Zhong Yujie, Cook Richard J
School of Statistics and Management, Shanghai University of Finance and Economics, Shanghai, P.R. China.
Department of Statistics and Actuarial Science, University of Waterloo, Waterloo, Ontario, Canada.
Stat Med. 2021 Jan 30;40(2):254-270. doi: 10.1002/sim.8772. Epub 2020 Oct 17.
Family studies routinely employ biased sampling schemes in which individuals are randomly chosen from a disease registry and genetic and phenotypic data are obtained from their consenting relatives. We view this as a two-phase study and propose the use of an efficient selection model for the recruitment of families to form a phase II sample subject to budgetary constraints. Simple random sampling, balanced sampling and use of an approximately optimal selection model are considered where the latter is chosen to minimize the variance of parameters of interest. We consider the setting where family members provide current status data with respect to the disease and use copula models to address within-family dependence. The efficiency gains from the use of an optimal selection model over simple random sampling and balanced sampling schemes are investigated as is the robustness of optimal sampling to model misspecification. An application to a family study on psoriatic arthritis is given for illustration.
家族研究通常采用有偏抽样方案，即从疾病登记处随机选取个体，并从其同意参与的亲属那里获取基因和表型数据。我们将此视为一个两阶段研究，并提出使用一种有效的选择模型来招募家族，以便在预算限制下形成第二阶段样本。考虑了简单随机抽样、平衡抽样以及使用近似最优选择模型，其中选择后者是为了使感兴趣参数的方差最小化。我们考虑家庭成员提供疾病当前状态数据的情况，并使用copula模型来处理家庭内部的相关性。研究了使用最优选择模型相对于简单随机抽样和平衡抽样方案的效率提升，以及最优抽样对模型误设的稳健性。给出了一个银屑病关节炎家族研究的应用示例。
