43: The Rumor of One Trial for Substantial Evidence

In this episode of "In the Interim…", host Dr. Scott Berry and frequent co-host Dr. Kert Viele, Senior Statistical Scientist at Berry Consultants, analyze the potential shift in FDA regulatory policy from requiring two independent trials to accepting a single trial as sufficient for “substantial evidence” in drug approvals. Reflecting on the statutory and regulatory definitions originating with the 1962 Federal Food, Drug, and Cosmetic Act and 21 CFR 314.126, they dissect current and emerging interpretations, referencing recent statements by Dr. Martin Makary and coverage described in a STAT article. The conversation focuses on the scientific and statistical foundations of the two-trial threshold, challenges with dichotomous results, and how pooled evidence might increase efficiency and rigor. They discuss statistical implications including alpha thresholds, sample size effects, program power, and the consequences for clinical labeling. The episode also introduces Bayesian approaches as a method for integrating totality of evidence. Attention is given to both population breadth and the possible risks of a narrowed evidentiary base under a single-trial standard.

Key Highlights

• Regulatory and historical context of “substantial evidence” since 1962 and current FDA directives.

• Industry practice: simultaneous Phase III trials, statistical power, and evidentiary replication.

• Criticism of binary, trial-level significance thresholds; merits of pooling or meta-analysis.

• Potential efficiency gains and tradeoffs with a more stringent alpha requirement for single trials.

• Strategic and operational effects on trial design, sample size, and label indications.

• Bayesian statistical approaches for full evidence integration, discussed as an analytical viewpoint.