Statistical estimation and comparison of group-specific bivariate correlation coefficients in family-type clustered studies.

ABSTRACT: Bivariate correlation coefficients (BCCs) are often calculated to gauge the relationship between two variables in medical research. In a family-type clustered design where multiple participants from same units/families are enrolled, BCCs can be defined and estimated at various hierarchical levels (subject level, family level and marginal BCC). Heterogeneity usually exists between subject groups and, as a result, subject level BCCs may differ between subject groups. In the framework of bivariate linear mixed effects modeling, we define and estimate BCCs at various hierarchical levels in a family-type clustered design, accommodating subject group heterogeneity. Simplified and modified asymptotic confidence intervals are constructed to the BCC differences and Wald type tests are conducted. A

SUBMITTER: Luo J 

PROVIDER: S-EPMC9225315 | biostudies-literature | 2022

REPOSITORIES: biostudies-literature