Methods to apply results from randomized trials to patients in clinical practice / Issa J. Dahabreh and Elizabeth A. Stuart.

Format:

Book

Author/Creator:

Dahabreh, Issa J., author.

Stuart, Elizabeth A., author.

Series:

Final research report (Patient-Centered Outcomes Research Institute (U.S.))

Final research report

Language:

English

Subjects (All):

Clinical medicine.

Randomized controlled trials.

Physical Description:

1 online resource.

Place of Publication:

Washington, DC : Patient-Centered Outcomes Research Institute (PCORI), 2021.

Summary:

TITLE AS IT APPEARS ON THE CONTRACT: Making Better Use of Randomized Trials: Assessing Applicability and Transporting Causal Effects ABSTRACT: BACKGROUND: Concerns that the results of randomized trials do not apply to patient populations seen in clinical practice are commonplace. OBJECTIVES: To provide methods for extending - generalizing or transporting - causal inferences from one or more randomized trials to a target population. METHODS AND RESULTS: This project developed methods for study design, identification (with graphical causal models or potential outcomes), estimation, and sensitivity analysis for counterfactual means and average treatment effects in a target population in which no experimentation is possible by extending (generalizing or transporting) inferences from one or more completed randomized trials. In this report we review the project's rationale, provide an accessible description of key results when extending inferences from a single trial to a new target population, and briefly describe connections across other project components. LIMITATIONS: Further development of the methods is needed to address complications such as missing data, measurement error, clustering, or unidentifiable overlap between randomized trials and observational datasets used as samples from target populations. Future applications of the methods will require careful specification of the target population of substantive interest, and the collection of data from that population. CONCLUSIONS: We provide a toolbox that can be used to reason about and perform statistical analyses that extend causal inferences from one or more randomized trials to a target population. Additional work is needed to expand this toolbox to address more complicated study designs and data structures.

Notes:

Description based on publisher supplied metadata and other sources.