Online Program

Return to main conference page
Tuesday, September 24
Tue, Sep 24, 11:45 AM - 1:00 PM
Various Rooms
Roundtable Discussions

TL10: How to Statistically Test Up-Titration Effect on Efficacy of Binary Endpoints for Post-Randomization Dose Up-Titration in Phase 3/4 Clinical Trials: Simulation and Implementation (300818)

*HONG DING, Novartis 

Keywords: dose up-titration, phase 3/4 randomized clinical trials, conditional logistic regression,

Room: McKinley

In long term randomized clinical trials with multiple doses of investigative drug, it is often found later that one dose has better efficacy than another one. So it is often a decision after primary analysis time point that one dose can be up-titrated to a higher dose for subjects whose signs and symptoms are not fully controlled with the current dose, and may improve with higher dose as judged by investigator for the benefit of patients.

Statistically testing if there is a significant effect on efficacy over time from this up-titration and if so, how much is a challenge that has not been addressed widely, particularly for binary endpoints, but has been of great interest from clinical team and regulatory agency as it is closely related to label claim for a particular dose for confirmatory clinical trials.

Currently in Novartis, the efficacy effect of post-randomization dose up-titration for binary endpoints in phase 3 or 4 clinical trials is illustrated by Sankey plot , a data visualization method, as an evidence of efficacy improvement for dose claim to regulatory agencies. Is there a rigorous statistical method that can test this effect? This study uses two simulations of correlated longitudinal data to prove that Conditional Logistic Regression is a valid method for this purpose and provies implementation guide for the application of this method in Phase 3 /4 clinical trials and a real example.