Biostatistics Weekly Seminar

Design and Analysis of a Two-Phase Study for Multivariate Longitudinal Outcomes

Chiara Di Gravio, PhD
Vanderbilt University

Electronic health records and existing cohort studies provide easily accessible data on phenotype (e.g., disease status, vital signs, gender, age) for most or all study subjects. However, many scientific questions require additional collection of costly variables that are the primary exposures of interest (e.g., DNA genotype, gut microbiome). When the exposure of interest is expensive, researchers might not be able to collect information on every subjects due to resource constraints. Since a limited sample size negatively affects the precision and power of a study, it becomes important to utilize available data in order to identify the most informative subjects for whom the expensive covariate can be collected. Two-phase outcome dependent sampling (ODS) designs are cost-efficient pragmatic solutions that can be used to individuate the most informative subjects in situations where outcome and inexpensive confounders are available, but the exposure of interest needs to be collected.

Motivated by data provided by the Lung Health Study, we will introduce the two-phase ODS designs for scenarios where we have information on two or more longitudinal outcomes. Throughout the talk, we will explore different designs and inference procedures with the aim of estimating genetic associations with lung function decline over time. We will present two inference approaches: a complete-case approach that accounts for subjects with complete data on outcome, exposure and confounders, and an imputation approach that combines information on subjects with complete data with partial information of subjects whose exposure was not measured. Finally, we will show how both approaches can be used I settings where we wish to reuse data previously collected in a two-phase ODS design to study the association between the expensive exposure that had been obtained and a novel longitudinal outcome.

Zoom (Link to Follow)
29 September 2021

Topic revision: r1 - 30 Aug 2021, SimonVandekar

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