A conditional likelihood approach for regression analysis using biomarkers measured with batch-specific error

Ming Wang, W. Dana Flanders, Roberd M. Bostick, Qi Long

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Measurement error is common in epidemiological and biomedical studies. When biomarkers are measured in batches or groups, measurement error is potentially correlated within each batch or group. In regression analysis, most existing methods are not applicable in the presence of batch-specific measurement error in predictors. We propose a robust conditional likelihood approach to account for batch-specific error in predictors when batch effect is additive and the predominant source of error, which requires no assumptions on the distribution of measurement error. Although a regression model with batch as a categorical covariable yields the same parameter estimates as the proposed conditional likelihood approach for linear regression, this result does not hold in general for all generalized linear models, in particular, logistic regression. Our simulation studies show that the conditional likelihood approach achieves better finite sample performance than the regression calibration approach or a naive approach without adjustment for measurement error. In the case of logistic regression, our proposed approach is shown to also outperform the regression approach with batch as a categorical covariate. In addition, we also examine a 'hybrid' approach combining the conditional likelihood method and the regression calibration method, which is shown in simulations to achieve good performance in the presence of both batch-specific and measurement-specific errors. We illustrate our method by using data from a colorectal adenoma study.

Original languageEnglish (US)
Pages (from-to)3896-3906
Number of pages11
JournalStatistics in Medicine
Volume31
Issue number29
DOIs
StatePublished - Dec 20 2012

Fingerprint

Conditional Likelihood
Biomarkers
Regression Analysis
Batch
Measurement Error
Calibration
Linear Models
Logistic Models
Regression Calibration
Logistic Regression
Categorical
Adenoma
Epidemiologic Studies
Predictors
Research Design
Likelihood Methods
Generalized Linear Model
Hybrid Approach
Linear regression
Covariates

All Science Journal Classification (ASJC) codes

  • Epidemiology
  • Statistics and Probability

Cite this

Wang, Ming ; Dana Flanders, W. ; Bostick, Roberd M. ; Long, Qi. / A conditional likelihood approach for regression analysis using biomarkers measured with batch-specific error. In: Statistics in Medicine. 2012 ; Vol. 31, No. 29. pp. 3896-3906.
@article{cc729d59a5134e37a2795ce4798ef391,
title = "A conditional likelihood approach for regression analysis using biomarkers measured with batch-specific error",
abstract = "Measurement error is common in epidemiological and biomedical studies. When biomarkers are measured in batches or groups, measurement error is potentially correlated within each batch or group. In regression analysis, most existing methods are not applicable in the presence of batch-specific measurement error in predictors. We propose a robust conditional likelihood approach to account for batch-specific error in predictors when batch effect is additive and the predominant source of error, which requires no assumptions on the distribution of measurement error. Although a regression model with batch as a categorical covariable yields the same parameter estimates as the proposed conditional likelihood approach for linear regression, this result does not hold in general for all generalized linear models, in particular, logistic regression. Our simulation studies show that the conditional likelihood approach achieves better finite sample performance than the regression calibration approach or a naive approach without adjustment for measurement error. In the case of logistic regression, our proposed approach is shown to also outperform the regression approach with batch as a categorical covariate. In addition, we also examine a 'hybrid' approach combining the conditional likelihood method and the regression calibration method, which is shown in simulations to achieve good performance in the presence of both batch-specific and measurement-specific errors. We illustrate our method by using data from a colorectal adenoma study.",
author = "Ming Wang and {Dana Flanders}, W. and Bostick, {Roberd M.} and Qi Long",
year = "2012",
month = "12",
day = "20",
doi = "10.1002/sim.5473",
language = "English (US)",
volume = "31",
pages = "3896--3906",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",
number = "29",

}

A conditional likelihood approach for regression analysis using biomarkers measured with batch-specific error. / Wang, Ming; Dana Flanders, W.; Bostick, Roberd M.; Long, Qi.

In: Statistics in Medicine, Vol. 31, No. 29, 20.12.2012, p. 3896-3906.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A conditional likelihood approach for regression analysis using biomarkers measured with batch-specific error

AU - Wang, Ming

AU - Dana Flanders, W.

AU - Bostick, Roberd M.

AU - Long, Qi

PY - 2012/12/20

Y1 - 2012/12/20

N2 - Measurement error is common in epidemiological and biomedical studies. When biomarkers are measured in batches or groups, measurement error is potentially correlated within each batch or group. In regression analysis, most existing methods are not applicable in the presence of batch-specific measurement error in predictors. We propose a robust conditional likelihood approach to account for batch-specific error in predictors when batch effect is additive and the predominant source of error, which requires no assumptions on the distribution of measurement error. Although a regression model with batch as a categorical covariable yields the same parameter estimates as the proposed conditional likelihood approach for linear regression, this result does not hold in general for all generalized linear models, in particular, logistic regression. Our simulation studies show that the conditional likelihood approach achieves better finite sample performance than the regression calibration approach or a naive approach without adjustment for measurement error. In the case of logistic regression, our proposed approach is shown to also outperform the regression approach with batch as a categorical covariate. In addition, we also examine a 'hybrid' approach combining the conditional likelihood method and the regression calibration method, which is shown in simulations to achieve good performance in the presence of both batch-specific and measurement-specific errors. We illustrate our method by using data from a colorectal adenoma study.

AB - Measurement error is common in epidemiological and biomedical studies. When biomarkers are measured in batches or groups, measurement error is potentially correlated within each batch or group. In regression analysis, most existing methods are not applicable in the presence of batch-specific measurement error in predictors. We propose a robust conditional likelihood approach to account for batch-specific error in predictors when batch effect is additive and the predominant source of error, which requires no assumptions on the distribution of measurement error. Although a regression model with batch as a categorical covariable yields the same parameter estimates as the proposed conditional likelihood approach for linear regression, this result does not hold in general for all generalized linear models, in particular, logistic regression. Our simulation studies show that the conditional likelihood approach achieves better finite sample performance than the regression calibration approach or a naive approach without adjustment for measurement error. In the case of logistic regression, our proposed approach is shown to also outperform the regression approach with batch as a categorical covariate. In addition, we also examine a 'hybrid' approach combining the conditional likelihood method and the regression calibration method, which is shown in simulations to achieve good performance in the presence of both batch-specific and measurement-specific errors. We illustrate our method by using data from a colorectal adenoma study.

UR - http://www.scopus.com/inward/record.url?scp=84870066493&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84870066493&partnerID=8YFLogxK

U2 - 10.1002/sim.5473

DO - 10.1002/sim.5473

M3 - Article

VL - 31

SP - 3896

EP - 3906

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 29

ER -