A Bayesian nonparametric approach to marginal structural models for point treatments and a continuous or survival outcome

Jason Roy, Kirsten J. Lum, Michael J. Daniels

Research output: Contribution to journalArticle

6 Citations (Scopus)

Abstract

Marginal structural models (MSMs) are a general class of causal models for specifying the average effect of treatment on an outcome. These models can accommodate discrete or continuous treatments, as well as treatment effect heterogeneity (causal effect modification). The literature on estimation of MSM parameters has been dominated by semiparametric estimation methods, such as inverse probability of treatment weighted (IPTW). Likelihood-based methods have received little development, probably in part due to the need to integrate out confounders from the likelihood and due to reluctance to make parametric modeling assumptions. In this article we develop a fully Bayesian MSM for continuous and survival outcomes. In particular, we take a Bayesian nonparametric (BNP) approach, using a combination of a dependent Dirichlet process and Gaussian process to model the observed data. The BNP approach, like semiparametric methods such as IPTW, does not require specifying a parametric outcome distribution. Moreover, by using a likelihood-based method, there are potential gains in efficiency over semiparametric methods. An additional advantage of taking a fully Bayesian approach is the ability to account for uncertainty in our (uncheckable) identifying assumption. To this end, we propose informative prior distributions that can be used to capture uncertainty about the identifying "no unmeasured confounders" assumption. Thus, posterior inference about the causal effect parameters can reflect the degree of uncertainty about this assumption. The performance of the methodology is evaluated in several simulation studies. The results show substantial efficiency gains over semiparametric methods, and very little efficiency loss over correctly specified maximum likelihood estimates. The method is also applied to data from a study on neurocognitive performance in HIV-infected women and a study of the comparative effectiveness of antihypertensive drug classes.

Original languageEnglish (US)
Pages (from-to)32-47
Number of pages16
JournalBiostatistics
Volume18
Issue number1
DOIs
StatePublished - Jan 1 2017
Externally publishedYes

Fingerprint

Marginal Structural Models
Bayesian Nonparametrics
Semiparametric Methods
Likelihood
Causal Effect
Uncertainty
Parametric Modeling
Causal Model
Semiparametric Estimation
Dirichlet Process
Treatment Effects
Prior distribution
Maximum Likelihood Estimate
Gaussian Process
Bayesian Approach
Drugs
Integrate
Simulation Study
Structural model
Methodology

All Science Journal Classification (ASJC) codes

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Cite this

@article{3b9778cc26264040b2a4260804f4c33a,
title = "A Bayesian nonparametric approach to marginal structural models for point treatments and a continuous or survival outcome",
abstract = "Marginal structural models (MSMs) are a general class of causal models for specifying the average effect of treatment on an outcome. These models can accommodate discrete or continuous treatments, as well as treatment effect heterogeneity (causal effect modification). The literature on estimation of MSM parameters has been dominated by semiparametric estimation methods, such as inverse probability of treatment weighted (IPTW). Likelihood-based methods have received little development, probably in part due to the need to integrate out confounders from the likelihood and due to reluctance to make parametric modeling assumptions. In this article we develop a fully Bayesian MSM for continuous and survival outcomes. In particular, we take a Bayesian nonparametric (BNP) approach, using a combination of a dependent Dirichlet process and Gaussian process to model the observed data. The BNP approach, like semiparametric methods such as IPTW, does not require specifying a parametric outcome distribution. Moreover, by using a likelihood-based method, there are potential gains in efficiency over semiparametric methods. An additional advantage of taking a fully Bayesian approach is the ability to account for uncertainty in our (uncheckable) identifying assumption. To this end, we propose informative prior distributions that can be used to capture uncertainty about the identifying {"}no unmeasured confounders{"} assumption. Thus, posterior inference about the causal effect parameters can reflect the degree of uncertainty about this assumption. The performance of the methodology is evaluated in several simulation studies. The results show substantial efficiency gains over semiparametric methods, and very little efficiency loss over correctly specified maximum likelihood estimates. The method is also applied to data from a study on neurocognitive performance in HIV-infected women and a study of the comparative effectiveness of antihypertensive drug classes.",
author = "Jason Roy and Lum, {Kirsten J.} and Daniels, {Michael J.}",
year = "2017",
month = "1",
day = "1",
doi = "10.1093/biostatistics/kxw029",
language = "English (US)",
volume = "18",
pages = "32--47",
journal = "Biostatistics",
issn = "1465-4644",
publisher = "Oxford University Press",
number = "1",

}

A Bayesian nonparametric approach to marginal structural models for point treatments and a continuous or survival outcome. / Roy, Jason; Lum, Kirsten J.; Daniels, Michael J.

