A method for sample size calculation via E-value in the planning of observational studies

Area of Science:

• Epidemiology
• Biostatistics
• Observational Research Methods

Background:

• Confounding adjustment is critical for valid observational studies (cross-sectional, case-control, cohort).
• Existing sample size calculations may not adequately address potential confounding.
• Quantifying and adjusting for confounders is essential for accurate effect size estimation.

Purpose of the Study:

• To propose a simple, E-value-inspired method for sample size calculation in observational studies with confounding.
• To provide a versatile method applicable to binary, continuous, and time-to-event outcomes.
• To facilitate straightforward implementation using statistical software.

Main Methods:

• The proposed method utilizes a bounding factor, akin to the E-value, to quantify confounder impact.
• Sample size calculations are demonstrated for different outcome variable types.
• The method's performance is evaluated through numerical examples, simulations, and a real-world case study.

Main Results:

• The proposed method provides a conservative estimate, suggesting a slightly larger sample size than minimally required.
• The approach is adaptable for various observational study designs and outcome variable types.
• Demonstrations confirm the method's practicality and effectiveness.

Conclusions:

• A simple and effective method for sample size calculation in the presence of confounding is presented.
• The E-value-inspired approach offers a robust way to ensure adequate power in observational research.
• The method is practical for researchers using standard statistical software.