Joint GEEs for multivariate correlated data with incomplete binary outcomes

İnan, Gül; Yucel, Recai

doi:10.1080/02664763.2016.1238049

Cited by 4 publications

(9 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on the joint GEE model formulation [ 40 ], we construct the multivariate linear model for describing the association relationship between K correlated traits and genetic variants, which is given as follows:

where g −1 (•) is the inverse function of g (•) and is a response-specific link function [ 40 ], μ i = ( μ i 1 T , μ i 2 T ,⋯, μ iK T ) T is the ( n i × K ) × 1 vector of the expected mean of the multivariate traits y i = ( y i 1 T , y i 2 T ,⋯, y iK T ) T , α = ( α 1 T , α 2 T ,⋯, α K T ) T is the (( q + 1) × K ) × 1 vector of regression coefficients of the ( q + 1) nongenetic covariates for the K correlated traits, and β = ( β 1 T , β 2 T ,⋯, β K T ) T is the ( p × K ) × 1 vector of regression coefficients of the p genetic variants for the K correlated traits.…”

Section: Methodsmentioning

confidence: 99%

“…Let R n i ( φ ) and R K ( γ ) be the n i × n i within-in cluster correlation matrix and the K × K multivariate-response cluster correlation matrix, which depend on a vector of parameters φ and γ , respectively. The ( n i × K ) × ( n i × K ) working (or approximate) covariance matrix of y i is given by [ 40 ].

where A i = diag( A i 1 , A i 2 , ⋯, A iK ) is a ( n i × K ) × ( n i × K ) block diagonal matrix with the components A ik = diag( ∂μ i 1 k / ∂θ i 1 k , ∂μ i 2 k / ∂θ i 2 k , ⋯, ∂μ in i k / ∂θ in i k ), k = 1, 2, ⋯, K being the n i × n i diagonal matrices.…”

Section: Methodsmentioning

confidence: 99%

“…The rest of the article is organized as follows. In the materials and methods section, we construct the multivariate effect model using the joint GEE model formulation (JGEE) [ 40 ]. We apply the JGEE to pedigree- or population-structured data and introduce a retrospective framework for analyzing multivariate traits in genetic association studies.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Associating Multivariate Traits with Genetic Variants Using Collapsing and Kernel Methods with Pedigree- or Population-Based Studies

Chien

2021

Computational and Mathematical Methods in Medicine

View full text Add to dashboard Cite

In genetic association analysis, several relevant phenotypes or multivariate traits with different types of components are usually collected to study complex or multifactorial diseases. Over the past few years, jointly testing for association between multivariate traits and multiple genetic variants has become more popular because it can increase statistical power to identify causal genes in pedigree- or population-based studies. However, most of the existing methods mainly focus on testing genetic variants associated with multiple continuous phenotypes. In this investigation, we develop a framework for identifying the pleiotropic effects of genetic variants on multivariate traits by using collapsing and kernel methods with pedigree- or population-structured data. The proposed framework is applicable to the burden test, the kernel test, and the omnibus test for autosomes and the X chromosome. The proposed multivariate trait association methods can accommodate continuous phenotypes or binary phenotypes and further can adjust for covariates. Simulation studies show that the performance of our methods is satisfactory with respect to the empirical type I error rates and power rates in comparison with the existing methods.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Associating Multivariate Traits with Genetic Variants Using Collapsing and Kernel Methods with Pedigree- or Population-Based Studies

Chien

2021

Computational and Mathematical Methods in Medicine

View full text Add to dashboard Cite

show abstract

“…A joint modeling of multiple response variables is based on straightforward extension of univariate GEEs with correlation structure across responses which provides separate set of regression parameters for each response variable. 19 , 20 GEEs are less sensitive to covariance misspecification compared to mixed effects models. 18 , 21 …”

Section: Introductionmentioning

confidence: 99%

Repeated measures discriminant analysis using multivariate generalized estimation equations

