Analyses for continuous, correlated outcomes in cross-sectional studies
Gui-Shuang Ying, Scheie Eye Institute, Wynnewood, Pennsylvania, United States
DisclosureBlock: Gui-Shuang Ying, Chengdu Kanghong Biotechnology Co. Ltd.Ltd Code C (Consultant) , Ziemer Ophthalmic Systems AG Code C (Consultant)
Description
This presentation will start with the non-model-based analysis approaches (paired t-test, Wilcoxon signed rank test) for correlated continuous eye data under paired design (e.g., two eyes in two different comparison groups) and analysis approaches (e.g., clustered Wilcoxon rank test) for data from unpaired design (e.g., two eyes in the same comparison groups). The presentation will then describe several model-based analysis approaches for continuous correlated eye data from cross-sectional studies, including mixed effects models and marginal models under various covariance structures to account for inter-eye correlation and other covariates (person-specific or eye-specific). The presentation will demonstrate, with SAS statistical software, applications in a study comparing baseline refractive error between one eye with choroidal neovascularization and the unaffected fellow eye, and in a study determining factors associated with visual field in the elderly. The comparisons between these appropriate analysis approaches vs. the inappropriate yet commonly used analysis approaches (e.g., ignore inter-eye correlation, or one-eye analysis) will be made and discussed.