Minimax two-stage-designs with applications to tissue banking case-control studies

Area of Science:

• Biostatistics
• Medical Informatics
• Genomic Medicine

Background:

• Tissue banks preserve valuable patient samples for disease research.
• Case-control studies efficiently assess prognostic markers using banked samples.
• Existing group sequential methods are underutilized in case-control studies.

Purpose of the Study:

• To develop an efficient two-stage design for case-control studies.
• To minimize the maximum expected sample size in prognostic marker research.
• To provide practical tools for implementing optimized study designs.

Main Methods:

• A minimax strategy was employed to minimize the maximum expected sample size for two-stage designs.
• The method determines optimal designs based on specified type I and type II errors.
• Simple tables are provided for practical application, derived from single-stage design calculations.

Main Results:

• The proposed strategy yields a two-stage design that minimizes the maximum expected sample size.
• The method allows for study redesign after the first stage using variance estimates.
• A modified 'two-plus' stage design offers superior efficiency for matched pair studies.

Conclusions:

• The developed minimax approach offers an efficient strategy for two-stage case-control studies.
• This method optimizes sample size in prognostic marker research, enhancing resource utilization.
• The findings facilitate the application of advanced statistical designs in biomedical research using tissue banks.