate use cases: FDA clarifies when Bayesian approaches may be suitable and when they may be less appropriate depending on trial objectives and data availability.
- Selection of prior distributions: Sponsors are expected to justify prior assumptions, including the relevance and quality of external data used for borrowing.
- Handling estimands and missing data: The guidance aligns Bayesian trial planning with existing principles for estimands and data handling.
- Software and computation: FDA encourages transparency around analytical tools and model implementation.
- Documentation and reporting: Clear, detailed documentation is required to support regulatory review.
FDA emphasizes that Bayesian methods should be prospectively planned and fully described in protocols and statistical analysis plans.
Illustrative Case Example
The guidance highlights an example in which Bayesian methods were used to incorporate data from an earlier study into a phase 3 trial evaluating a therapy for recurrent Clostridioides difficile infection. By formally borrowing prior information, the sponsor was able to support primary effectiveness analyses while maintaining appropriate control of uncertainty. The product ultimately received FDA approval, illustrating how Bayesian approaches can support regulatory decision-making when applied rigorously.
Early Engagement Remains Critical
FDA strongly encourages sponsors to engage early with the agency when considering Bayesian designs. Submitting protocols, simulation reports, and statistical plans during the design phase allows FDA reviewers to provide feedback before trials are initiated, reducing the risk of misalignment later in development.
How EMMA International Supports Bayesian Trial Strategy
EMMA International supports life sciences organizations in designing, evaluating, and operationalizing innovative clinical trial approaches, including Bayesian and adaptive designs. Our teams help sponsors assess regulatory feasibility, develop statistically and scientifically sound trial strategies, prepare documentation for FDA engagement, and align development programs with evolving regulatory expectations.
As FDA continues to promote advanced trial methodologies, EMMA partners with organizations to ensure innovation is implemented in a way that supports regulatory confidence and long-term program success.
