A simple and robust method for multivariate meta-analysis of diagnostic test accuracy

Yong Chen, Yulun Liu, Haitao Chu, Mei Ling Ting Lee, Christopher H. Schmid

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Meta-analysis of diagnostic test accuracy often involves mixture of case–control and cohort studies. The existing bivariate random-effects models, which jointly model bivariate accuracy indices (e.g., sensitivity and specificity), do not differentiate cohort studies from case–control studies and thus do not utilize the prevalence information contained in the cohort studies. The recently proposed trivariate generalized linear mixed-effects models are only applicable to cohort studies, and more importantly, they assume a common correlation structure across studies and trivariate normality on disease prevalence, test sensitivity, and specificity after transformation by some pre-specified link functions. In practice, very few studies provide justifications of these assumptions, and sometimes these assumptions are violated. In this paper, we evaluate the performance of the commonly used random-effects model under violations of these assumptions and propose a simple and robust method to fully utilize the information contained in case–control and cohort studies. The proposed method avoids making the aforementioned assumptions and can provide valid joint inferences for any functions of overall summary measures of diagnostic accuracy. Through simulation studies, we find that the proposed method is more robust to model misspecifications than the existing methods. We apply the proposed method to a meta-analysis of diagnostic test accuracy for the detection of recurrent ovarian carcinoma.

Original languageEnglish (US)
Pages (from-to)105-121
Number of pages17
JournalStatistics in Medicine
Volume36
Issue number1
DOIs
StatePublished - Jan 15 2017
Externally publishedYes

Fingerprint

Cohort Study
Diagnostic Tests
Robust Methods
Routine Diagnostic Tests
Meta-Analysis
Cohort Studies
Multivariate Analysis
Case-control Study
Trivariate
Random Effects Model
Specificity
Linear Mixed Effects Model
Diagnostic Accuracy
Sensitivity and Specificity
Model Misspecification
Link Function
Correlation Structure
Differentiate
Justification
Normality

Keywords

  • composite likelihood
  • diagnostic accuracy study
  • diagnostic review
  • meta-analysis
  • multivariate beta-binomial model
  • Sarmanov family

ASJC Scopus subject areas

  • Epidemiology
  • Statistics and Probability

Cite this

A simple and robust method for multivariate meta-analysis of diagnostic test accuracy. / Chen, Yong; Liu, Yulun; Chu, Haitao; Ting Lee, Mei Ling; Schmid, Christopher H.

In: Statistics in Medicine, Vol. 36, No. 1, 15.01.2017, p. 105-121.

Research output: Contribution to journalArticle

Chen, Yong ; Liu, Yulun ; Chu, Haitao ; Ting Lee, Mei Ling ; Schmid, Christopher H. / A simple and robust method for multivariate meta-analysis of diagnostic test accuracy. In: Statistics in Medicine. 2017 ; Vol. 36, No. 1. pp. 105-121.
@article{175db191028e4197b8f06ceefd0d7961,
title = "A simple and robust method for multivariate meta-analysis of diagnostic test accuracy",
abstract = "Meta-analysis of diagnostic test accuracy often involves mixture of case–control and cohort studies. The existing bivariate random-effects models, which jointly model bivariate accuracy indices (e.g., sensitivity and specificity), do not differentiate cohort studies from case–control studies and thus do not utilize the prevalence information contained in the cohort studies. The recently proposed trivariate generalized linear mixed-effects models are only applicable to cohort studies, and more importantly, they assume a common correlation structure across studies and trivariate normality on disease prevalence, test sensitivity, and specificity after transformation by some pre-specified link functions. In practice, very few studies provide justifications of these assumptions, and sometimes these assumptions are violated. In this paper, we evaluate the performance of the commonly used random-effects model under violations of these assumptions and propose a simple and robust method to fully utilize the information contained in case–control and cohort studies. The proposed method avoids making the aforementioned assumptions and can provide valid joint inferences for any functions of overall summary measures of diagnostic accuracy. Through simulation studies, we find that the proposed method is more robust to model misspecifications than the existing methods. We apply the proposed method to a meta-analysis of diagnostic test accuracy for the detection of recurrent ovarian carcinoma.",
keywords = "composite likelihood, diagnostic accuracy study, diagnostic review, meta-analysis, multivariate beta-binomial model, Sarmanov family",
author = "Yong Chen and Yulun Liu and Haitao Chu and {Ting Lee}, {Mei Ling} and Schmid, {Christopher H.}",
year = "2017",
month = "1",
day = "15",
doi = "10.1002/sim.7093",
language = "English (US)",
volume = "36",
pages = "105--121",
journal = "Statistics in Medicine",
issn = "0277-6715",
publisher = "John Wiley and Sons Ltd",
number = "1",

