Commensurate priors for incorporating historical information in clinical trials using general and generalized linear models

Brian P. Hobbs, Daniel J. Sargent, Bradley P. Carlin

Research output: Contribution to journalArticle

58 Citations (Scopus)

Abstract

Assessing between-study variability in the context of conventional random-effects meta-analysis is notoriously diffcult when incorporating data from only a small number of historical studies. In order to borrow strength, historical and current data are often assumed to be fully homogeneous, but this can have drastic consequences for power and Type I error if the historical information is biased. In this paper, we propose empirical and fully Bayesian modifications of the commensurate prior model (Hobbs et al. 2011) extending Pocock (1976), and evaluate their frequentist and Bayesian properties for incorporating patient-level historical data using general and generalized linear mixed regression models. Our proposed commensurate prior models lead to preposterior admissible estimators that facilitate alternative bias-variance trade-offs than those offered by pre-existing methodologies for incorporating historical data from a small number of historical studies. We also provide a sample analysis of a colon cancer trial comparing time-to-disease progression using a Weibull regression model.

Original languageEnglish (US)
Pages (from-to)639-674
Number of pages36
JournalBayesian Analysis
Volume7
Issue number3
DOIs
StatePublished - 2012

Fingerprint

Historical Data
Generalized Linear Model
Clinical Trials
Regression Model
Weibull Model
Type I error
Mixed Model
Random Effects
Progression
Biased
Cancer
Trade-offs
Estimator
Methodology
Evaluate
Alternatives
Model
Context

Keywords

  • Bayesian analysis
  • Clinical trials
  • Correlated data
  • Historical controls
  • Meta-analysis
  • Survival analysis

ASJC Scopus subject areas

  • Applied Mathematics
  • Statistics and Probability

Cite this

Commensurate priors for incorporating historical information in clinical trials using general and generalized linear models. / Hobbs, Brian P.; Sargent, Daniel J.; Carlin, Bradley P.

In: Bayesian Analysis, Vol. 7, No. 3, 2012, p. 639-674.

Research output: Contribution to journalArticle

Hobbs, Brian P. ; Sargent, Daniel J. ; Carlin, Bradley P. / Commensurate priors for incorporating historical information in clinical trials using general and generalized linear models. In: Bayesian Analysis. 2012 ; Vol. 7, No. 3. pp. 639-674.
@article{f67965ec97744846af54020096307133,
title = "Commensurate priors for incorporating historical information in clinical trials using general and generalized linear models",
abstract = "Assessing between-study variability in the context of conventional random-effects meta-analysis is notoriously diffcult when incorporating data from only a small number of historical studies. In order to borrow strength, historical and current data are often assumed to be fully homogeneous, but this can have drastic consequences for power and Type I error if the historical information is biased. In this paper, we propose empirical and fully Bayesian modifications of the commensurate prior model (Hobbs et al. 2011) extending Pocock (1976), and evaluate their frequentist and Bayesian properties for incorporating patient-level historical data using general and generalized linear mixed regression models. Our proposed commensurate prior models lead to preposterior admissible estimators that facilitate alternative bias-variance trade-offs than those offered by pre-existing methodologies for incorporating historical data from a small number of historical studies. We also provide a sample analysis of a colon cancer trial comparing time-to-disease progression using a Weibull regression model.",
keywords = "Bayesian analysis, Clinical trials, Correlated data, Historical controls, Meta-analysis, Survival analysis",
author = "Hobbs, {Brian P.} and Sargent, {Daniel J.} and Carlin, {Bradley P.}",
year = "2012",
doi = "10.1214/12-BA722",
language = "English (US)",
volume = "7",
pages = "639--674",
journal = "Bayesian Analysis",
issn = "1936-0975",
publisher = "Carnegie Mellon University",
number = "3",

}

TY - JOUR

T1 - Commensurate priors for incorporating historical information in clinical trials using general and generalized linear models

AU - Hobbs, Brian P.

AU - Sargent, Daniel J.

AU - Carlin, Bradley P.

PY - 2012

Y1 - 2012

N2 - Assessing between-study variability in the context of conventional random-effects meta-analysis is notoriously diffcult when incorporating data from only a small number of historical studies. In order to borrow strength, historical and current data are often assumed to be fully homogeneous, but this can have drastic consequences for power and Type I error if the historical information is biased. In this paper, we propose empirical and fully Bayesian modifications of the commensurate prior model (Hobbs et al. 2011) extending Pocock (1976), and evaluate their frequentist and Bayesian properties for incorporating patient-level historical data using general and generalized linear mixed regression models. Our proposed commensurate prior models lead to preposterior admissible estimators that facilitate alternative bias-variance trade-offs than those offered by pre-existing methodologies for incorporating historical data from a small number of historical studies. We also provide a sample analysis of a colon cancer trial comparing time-to-disease progression using a Weibull regression model.

AB - Assessing between-study variability in the context of conventional random-effects meta-analysis is notoriously diffcult when incorporating data from only a small number of historical studies. In order to borrow strength, historical and current data are often assumed to be fully homogeneous, but this can have drastic consequences for power and Type I error if the historical information is biased. In this paper, we propose empirical and fully Bayesian modifications of the commensurate prior model (Hobbs et al. 2011) extending Pocock (1976), and evaluate their frequentist and Bayesian properties for incorporating patient-level historical data using general and generalized linear mixed regression models. Our proposed commensurate prior models lead to preposterior admissible estimators that facilitate alternative bias-variance trade-offs than those offered by pre-existing methodologies for incorporating historical data from a small number of historical studies. We also provide a sample analysis of a colon cancer trial comparing time-to-disease progression using a Weibull regression model.

KW - Bayesian analysis

KW - Clinical trials

KW - Correlated data

KW - Historical controls

KW - Meta-analysis

KW - Survival analysis

UR - http://www.scopus.com/inward/record.url?scp=84865835107&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84865835107&partnerID=8YFLogxK

U2 - 10.1214/12-BA722

DO - 10.1214/12-BA722

M3 - Article

AN - SCOPUS:84865835107

VL - 7

SP - 639

EP - 674

JO - Bayesian Analysis

JF - Bayesian Analysis

SN - 1936-0975

IS - 3

ER -