|
Abstract:
|
In pharmacogenetic analyses, and other genetic analyses in early clinical development, independent studies are often pooled, at the individual level, to improve the power of detecting small to moderate genetic effects with an increased sample size. Alternatively, meta-analysis methods are used to aggregate results, at the summary level, across independent studies. Both pooled and meta-analysis methods rely on the assumption of common genetic effects across studies. Though, due to diverse study designs, considerable heterogeneity between studies often exists. While pooled and meta-analyses have been shown to estimate genetic effect sizes with comparable efficiencies, method choice in the presence of between-study heterogeneity is unclear. Here we use simulations to understand how between-study heterogeneity, mirroring that observed in pharmacogenetic analyses, impacts the choice of pooled versus meta-analysis methods.
|
