FDA draft guidance clarifies use of Bayesian methods in clinical trials

The US Food and Drug Administration (FDA) has published draft guidance outlining how Bayesian statistical methods can be used in clinical trials of drugs and biological products. The guidance focuses on their application to primary inference and the appropriate use of prior information, and has been released for public consultation to support more efficient trial designs while maintaining regulatory standards for evidence.

Titled ‘Use of Bayesian Methodology in Clinical Trials of Drug and Biological Products; Draft Guidance for Industry,’ the document is intended for sponsors and applicants submitting investigational new drug applications, new drug applications, biologics license applications, and supplemental applications. Announcing the publication, Marty Makary, Commissioner of the FDA, framed the guidance in terms of development efficiency, stating that:

“Bayesian methodologies help address two of the biggest problems of drug development: high costs and long timelines.”

He added that greater clarity on modern statistical approaches would help sponsors “bring more cures and meaningful treatments to patients faster and more affordably.”

The guidance explains Bayesian statistics as an approach in which trial data are combined with relevant prior information to generate a posterior distribution for inference on efficacy or safety. While Bayesian methods may be used across development, the FDA emphasizes that:

“The primary focus of this guidance is on the use of Bayesian methods to support primary inference in clinical trials intended to support the effectiveness and safety of drugs.”

The draft highlights applications including adaptive designs, borrowing from prior trials or external and nonconcurrent controls, pediatric extrapolation, subgroup analyses, and oncology dose-finding, noting that Bayesian methods “may be especially valuable for sponsors targeting rare or pediatric indications, where patient populations are smaller,” and where conventional trial designs can be challenging to implement.

A central theme of the guidance is the expectation that prior distributions, particularly informative priors used to incorporate external information, are systematically constructed and transparently justified. Sponsors are expected to pre-specify priors in the protocol, explain their influence on trial conclusions, and evaluate operating characteristics through simulation. The FDA also highlights the need to assess potential “prior-data conflict,” where observed data are inconsistent with assumptions encoded in the prior, and recommends sensitivity analyses that explore alternative plausible priors and degrees of borrowing. The publication of the draft guidance also fulfils a commitment made under the Prescription Drug User Fee Act VII (PDUFA VII) to enhance the FDA’s capacity to review complex and innovative clinical trial designs, including the use of Bayesian methodologies.

Zhaohui Su, Vice President of Biostatistics at Ontada, described the draft as:

“A significant advancement in the integration of real-world evidence within clinical trials,”

highlighting its support for external and non-concurrent controls when data sources are transparently justified. He added that Bayesian augmented-control designs allow sponsors to borrow from real-world sources while “automatically down-weighting non-exchangeable information,” which he described as particularly important “in the context of rare diseases and small populations.”