Marginal and conditional approaches to multivariate variables subject to limit of detection

L. Nie; H. Chu; Y. Cheng; C. Spurney; K. Nagaraju; J. Chen

doi:10.1080/10543400903243033

Marginal and conditional approaches to multivariate variables subject to limit of detection

L. Nie, H. Chu, Y. Cheng, C. Spurney, K. Nagaraju, J. Chen

Biostatistics

Research output: Contribution to journal › Article › peer-review

5 Scopus citations

Abstract

A marginal approach and a variance-component mixed effect model approach (here called a conditional approach) are commonly used to analyze variables that are subject to limit of detection. We examine the theoretical relationship and investigate the numerical performance of these two approaches. We make some recommendations based on our results. The marginal approach is recommended for bivariate normal variables, and the variance-component mixed effect model is preferable for other multivariate analysis in most circumstances. Two approaches are illustrated through one case study from a preclinical experiment.

Original language	English (US)
Pages (from-to)	1151-1161
Number of pages	11
Journal	Journal of Biopharmaceutical Statistics
Volume	19
Issue number	6
DOIs	https://doi.org/10.1080/10543400903243033
State	Published - Nov 2009

Bibliographical note

Funding Information:
We thank the editor and two referees for their very insightful and constructive comments, which have greatly improved the quality of the article. Dr. Chu was supported in part by the Lineberger Cancer Center Core Grant CA16086 from the U.S. National Cancer Institute and P30-AI-50410 from the U.S. National Institutes of Health.

Keywords

Limit of detection
Marginal approach
Maximum likelihood estimate
Variance-component mixed effect model

Publisher link

10.1080/10543400903243033

Find It

Find It at U of M Libraries

Cite this

@article{07bd1cedbcfb43b6b2a9deb330e90f5e,

title = "Marginal and conditional approaches to multivariate variables subject to limit of detection",

abstract = "A marginal approach and a variance-component mixed effect model approach (here called a conditional approach) are commonly used to analyze variables that are subject to limit of detection. We examine the theoretical relationship and investigate the numerical performance of these two approaches. We make some recommendations based on our results. The marginal approach is recommended for bivariate normal variables, and the variance-component mixed effect model is preferable for other multivariate analysis in most circumstances. Two approaches are illustrated through one case study from a preclinical experiment.",

keywords = "Limit of detection, Marginal approach, Maximum likelihood estimate, Variance-component mixed effect model",

author = "L. Nie and H. Chu and Y. Cheng and C. Spurney and K. Nagaraju and J. Chen",

note = "Funding Information: We thank the editor and two referees for their very insightful and constructive comments, which have greatly improved the quality of the article. Dr. Chu was supported in part by the Lineberger Cancer Center Core Grant CA16086 from the U.S. National Cancer Institute and P30-AI-50410 from the U.S. National Institutes of Health.",

year = "2009",

month = nov,

doi = "10.1080/10543400903243033",

language = "English (US)",

volume = "19",

pages = "1151--1161",

journal = "Journal of Biopharmaceutical Statistics",

issn = "1054-3406",

publisher = "Taylor and Francis Ltd.",

number = "6",

}

TY - JOUR

T1 - Marginal and conditional approaches to multivariate variables subject to limit of detection

AU - Nie, L.

AU - Chu, H.

AU - Cheng, Y.

AU - Spurney, C.

AU - Nagaraju, K.

AU - Chen, J.

N1 - Funding Information: We thank the editor and two referees for their very insightful and constructive comments, which have greatly improved the quality of the article. Dr. Chu was supported in part by the Lineberger Cancer Center Core Grant CA16086 from the U.S. National Cancer Institute and P30-AI-50410 from the U.S. National Institutes of Health.

PY - 2009/11

Y1 - 2009/11

N2 - A marginal approach and a variance-component mixed effect model approach (here called a conditional approach) are commonly used to analyze variables that are subject to limit of detection. We examine the theoretical relationship and investigate the numerical performance of these two approaches. We make some recommendations based on our results. The marginal approach is recommended for bivariate normal variables, and the variance-component mixed effect model is preferable for other multivariate analysis in most circumstances. Two approaches are illustrated through one case study from a preclinical experiment.

AB - A marginal approach and a variance-component mixed effect model approach (here called a conditional approach) are commonly used to analyze variables that are subject to limit of detection. We examine the theoretical relationship and investigate the numerical performance of these two approaches. We make some recommendations based on our results. The marginal approach is recommended for bivariate normal variables, and the variance-component mixed effect model is preferable for other multivariate analysis in most circumstances. Two approaches are illustrated through one case study from a preclinical experiment.

KW - Limit of detection

KW - Marginal approach

KW - Maximum likelihood estimate

KW - Variance-component mixed effect model

UR - http://www.scopus.com/inward/record.url?scp=75749106258&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=75749106258&partnerID=8YFLogxK

U2 - 10.1080/10543400903243033

DO - 10.1080/10543400903243033

M3 - Article

C2 - 20183469

AN - SCOPUS:75749106258

SN - 1054-3406

VL - 19

SP - 1151

EP - 1161

JO - Journal of Biopharmaceutical Statistics

JF - Journal of Biopharmaceutical Statistics

IS - 6

ER -

Marginal and conditional approaches to multivariate variables subject to limit of detection

Abstract

Bibliographical note

Keywords

Publisher link

Find It

Other files and links

Fingerprint

Cite this