In: Biostatistics, Vol. 18, No. 1, 01.01.2017, p. 32-47.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A Bayesian nonparametric approach to marginal structural models for point treatments and a continuous or survival outcome

AU - Roy, Jason

AU - Lum, Kirsten J.

AU - Daniels, Michael J.

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Marginal structural models (MSMs) are a general class of causal models for specifying the average effect of treatment on an outcome. These models can accommodate discrete or continuous treatments, as well as treatment effect heterogeneity (causal effect modification). The literature on estimation of MSM parameters has been dominated by semiparametric estimation methods, such as inverse probability of treatment weighted (IPTW). Likelihood-based methods have received little development, probably in part due to the need to integrate out confounders from the likelihood and due to reluctance to make parametric modeling assumptions. In this article we develop a fully Bayesian MSM for continuous and survival outcomes. In particular, we take a Bayesian nonparametric (BNP) approach, using a combination of a dependent Dirichlet process and Gaussian process to model the observed data. The BNP approach, like semiparametric methods such as IPTW, does not require specifying a parametric outcome distribution. Moreover, by using a likelihood-based method, there are potential gains in efficiency over semiparametric methods. An additional advantage of taking a fully Bayesian approach is the ability to account for uncertainty in our (uncheckable) identifying assumption. To this end, we propose informative prior distributions that can be used to capture uncertainty about the identifying "no unmeasured confounders" assumption. Thus, posterior inference about the causal effect parameters can reflect the degree of uncertainty about this assumption. The performance of the methodology is evaluated in several simulation studies. The results show substantial efficiency gains over semiparametric methods, and very little efficiency loss over correctly specified maximum likelihood estimates. The method is also applied to data from a study on neurocognitive performance in HIV-infected women and a study of the comparative effectiveness of antihypertensive drug classes.

AB - Marginal structural models (MSMs) are a general class of causal models for specifying the average effect of treatment on an outcome. These models can accommodate discrete or continuous treatments, as well as treatment effect heterogeneity (causal effect modification). The literature on estimation of MSM parameters has been dominated by semiparametric estimation methods, such as inverse probability of treatment weighted (IPTW). Likelihood-based methods have received little development, probably in part due to the need to integrate out confounders from the likelihood and due to reluctance to make parametric modeling assumptions. In this article we develop a fully Bayesian MSM for continuous and survival outcomes. In particular, we take a Bayesian nonparametric (BNP) approach, using a combination of a dependent Dirichlet process and Gaussian process to model the observed data. The BNP approach, like semiparametric methods such as IPTW, does not require specifying a parametric outcome distribution. Moreover, by using a likelihood-based method, there are potential gains in efficiency over semiparametric methods. An additional advantage of taking a fully Bayesian approach is the ability to account for uncertainty in our (uncheckable) identifying assumption. To this end, we propose informative prior distributions that can be used to capture uncertainty about the identifying "no unmeasured confounders" assumption. Thus, posterior inference about the causal effect parameters can reflect the degree of uncertainty about this assumption. The performance of the methodology is evaluated in several simulation studies. The results show substantial efficiency gains over semiparametric methods, and very little efficiency loss over correctly specified maximum likelihood estimates. The method is also applied to data from a study on neurocognitive performance in HIV-infected women and a study of the comparative effectiveness of antihypertensive drug classes.

UR - http://www.scopus.com/inward/record.url?scp=85021314425&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85021314425&partnerID=8YFLogxK

U2 - 10.1093/biostatistics/kxw029

DO - 10.1093/biostatistics/kxw029

M3 - Article

C2 - 27345532

AN - SCOPUS:85021314425

VL - 18

SP - 32

EP - 47

JO - Biostatistics

JF - Biostatistics

SN - 1465-4644

IS - 1

ER -