Online Program Home
  My Program

Abstract Details

Activity Number: 607 - Recent Advances in Missing Data Methods: From Estimands to Assumptions for Primary and Sensitivity Analyses
Type: Topic Contributed
Date/Time: Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
Sponsor: Biopharmaceutical Section
Abstract #324025
Title: Likelihood Based Approaches for Missing Data with Imputation Guided by the Worst Observed Responses
Author(s): Frank Liu* and Fang Liu and Devan Mehrotra
Companies: Merck & Co. Inc. and Merck & Co., Inc. and Merck & Co. Inc.
Keywords: Dropout ; Missing data ; control-based imputation ; trimmed mean
Abstract:

Despite efforts undertaken to prevent missing data in clinical trials, it is still inevitable to see some early dropouts due to various reasons. The 'true' but unobserved end-of-study outcomes for such dropouts may usually be worse than any observed outcomes because the dropouts have stopped taking their assigned therapy. For a binary endpoint, the 'dropout equals failure' approach has been widely applied. However, there is no similar approach for a continuous endpoint. The conventional baseline carried forward approaches suffer from drawbacks associated with single imputation. Commonly used mixed model repeated measures (MMRM) analysis requires a missing at random assumption which may not realistically reflect the poor response for dropouts. We propose a likelihood based method to assess treatment effect which assumes that all missing outcomes are worse than the observed responses for each treatment group. Simulations are used to compare this method with the trimmed mean approach using quantile regression, control-based patient-level imputation and control-based mean imputation. Applications to real clinical trial examples are presented for illustration.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association