Fitting marginal structural models: estimating covariate-treatment associations in the reweighted data set can guide model fitting

Eleanor M. Pullenayegum, Catherine Lam, Cedric Manlhiot, Brian M. Feldman

Research output: Contribution to journalArticle

Abstract

Objectives: Marginal structural models (MSMs) are an increasingly popular framework in which observational data are reweighted to draw causal inferences. The ability of an MSM to accomplish this rests crucially on treatment being unconfounded by covariates in the inverse probability-of-treatment weighted data set. This paper discusses how this important property of the weights can be evaluated when both the treatment and the covariates are time dependent. Often treatment effect estimates will be sensitive to the choice of weights. A framework within which to explore reasons for this sensitivity and to select a good weighting scheme is suggested. Study Design and Setting: The methods are illustrated using an observational study of intravenous immunoglobulin for the treatment of juvenile dermatomyositis. Results: Using traditional methods for fitting the probability-of-treatment model leaves important associations between treatment and covariates. Augmenting the probability-of-treatment model accordingly both reduces the confounding and alters the treatment effect estimate. Conclusions: Traditional model-fitting strategies for the probability-of-treatment model may leave important associations between treatment and covariates in the reweighted data set. The framework described in this article can help both to detect this and to formulate a better probability-of-treatment model.

Original languageEnglish (US)
Pages (from-to)875-881
Number of pages7
JournalJournal of Clinical Epidemiology
Volume61
Issue number9
DOIs
StatePublished - Sep 1 2008
Externally publishedYes

Fingerprint

Structural Models
Weights and Measures
Intravenous Immunoglobulins
Observational Studies
Datasets

Keywords

  • Causal inference
  • Confounding
  • Goodness-of-fit
  • juvenile dermatomyositis
  • Marginal structural models
  • Sensitivity analysis

ASJC Scopus subject areas

  • Epidemiology

Cite this

Fitting marginal structural models : estimating covariate-treatment associations in the reweighted data set can guide model fitting. / Pullenayegum, Eleanor M.; Lam, Catherine; Manlhiot, Cedric; Feldman, Brian M.

In: Journal of Clinical Epidemiology, Vol. 61, No. 9, 01.09.2008, p. 875-881.

Research output: Contribution to journalArticle

@article{2b73fc0682d94062a92507afe25c9fd2,
title = "Fitting marginal structural models: estimating covariate-treatment associations in the reweighted data set can guide model fitting",
abstract = "Objectives: Marginal structural models (MSMs) are an increasingly popular framework in which observational data are reweighted to draw causal inferences. The ability of an MSM to accomplish this rests crucially on treatment being unconfounded by covariates in the inverse probability-of-treatment weighted data set. This paper discusses how this important property of the weights can be evaluated when both the treatment and the covariates are time dependent. Often treatment effect estimates will be sensitive to the choice of weights. A framework within which to explore reasons for this sensitivity and to select a good weighting scheme is suggested. Study Design and Setting: The methods are illustrated using an observational study of intravenous immunoglobulin for the treatment of juvenile dermatomyositis. Results: Using traditional methods for fitting the probability-of-treatment model leaves important associations between treatment and covariates. Augmenting the probability-of-treatment model accordingly both reduces the confounding and alters the treatment effect estimate. Conclusions: Traditional model-fitting strategies for the probability-of-treatment model may leave important associations between treatment and covariates in the reweighted data set. The framework described in this article can help both to detect this and to formulate a better probability-of-treatment model.",
keywords = "Causal inference, Confounding, Goodness-of-fit, juvenile dermatomyositis, Marginal structural models, Sensitivity analysis",
author = "Pullenayegum, {Eleanor M.} and Catherine Lam and Cedric Manlhiot and Feldman, {Brian M.}",
year = "2008",
month = "9",
day = "1",
doi = "10.1016/j.jclinepi.2007.10.024",
language = "English (US)",
volume = "61",
pages = "875--881",
journal = "Journal of Clinical Epidemiology",
issn = "0895-4356",
publisher = "Elsevier USA",
number = "9",

}

TY - JOUR

T1 - Fitting marginal structural models

T2 - estimating covariate-treatment associations in the reweighted data set can guide model fitting

AU - Pullenayegum, Eleanor M.

AU - Lam, Catherine

AU - Manlhiot, Cedric

AU - Feldman, Brian M.

PY - 2008/9/1

Y1 - 2008/9/1

N2 - Objectives: Marginal structural models (MSMs) are an increasingly popular framework in which observational data are reweighted to draw causal inferences. The ability of an MSM to accomplish this rests crucially on treatment being unconfounded by covariates in the inverse probability-of-treatment weighted data set. This paper discusses how this important property of the weights can be evaluated when both the treatment and the covariates are time dependent. Often treatment effect estimates will be sensitive to the choice of weights. A framework within which to explore reasons for this sensitivity and to select a good weighting scheme is suggested. Study Design and Setting: The methods are illustrated using an observational study of intravenous immunoglobulin for the treatment of juvenile dermatomyositis. Results: Using traditional methods for fitting the probability-of-treatment model leaves important associations between treatment and covariates. Augmenting the probability-of-treatment model accordingly both reduces the confounding and alters the treatment effect estimate. Conclusions: Traditional model-fitting strategies for the probability-of-treatment model may leave important associations between treatment and covariates in the reweighted data set. The framework described in this article can help both to detect this and to formulate a better probability-of-treatment model.

AB - Objectives: Marginal structural models (MSMs) are an increasingly popular framework in which observational data are reweighted to draw causal inferences. The ability of an MSM to accomplish this rests crucially on treatment being unconfounded by covariates in the inverse probability-of-treatment weighted data set. This paper discusses how this important property of the weights can be evaluated when both the treatment and the covariates are time dependent. Often treatment effect estimates will be sensitive to the choice of weights. A framework within which to explore reasons for this sensitivity and to select a good weighting scheme is suggested. Study Design and Setting: The methods are illustrated using an observational study of intravenous immunoglobulin for the treatment of juvenile dermatomyositis. Results: Using traditional methods for fitting the probability-of-treatment model leaves important associations between treatment and covariates. Augmenting the probability-of-treatment model accordingly both reduces the confounding and alters the treatment effect estimate. Conclusions: Traditional model-fitting strategies for the probability-of-treatment model may leave important associations between treatment and covariates in the reweighted data set. The framework described in this article can help both to detect this and to formulate a better probability-of-treatment model.

KW - Causal inference

KW - Confounding

KW - Goodness-of-fit

KW - juvenile dermatomyositis

KW - Marginal structural models

KW - Sensitivity analysis

UR - http://www.scopus.com/inward/record.url?scp=48349086125&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=48349086125&partnerID=8YFLogxK

U2 - 10.1016/j.jclinepi.2007.10.024

DO - 10.1016/j.jclinepi.2007.10.024

M3 - Article

C2 - 18486447

AN - SCOPUS:48349086125

VL - 61

SP - 875

EP - 881

JO - Journal of Clinical Epidemiology

JF - Journal of Clinical Epidemiology

SN - 0895-4356

IS - 9

ER -