Description
- Abstract:
- Individual-Level Data (IPD) approach has become increasingly popular as it does not depend on published results. Combining IPD across multiple studies usually adopts a multilevel framework with random study effects and intercepts. Such models include two variance components for the random effects. Our study delves into various modeling considerations in Bayesian estimations. Chapter 1 focuses on the meta-analysis of randomized control trials (RCTs), while Chapter 2 concentrates on aggregating N-of-1 trials. Previous studies on meta-analyses of RCTs have suggested that likelihood-based estimates of these variances may be biased by an amount that might depend on how the treatment variable is coded and estimated in the frequentist framework. Whether this bias carries over to a Bayesian formulation or to the application of N-of-1 trials is unknown. Through extensive simulations, we explore the performance of variance estimation (bias, mean squared error, coverage and precision) under a variety of different models (fixed and random intercepts, different codings of treatment, auto-correlations), sample sizes (numbers of trials and lengths of each trial), amounts of within and between-study variance and formulations of prior distributions when outcomes are continuous. We conclude that careful choice of model can improve the accuracy of variance estimation and that accuracy varies substantially depending on the underlying variation as well as the number and size of studies. In Chapter 3, we propose an instrumental variable approach to address the noncompliance issue in N-of-1 trials. We describe a causal framework using potential treatment selection paths and potential outcome paths. Two estimands of individual causal effect are defined: 1) the effect of continuous exposure to alcohol or daily drinking, and 2) the effect of an individual’s usual drinking behavior. Our Bayesian IV structural model accounts for confounding due to self-selection of exposure (noncompliance), auto-correlation in study outcomes (interference among study outcomes), and is able to estimate functions of model parameters to solve the issue of non-collapsibility. The simulation study showed our method largely reduced bias and greatly improved the coverage of the estimated causal effect, compared to existing methods (ITT, PP, and AT). We apply the method to I-STOP-AFib Study to estimate the individual effect of alcohol on AF occurrence.
- Notes:
- Thesis (Ph. D.)--Brown University, 2024
Citation
Qu, Kexin,
"Bayesian Methods for Analyzing Individual-Level Data from Multiple Trials Considering Correlation, Non-compliance and N-of-1 Designs"
(2024).
Biostatistics Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://repository.library.brown.edu/studio/item/bdr:jyvyk52t/
Relations
Collection:
-
Biostatistics Theses and Dissertations
Theses and Dissertations for the Biostatistics department....