Meta-analytical methods and their applications to biomedical studies
Meta-analysis is an important statistical tool, which can synthesize the available evidence and integrate the findings from individual studies. The results from a meta-analysis can provide more precise estimates of intervention effects or other outcomes than the results from a single study. A variety of meta-analytic methodology has long been applied in research fields of genetics and genomics, clinical medicine, behavioral science, and other fields.^ In the field of genomics, combining information across multiple genome-wide association studies (GWAS) has become very popular for its ability in gaining statistical power and providing validated conclusions. Meta-analysis can be conducted where summary statistics from multiple GWAS are integrated together. In this dissertation, we propose a new procedure for detecting gene-by-gene interactions through heterogeneity in estimated low-order (e.g. marginal) effect sizes by leveraging population structure, or ancestral differences, among studies in which the same phenotypes were measured. We implement this approach in a meta-analytic framework, which offers numerous advantages, such as robustness and computational eciency, and is necessary when data-sharing limitations restrict joint analysis.^ Sharing the genomic data is increasingly common. The benefits of such data sharing have been widely recognized. However, some potential issues such as the privacy-preserving at individual patient-level, heterogeneous data sources and the other practical factors are raised by data sharing. This leads to the demand of new statistical approaches for the distributed analyses of GWAS databases. In this dissertation, we proposed a new two-stage testing procedure for detecting gene-gene interactions based on the meta-analytic framework; it is a computationally fast algorithm for combining multiple and distributed GWAS databases. Such the proposed method can not only be applied to distributed GWAS databases without shared individual patient-level information but also can be used to leverage site-specific heterogeneity among sites in which the same phenotypes were measured.^ In the evidence-based medicine, comparison of relevant treatments is recognized as an important part for decision making process. Mixed treatment comparisons (MTC) meta-analysis have focused on univariate or very recently, bivariate outcomes. To broaden their application, we propose a framework for MTC meta-analysis of multivariate outcomes where the correlations among multivariate outcomes within- and between-studies are accounted for through copulas, and the joint modeling of multivariate random effects, respectively. We consider a Bayesian hierarchical model using Markov chain Monte Carlo methods for estimation. An important feature of the proposed framework is that it allows for borrowing of information across correlated outcomes. We show via simulation that our approach reduces the impact of outcome reporting bias (ORB) in a variety of missing outcome scenarios. We apply the method to a systematic review of randomized controlled trials of pharmacological treatments for alcohol dependence, which tend to report multiple outcomes subject to ORB.^
Liu, Yulun, "Meta-analytical methods and their applications to biomedical studies" (2016). Texas Medical Center Dissertations (via ProQuest). AAI10127433.