Back to search

Boston University

Statistical methods for index trait bias in genetic association studies of subsequent traits

Abstract

dc:description.abstract

Genome-Wide Association Studies (GWAS) are widely used to obtain genetic associations for disease or trait outcomes. GWAS of subsequent traits can be useful in understanding pathways of disease progression, however, such analyses are affected by Index Trait Bias. Index-Trait bias (also called Index-Event Bias) can occur in genetic studies due to conditioning on an incident trait, which can bias genetic associations with subsequent traits. In this dissertation I develop novel methods to correct for index trait bias, analyze how these novel methods fare against existing methods, and observe behaviors of bias correction methods in differing simulation scenarios. First, I adapt two Bayesian Mendelian Randomization (MR) methods, Bayesian Weighted MR and MR-HORSE, for index trait bias correction, and compare performance of these methods with existing bias correction methods through simulation. Bayesian methods can be more robust in their parameter estimation and could provide reasonable bias corrected estimates across differing simulation scenarios. I then applied these methods to a real dataset for incident Acute Ischemic Stroke (AIS) and subsequent AIS or subsequent Major Adverse Cardiovascular Event (MACE), as well as for Body Mass Index (BMI) and BMI adjusted for Fasting Insulin (FI). Second, I use Bayesian Shrinkage methods that estimate bias at a Single Nucleotide Polymorphism (SNP) level, since not all SNPs are affected by index trait bias. I also test performance of these methods in the instance when there is less or zero index trait bias, which can be controlled in simulation using the variance of the shared non-genetic confounder between incident and subsequent event. I evaluate these methods, alongside existing methods, in a simulation framework to assess the performance of existing index trait bias correction methods in samples with varying amounts of bias. Finally, since index trait bias methods have typically been evaluated using only continuous traits, I evaluate all methods in simulation settings where either the incident or both the incident and the subsequent traits are binary. This dissertation provides insights into the performance of index trait bias correction under different scenarios and provides guidance on index trait bias methods to correct for it to practitioners looking to study the genetics of subsequent traits or events.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy
Level thesis:degree_level
doctoral
Discipline thesis:degree_discipline
Biostatistics
Grantor
Boston University
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Adhikari, Nimish
Advisor dc:contributor.advisor
  • Peloso, Gina M
ORCID dc:identifier.orcid

Subjects

dc:subject × 1

Rights

dc:rights
Statement dc:rights
  • Attribution 4.0 International
Language dc:language.iso
en_US

Identifiers

dc:identifier.*
Handle dc:identifier.uri
https://hdl.handle.net/2144/53784

Chain of custody

source
Harvested from
Boston University
Base URL
open.bu.edu/oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
related terms
citation

Adhikari, Nimish. Statistical methods for index trait bias in genetic association studies of subsequent traits. doctoral thesis, Boston University, 2026. https://hdl.handle.net/2144/53784