A comparison of approaches to control for confounding factors by regression models

Guan Xing, Chang Yun Lin, Chao Xing

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

A common technique to control for confounding factors in practice is by regression adjustment. There are various versions of regression modeling in the literature, and in this paper we considered four approaches often seen in genetic association studies. We carried out both analytical and simulation studies comparing the bias of effect size estimates and examining the test sizes under the null hypothesis of no association between an outcome and an exposure. Further, we compared the methods in a nonsynonymous genome-wide scan for plasma lipoprotein(a) levels using a dataset from the Dallas Heart Study. We found that a widely employed approach that models the covariate-adjusted outcome and the exposure leads to an infranominal test size and underestimation of the exposure effect size. In conclusion, we recommend either using multiple regression models or modeling the covariate-adjusted outcome and the covariate-adjusted exposure to control for confounding factors.

Original languageEnglish (US)
Pages (from-to)194-205
Number of pages12
JournalHuman Heredity
Volume72
Issue number3
DOIs
StatePublished - Nov 2011

Fingerprint

Lipoprotein(a)
Genetic Association Studies
Genome
Datasets

Keywords

  • Adjustment
  • Confounding factor
  • Linear regression

ASJC Scopus subject areas

  • Genetics(clinical)
  • Genetics

Cite this

A comparison of approaches to control for confounding factors by regression models. / Xing, Guan; Lin, Chang Yun; Xing, Chao.

In: Human Heredity, Vol. 72, No. 3, 11.2011, p. 194-205.

Research output: Contribution to journalArticle

Xing, Guan ; Lin, Chang Yun ; Xing, Chao. / A comparison of approaches to control for confounding factors by regression models. In: Human Heredity. 2011 ; Vol. 72, No. 3. pp. 194-205.
@article{d34ff4940fe14b47aa6486b129b39560,
title = "A comparison of approaches to control for confounding factors by regression models",
abstract = "A common technique to control for confounding factors in practice is by regression adjustment. There are various versions of regression modeling in the literature, and in this paper we considered four approaches often seen in genetic association studies. We carried out both analytical and simulation studies comparing the bias of effect size estimates and examining the test sizes under the null hypothesis of no association between an outcome and an exposure. Further, we compared the methods in a nonsynonymous genome-wide scan for plasma lipoprotein(a) levels using a dataset from the Dallas Heart Study. We found that a widely employed approach that models the covariate-adjusted outcome and the exposure leads to an infranominal test size and underestimation of the exposure effect size. In conclusion, we recommend either using multiple regression models or modeling the covariate-adjusted outcome and the covariate-adjusted exposure to control for confounding factors.",
keywords = "Adjustment, Confounding factor, Linear regression",
author = "Guan Xing and Lin, {Chang Yun} and Chao Xing",
year = "2011",
month = "11",
doi = "10.1159/000332743",
language = "English (US)",
volume = "72",
pages = "194--205",
journal = "Human Heredity",
issn = "0001-5652",
publisher = "S. Karger AG",
number = "3",

}

TY - JOUR

T1 - A comparison of approaches to control for confounding factors by regression models

AU - Xing, Guan

AU - Lin, Chang Yun

AU - Xing, Chao

PY - 2011/11

Y1 - 2011/11

N2 - A common technique to control for confounding factors in practice is by regression adjustment. There are various versions of regression modeling in the literature, and in this paper we considered four approaches often seen in genetic association studies. We carried out both analytical and simulation studies comparing the bias of effect size estimates and examining the test sizes under the null hypothesis of no association between an outcome and an exposure. Further, we compared the methods in a nonsynonymous genome-wide scan for plasma lipoprotein(a) levels using a dataset from the Dallas Heart Study. We found that a widely employed approach that models the covariate-adjusted outcome and the exposure leads to an infranominal test size and underestimation of the exposure effect size. In conclusion, we recommend either using multiple regression models or modeling the covariate-adjusted outcome and the covariate-adjusted exposure to control for confounding factors.

AB - A common technique to control for confounding factors in practice is by regression adjustment. There are various versions of regression modeling in the literature, and in this paper we considered four approaches often seen in genetic association studies. We carried out both analytical and simulation studies comparing the bias of effect size estimates and examining the test sizes under the null hypothesis of no association between an outcome and an exposure. Further, we compared the methods in a nonsynonymous genome-wide scan for plasma lipoprotein(a) levels using a dataset from the Dallas Heart Study. We found that a widely employed approach that models the covariate-adjusted outcome and the exposure leads to an infranominal test size and underestimation of the exposure effect size. In conclusion, we recommend either using multiple regression models or modeling the covariate-adjusted outcome and the covariate-adjusted exposure to control for confounding factors.

KW - Adjustment

KW - Confounding factor

KW - Linear regression

UR - http://www.scopus.com/inward/record.url?scp=80855143822&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=80855143822&partnerID=8YFLogxK

U2 - 10.1159/000332743

DO - 10.1159/000332743

M3 - Article

C2 - 22086357

AN - SCOPUS:80855143822

VL - 72

SP - 194

EP - 205

JO - Human Heredity

JF - Human Heredity

SN - 0001-5652

IS - 3

ER -