Computationally simple analysis of matched, outcome-based studies of ordinal disease states

Rebecca Betensky, Jackie Szymonifka, Eudocia Q. Lee, Catherine L. Nutt, Tracy T. Batchelor

Research output: Contribution to journalArticle

Abstract

Outcome-based sampling is an efficient study design for rare conditions, such as glioblastoma. It is often used in conjunction with matching, for increased efficiency and to potentially avoid bias due to confounding. A study was conducted at the Massachusetts General Hospital that involved retrospective sampling of glioblastoma patients with respect to multiple-ordered disease states, as defined by three categories of overall survival time. To analyze such studies, we posit an adjacent categories logit model and exploit its allowance for prospective analysis of a retrospectively sampled study and its advantageous removal of set and level specific nuisance parameters through conditioning on sufficient statistics. This framework allows for any sampling design and is not limited to one level of disease within each set, such as in previous publications. We describe how this ordinal conditional model can be fit using standard conditional logistic regression procedures. We consider an alternative pseudo-likelihood approach that potentially offers robustness under partial model misspecification at the expense of slight loss of efficiency under correct model specification for small sample sizes. We apply our methods to the Massachusetts General Hospital glioblastoma study.

Original languageEnglish (US)
Pages (from-to)2514-2527
Number of pages14
JournalStatistics in Medicine
Volume34
Issue number17
DOIs
StatePublished - Jul 30 2015

Fingerprint

Glioblastoma
Conditional Logistic Regression
Outcome Assessment (Health Care)
General Hospitals
Pseudo-likelihood
Logit Model
Conditional Model
Model Misspecification
Sufficient Statistics
Sampling Design
Model Specification
Confounding
Survival Time
Nuisance Parameter
Small Sample Size
Logistic Models
Conditioning
Adjacent
Robustness
Partial

Keywords

  • Adjacent categories model
  • Conditional logistic regression
  • Glioblastoma
  • Outcome-based sampling
  • Pseudo-likelihood

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability

Cite this

Computationally simple analysis of matched, outcome-based studies of ordinal disease states. / Betensky, Rebecca; Szymonifka, Jackie; Lee, Eudocia Q.; Nutt, Catherine L.; Batchelor, Tracy T.

In: Statistics in Medicine, Vol. 34, No. 17, 30.07.2015, p. 2514-2527.

Research output: Contribution to journalArticle

Betensky, Rebecca ; Szymonifka, Jackie ; Lee, Eudocia Q. ; Nutt, Catherine L. ; Batchelor, Tracy T. / Computationally simple analysis of matched, outcome-based studies of ordinal disease states. In: Statistics in Medicine. 2015 ; Vol. 34, No. 17. pp. 2514-2527.
@article{d3dd5f01b13f40b3b5c7e1c0a961bbc1,
title = "Computationally simple analysis of matched, outcome-based studies of ordinal disease states",
abstract = "Outcome-based sampling is an efficient study design for rare conditions, such as glioblastoma. It is often used in conjunction with matching, for increased efficiency and to potentially avoid bias due to confounding. A study was conducted at the Massachusetts General Hospital that involved retrospective sampling of glioblastoma patients with respect to multiple-ordered disease states, as defined by three categories of overall survival time. To analyze such studies, we posit an adjacent categories logit model and exploit its allowance for prospective analysis of a retrospectively sampled study and its advantageous removal of set and level specific nuisance parameters through conditioning on sufficient statistics. This framework allows for any sampling design and is not limited to one level of disease within each set, such as in previous publications. We describe how this ordinal conditional model can be fit using standard conditional logistic regression procedures. We consider an alternative pseudo-likelihood approach that potentially offers robustness under partial model misspecification at the expense of slight loss of efficiency under correct model specification for small sample sizes. We apply our methods to the Massachusetts General Hospital glioblastoma study.",
keywords = "Adjacent categories model, Conditional logistic regression, Glioblastoma, Outcome-based sampling, Pseudo-likelihood",
author = "Rebecca Betensky and Jackie Szymonifka and Lee, {Eudocia Q.} and Nutt, {Catherine L.} and Batchelor, {Tracy T.}",
year = "2015",
month = "7",
day = "30",
doi = "10.1002/sim.6503",
language = "English (US)",
volume = "34",
pages = "2514--2527",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",
number = "17",

}

TY - JOUR

T1 - Computationally simple analysis of matched, outcome-based studies of ordinal disease states

AU - Betensky, Rebecca

AU - Szymonifka, Jackie

AU - Lee, Eudocia Q.

AU - Nutt, Catherine L.

AU - Batchelor, Tracy T.

PY - 2015/7/30

Y1 - 2015/7/30

N2 - Outcome-based sampling is an efficient study design for rare conditions, such as glioblastoma. It is often used in conjunction with matching, for increased efficiency and to potentially avoid bias due to confounding. A study was conducted at the Massachusetts General Hospital that involved retrospective sampling of glioblastoma patients with respect to multiple-ordered disease states, as defined by three categories of overall survival time. To analyze such studies, we posit an adjacent categories logit model and exploit its allowance for prospective analysis of a retrospectively sampled study and its advantageous removal of set and level specific nuisance parameters through conditioning on sufficient statistics. This framework allows for any sampling design and is not limited to one level of disease within each set, such as in previous publications. We describe how this ordinal conditional model can be fit using standard conditional logistic regression procedures. We consider an alternative pseudo-likelihood approach that potentially offers robustness under partial model misspecification at the expense of slight loss of efficiency under correct model specification for small sample sizes. We apply our methods to the Massachusetts General Hospital glioblastoma study.

AB - Outcome-based sampling is an efficient study design for rare conditions, such as glioblastoma. It is often used in conjunction with matching, for increased efficiency and to potentially avoid bias due to confounding. A study was conducted at the Massachusetts General Hospital that involved retrospective sampling of glioblastoma patients with respect to multiple-ordered disease states, as defined by three categories of overall survival time. To analyze such studies, we posit an adjacent categories logit model and exploit its allowance for prospective analysis of a retrospectively sampled study and its advantageous removal of set and level specific nuisance parameters through conditioning on sufficient statistics. This framework allows for any sampling design and is not limited to one level of disease within each set, such as in previous publications. We describe how this ordinal conditional model can be fit using standard conditional logistic regression procedures. We consider an alternative pseudo-likelihood approach that potentially offers robustness under partial model misspecification at the expense of slight loss of efficiency under correct model specification for small sample sizes. We apply our methods to the Massachusetts General Hospital glioblastoma study.

KW - Adjacent categories model

KW - Conditional logistic regression

KW - Glioblastoma

KW - Outcome-based sampling

KW - Pseudo-likelihood

UR - http://www.scopus.com/inward/record.url?scp=84933675001&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84933675001&partnerID=8YFLogxK

U2 - 10.1002/sim.6503

DO - 10.1002/sim.6503

M3 - Article

VL - 34

SP - 2514

EP - 2527

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 17

ER -