Biostatistics Weekly Seminar

Evidence-based analysis of longitudinal data from randomized clinical trials

Patrick Heagerty, PhD
University of Washington

Prospective randomized longitudinal studies are the foundation for rigorous generation of medical evidence. Increasing focus on patient-reported outcomes and serial measures of health status imply that quantitative outcomes are common primary outcomes in contemporary trials. Frequently the study design and protocol specify the collection of outcomes both at baseline and select follow-up times, with a primary analysis that often considers one key follow-up time. The goal of this talk is to review the statistical evidence that informs the selection of a primary longitudinal analysis strategy. Seminal results of Frison and Pocock (1992) and Yang and Tsiatis (2001) form the basis for leveraging model-assisted inference with repeated measures. We review these fundamental results and generalize to multiple follow-up measures. We emphasize the advantage of using known assumptions such as randomization to enhance precision while not making additional assumptions that can threaten validity of model-based or model-assisted inference. We illustrate ideas using a recently published trial of depression treatment and highlight how existing software can be adopted for evidence-based analysis.

Zoom (Link to Follow)
19 May 2021

Speaker Itinerary

Topic revision: r3 - 14 May 2021, AndrewSpieker

This site is powered by FoswikiCopyright © 2013-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback