Drug Evaluation Committee Current Status and Future Issues of Multiplicity Problems in Clinical Trials Explanation for non-statisticians
Data Science Subcommittee
June 2020
In statistical hypothesis testing in drug development, there are two types of errors: the error of judging a test drug as "effective" when it is not, and the error of judging a test drug as "ineffective" when it is effective. Although relatively simple multiplicity adjustment methods such as the Bonferroni method have been used in the past, as the complexity of the hypothesis structure increases and the demand for more powerful methods increases, multiplicity adjustment methods for more complex hypothesis structures, such as the gatekeeping method, have been proposed. As multiplicity adjustment methods become more sophisticated, a deep understanding of the methods is required to use the appropriate method that matches the test objectives and hypothesis structure, and a high level of expertise is required.
Discussion of appropriate hypothesis structures and associated decision-making frameworks that are consistent with the objectives is an extremely important issue that concerns not only statistical analysts but also many other functions related to statistical analysis.
This paper is intended to provide Explanation of the multiplicity issues associated with testing to those involved in drug development who are not specialists in statistics, and we hope that you will find it useful.
Japan Pharmaceutical Manufacturers Association Drug Evaluation Committee Data Science Subcommittee
2019 Task Force 4 Multiplicity in Clinical Trials Coordination Team