Online Program

Thursday, February 21
PS1 Poster Session 1 & Opening Mixer Thu, Feb 21, 6:30 PM - 8:00 PM
Napoleon Ballroom

Optimal Longitudinal Data Utilization in a Seamless Adaptive (SA) Phase II/III Clinical Trial – Use of the conditional test statistic to identify patient allocation ratio at each trial phase (302453)

View Presentation View Presentation

*Caitlyn Nicole Ellerbe, Medical University of South Carolina 
Jordan Elm, Medical University of South Carolina 
Viswanathan Ramakrishnan , Medical University of South Carolina 
Bruce Turnbull, Cornell University 
Stacia DeSantis, Medical University of South Carolina 
Edward Jauch, Medical University of South Carolina 
Valerie Durkalski, Medical University of South Carolina 

Keywords: Adaptive, Seamless Design, Clinical Trial, Dose Selection, Longitudinal Data, Closed Testing Procedure

In the drug development pipeline there are too few resources to undergo traditional separate phased trials. In response to the FDA’s critical path initiative, SA trials attempt to address this concern by eliminating the time between trials and using information from both stages. Consider a phase II/III trial where the first stage selects the best treatment and the second stage compares it to a control. Existing designs allow for the use of only short term endpoints or a single short term and long term endpoint for stage I and II, respectively. These designs are limited in that they do not utilize intermediate followup data. We propose an SA phase II/III design that conditions (GEE) on all previous observations to improve decisions. This is desirable because the characteristics of GEE are known, decreasing the need for simulation. We also use the proposed approach to achieve an optimal balance (selection vs. testing) in the sample size between the two stages. This approach can improve statistical practice in fields with limited resources for controlled trials where the number of possible candidates to test is large and the outcome is measured over a long duration.