Department of Biostatistics
Adaptive Trials Working Group

2014 - 2015

Organizer: Lorenzo Trippa

Schedule: Mondays, 12:00-1:00 p.m.
Center for Life Sciences Building, Conference Room 11/F

Contract All | Expand All
Seminar Description
This working group is a regular monthly meeting for the faculty and students who are actively working on adaptive trials. The goal would be to share updates on ongoing work, get feedback, potentially initiate new collaborations or projects. All students and faculty are invited to attend and participate.


September 29

Dan Gillen, Ph.D.
Professor, Department of Statistics, University of California - Irvine

"A modified continual reassessment method for balancing individual- and population-ethics in phase I clinical trials"
ABSTRACT: The typical objective of a Phase I clinical trial is to study the toxicity of a treatment and to determine a maximum tolerable dose (MTD) for future patients. O'Quigley, Pepe and Fisher (1990) proposed the continual reassessment method (CRM), a Bayesian adaptive design, to overcome many of the ethical and statistical concerns confronted in dose finding studies where limited data are available. The CRM seeks to treat newly enrolled patients at the posterior mean of the dose corresponding to a pre-specified risk level. Taking a different perspective from the CRM, Whitehead and Brunier (1995) applied Bayesian decision theory to phase I clinical trials wherein the gain function was defined as the inverse of the asymptotic variance of the maximum likelihood estimator for the MTD. The objective of this strategy was hence to optimize dose allocation to maximize the precision of the resulting MTD estimate at the conclusion of the trial. As such, the information gain method of Whitehead and Brunier can be viewed as an allocation procedure concerned with treating future patients (outside of the current trial) by focusing on precision, while the CRM approach can be viewed as an allocation procedure concerned with treating the next trial participant at the current best guess of the MTD even though such an allocation scheme may not yield the most precise estimate of the MTD that will be utilized in future studies. This talk will present a new decision theoretic approach to dose finding in early phase clinical trials that balances individual (current trial participants) and population (future treated patients) ethics when allocating subjects to new dose levels. The work is primarily motivated by our design for a phase I trial in the setting of hyperthermic intraoperative intraperitoneal chemotherapy (HIPEC). Operating characteristics of the proposed methods will be evaluated in the context of the HIPEC trial.



Back to HSPH Biostatistics Maintained by the Biostatistics Webmaster
Last Update: September 23, 2014