TITLE: Beyond Bonferroni Correction – Innovation, Intuition and Common Sense
INSTRUCTOR: Qian Helen Li, BMS
In drug development and evaluation, the need for multiplicity adjustment exists in almost all phase III studies. Despite the abundant choices of statistical methods, questions remain. For example, when only one dose is included in a Phase III study, a positive study can be claimed if the 1-side p-value of the primary endpoint is 0.021. However, when two doses are included in a study and the multiplicity adjustment procedure is applied, a result such that the 1-sided p-value of the low dose is 0.021 and that of the high dose is 0.026 may not be considered as a positive study. The counterintuitive conclusions lead to a gap in common sense, but nonetheless open opportunities for innovation. An innovated statistical method for multiplicity adjustment that can be used to address such questions will be introduced and discussed. Centered on the new method, the concept of consistent and collective evidence is introduced. Cases of application in clinical trials will be illustrated. In particular, a case of successful labeling an indication using a secondary endpoint which is in the hierarchy that the primary endpoint was failed.
Dr. Li has over 20 years of experience in the field of clinical statistics and worked on a range of therapeutic areas including oncology, cardiovascular, pulmonary, pain, ophthalmic, anti-inflammatory and anti-infective drug products. Her publications cover innovative statistical methods including the area of multiplicity and survival analyses. She currently works for Bristol Myers Squibb and was an experienced statistical reviewer in FDA and NIH. She received her doctoral degree in Biostatistics from Harvard School of Public Health, had a master degree from Purdue University and undergraduate degree from Tsinghua University.