Communicating statistical concerns to the FDA

Does anyone have insight into or experience with reporting concerns to the FDA regarding their approval of drugs based on inappropriate statistical methods and endpoints, such as but not limited to comparing median percent change from baseline (PCHG) in seizures between groups, and comparing the responder rate derived from PCHG between groups?

A paper discussing the lack of a breakthrough in seizure control despite an exponential growth of approved anti-epileptic drugs provides evidence that the use of the above poor methods and endpoints has persisted for over 30 years and remains the golden standard for epilepsy drug trials to this day. Here are a few recent studies of the sort (L-DEE-2024, E-FS-2023, Z-FS-2021, O-MS-2020). Iā€™m afraid this trend will persist for as long as the regulatory agencies deem these acceptable. I hope to see real breakthroughs in this field, which would require applying the appropriate methodology and endpoint for these studies, and we need this message to come directly from the regulatory agencies.

1 Like

Extra effort is needed to budget from accepted practice. Real examples showing how analyses are misleading would help. Also, pharma would do well to learn that poor methods require larger sample sizes so waste $.


Thank you. A few thoughts from one patient advocate to another. Explain your concerns also to patient advocacy groups - such as Identify the patient advocates who consult FDA or are represented at ODAC hearings as advisors to FDA. Attend conferences and speak at venues that present on clinical research findings for the disease.

A starting point for how to communicate with the FDA provided by the FDA.


Thank you both for your feedback. I will look into these and share any learnings/progress along the way.