}

TY - JOUR

T1 - A simple and robust method for multivariate meta-analysis of diagnostic test accuracy

AU - Chen, Yong

AU - Liu, Yulun

AU - Chu, Haitao

AU - Ting Lee, Mei Ling

AU - Schmid, Christopher H.

PY - 2017/1/15

Y1 - 2017/1/15

N2 - Meta-analysis of diagnostic test accuracy often involves mixture of case–control and cohort studies. The existing bivariate random-effects models, which jointly model bivariate accuracy indices (e.g., sensitivity and specificity), do not differentiate cohort studies from case–control studies and thus do not utilize the prevalence information contained in the cohort studies. The recently proposed trivariate generalized linear mixed-effects models are only applicable to cohort studies, and more importantly, they assume a common correlation structure across studies and trivariate normality on disease prevalence, test sensitivity, and specificity after transformation by some pre-specified link functions. In practice, very few studies provide justifications of these assumptions, and sometimes these assumptions are violated. In this paper, we evaluate the performance of the commonly used random-effects model under violations of these assumptions and propose a simple and robust method to fully utilize the information contained in case–control and cohort studies. The proposed method avoids making the aforementioned assumptions and can provide valid joint inferences for any functions of overall summary measures of diagnostic accuracy. Through simulation studies, we find that the proposed method is more robust to model misspecifications than the existing methods. We apply the proposed method to a meta-analysis of diagnostic test accuracy for the detection of recurrent ovarian carcinoma.

AB - Meta-analysis of diagnostic test accuracy often involves mixture of case–control and cohort studies. The existing bivariate random-effects models, which jointly model bivariate accuracy indices (e.g., sensitivity and specificity), do not differentiate cohort studies from case–control studies and thus do not utilize the prevalence information contained in the cohort studies. The recently proposed trivariate generalized linear mixed-effects models are only applicable to cohort studies, and more importantly, they assume a common correlation structure across studies and trivariate normality on disease prevalence, test sensitivity, and specificity after transformation by some pre-specified link functions. In practice, very few studies provide justifications of these assumptions, and sometimes these assumptions are violated. In this paper, we evaluate the performance of the commonly used random-effects model under violations of these assumptions and propose a simple and robust method to fully utilize the information contained in case–control and cohort studies. The proposed method avoids making the aforementioned assumptions and can provide valid joint inferences for any functions of overall summary measures of diagnostic accuracy. Through simulation studies, we find that the proposed method is more robust to model misspecifications than the existing methods. We apply the proposed method to a meta-analysis of diagnostic test accuracy for the detection of recurrent ovarian carcinoma.

KW - composite likelihood

KW - diagnostic accuracy study

KW - diagnostic review

KW - meta-analysis

KW - multivariate beta-binomial model

KW - Sarmanov family

UR - http://www.scopus.com/inward/record.url?scp=84984674460&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84984674460&partnerID=8YFLogxK

U2 - 10.1002/sim.7093

DO - 10.1002/sim.7093

M3 - Article

C2 - 27580758

AN - SCOPUS:84984674460

VL - 36

SP - 105

EP - 121

JO - Statistics in Medicine

JF - Statistics in Medicine

SN - 0277-6715

IS - 1

ER -