Related Experiment Videos
Variance components analysis for pedigree-based censored survival data using generalized linear mixed models (GLMMs)
K J Scurrah1, L J Palmer, P R Burton
1Division of Biostatistics and Genetic Epidemiology, TVW Telethon Institute for Child Health Research, Perth, Australia.
Genetic Epidemiology
|August 30, 2000
Summary
Generalized linear mixed models (GLMMs) can analyze complex human diseases with censored survival data. This approach, using Bayesian inference via Gibbs sampling, aids genetic linkage analysis in families.
Area of Science:
- Genetics
- Biostatistics
- Epidemiology
Background:
- Complex human diseases present significant genetic research challenges due to unknown determinants and residual familial covariance.
- Accurate modeling of these factors is crucial for both association and covariance structure analyses.
- Standard analysis methods are limited for non-normally distributed traits, necessitating advanced statistical approaches.
Purpose of the Study:
- To present a unifying approach for analyzing diverse phenotypes, including right-censored survival times, using generalized linear mixed models (GLMMs).
- To demonstrate the application of Markov chain Monte Carlo (MCMC) methods, specifically Gibbs sampling, for fitting GLMMs in family studies.
- To introduce a novel method for linkage analysis by treating GLMM random effects as adjusted phenotypes.
Main Methods:
- Generalized linear mixed models (GLMMs) were employed to analyze right-censored survival data (age-at-onset/death) in nuclear and extended families.
- Bayesian inference using Gibbs sampling (BUGS) was utilized as the computational framework for fitting the GLMMs.
- Simulated data were used to validate model parameter consistency, and real data from a cohort study were used for illustration.
Main Results:
- The study successfully fitted GLMMs for right-censored survival times in family structures using MCMC methods.
- Model parameters were shown to be consistent using simulated data.
- The proposed method for using GLMM random effects in linkage analysis was demonstrated.
Conclusions:
- GLMMs provide a flexible and unifying framework for analyzing complex traits, including censored survival data, in genetic research.
- MCMC methods, particularly Gibbs sampling, offer a practical approach for fitting these complex models.
- Treating GLMM random effects as adjusted phenotypes simplifies and enhances linkage analysis for survival traits, accounting for various confounders.