|
Abstract:
|
Many clinical studies in Inflammation and Immunology space include a binary outcome such as response/remission and the primary objective is to estimate the treatment effect at the end of the treatment period. In an ideal scenario, when we do not have any missing data the treatment effect will be the difference in success probabilities of the binary responses between treatment and control groups. However in practice, some patients drop out before the trial ends, and consequently the response is not observed for all the enrolled subjects. Currently in such a scenario, the method primarily used is known as the non-responder imputation (NRI) where missing response is treated as a non-responder or failure. Since this standard technique arbitrarily assumes missing responses as failures, we aim to investigate benefit after incorporating supplementary or surrogate information collected throughout the treatment period. In fact, throughout the treatment period, we collect longitudinal information on several biomarkers or covariates. The objective is to estimate the treatment effect for a binary response utilizing this surrogate information and compare with the NRI method.
|
