Activity Number:
|
77
- Contributed Poster Presentations: Biopharmaceutical Section
|
Type:
|
Contributed
|
Date/Time:
|
Monday, August 3, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Biopharmaceutical Section
|
Abstract #314089
|
|
Title:
|
Imputation Strategies Within the Estimand Framework to Evaluate the Overall Likelihood of Patient Improvement in Longitudinal Trails
|
Author(s):
|
Lysbeth Floden*
|
Companies:
|
Clinical Outcomes Solutions
|
Keywords:
|
Responder analysis;
Clinical trials;
Multiple imputation;
Missing data;
Patient reported outcomes;
Missing at random
|
Abstract:
|
Introduction: In clinical trials continuous outcomes can be dichotomized to compare proportions of patients who achieve meaningful improvement. We defined estimands to evaluate the likelihood of response of patient-reported outcomes (PROs) and demonstrated the use of missing-at-random (MAR) and missing-not-at-random (MNAR) imputation methods within an estimand framework. Methods: We simulated data including adherence indicators and dropout from a two-arm trial measuring a PRO score which was then dichotomized and modeled using a generalized estimating equation. We defined and evaluated a de jure and de facto estmand, considering scenarios with dropout and protocol non-compliance. We evaluated bias relative to the true value and linked the imputation method to the estimand. Results: The estimates from multiple imputation and control-based multiple imputation were similar within all scenarios with dropout (regardless of the true value) but differed from each other. These results suggest that the mechanism of missingness may be less important than defining the estimand and using an appropriate imputation approach.
|
Authors who are presenting talks have a * after their name.