Brobbey

Wiebe

Nettel-Aguirre

et al. 2021

Stat Methods Med Res

View full text Add to dashboard Cite

Discriminant analysis procedures that assume parsimonious covariance and/or means structures have been proposed for distinguishing between two or more populations in multivariate repeated measures designs. However, these procedures rely on the assumptions of multivariate normality which is not tenable in multivariate repeated measures designs which are characterized by binary, ordinal, or mixed types of response distributions. This study investigates the accuracy of repeated measures discriminant analysis (RMDA) based on the multivariate generalized estimating equations (GEE) framework for classification in multivariate repeated measures designs with the same or different types of responses repeatedly measured over time. Monte Carlo methods were used to compare the accuracy of RMDA procedures based on GEE, and RMDA based on maximum likelihood estimators (MLE) under diverse simulation conditions, which included number of repeated measure occasions, number of responses, sample size, correlation structures, and type of response distribution. RMDA based on GEE exhibited higher average classification accuracy than RMDA based on MLE especially in multivariate non-normal distributions. Three repeatedly measured responses namely severity of epilepsy, current number of anti-epileptic drugs, and parent-reported quality of life in children with epilepsy were used to demonstrate the application of these procedures.

show abstract

“…Moreover, all of these articles considered the CDM setting with fully observed baseline covariates and all specifically considered no clustering in the missing process, except for Caille et al 20 Furthermore, this literature fits in the broader literature on imputation for missing outcomes for correlated data in which it is also well recognized that the multiple imputation strategy should use a multilevel structure in order to reflect the multilevel data structure. 25,26 For the alternative approach of weighting to account for missing outcomes, although there are methodological descriptions of general weighting approaches, 11 of weighted GEE for longitudinal data, 25,[27][28][29] and methods for general correlated data, 25,30,31 we have found no articles that address weighted GEE specifically for clustered outcome data that arise in a CRT. This gap in the literature could explain why few trials implemented this approach in practice.…”

Section: Introductionmentioning

confidence: 99%

Properties and pitfalls of weighting as an alternative to multilevel multiple imputation in cluster randomized trials with missing binary outcomes under covariate-dependent missingness

Turner

Yao

et al. 2019

Stat Methods Med Res

View full text Add to dashboard Cite

The generalized estimating equation (GEE) approach can be used to analyze cluster randomized trial data to obtain population-averaged intervention effects. However, most cluster randomized trials have some missing outcome data and a GEE analysis of available data may be biased when outcome data are not missing completely at random. Although multilevel multiple imputation for GEE (MMI-GEE) has been widely used, alternative approaches such as weighted GEE are less common in practice. Using both simulations and a real data example, we evaluate the performance of inverse probability weighted GEE vs. MMI-GEE for binary outcomes. Simulated data are generated assuming a covariate-dependent missing data pattern across a range of missingness clustering (from none to high), where all covariates are measured at baseline and are fully observed (i.e. a type of missing-at-random mechanism). Two types of weights are estimated and used in the weighted GEE: (1) assuming no clustering of missingness (W-GEE) and (2) accounting for such clustering (CW-GEE). Results show that, even in settings with high missingness clustering, CW-GEE can lead to more bias and lower coverage than W-GEE, whereas W-GEE and MMI-GEE provide comparable results. W-GEE should be considered a viable strategy to account for missing outcomes in cluster randomized trials.

show abstract

Joint GEEs for multivariate correlated data with incomplete binary outcomes

Cited by 4 publications

References 29 publications

Associating Multivariate Traits with Genetic Variants Using Collapsing and Kernel Methods with Pedigree- or Population-Based Studies

Associating Multivariate Traits with Genetic Variants Using Collapsing and Kernel Methods with Pedigree- or Population-Based Studies

Repeated measures discriminant analysis using multivariate generalized estimation equations

Properties and pitfalls of weighting as an alternative to multilevel multiple imputation in cluster randomized trials with missing binary outcomes under covariate-dependent missingness

Contact Info

Product

Resources

About