Semiparametric inference for a 2-stage outcome-auxiliary-dependent sampling design with continuous outcome

Area of Science:

• Biostatistics
• Epidemiology
• Clinical Trial Design

Background:

• Two-stage designs are cost-effective for biomedical studies.
• Auxiliary covariate information can improve study efficiency.
• Existing methods may not fully exploit covariate data in two-stage designs.

Purpose of the Study:

• To propose a novel two-stage design incorporating an outcome-auxiliary-dependent sampling (OADS) scheme for continuous outcomes.
• To develop a consistent and asymptotically normal estimator for this design.
• To demonstrate the efficiency gains of the proposed OADS design.

Main Methods:

• Development of a two-stage design with OADS for continuous outcomes.
• Proposal of an estimator based on maximizing an estimated likelihood function.
• Theoretical analysis of estimator consistency and asymptotic normality.
• Simulation studies comparing OADS with alternative sampling schemes.

Main Results:

• The proposed estimator is shown to be consistent and asymptotically normal.
• Simulation results indicate significant efficiency gains using the 2-stage OADS design.
• The OADS approach effectively utilizes auxiliary covariate information compared to other schemes.

Conclusions:

• The proposed two-stage OADS design offers enhanced efficiency in biomedical studies.
• This method provides a valuable tool for utilizing auxiliary data in clinical trial design.
• The approach was successfully illustrated using an environmental epidemiology study dataset.