|
Activity Number:
|
402
- Integrative Inference with Data from Multiple Sources: Challenges and New Developments
|
|
Type:
|
Topic-Contributed
|
|
Date/Time:
|
Thursday, August 12, 2021 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
ENAR
|
|
Abstract #317224
|
|
|
Title:
|
Causal Inference for Combining RCTs and Observational Studies: Methods Comparison and Medical Application
|
|
Author(s):
|
Bénédicte Colnet* and Imke Mayer and Guanhua Chen and Awa Dieng and Ruohong Li and Gaël Varoquaux and Jean-Philippe Vert and Julie Josse and Shu Yang
|
|
Companies:
|
INRIA Saclay and Centre d'Analyse et de Mathématiques Sociales, EHESS and Department of Biostatistics and Medical Informatics, University of Winsconsin-Madison and Google Research and Indiana University School of Medicine and INRIA Saclay and Google Research and INRIA Sophia-Antipolis and North Carolina State University
|
|
Keywords:
|
causal effect generalization;
transportability;
external validity;
data integration;
distributional shift;
double robustness
|
|
Abstract:
|
The simultaneous availability of observational and experimental data for the same medical question of a treatment effect is both an opportunity and a theoretical and methodological challenge. In this work we address the question of how to leverage the advantages and information and how to address the shortcomings of both data sources to improve the validity and scope of the treatment effect estimates. This work is motivated by the analysis of a large prospective French database on severely traumatized patients and an international randomized controlled trial (RCT) studying the effect of tranexamic acid administration on mortality among patients with traumatic brain injury. We first discuss identification and estimation methods that improve generalizability of RCTs using the representativeness of observational data. We then discuss methods that combine RCTs and observational data to improve the average treatment effect estimation, handling possible unmeasured confounding in the observational data. We compare the methods with extensive simulations and propose an implementation of the different methods to provide analysis pipelines for reproducible data analyses.
|
Authors who are presenting talks have a * after their name.
