Overadjustment bias and unnecessary adjustment in epidemiologic studies

Enrique F. Schisterman, Stephen R. Cole, Robert W. Platf

Research output: Contribution to journalArticle

Abstract

Overadjustment is defined inconsistently. This term is meant to describe control (eg, by regression adjustment, stratification, or restriction) for a variable that either increases net bias or decreases precision without affecting bias. We define overadjust-ment bias as control for an intermediate variable (or a descending proxy for an intermediate variable) on a causal path from exposure to outcome. We define unnecessary adjustment as control for a variable that does not affect bias of the causal relation between exposure and outcome but may affect its precision. We use causal diagrams and an empirical example (the effect of maternal smoking on neonatal mortality) to illustrate and clarify the definition of overadjustment bias, and to distinguish overadjustment bias from unnecessary adjustment. Using simulations, we quantify the amount of bias associated with overadjustment. Moreover, we show that this bias is based on a different causal structure from confounding or selection biases. Overadjustment bias is not a finite sample bias, while inefficiencies due to control for unnecessary variables are a function of sample size.

Original languageEnglish (US)
Pages (from-to)488-495
Number of pages8
JournalEpidemiology
Volume20
Issue number4
DOIs
StatePublished - Jul 2009
Externally publishedYes

Fingerprint

Social Adjustment
Epidemiologic Studies
Selection Bias
Infant Mortality
Proxy
Sample Size
Smoking
Mothers

ASJC Scopus subject areas

  • Epidemiology

Cite this

Overadjustment bias and unnecessary adjustment in epidemiologic studies. / Schisterman, Enrique F.; Cole, Stephen R.; Platf, Robert W.

In: Epidemiology, Vol. 20, No. 4, 07.2009, p. 488-495.

Research output: Contribution to journalArticle

Schisterman, Enrique F. ; Cole, Stephen R. ; Platf, Robert W. / Overadjustment bias and unnecessary adjustment in epidemiologic studies. In: Epidemiology. 2009 ; Vol. 20, No. 4. pp. 488-495.
@article{51343de2c92a42caa2cd40f7d71698a9,
title = "Overadjustment bias and unnecessary adjustment in epidemiologic studies",
abstract = "Overadjustment is defined inconsistently. This term is meant to describe control (eg, by regression adjustment, stratification, or restriction) for a variable that either increases net bias or decreases precision without affecting bias. We define overadjust-ment bias as control for an intermediate variable (or a descending proxy for an intermediate variable) on a causal path from exposure to outcome. We define unnecessary adjustment as control for a variable that does not affect bias of the causal relation between exposure and outcome but may affect its precision. We use causal diagrams and an empirical example (the effect of maternal smoking on neonatal mortality) to illustrate and clarify the definition of overadjustment bias, and to distinguish overadjustment bias from unnecessary adjustment. Using simulations, we quantify the amount of bias associated with overadjustment. Moreover, we show that this bias is based on a different causal structure from confounding or selection biases. Overadjustment bias is not a finite sample bias, while inefficiencies due to control for unnecessary variables are a function of sample size.",
author = "Schisterman, {Enrique F.} and Cole, {Stephen R.} and Platf, {Robert W.}",
year = "2009",
month = "7",
doi = "10.1097/EDE.0b013e3181a819a1",
language = "English (US)",
volume = "20",
pages = "488--495",
journal = "Epidemiology",
issn = "1044-3983",
publisher = "Lippincott Williams and Wilkins",
number = "4",

}

TY - JOUR

T1 - Overadjustment bias and unnecessary adjustment in epidemiologic studies

AU - Schisterman, Enrique F.

AU - Cole, Stephen R.

AU - Platf, Robert W.

PY - 2009/7

Y1 - 2009/7

N2 - Overadjustment is defined inconsistently. This term is meant to describe control (eg, by regression adjustment, stratification, or restriction) for a variable that either increases net bias or decreases precision without affecting bias. We define overadjust-ment bias as control for an intermediate variable (or a descending proxy for an intermediate variable) on a causal path from exposure to outcome. We define unnecessary adjustment as control for a variable that does not affect bias of the causal relation between exposure and outcome but may affect its precision. We use causal diagrams and an empirical example (the effect of maternal smoking on neonatal mortality) to illustrate and clarify the definition of overadjustment bias, and to distinguish overadjustment bias from unnecessary adjustment. Using simulations, we quantify the amount of bias associated with overadjustment. Moreover, we show that this bias is based on a different causal structure from confounding or selection biases. Overadjustment bias is not a finite sample bias, while inefficiencies due to control for unnecessary variables are a function of sample size.

AB - Overadjustment is defined inconsistently. This term is meant to describe control (eg, by regression adjustment, stratification, or restriction) for a variable that either increases net bias or decreases precision without affecting bias. We define overadjust-ment bias as control for an intermediate variable (or a descending proxy for an intermediate variable) on a causal path from exposure to outcome. We define unnecessary adjustment as control for a variable that does not affect bias of the causal relation between exposure and outcome but may affect its precision. We use causal diagrams and an empirical example (the effect of maternal smoking on neonatal mortality) to illustrate and clarify the definition of overadjustment bias, and to distinguish overadjustment bias from unnecessary adjustment. Using simulations, we quantify the amount of bias associated with overadjustment. Moreover, we show that this bias is based on a different causal structure from confounding or selection biases. Overadjustment bias is not a finite sample bias, while inefficiencies due to control for unnecessary variables are a function of sample size.

UR - http://www.scopus.com/inward/record.url?scp=67651021231&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=67651021231&partnerID=8YFLogxK

U2 - 10.1097/EDE.0b013e3181a819a1

DO - 10.1097/EDE.0b013e3181a819a1

M3 - Article

VL - 20

SP - 488

EP - 495

JO - Epidemiology

JF - Epidemiology

SN - 1044-3983

IS - 4

ER -