Meta-analysis of studies with bivariate binary outcomes: a marginal beta-binomial model approach

被引:16
作者
Chen, Yong [1 ]
Hong, Chuan [2 ]
Ning, Yang [3 ]
Su, Xiao [2 ]
机构
[1] Univ Penn, Dept Biostat & Epidemiol, Philadelphia, PA 19104 USA
[2] Univ Texas Houston, Sch Publ Hlth, Div Biostat, Houston, TX 77030 USA
[3] Princeton Univ, Dept Operat Res & Financial Engn, Princeton, NJ 08544 USA
基金
美国医疗保健研究与质量局;
关键词
bivariate beta-binomial model; composite likelihood; marginal model; meta-analysis; Sarmanov family; PAIRWISE LIKELIHOOD INFERENCE; DIAGNOSTIC-ACCURACY; MIXED MODELS; DISTRIBUTIONS; SPECIFICITY; SENSITIVITY; ESTIMATORS; TESTS;
D O I
10.1002/sim.6620
中图分类号
Q [生物科学];
学科分类号
07 ; 0710 ; 09 ;
摘要
When conducting a meta-analysis of studies with bivariate binary outcomes, challenges arise when the within-study correlation and between-study heterogeneity should be taken into account. In this paper, we propose a marginal beta-binomial model for the meta-analysis of studies with binary outcomes. This model is based on the composite likelihood approach and has several attractive features compared with the existing models such as bivariate generalized linear mixed model (Chu and Cole, 2006) and Sarmanov beta-binomial model (Chen et al., 2012). The advantages of the proposed marginal model include modeling the probabilities in the original scale, not requiring any transformation of probabilities or any link function, having closed-form expression of likelihood function, and no constraints on the correlation parameter. More importantly, because the marginal beta-binomial model is only based on the marginal distributions, it does not suffer from potential misspecification of the joint distribution of bivariate study-specific probabilities. Such misspecification is difficult to detect and can lead to biased inference using currents methods. We compare the performance of the marginal beta-binomial model with the bivariate generalized linear mixed model and the Sarmanov beta-binomial model by simulation studies. Interestingly, the results show that the marginal beta-binomial model performs better than the Sarmanov beta-binomial model, whether or not the true model is Sarmanov beta-binomial, and the marginal beta-binomial model is more robust than the bivariate generalized linear mixed model under model misspecifications. Two meta-analyses of diagnostic accuracy studies and a meta-analysis of case-control studies are conducted for illustration. Copyright (C) 2015 John Wiley & Sons, Ltd.
引用
收藏
页码:21 / 40
页数:20
相关论文
共 69 条
[1]  
[Anonymous], 1996, DISTRIBUTIONS FIXED, DOI DOI 10.1214/LNMS/1215452606
[2]  
[Anonymous], AOD ANAL OVERDISPERS
[3]  
[Anonymous], 1999, INTRO COPULAS
[4]  
[Anonymous], 1966, DOKLADY SOVIET MATH
[5]  
[Anonymous], 2021, INTRO META ANAL
[6]  
[Anonymous], 1988, Statistical Inference from Stochastic Processes
[7]  
[Anonymous], STAT METHOD IN PRESS
[8]   Aberrant crypt foci and semiparametric modeling of correlated binary data [J].
Apanasovich, Tatiyana V. ;
Ruppert, David ;
Lupton, Joanne R. ;
Popovic, Natasa ;
Turner, Nancy D. ;
Chapkin, Robert S. ;
Carroll, Raymond J. .
BIOMETRICS, 2008, 64 (02) :490-500
[9]   Bivariate random effects meta-analysis of ROC curves [J].
Arends, L. R. ;
Hamza, T. H. ;
van Houwelingen, J. C. ;
Heijenbrok-Kal, M. H. ;
Hunink, M. G. M. ;
Stijnen, T. .
MEDICAL DECISION MAKING, 2008, 28 (05) :621-638
[10]   Linear mixed models for longitudinal shape data with applications to facial modeling [J].
Barry, Sarah J. E. ;
Bowman, Adrian W. .
BIOSTATISTICS, 2008, 9 (03) :555-565