Bayesian trials, with frequentist training wheels

The Food and Drug Administration has released a draft “Guidance for Industry” on using Bayesian methodology in clinical trials for drugs and biologics, which has been in the works for several years. It lays out practical use cases (largely forms of “borrowing” information across phases, subgroups, ages/pediatrics, rare diseases, and even across diseases) and discusses priors in some detail, including discounting approaches. But as Witold Wieçek points out, the document reads like Bayesianism with frequentist guardrails: by emphasizing calibration to Type I error rates, it effectively reintroduces frequentist control criteria, and it still keeps applications case by case. Time will tell how it plays out in practice.