The hidden factor: accounting for covariate effects in power and sample size computation for a binary trait

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Accurate power and sample size estimation is critical for genetic association studies.
• Covariate effects (e.g., age, sex) are often included in logistic regression models for binary traits but are frequently overlooked in power calculations.
• Existing methods for power and sample size computation in genetic association studies lack flexibility in handling covariate effects and gene-environment relationships.

Purpose of the Study:

• To propose and implement a generalized method for estimating power and sample size in genetic association studies of binary traits.
• To develop a flexible method that accommodates various nongenetic covariates and gene-environment relationships.
• To ensure computational efficiency in power and sample size calculations.

Main Methods:

• Developed a generalized statistical method for power and sample size estimation.
• Implemented the method in an R package (SPCompute).
• Validated the method using extensive simulation studies for prospective and retrospective sampling designs.

Main Results:

• The proposed method accurately estimates power and sample size and is computationally efficient.
• A proof-of-principle application using UK Biobank data demonstrated that ignoring covariate effects (age, sex) for binary hypertension traits leads to overestimated power and underestimated replication sample size.
• Results contrast with analyses of continuous traits, highlighting the importance of covariate consideration for binary traits.

Conclusions:

• The generalized method provides accurate and efficient power and sample size estimations for genetic association studies of binary traits.
• Properly accounting for covariate effects is crucial for reliable power and sample size calculations in binary trait genetic association studies.
• The developed R package (SPCompute) facilitates the application of this method in study planning.