Assessing the use of GEE methods for analyzing binary outcomes in family studies: the Strong Heart Family Study
Xi Chen,Ying Zhang,Amanda M. Fretts,Tauqeer Ali,Jason G. Umans,Richard B. Devereux,Elisa T. Lee,Shelley A. Cole,Yan Daniel Zhao,Xi ChenYing ZhangAmanda M. FrettsTauqeer AliJason G. UmansRichard B. DevereuxElisa T. LeeShelley A. ColeYan Daniel Zhaoa KBR Inc,Houston,Texas,USAb Department of Biostatistics and Epidemiology (BSE),The University of Oklahoma Health Sciences Center (OUHSC),Oklahoma City,Oklahoma,USAc Cardiovascular Health Research Unit,Department of Epidemiology. University of Washington,Seattle,Washington,USAd Center for American Indian Health Research,BSE,OUHSC,Oklahoma City,Oklahoma,USAe MedStar Health Research Institute,Hyattsville,Maryland,USAf Georgetown-Howard University,Washington,DC,USAg Cardiology,Weill Cornell Medical Center,Starr Pavilion,New York,New York,USAh Population Health,Texas Biomedical Research Institute,San Antonio,Texas,USA
DOI: https://doi.org/10.1080/10543406.2024.2333516
2024-03-30
Journal of Biopharmaceutical Statistics
Abstract:The generalized estimating equations method (GEE) is commonly applied to analyze data obtained from family studies. GEE is well known for its robustness on misspecification of correlation structure. However, the unbalanced distribution of family sizes and complicated genetic relatedness structure within each family may challenge GEE performance. We focused our research on binary outcomes. To evaluate the performance of GEE, we conducted a series of simulations, on data generated adopting the kinship matrix (correlation structure within each family) from the Strong Heart Family Study (SHFS). We performed a fivefold cross-validation to further evaluate the GEE predictive power on data from the SHFS. A Bayesian modeling approach, with direct integration of the kinship matrix, was also included to contrast with GEE. Our simulation studies revealed that GEE performs well on a binary outcome from families having a relatively simple kinship structure. However, data with a binary outcome generated from families with complex kinship structures, especially with a large genetic variance, can challenge the performance of GEE.
pharmacology & pharmacy,statistics & probability