145 – Adaptive Design II
Describing Patient-Reported Outcomes from Oncology Trials with Informative Dropouts Due to Adverse Events
Chia-Wen Ko
FDA
Paul Kluetz
FDA
Rajeshwari Sridhara
FDA
Patient-reported outcome (PRO) data are rarely used as primary or secondary endpoints in oncology clinical trials due to the availability of radiographic endpoints as well as trial design characteristics unique to oncology that subject PRO to potential bias, including open label studies and high degrees of missing data. Nevertheless, most oncology pivotal trials have collected PRO data that can be used as part of the overall risk:benefit assessment. It is critical that we find ways to describe and mitigate the challenges associated with PRO data in cancer clinical trials. One important issue when analyzing PRO data is how to take into account informative missing data as a result of treatment discontinuations due to drug-related toxicities. The goal of this work is to compare statistical methods for describing PRO data obtained from oncology patients in the presence of imbalanced treatment discontinuation between study arms. We will apply imputation strategies and modeling to PRO data from oncology randomized pivotal trials with different degrees of imbalance between study arms in the incidence of adverse events leading to treatment discontinuation.